diff --git a/Makefile b/Makefile index 88de5d0d2..9ff867fbb 100644 --- a/Makefile +++ b/Makefile @@ -36,6 +36,7 @@ PRGDIR = programs ZSTDDIR = lib +DICTDIR = dictBuilder # Define nul output ifneq (,$(filter Windows%,$(OS))) @@ -51,6 +52,7 @@ default: zstdprogram all: $(MAKE) -C $(ZSTDDIR) $@ $(MAKE) -C $(PRGDIR) $@ + $(MAKE) -C $(DICTDIR) $@ zstdprogram: $(MAKE) -C $(PRGDIR) @@ -58,6 +60,7 @@ zstdprogram: clean: @$(MAKE) -C $(ZSTDDIR) $@ > $(VOID) @$(MAKE) -C $(PRGDIR) $@ > $(VOID) + @$(MAKE) -C $(DICTDIR) $@ > $(VOID) @echo Cleaning completed @@ -78,6 +81,7 @@ travis-install: test: $(MAKE) -C $(PRGDIR) $@ + $(MAKE) -C $(DICTDIR) $@ cmaketest: cd contrib/cmake ; cmake . ; $(MAKE) diff --git a/NEWS b/NEWS index 8aaf02d99..46b170a10 100644 --- a/NEWS +++ b/NEWS @@ -1,3 +1,8 @@ +v0.5.0 +New : dictionary builder utility +Changed : streaming & dictionary API +Improved : better compression of small data + v0.4.7 Improved : small compression speed improvement in HC mode Changed : `zstd_decompress.c` has ZSTD_LEGACY_SUPPORT to 0 by default diff --git a/README.md b/README.md index 4a339ccd2..11e41a4c8 100644 --- a/README.md +++ b/README.md @@ -7,38 +7,83 @@ It is provided as a BSD-license package, hosted on Github. |master | [![Build Status](https://travis-ci.org/Cyan4973/zstd.svg?branch=master)](https://travis-ci.org/Cyan4973/zstd) | |dev | [![Build Status](https://travis-ci.org/Cyan4973/zstd.svg?branch=dev)](https://travis-ci.org/Cyan4973/zstd) | -For a taste of its performance, here are a few benchmark numbers from a number of compression codecs suitable for real-time. The test was completed on a Core i7-5600U @ 2.6 GHz, using m^2's [fsbench 0.14.3](http://encode.ru/threads/1371-Filesystem-benchmark?p=34029&viewfull=1#post34029) compiled with gcc 4.8.4, on the [Silesia compression corpus](http://sun.aei.polsl.pl/~sdeor/index.php?page=silesia). +As a reference, several fast compression algorithms were tested and compared to [zlib] on a Core i7-3930K CPU @ 4.5GHz, using [lzbench], an open-source in-memory benchmark by @inikep compiled with gcc 5.2.1, on the [Silesia compression corpus]. + +[lzbench]: https://github.com/inikep/lzbench +[Silesia compression corpus]: http://sun.aei.polsl.pl/~sdeor/index.php?page=silesia + |Name | Ratio | C.speed | D.speed | |-----------------|-------|--------:|--------:| | | | MB/s | MB/s | -| **zstd 0.4** |**2.872**|**280**| **670** | -| [zlib] 1.2.8 -1 | 2.730 | 70 | 300 | -| QuickLZ 1.5.1b6 | 2.237 | 370 | 415 | -| LZO 2.06 | 2.106 | 400 | 580 | -| [LZ4] r131 | 2.101 | 450 | 2100 | -| Snappy 1.1.0 | 2.091 | 330 | 1100 | -| LZF 3.6 | 2.077 | 200 | 560 | +|**zstd 0.4.7 -1**|**2.875**|**330**| **890** | +| [zlib] 1.2.8 -1 | 2.730 | 95 | 360 | +| brotli -0 | 2.708 | 220 | 430 | +| QuickLZ 1.5 | 2.237 | 510 | 605 | +| LZO 2.09 | 2.106 | 610 | 870 | +| [LZ4] r131 | 2.101 | 620 | 3100 | +| Snappy 1.1.3 | 2.091 | 480 | 1600 | +| LZF 3.6 | 2.077 | 375 | 790 | [zlib]:http://www.zlib.net/ -[LZ4]:http://www.lz4.org/ +[LZ4]: http://www.lz4.org/ -Zstd can also offer stronger compression ratio at the cost of compression speed. Speed / Ratio trade-off is configurable by small increment, to fit different situations. Note however that decompression speed is preserved and remain roughly the same at all settings, a property shared by most LZ compression algorithms, such as [zlib]. The following test is run on a Core i7-3930K CPU @ 4.5GHz, using [lzbench], an open-source in-memory benchmark by inikep compiled with gcc 5.2.1, on the [Silesia compression corpus](http://sun.aei.polsl.pl/~sdeor/index.php?page=silesia). +Zstd can also offer stronger compression ratio at the cost of compression speed. +Speed vs Compression trade-off is configurable by small increment. Decompression speed is preserved and remain roughly the same at all settings, a property shared by most LZ compression algorithms, such as [zlib]. -[lzbench]:https://github.com/inikep/lzbench +The following test is run on a Core i7-3930K CPU @ 4.5GHz, using [lzbench], an open-source in-memory benchmark by @inikep compiled with gcc 5.2.1, on the [Silesia compression corpus]. Compression Speed vs Ratio | Decompression Speed ---------------------------|-------------------- ![Compression Speed vs Ratio](images/CSpeed.png "Compression Speed vs Ratio") | ![Decompression Speed](images/DSpeed.png "Decompression Speed") +### The case for Small Data compression + +The above chart is applicable to large files or large streams scenarios (200 MB in this case). +Small data (< 64 KB) come with different perspectives. +The smaller the amount of data to compress, the more difficult it is to achieve any significant compression. +On reaching the 1 KB region, it becomes almost impossible to compress anything. +This problem is common to all compression algorithms, and throwing CPU power at it achieves no significant gains. + +The reason is, compression algorithms learn from past data how to compress future data. +But at the beginning of a new file, there is no "past" to build upon. + +[Starting with 0.5](https://github.com/Cyan4973/zstd/releases), Zstd now offers [a _Dictionary Builder_ tool](https://github.com/Cyan4973/zstd/tree/master/dictBuilder). +It can be used to train the algorithm to fit a selected type of data, by providing it with some samples. +The result is a file (or a byte buffer) called "dictionary", which can be loaded before compression and decompression. +By using this dictionary, the compression ratio achievable on small data improves dramatically : + +| Collection Name | Direct compression | Dictionary Compression | Gains | Average unit | Range | +| --------------- | ------------------ | ---------------------- | ----- | ------------:| ----- | +| Small JSON records | x1.331 - x1.366 | x5.860 - x6.830 | ~ x4.7 | 300 | 200 - 400 | +| Mercurial events | x2.322 - x2.538 | x3.377 - x4.462 | ~ x1.5 | 1.5 KB | 20 - 200 KB | +| Large JSON docs | x3.813 - x4.043 | x8.935 - x13.366 | ~ x2.8 | 6 KB | 800 - 20 KB | + +It has to be noted that these compression gains are achieved without any speed loss, and even some faster decompression processing. + +Dictionary work if there is some correlation in a family of small data (there is no _universal dictionary_). +Hence, deploying one dictionary per type of data will provide the greater benefits. + +Large documents will benefit proportionally less, since dictionary gains are mostly effective in the first few KB. +Then there is enough history to build upon, and the compression algorithm can rely on it to compress the rest of the file. + + +### Status + +Zstd has not yet reached "stable format" status. It doesn't guarantee yet that its current compression format will remain stable in future versions. During this period, it can still change to adapt new optimizations still being investigated. "Stable Format" is projected H1 2016, and will be tagged `v1.0`. + +That being said, the library is now fairly robust, able to withstand hazards situations, including invalid inputs. It also features legacy support, so that documents compressed with current and previous version of zstd can still be decoded in the future. +Library reliability has been tested using [Fuzz Testing](https://en.wikipedia.org/wiki/Fuzz_testing), with both [internal tools](programs/fuzzer.c) and [external ones](http://lcamtuf.coredump.cx/afl). Therefore, Zstandard is not considered safe for testings, even within production environments. + +### Branch Policy + +The "dev" branch is the one where all contributions will be merged before reaching "master". If you plan to propose a patch, please commit into the "dev" branch or its own feature branch. Direct commit to "master" are not permitted. + + +### Trivia + Zstd entropy stage is provided by [Huff0 and FSE, from Finite State Entropy library](https://github.com/Cyan4973/FiniteStateEntropy). Its memory requirement can be configured to fit into low-memory hardware configurations, or servers handling multiple connections/contexts in parallel. -Zstd has not yet reached "stable format" status. It doesn't guarantee yet that its current compressed format will remain stable and supported in future versions. During this period, it can still change to adapt new optimizations still being investigated. "Stable Format" is projected sometimes early 2016. - -That being said, the library is now fairly robust, able to withstand hazards situations, including invalid inputs. The library reliability has been tested using [Fuzz Testing](https://en.wikipedia.org/wiki/Fuzz_testing), with both [internal tools](programs/fuzzer.c) and [external ones](http://lcamtuf.coredump.cx/afl). Therefore, it seems now safe to test Zstandard even within production environments. - -### Branch Policy -The "dev" branch is the one where all contributions will be merged before reaching "master". If you plan to propose a patch, please commit into the "dev" branch or its own feature branch. Direct commit to "master" are not permitted. diff --git a/dictBuilder/COPYING b/dictBuilder/COPYING new file mode 100644 index 000000000..d159169d1 --- /dev/null +++ b/dictBuilder/COPYING @@ -0,0 +1,339 @@ + GNU GENERAL PUBLIC LICENSE + Version 2, June 1991 + + Copyright (C) 1989, 1991 Free Software Foundation, Inc., + 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + Everyone is permitted to copy and distribute verbatim copies + of this license document, but changing it is not allowed. + + Preamble + + The licenses for most software are designed to take away your +freedom to share and change it. By contrast, the GNU General Public +License is intended to guarantee your freedom to share and change free +software--to make sure the software is free for all its users. This +General Public License applies to most of the Free Software +Foundation's software and to any other program whose authors commit to +using it. (Some other Free Software Foundation software is covered by +the GNU Lesser General Public License instead.) You can apply it to +your programs, too. + + When we speak of free software, we are referring to freedom, not +price. Our General Public Licenses are designed to make sure that you +have the freedom to distribute copies of free software (and charge for +this service if you wish), that you receive source code or can get it +if you want it, that you can change the software or use pieces of it +in new free programs; and that you know you can do these things. + + To protect your rights, we need to make restrictions that forbid +anyone to deny you these rights or to ask you to surrender the rights. +These restrictions translate to certain responsibilities for you if you +distribute copies of the software, or if you modify it. + + For example, if you distribute copies of such a program, whether +gratis or for a fee, you must give the recipients all the rights that +you have. You must make sure that they, too, receive or can get the +source code. And you must show them these terms so they know their +rights. + + We protect your rights with two steps: (1) copyright the software, and +(2) offer you this license which gives you legal permission to copy, +distribute and/or modify the software. + + Also, for each author's protection and ours, we want to make certain +that everyone understands that there is no warranty for this free +software. If the software is modified by someone else and passed on, we +want its recipients to know that what they have is not the original, so +that any problems introduced by others will not reflect on the original +authors' reputations. + + Finally, any free program is threatened constantly by software +patents. We wish to avoid the danger that redistributors of a free +program will individually obtain patent licenses, in effect making the +program proprietary. To prevent this, we have made it clear that any +patent must be licensed for everyone's free use or not licensed at all. + + The precise terms and conditions for copying, distribution and +modification follow. + + GNU GENERAL PUBLIC LICENSE + TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION + + 0. This License applies to any program or other work which contains +a notice placed by the copyright holder saying it may be distributed +under the terms of this General Public License. The "Program", below, +refers to any such program or work, and a "work based on the Program" +means either the Program or any derivative work under copyright law: +that is to say, a work containing the Program or a portion of it, +either verbatim or with modifications and/or translated into another +language. (Hereinafter, translation is included without limitation in +the term "modification".) Each licensee is addressed as "you". + +Activities other than copying, distribution and modification are not +covered by this License; they are outside its scope. The act of +running the Program is not restricted, and the output from the Program +is covered only if its contents constitute a work based on the +Program (independent of having been made by running the Program). +Whether that is true depends on what the Program does. + + 1. You may copy and distribute verbatim copies of the Program's +source code as you receive it, in any medium, provided that you +conspicuously and appropriately publish on each copy an appropriate +copyright notice and disclaimer of warranty; keep intact all the +notices that refer to this License and to the absence of any warranty; +and give any other recipients of the Program a copy of this License +along with the Program. + +You may charge a fee for the physical act of transferring a copy, and +you may at your option offer warranty protection in exchange for a fee. + + 2. You may modify your copy or copies of the Program or any portion +of it, thus forming a work based on the Program, and copy and +distribute such modifications or work under the terms of Section 1 +above, provided that you also meet all of these conditions: + + a) You must cause the modified files to carry prominent notices + stating that you changed the files and the date of any change. + + b) You must cause any work that you distribute or publish, that in + whole or in part contains or is derived from the Program or any + part thereof, to be licensed as a whole at no charge to all third + parties under the terms of this License. + + c) If the modified program normally reads commands interactively + when run, you must cause it, when started running for such + interactive use in the most ordinary way, to print or display an + announcement including an appropriate copyright notice and a + notice that there is no warranty (or else, saying that you provide + a warranty) and that users may redistribute the program under + these conditions, and telling the user how to view a copy of this + License. (Exception: if the Program itself is interactive but + does not normally print such an announcement, your work based on + the Program is not required to print an announcement.) + +These requirements apply to the modified work as a whole. If +identifiable sections of that work are not derived from the Program, +and can be reasonably considered independent and separate works in +themselves, then this License, and its terms, do not apply to those +sections when you distribute them as separate works. But when you +distribute the same sections as part of a whole which is a work based +on the Program, the distribution of the whole must be on the terms of +this License, whose permissions for other licensees extend to the +entire whole, and thus to each and every part regardless of who wrote it. + +Thus, it is not the intent of this section to claim rights or contest +your rights to work written entirely by you; rather, the intent is to +exercise the right to control the distribution of derivative or +collective works based on the Program. + +In addition, mere aggregation of another work not based on the Program +with the Program (or with a work based on the Program) on a volume of +a storage or distribution medium does not bring the other work under +the scope of this License. + + 3. You may copy and distribute the Program (or a work based on it, +under Section 2) in object code or executable form under the terms of +Sections 1 and 2 above provided that you also do one of the following: + + a) Accompany it with the complete corresponding machine-readable + source code, which must be distributed under the terms of Sections + 1 and 2 above on a medium customarily used for software interchange; or, + + b) Accompany it with a written offer, valid for at least three + years, to give any third party, for a charge no more than your + cost of physically performing source distribution, a complete + machine-readable copy of the corresponding source code, to be + distributed under the terms of Sections 1 and 2 above on a medium + customarily used for software interchange; or, + + c) Accompany it with the information you received as to the offer + to distribute corresponding source code. (This alternative is + allowed only for noncommercial distribution and only if you + received the program in object code or executable form with such + an offer, in accord with Subsection b above.) + +The source code for a work means the preferred form of the work for +making modifications to it. For an executable work, complete source +code means all the source code for all modules it contains, plus any +associated interface definition files, plus the scripts used to +control compilation and installation of the executable. However, as a +special exception, the source code distributed need not include +anything that is normally distributed (in either source or binary +form) with the major components (compiler, kernel, and so on) of the +operating system on which the executable runs, unless that component +itself accompanies the executable. + +If distribution of executable or object code is made by offering +access to copy from a designated place, then offering equivalent +access to copy the source code from the same place counts as +distribution of the source code, even though third parties are not +compelled to copy the source along with the object code. + + 4. You may not copy, modify, sublicense, or distribute the Program +except as expressly provided under this License. Any attempt +otherwise to copy, modify, sublicense or distribute the Program is +void, and will automatically terminate your rights under this License. +However, parties who have received copies, or rights, from you under +this License will not have their licenses terminated so long as such +parties remain in full compliance. + + 5. You are not required to accept this License, since you have not +signed it. However, nothing else grants you permission to modify or +distribute the Program or its derivative works. These actions are +prohibited by law if you do not accept this License. Therefore, by +modifying or distributing the Program (or any work based on the +Program), you indicate your acceptance of this License to do so, and +all its terms and conditions for copying, distributing or modifying +the Program or works based on it. + + 6. Each time you redistribute the Program (or any work based on the +Program), the recipient automatically receives a license from the +original licensor to copy, distribute or modify the Program subject to +these terms and conditions. You may not impose any further +restrictions on the recipients' exercise of the rights granted herein. +You are not responsible for enforcing compliance by third parties to +this License. + + 7. If, as a consequence of a court judgment or allegation of patent +infringement or for any other reason (not limited to patent issues), +conditions are imposed on you (whether by court order, agreement or +otherwise) that contradict the conditions of this License, they do not +excuse you from the conditions of this License. If you cannot +distribute so as to satisfy simultaneously your obligations under this +License and any other pertinent obligations, then as a consequence you +may not distribute the Program at all. For example, if a patent +license would not permit royalty-free redistribution of the Program by +all those who receive copies directly or indirectly through you, then +the only way you could satisfy both it and this License would be to +refrain entirely from distribution of the Program. + +If any portion of this section is held invalid or unenforceable under +any particular circumstance, the balance of the section is intended to +apply and the section as a whole is intended to apply in other +circumstances. + +It is not the purpose of this section to induce you to infringe any +patents or other property right claims or to contest validity of any +such claims; this section has the sole purpose of protecting the +integrity of the free software distribution system, which is +implemented by public license practices. Many people have made +generous contributions to the wide range of software distributed +through that system in reliance on consistent application of that +system; it is up to the author/donor to decide if he or she is willing +to distribute software through any other system and a licensee cannot +impose that choice. + +This section is intended to make thoroughly clear what is believed to +be a consequence of the rest of this License. + + 8. If the distribution and/or use of the Program is restricted in +certain countries either by patents or by copyrighted interfaces, the +original copyright holder who places the Program under this License +may add an explicit geographical distribution limitation excluding +those countries, so that distribution is permitted only in or among +countries not thus excluded. In such case, this License incorporates +the limitation as if written in the body of this License. + + 9. The Free Software Foundation may publish revised and/or new versions +of the General Public License from time to time. Such new versions will +be similar in spirit to the present version, but may differ in detail to +address new problems or concerns. + +Each version is given a distinguishing version number. If the Program +specifies a version number of this License which applies to it and "any +later version", you have the option of following the terms and conditions +either of that version or of any later version published by the Free +Software Foundation. If the Program does not specify a version number of +this License, you may choose any version ever published by the Free Software +Foundation. + + 10. If you wish to incorporate parts of the Program into other free +programs whose distribution conditions are different, write to the author +to ask for permission. For software which is copyrighted by the Free +Software Foundation, write to the Free Software Foundation; we sometimes +make exceptions for this. Our decision will be guided by the two goals +of preserving the free status of all derivatives of our free software and +of promoting the sharing and reuse of software generally. + + NO WARRANTY + + 11. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY +FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN +OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES +PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED +OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF +MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS +TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE +PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, +REPAIR OR CORRECTION. + + 12. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING +WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR +REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, +INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING +OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED +TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY +YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER +PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE +POSSIBILITY OF SUCH DAMAGES. + + END OF TERMS AND CONDITIONS + + How to Apply These Terms to Your New Programs + + If you develop a new program, and you want it to be of the greatest +possible use to the public, the best way to achieve this is to make it +free software which everyone can redistribute and change under these terms. + + To do so, attach the following notices to the program. It is safest +to attach them to the start of each source file to most effectively +convey the exclusion of warranty; and each file should have at least +the "copyright" line and a pointer to where the full notice is found. + + + Copyright (C) + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 2 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License along + with this program; if not, write to the Free Software Foundation, Inc., + 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. + +Also add information on how to contact you by electronic and paper mail. + +If the program is interactive, make it output a short notice like this +when it starts in an interactive mode: + + Gnomovision version 69, Copyright (C) year name of author + Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'. + This is free software, and you are welcome to redistribute it + under certain conditions; type `show c' for details. + +The hypothetical commands `show w' and `show c' should show the appropriate +parts of the General Public License. Of course, the commands you use may +be called something other than `show w' and `show c'; they could even be +mouse-clicks or menu items--whatever suits your program. + +You should also get your employer (if you work as a programmer) or your +school, if any, to sign a "copyright disclaimer" for the program, if +necessary. Here is a sample; alter the names: + + Yoyodyne, Inc., hereby disclaims all copyright interest in the program + `Gnomovision' (which makes passes at compilers) written by James Hacker. + + , 1 April 1989 + Ty Coon, President of Vice + +This General Public License does not permit incorporating your program into +proprietary programs. If your program is a subroutine library, you may +consider it more useful to permit linking proprietary applications with the +library. If this is what you want to do, use the GNU Lesser General +Public License instead of this License. diff --git a/dictBuilder/Makefile b/dictBuilder/Makefile new file mode 100644 index 000000000..e5a4f1aa7 --- /dev/null +++ b/dictBuilder/Makefile @@ -0,0 +1,69 @@ +# ########################################################################## +# Dict Builder - Makefile +# Copyright (C) Yann Collet 2015 +# +# GPL v2 License +# +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU General Public License as published by +# the Free Software Foundation; either version 2 of the License, or +# (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License along +# with this program; if not, write to the Free Software Foundation, Inc., +# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. +# +# You can contact the author at : +# - ZSTD source repository : http://code.google.com/p/zstd/ +# - Public forum : https://groups.google.com/forum/#!forum/lz4c +# ########################################################################## + +CPPFLAGS= -I../lib +CFLAGS ?= -O3 +CFLAGS += -std=c99 -Wall -Wextra -Wshadow -Wcast-qual -Wcast-align -Wundef -Wstrict-prototypes -Wstrict-aliasing=1 +FLAGS = $(CPPFLAGS) $(CFLAGS) $(LDFLAGS) $(MOREFLAGS) + +ZSTDDIR = ../lib + + +# Define *.exe as extension for Windows systems +ifneq (,$(filter Windows%,$(OS))) +EXT =.exe +VOID = nul +else +EXT = +VOID = /dev/null +endif + + +.PHONY: default all test + +default: dictBuilder + +all: dictBuilder + +dictBuilder: dictBuilder.c dibcli.c divsufsort.c sssort.c trsort.c $(ZSTDDIR)/huff0.c $(ZSTDDIR)/fse.c $(ZSTDDIR)/zstd_decompress.c + $(CC) $(FLAGS) $^ -o $@$(EXT) + +clean: + @rm -f core *.o tmp* result* *.gcda \ + dictBuilder$(EXT) + @echo Cleaning completed + +test: dictBuilder + ./dictBuilder * + @rm dictionary + +clangtest: CC = clang +clangtest: CFLAGS += -Werror +clangtest: clean dictBuilder + +gpptest: CC = g++ +gpptest: CFLAGS=-O3 -Wall -Wextra -Wshadow -Wcast-align -Wcast-qual -Wundef -Werror +gpptest: clean dictBuilder + diff --git a/dictBuilder/config.h b/dictBuilder/config.h new file mode 100644 index 000000000..c2925d335 --- /dev/null +++ b/dictBuilder/config.h @@ -0,0 +1,83 @@ +/* + * config.h for libdivsufsort + * Copyright (c) 2003-2008 Yuta Mori All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, + * copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following + * conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES + * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#ifndef _CONFIG_H +#define _CONFIG_H 1 + +#ifdef __cplusplus +extern "C" { +#endif /* __cplusplus */ + +/** Define to the version of this package. **/ +#define PROJECT_VERSION_FULL "2.0.1" + +/** Define to 1 if you have the header files. **/ +#define HAVE_INTTYPES_H 1 +#define HAVE_STDDEF_H 1 +#define HAVE_STDINT_H 1 +#define HAVE_STDLIB_H 1 +#define HAVE_STRING_H 1 +#define HAVE_STRINGS_H 1 +#define HAVE_MEMORY_H 1 +#define HAVE_SYS_TYPES_H 1 + +/** for WinIO **/ +/* #undef HAVE_IO_H */ +/* #undef HAVE_FCNTL_H */ +/* #undef HAVE__SETMODE */ +/* #undef HAVE_SETMODE */ +/* #undef HAVE__FILENO */ +/* #undef HAVE_FOPEN_S */ +/* #undef HAVE__O_BINARY */ +/* +#ifndef HAVE__SETMODE +# if HAVE_SETMODE +# define _setmode setmode +# define HAVE__SETMODE 1 +# endif +# if HAVE__SETMODE && !HAVE__O_BINARY +# define _O_BINARY 0 +# define HAVE__O_BINARY 1 +# endif +#endif +*/ + +/** for inline **/ +#ifndef INLINE +# define INLINE inline +#endif + +/** for VC++ warning **/ +#ifdef _MSC_VER +#pragma warning(disable: 4127) +#endif + + +#ifdef __cplusplus +} /* extern "C" */ +#endif /* __cplusplus */ + +#endif /* _CONFIG_H */ diff --git a/dictBuilder/dibcli.c b/dictBuilder/dibcli.c new file mode 100644 index 000000000..8566c6245 --- /dev/null +++ b/dictBuilder/dibcli.c @@ -0,0 +1,263 @@ +/* + dibcli - Command Line Interface (cli) for Dictionary Builder + Copyright (C) Yann Collet 2016 + + GPL v2 License + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 2 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License along + with this program; if not, write to the Free Software Foundation, Inc., + 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. + + You can contact the author at : + - zstd source repository : https://github.com/Cyan4973/zstd +*/ + +/* ************************************** +* Compiler Specifics +****************************************/ +/* Disable some Visual warning messages */ +#ifdef _MSC_VER +# pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */ +#endif + + +/*-************************************ +* Includes +**************************************/ +#include /* exit, calloc, free */ +#include /* strcmp, strlen */ +#include /* fprintf, getchar */ + +#include "dictBuilder.h" + + +/*-************************************ +* Constants +**************************************/ +#define PROGRAM_DESCRIPTION "Dictionary builder" +#ifndef PROGRAM_VERSION +# define QUOTE(str) #str +# define EXP_Q(str) QUOTE(str) +# define PROGRAM_VERSION "v" EXP_Q(DiB_VERSION_MAJOR) "." EXP_Q(DiB_VERSION_MINOR) "." EXP_Q(DiB_VERSION_RELEASE) +#endif +#define AUTHOR "Yann Collet" +#define WELCOME_MESSAGE "*** %s %s %i-bits, by %s ***\n", PROGRAM_DESCRIPTION, PROGRAM_VERSION, (int)(sizeof(void*)*8), AUTHOR + +#define KB *(1 <<10) +#define MB *(1 <<20) +#define GB *(1U<<30) + +static const unsigned compressionLevelDefault = 5; +static const unsigned selectionLevelDefault = 9; /* determined experimentally */ +static const unsigned maxDictSizeDefault = 110 KB; +static const char* dictFileNameDefault = "dictionary"; + + +/*-************************************ +* Display Macros +**************************************/ +#define DISPLAY(...) fprintf(g_displayOut, __VA_ARGS__) +#define DISPLAYLEVEL(l, ...) if (g_displayLevel>=l) { DISPLAY(__VA_ARGS__); } +static FILE* g_displayOut; +static unsigned g_displayLevel = 2; // 0 : no display // 1: errors // 2 : + result + interaction + warnings ; // 3 : + progression; // 4 : + information + + +/*-************************************ +* Exceptions +**************************************/ +#define DEBUG 0 +#define DEBUGOUTPUT(...) if (DEBUG) DISPLAY(__VA_ARGS__); +#define EXM_THROW(error, ...) \ +{ \ + DEBUGOUTPUT("Error defined at %s, line %i : \n", __FILE__, __LINE__); \ + DISPLAYLEVEL(1, "Error %i : ", error); \ + DISPLAYLEVEL(1, __VA_ARGS__); \ + DISPLAYLEVEL(1, "\n"); \ + exit(error); \ +} + + +/*-************************************ +* Command Line +**************************************/ +static int usage(const char* programName) +{ + DISPLAY( "Usage :\n"); + DISPLAY( " %s [arg] [filenames]\n", programName); + DISPLAY( "\n"); + DISPLAY( "Arguments :\n"); + DISPLAY( " -o : name of dictionary file (default: %s) \n", dictFileNameDefault); + DISPLAY( "--maxdict : limit dictionary to specified size (default : %u) \n", maxDictSizeDefault); + DISPLAY( " -h/-H : display help/long help and exit\n"); + return 0; +} + +static int usage_advanced(const char* programName) +{ + DISPLAY(WELCOME_MESSAGE); + usage(programName); + DISPLAY( "\n"); + DISPLAY( "Advanced arguments :\n"); + DISPLAY( " -V : display Version number and exit\n"); + DISPLAY( "--fast : fast sampling mode\n"); + DISPLAY( " -L# : target compression level (default: %u)\n", compressionLevelDefault); + DISPLAY( " -S# : dictionary selectivity level (default: %u)\n", selectionLevelDefault); + DISPLAY( " -v : verbose mode\n"); + DISPLAY( " -q : suppress notifications; specify twice to suppress errors too\n"); + return 0; +} + +static int badusage(const char* programName) +{ + DISPLAYLEVEL(1, "Incorrect parameters\n"); + if (g_displayLevel >= 1) usage(programName); + return 1; +} + + +static void waitEnter(void) +{ + int unused; + DISPLAY("Press enter to continue...\n"); + unused = getchar(); + (void)unused; +} + + +int main(int argCount, const char** argv) +{ + int i, + main_pause=0, + operationResult=0, + nextArgumentIsMaxDict=0, + nextArgumentIsDictFileName=0; + unsigned cLevel = compressionLevelDefault; + unsigned maxDictSize = maxDictSizeDefault; + unsigned selectionLevel = selectionLevelDefault; + const char** filenameTable = (const char**)malloc(argCount * sizeof(const char*)); /* argCount >= 1 */ + unsigned filenameIdx = 0; + const char* programName = argv[0]; + const char* dictFileName = dictFileNameDefault; + + /* init */ + g_displayOut = stderr; /* unfortunately, cannot be set at declaration */ + if (filenameTable==NULL) EXM_THROW(1, "not enough memory\n"); + /* Pick out program name from path. Don't rely on stdlib because of conflicting behavior */ + for (i = (int)strlen(programName); i > 0; i--) { if ((programName[i] == '/') || (programName[i] == '\\')) { i++; break; } } + programName += i; + + /* command switches */ + for(i=1; i='0') && (*argument<='9')) + maxDictSize = maxDictSize * 10 + (*argument - '0'), argument++; + if (*argument=='k' || *argument=='K') + maxDictSize <<= 10; + continue; + } + + /* long commands (--long-word) */ + if (!strcmp(argument, "--version")) { g_displayOut=stdout; DISPLAY(WELCOME_MESSAGE); return 0; } + if (!strcmp(argument, "--help")) { g_displayOut=stdout; return usage_advanced(programName); } + if (!strcmp(argument, "--verbose")) { g_displayLevel++; if (g_displayLevel<3) g_displayLevel=3; continue; } + if (!strcmp(argument, "--quiet")) { g_displayLevel--; continue; } + if (!strcmp(argument, "--maxdict")) { nextArgumentIsMaxDict=1; continue; } + if (!strcmp(argument, "--fast")) { selectionLevel=1; cLevel=1; continue; } + + /* Decode commands (note : aggregated commands are allowed) */ + if (argument[0]=='-') { + argument++; + + while (argument[0]!=0) { + switch(argument[0]) + { + /* Display help */ + case 'V': g_displayOut=stdout; DISPLAY(WELCOME_MESSAGE); return 0; /* Version Only */ + case 'H': + case 'h': g_displayOut=stdout; return usage_advanced(programName); + + /* Selection level */ + case 'S': argument++; + selectionLevel = 0; + while ((*argument >= '0') && (*argument <= '9')) + selectionLevel *= 10, selectionLevel += *argument++ - '0'; + break; + + /* Selection level */ + case 'L': argument++; + cLevel = 0; + while ((*argument >= '0') && (*argument <= '9')) + cLevel *= 10, cLevel += *argument++ - '0'; + break; + + /* Verbose mode */ + case 'v': g_displayLevel++; if (g_displayLevel<3) g_displayLevel=3; argument++; break; + + /* Quiet mode */ + case 'q': g_displayLevel--; argument++; break; + + /* dictionary name */ + case 'o': nextArgumentIsDictFileName=1; argument++; break; + + /* Pause at the end (hidden option) */ + case 'p': main_pause=1; argument++; break; + + /* unknown command */ + default : return badusage(programName); + } } + continue; + } + + /* add filename to list */ + filenameTable[filenameIdx++] = argument; + } + + /* Welcome message (if verbose) */ + DISPLAYLEVEL(3, WELCOME_MESSAGE); + + /* check nb files */ + if (filenameIdx==0) return badusage(programName); + if (filenameIdx < 100) + { + DISPLAYLEVEL(2, "Warning : set contains only %u files ... \n", filenameIdx); + DISPLAYLEVEL(3, "!! For better results, consider providing > 1.000 samples !!\n"); + DISPLAYLEVEL(3, "!! Each sample should preferably be stored as a separate file !!\n"); + } + + /* building ... */ + { + DiB_params_t param; + param.selectivityLevel = selectionLevel; + param.compressionLevel = cLevel; + DiB_setNotificationLevel(g_displayLevel); + operationResult = DiB_trainFromFiles(dictFileName, maxDictSize, + filenameTable, filenameIdx, + param); + } + + if (main_pause) waitEnter(); + free((void*)filenameTable); + return operationResult; +} diff --git a/dictBuilder/dictBuilder.c b/dictBuilder/dictBuilder.c new file mode 100644 index 000000000..d8b2bdb30 --- /dev/null +++ b/dictBuilder/dictBuilder.c @@ -0,0 +1,1020 @@ +/* + dictBuilder - dictionary builder for LZ algorithms + Copyright (C) Yann Collet 2016 + + GPL v2 License + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 2 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License along + with this program; if not, write to the Free Software Foundation, Inc., + 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. + + You can contact the author at : + - zstd source repository : https://github.com/Cyan4973/zstd +*/ + +/* ************************************** +* Compiler Options +****************************************/ +/* Disable some Visual warning messages */ +#ifdef _MSC_VER +# define _CRT_SECURE_NO_WARNINGS /* fopen */ +# pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */ +#endif + +/* Unix Large Files support (>4GB) */ +#define _FILE_OFFSET_BITS 64 +#if (defined(__sun__) && (!defined(__LP64__))) /* Sun Solaris 32-bits requires specific definitions */ +# define _LARGEFILE_SOURCE +#elif ! defined(__LP64__) /* No point defining Large file for 64 bit */ +# define _LARGEFILE64_SOURCE +#endif + + +/*-************************************* +* Includes +***************************************/ +#include /* malloc, free */ +#include /* memset */ +#include /* fprintf, fopen, ftello64 */ +#include /* stat64 */ +#include /* stat64 */ +#include /* clock */ + +#include "mem.h" /* read */ +#include "error_private.h" +#include "divsufsort.h" +#include "dictBuilder.h" +#include "zstd_compress.c" +#include "huff0_static.h" + + +/*-************************************* +* Compiler specifics +***************************************/ +#if !defined(S_ISREG) +# define S_ISREG(x) (((x) & S_IFMT) == S_IFREG) +#endif + + +/*-************************************* +* Constants +***************************************/ +#define KB *(1 <<10) +#define MB *(1 <<20) +#define GB *(1U<<30) + +#define DICTLISTSIZE 10000 +#define MEMMULT 11 +static const size_t maxMemory = (sizeof(size_t) == 4) ? (2 GB - 64 MB) : ((size_t)(512 MB) << sizeof(size_t)); + +#define NOISELENGTH 32 +#define PRIME1 2654435761U +#define PRIME2 2246822519U + +#define MINRATIO 4 +static const U32 g_compressionLevel_default = 5; +static const U32 g_selectivity_default = 9; +static const size_t g_provision_entropySize = 200; +static const size_t g_min_fast_dictContent = 192; + + +/*-************************************* +* Console display +***************************************/ +#define DISPLAY(...) fprintf(stderr, __VA_ARGS__) +#define DISPLAYLEVEL(l, ...) if (g_displayLevel>=l) { DISPLAY(__VA_ARGS__); } +static unsigned g_displayLevel = 0; /* 0 : no display; 1: errors; 2: default; 4: full information */ +void DiB_setNotificationLevel(unsigned l) { g_displayLevel=l; } + +#define DISPLAYUPDATE(l, ...) if (g_displayLevel>=l) { \ + if (DiB_GetMilliSpan(g_time) > refreshRate) \ + { g_time = clock(); DISPLAY(__VA_ARGS__); \ + if (g_displayLevel>=4) fflush(stdout); } } +static const unsigned refreshRate = 300; +static clock_t g_time = 0; + +void DiB_printHex(U32 dlevel, const void* ptr, size_t length) +{ + const BYTE* const b = (const BYTE*)ptr; + size_t u; + for (u=0; u126) c = '.'; /* non-printable char */ + DISPLAYLEVEL(dlevel, "%c", c); + } +} + + +/*-************************************* +* Exceptions +***************************************/ +#ifndef DEBUG +# define DEBUG 0 +#endif +#define DEBUGOUTPUT(...) if (DEBUG) DISPLAY(__VA_ARGS__); +#define EXM_THROW(error, ...) \ +{ \ + DEBUGOUTPUT("Error defined at %s, line %i : \n", __FILE__, __LINE__); \ + DISPLAYLEVEL(1, "Error %i : ", error); \ + DISPLAYLEVEL(1, __VA_ARGS__); \ + DISPLAYLEVEL(1, "\n"); \ + exit(error); \ +} + + +/* ******************************************************** +* Helper functions +**********************************************************/ +unsigned DiB_versionNumber (void) { return DiB_VERSION_NUMBER; } + +static unsigned DiB_GetMilliSpan(clock_t nPrevious) +{ + clock_t nCurrent = clock(); + unsigned nSpan = (unsigned)(((nCurrent - nPrevious) * 1000) / CLOCKS_PER_SEC); + return nSpan; +} + +unsigned DiB_isError(size_t errorCode) { return ERR_isError(errorCode); } + +const char* DiB_getErrorName(size_t errorCode) { return ERR_getErrorName(errorCode); } + + +/* ******************************************************** +* File related operations +**********************************************************/ +static unsigned long long DiB_getFileSize(const char* infilename) +{ + int r; +#if defined(_MSC_VER) + struct _stat64 statbuf; + r = _stat64(infilename, &statbuf); +#else + struct stat statbuf; + r = stat(infilename, &statbuf); +#endif + if (r || !S_ISREG(statbuf.st_mode)) return 0; /* No good... */ + return (unsigned long long)statbuf.st_size; +} + + +static unsigned long long DiB_getTotalFileSize(const char** fileNamesTable, unsigned nbFiles) +{ + unsigned long long total = 0; + unsigned n; + for (n=0; n bufferSize-pos) fileSize = 0; /* stop there, not enough memory to load all files */ + readSize = fread(buff+pos, 1, (size_t)fileSize, f); + if (readSize != (size_t)fileSize) EXM_THROW(11, "could not read %s", fileNamesTable[n]); + pos += readSize; + fileSizes[n] = (size_t)fileSize; + fclose(f); + } +} + + +/*-******************************************************** +* Dictionary training functions +**********************************************************/ +static size_t DiB_read_ARCH(const void* p) { size_t r; memcpy(&r, p, sizeof(r)); return r; } + +static unsigned DiB_NbCommonBytes (register size_t val) +{ + if (MEM_isLittleEndian()) { + if (MEM_64bits()) { +# if defined(_MSC_VER) && defined(_WIN64) + unsigned long r = 0; + _BitScanForward64( &r, (U64)val ); + return (unsigned)(r>>3); +# elif defined(__GNUC__) && (__GNUC__ >= 3) + return (__builtin_ctzll((U64)val) >> 3); +# else + static const int DeBruijnBytePos[64] = { 0, 0, 0, 0, 0, 1, 1, 2, 0, 3, 1, 3, 1, 4, 2, 7, 0, 2, 3, 6, 1, 5, 3, 5, 1, 3, 4, 4, 2, 5, 6, 7, 7, 0, 1, 2, 3, 3, 4, 6, 2, 6, 5, 5, 3, 4, 5, 6, 7, 1, 2, 4, 6, 4, 4, 5, 7, 2, 6, 5, 7, 6, 7, 7 }; + return DeBruijnBytePos[((U64)((val & -(long long)val) * 0x0218A392CDABBD3FULL)) >> 58]; +# endif + } else { /* 32 bits */ +# if defined(_MSC_VER) + unsigned long r=0; + _BitScanForward( &r, (U32)val ); + return (unsigned)(r>>3); +# elif defined(__GNUC__) && (__GNUC__ >= 3) + return (__builtin_ctz((U32)val) >> 3); +# else + static const int DeBruijnBytePos[32] = { 0, 0, 3, 0, 3, 1, 3, 0, 3, 2, 2, 1, 3, 2, 0, 1, 3, 3, 1, 2, 2, 2, 2, 0, 3, 1, 2, 0, 1, 0, 1, 1 }; + return DeBruijnBytePos[((U32)((val & -(S32)val) * 0x077CB531U)) >> 27]; +# endif + } + } else { /* Big Endian CPU */ + if (MEM_64bits()) { +# if defined(_MSC_VER) && defined(_WIN64) + unsigned long r = 0; + _BitScanReverse64( &r, val ); + return (unsigned)(r>>3); +# elif defined(__GNUC__) && (__GNUC__ >= 3) + return (__builtin_clzll(val) >> 3); +# else + unsigned r; + const unsigned n32 = sizeof(size_t)*4; /* calculate this way due to compiler complaining in 32-bits mode */ + if (!(val>>n32)) { r=4; } else { r=0; val>>=n32; } + if (!(val>>16)) { r+=2; val>>=8; } else { val>>=24; } + r += (!val); + return r; +# endif + } else { /* 32 bits */ +# if defined(_MSC_VER) + unsigned long r = 0; + _BitScanReverse( &r, (unsigned long)val ); + return (unsigned)(r>>3); +# elif defined(__GNUC__) && (__GNUC__ >= 3) + return (__builtin_clz((U32)val) >> 3); +# else + unsigned r; + if (!(val>>16)) { r=2; val>>=8; } else { r=0; val>>=24; } + r += (!val); + return r; +# endif + } } +} + + +/*! DiB_count() : + Count the nb of common bytes between 2 pointers. + Note : this function presumes end of buffer followed by noisy guard band. +*/ +static size_t DiB_count(const void* pIn, const void* pMatch) +{ + const char* const pStart = (const char*)pIn; + for (;;) { + size_t diff = DiB_read_ARCH(pMatch) ^ DiB_read_ARCH(pIn); + if (!diff) { pIn = (const char*)pIn+sizeof(size_t); pMatch = (const char*)pMatch+sizeof(size_t); continue; } + pIn = (const char*)pIn+DiB_NbCommonBytes(diff); + return (size_t)((const char*)pIn - pStart); + } +} + + +typedef struct { + U32 pos; + U32 length; + U32 savings; +} dictItem; + +void DiB_initDictItem(dictItem* d) +{ + d->pos = 1; + d->length = 0; + d->savings = (U32)(-1); +} + + +#define LLIMIT 64 /* heuristic determined experimentally */ +#define MINMATCHLENGTH 7 /* heuristic determined experimentally */ +static dictItem DiB_analyzePos( + BYTE* doneMarks, + const saidx_t* suffix, U32 start, + const void* buffer, U32 minRatio) +{ + U32 lengthList[LLIMIT] = {0}; + U32 cumulLength[LLIMIT] = {0}; + U32 savings[LLIMIT] = {0}; + const BYTE* b = (const BYTE*)buffer; + size_t length; + size_t maxLength = LLIMIT; + size_t pos = suffix[start]; + U32 end = start; + dictItem solution; + + /* init */ + memset(&solution, 0, sizeof(solution)); + doneMarks[pos] = 1; + + /* trivial repetition cases */ + if ( (MEM_read16(b+pos+0) == MEM_read16(b+pos+2)) + ||(MEM_read16(b+pos+1) == MEM_read16(b+pos+3)) + ||(MEM_read16(b+pos+2) == MEM_read16(b+pos+4)) ) { + /* skip and mark segment */ + U16 u16 = MEM_read16(b+pos+4); + U32 u, e = 6; + while (MEM_read16(b+pos+e) == u16) e+=2 ; + if (b[pos+e] == b[pos+e-1]) e++; + for (u=1; u=MINMATCHLENGTH); + + /* look backward */ + do { + length = DiB_count(b + pos, b + *(suffix+start-1)); + if (length >=MINMATCHLENGTH) start--; + } while(length >= MINMATCHLENGTH); + + /* exit if not found a minimum nb of repetitions */ + if (end-start < minRatio) { + U32 idx; + for(idx=start; idx= %u at pos %7u ", (U32)(end-start), MINMATCHLENGTH, (U32)pos); + DISPLAYLEVEL(4, "\n"); + + for (searchLength = MINMATCHLENGTH ; ; searchLength++) { + BYTE currentChar = 0; + U32 currentCount = 0; + U32 currentID = refinedStart; + U32 id; + U32 selectedCount = 0; + U32 selectedID = currentID; + for (id =refinedStart; id < refinedEnd; id++) { + if (b[ suffix[id] + searchLength] != currentChar) { + if (currentCount > selectedCount) { + selectedCount = currentCount; + selectedID = currentID; + } + currentID = id; + currentChar = b[ suffix[id] + searchLength]; + currentCount = 0; + } + currentCount ++; + } + if (currentCount > selectedCount) { /* for last */ + selectedCount = currentCount; + selectedID = currentID; + } + + if (selectedCount < minRatio) + break; + refinedStart = selectedID; + refinedEnd = refinedStart + selectedCount; + } + + /* evaluate gain based on new ref */ + start = refinedStart; + pos = suffix[refinedStart]; + end = start; + memset(lengthList, 0, sizeof(lengthList)); + + /* look forward */ + do { + end++; + length = DiB_count(b + pos, b + suffix[end]); + if (length >= LLIMIT) length = LLIMIT-1; + lengthList[length]++; + } while (length >=MINMATCHLENGTH); + + /* look backward */ + do { + length = DiB_count(b + pos, b + suffix[start-1]); + if (length >= LLIMIT) length = LLIMIT-1; + lengthList[length]++; + if (length >=MINMATCHLENGTH) start--; + } while(length >= MINMATCHLENGTH); + + /* largest useful length */ + memset(cumulLength, 0, sizeof(cumulLength)); + cumulLength[maxLength-1] = lengthList[maxLength-1]; + for (i=(int)(maxLength-2); i>=0; i--) + cumulLength[i] = cumulLength[i+1] + lengthList[i]; + + for (i=LLIMIT-1; i>=MINMATCHLENGTH; i--) if (cumulLength[i]>=minRatio) break; + maxLength = i; + + /* reduce maxLength in case of final into repetitive data */ + { + U32 l = (U32)maxLength; + BYTE c = b[pos + maxLength-1]; + while (b[pos+l-2]==c) l--; + maxLength = l; + } + if (maxLength < MINMATCHLENGTH) return solution; /* skip : no long-enough solution */ + + /* calculate savings */ + savings[5] = 0; + for (i=MINMATCHLENGTH; i<=(int)maxLength; i++) + savings[i] = savings[i-1] + (lengthList[i] * (i-3)); + + DISPLAYLEVEL(4, "Selected ref at position %u, of length %u : saves %u (ratio: %.2f) \n", + (U32)pos, (U32)maxLength, savings[maxLength], (double)savings[maxLength] / maxLength); + + solution.pos = (U32)pos; + solution.length = (U32)maxLength; + solution.savings = savings[maxLength]; + + /* mark positions done */ + { + U32 id; + U32 testedPos; + for (id=start; id solution.length) length = solution.length; + } + pEnd = (U32)(testedPos + length); + for (p=testedPos; ppos; + const U32 max = elt.pos + (elt.length-1); + + /* tail overlap */ + U32 u; for (u=1; u elt.pos) && (table[u].pos < max)) { /* overlap */ + /* append */ + U32 addedLength = table[u].pos - elt.pos; + table[u].length += addedLength; + table[u].pos = elt.pos; + table[u].savings += elt.savings * addedLength / elt.length; /* rough approx */ + table[u].savings += elt.length / 8; /* rough approx */ + elt = table[u]; + while ((u>1) && (table[u-1].savings < elt.savings)) + table[u] = table[u-1], u--; + table[u] = elt; + return u; + } } + + /* front overlap */ + for (u=1; u elt.pos) && (table[u].pos < elt.pos)) { /* overlap */ + /* append */ + int addedLength = (elt.pos + elt.length) - (table[u].pos + table[u].length); + table[u].savings += elt.length / 8; /* rough approx */ + if (addedLength > 0) { /* otherwise, already included */ + table[u].length += addedLength; + table[u].savings += elt.savings * addedLength / elt.length; /* rough approx */ + } + elt = table[u]; + while ((u>1) && (table[u-1].savings < elt.savings)) + table[u] = table[u-1], u--; + table[u] = elt; + return u; + } } + + return 0; +} + + +static void DiB_removeDictItem(dictItem* table, U32 id) +{ + /* convention : first element is nb of elts */ + U32 max = table->pos; + U32 u; + if (!id) return; /* protection, should never happen */ + for (u=id; upos--; +} + + +static void DiB_insertDictItem(dictItem* table, U32 maxSize, dictItem elt) +{ + /* merge if possible */ + U32 mergeId = DiB_checkMerge(table, elt, 0); + if (mergeId) { + U32 newMerge = 1; + while (newMerge) { + newMerge = DiB_checkMerge(table, table[mergeId], mergeId); + if (newMerge) DiB_removeDictItem(table, mergeId); + mergeId = newMerge; + } + return; + } + + /* insert */ + { + U32 current; + U32 nextElt = table->pos; + if (nextElt >= maxSize) nextElt = maxSize-1; + current = nextElt-1; + while (table[current].savings < elt.savings) { + table[current+1] = table[current]; + current--; + } + table[current+1] = elt; + table->pos = nextElt+1; + } +} + + +static U32 DiB_dictSize(const dictItem* dictList) +{ + U32 u, dictSize = 0; + for (u=1; u> shiftRatio; + saint_t errorCode; + + /* init */ + DISPLAYLEVEL(2, "\r%70s\r", ""); /* clean display line */ + if (!suffix0 || !reverseSuffix || !doneMarks || !filePos) + EXM_THROW(1, "not enough memory for DiB_trainBuffer"); + if (minRatio < MINRATIO) minRatio = MINRATIO; + memset(doneMarks, 0, bufferSize+16); + + /* sort */ + DISPLAYLEVEL(2, "sorting %u files of total size %u MB ...\n", nbFiles, (U32)(bufferSize>>20)); + errorCode = divsufsort((const sauchar_t*)buffer, suffix, (saidx_t)bufferSize); + if (errorCode != 0) EXM_THROW(2, "sort failed"); + suffix[bufferSize] = (saidx_t)bufferSize; /* leads into noise */ + suffix0[0] = (saidx_t)bufferSize; /* leads into noise */ + { + /* build reverse suffix sort */ + size_t pos; + for (pos=0; pos < bufferSize; pos++) + reverseSuffix[suffix[pos]] = (U32)pos; + /* build file pos */ + filePos[0] = 0; + for (pos=1; pospos; /* convention : nb of useful elts within dictList */ + U32 currentSize = 0; + U32 n; for (n=1; n maxDictSize) break; + } + dictList->pos = n; + } + + free(suffix0); + free(reverseSuffix); + free(doneMarks); + free(filePos); +} + + +static size_t DiB_findMaxMem(unsigned long long requiredMem) +{ + size_t step = 8 MB; + void* testmem = NULL; + + requiredMem = (((requiredMem >> 23) + 1) << 23); + requiredMem += 2 * step; + if (requiredMem > maxMemory) requiredMem = maxMemory; + + while (!testmem) { + requiredMem -= step; + testmem = malloc((size_t)requiredMem); + } + + free(testmem); + return (size_t)(requiredMem - step); +} + + +static void DiB_fillNoise(void* buffer, size_t length) +{ + unsigned acc = PRIME1; + size_t p=0;; + + for (p=0; p> 21); + } +} + + +typedef struct +{ + ZSTD_CCtx* ref; + ZSTD_CCtx* zc; + void* workPlace; /* must be BLOCKSIZE allocated */ +} EStats_ress_t; + + +static void DiB_countEStats(EStats_ress_t esr, + U32* countLit, U32* offsetcodeCount, U32* matchlengthCount, U32* litlengthCount, + const void* src, size_t srcSize) +{ + const BYTE* bytePtr; + const U32* u32Ptr; + + if (srcSize > BLOCKSIZE) srcSize = BLOCKSIZE; /* protection vs large samples */ + ZSTD_copyCCtx(esr.zc, esr.ref); + ZSTD_compressBlock(esr.zc, esr.workPlace, BLOCKSIZE, src, srcSize); + + /* count stats */ + for(bytePtr = esr.zc->seqStore.litStart; bytePtr < esr.zc->seqStore.lit; bytePtr++) + countLit[*bytePtr]++; + for(u32Ptr = esr.zc->seqStore.offsetStart; u32Ptr < esr.zc->seqStore.offset; u32Ptr++) { + BYTE offcode = (BYTE)ZSTD_highbit(*u32Ptr) + 1; + if (*u32Ptr==0) offcode=0; + offsetcodeCount[offcode]++; + } + for(bytePtr = esr.zc->seqStore.matchLengthStart; bytePtr < esr.zc->seqStore.matchLength; bytePtr++) + matchlengthCount[*bytePtr]++; + for(bytePtr = esr.zc->seqStore.litLengthStart; bytePtr < esr.zc->seqStore.litLength; bytePtr++) + litlengthCount[*bytePtr]++; +} + + +#define OFFCODE_MAX 18 +static size_t DiB_analyzeEntropy(void* dstBuffer, size_t maxDstSize, + unsigned compressionLevel, + const void* srcBuffer, const size_t* fileSizes, unsigned nbFiles, + const void* dictBuffer, size_t dictBufferSize) +{ + U32 countLit[256]; + U32 offcodeCount[MaxOff+1]; + HUF_CREATE_STATIC_CTABLE(hufTable, 255); + short offcodeNCount[MaxOff+1]; + U32 matchLengthCount[MaxML+1]; + short matchLengthNCount[MaxML+1]; + U32 litlengthCount[MaxLL+1]; + short litlengthNCount[MaxLL+1]; + EStats_ress_t esr; + ZSTD_parameters params; + U32 u, huffLog = 12, Offlog = OffFSELog, mlLog = MLFSELog, llLog = LLFSELog, total; + size_t pos = 0, errorCode; + size_t eSize = 0; + + /* init */ + for (u=0; u<256; u++) countLit[u]=1; /* any character must be described */ + for (u=0; u<=OFFCODE_MAX; u++) offcodeCount[u]=1; + for (u=0; u<=MaxML; u++) matchLengthCount[u]=1; + for (u=0; u<=MaxLL; u++) litlengthCount[u]=1; + esr.ref = ZSTD_createCCtx(); + esr.zc = ZSTD_createCCtx(); + esr.workPlace = malloc(BLOCKSIZE); + if (!esr.ref || !esr.zc || !esr.workPlace) EXM_THROW(30, "Not enough memory"); + if (compressionLevel==0) compressionLevel=g_compressionLevel_default; + params = ZSTD_getParams(compressionLevel, dictBufferSize + 15 KB); + params.strategy = ZSTD_greedy; + ZSTD_compressBegin_advanced(esr.ref, dictBuffer, dictBufferSize, params); + + /* collect stats on all files */ + for (u=0; u1) { + DiB_trainBuffer(dictList, dictListSize, + samplesBuffer, sBuffSize, + sampleSizes, nbSamples, + selectivity, (U32)targetDictSize); + + /* display best matches */ + if (g_displayLevel>= 3) { + const U32 nb = 25; + U32 u; + U32 dictContentSize = DiB_dictSize(dictList); + DISPLAYLEVEL(3, "\n %u segments found, of total size %u \n", dictList[0].pos, dictContentSize); + DISPLAYLEVEL(3, "list %u best segments \n", nb); + for (u=1; u<=nb; u++) { + U32 p = dictList[u].pos; + U32 l = dictList[u].length; + U32 d = MIN(40, l); + DISPLAYLEVEL(3, "%3u:%3u bytes at pos %8u, savings %7u bytes |", + u, l, p, dictList[u].savings); + DiB_printHex(3, (const char*)samplesBuffer+p, d); + DISPLAYLEVEL(3, "| \n"); + } } } + + /* create dictionary */ + { + U32 dictContentSize = DiB_dictSize(dictList); + size_t hSize; + BYTE* ptr; + U32 u; + + /* build dict content */ + ptr = (BYTE*)dictBuffer + maxDictSize; + for (u=1; upos; u++) { + U32 l = dictList[u].length; + ptr -= l; + if (ptr<(BYTE*)dictBuffer) return ERROR(GENERIC); /* should not happen */ + memcpy(ptr, (const char*)samplesBuffer+dictList[u].pos, l); + } + + /* fast mode dict content */ + if (selectivity==1) { /* note could also be used to complete a dictionary, but not necessarily better */ + DISPLAYLEVEL(3, "\r%70s\r", ""); /* clean display line */ + DISPLAYLEVEL(3, "Adding %u KB with fast sampling \n", (U32)(targetDictSize>>10)); + dictContentSize = (U32)DiB_fastSampling((char*)dictBuffer + g_provision_entropySize, + targetDictSize, samplesBuffer, sBuffSize); + } + + /* dictionary header */ + MEM_writeLE32(dictBuffer, ZSTD_DICT_MAGIC); + hSize = 4; + + /* entropic tables */ + DISPLAYLEVEL(2, "\r%70s\r", ""); /* clean display line */ + DISPLAYLEVEL(2, "statistics ... \n"); + hSize += DiB_analyzeEntropy((char*)dictBuffer+4, maxDictSize-4, + compressionLevel, + samplesBuffer, sampleSizes, nbSamples, + (char*)dictBuffer + maxDictSize - dictContentSize, dictContentSize); + + if (hSize + dictContentSize < maxDictSize) + memmove((char*)dictBuffer + hSize, (char*)dictBuffer + maxDictSize - dictContentSize, dictContentSize); + dictSize = MIN(maxDictSize, hSize+dictContentSize); + } + + /* clean up */ + free(dictList); + return dictSize; +} + + +/* issue : samplesBuffer need to be followed by a noisy guard band. +* work around : duplicate the buffer, and add the noise ? */ +size_t DiB_trainFromBuffer(void* dictBuffer, size_t maxDictSize, + const void* samplesBuffer, const size_t* sampleSizes, unsigned nbSamples, + DiB_params_t params) +{ + size_t sBuffSize; + void* newBuff; + size_t result; + + { unsigned u; for (u=0, sBuffSize=0; u totalSizeToLoad) benchedSize = (size_t)totalSizeToLoad; + if (benchedSize < totalSizeToLoad) + DISPLAYLEVEL(1, "Not enough memory; training on %u MB only...\n", (unsigned)(benchedSize >> 20)); + + /* Memory allocation & restrictions */ + srcBuffer = malloc(benchedSize+NOISELENGTH); /* + noise */ + if ((!fileSizes) || (!srcBuffer) || (!dictBuffer)) EXM_THROW(12, "not enough memory for DiB_trainFiles"); /* should not happen */ + + /* Load input buffer */ + DiB_loadFiles(srcBuffer, benchedSize, fileSizes, fileNamesTable, nbFiles); + DiB_fillNoise((char*)srcBuffer + benchedSize, NOISELENGTH); /* guard band, for end of buffer condition */ + + /* call buffer version */ + dictSize = DiB_trainFromBuffer_internal(dictBuffer, maxDictSize, + srcBuffer, fileSizes, nbFiles, + params); + if (DiB_isError(dictSize)) + { + DISPLAYLEVEL(1, "dictionary training failed : %s", DiB_getErrorName(dictSize)); /* should not happen */ + result = 1; + goto _cleanup; + } + + /* save dict */ + DISPLAYLEVEL(2, "Save dictionary of size %u into file %s \n", (U32)dictSize, dictFileName); + DiB_saveDict(dictFileName, dictBuffer, dictSize); + + /* clean up */ +_cleanup: + free(srcBuffer); + free(dictBuffer); + free(fileSizes); + return result; +} diff --git a/dictBuilder/dictBuilder.h b/dictBuilder/dictBuilder.h new file mode 100644 index 000000000..a022583eb --- /dev/null +++ b/dictBuilder/dictBuilder.h @@ -0,0 +1,94 @@ +/* + dictBuilder.h + Copyright (C) Yann Collet 2016 + + GPL v2 License + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 2 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License along + with this program; if not, write to the Free Software Foundation, Inc., + 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. + + You can contact the author at : + - zstd source repository : https://github.com/Cyan4973/zstd + - ztsd public forum : https://groups.google.com/forum/#!forum/lz4c +*/ + +/* This library is designed for a single-threaded console application. +* It exit() and printf() into stderr when it encounters an error condition. */ + +#ifndef DICTBUILDER_H_001 +#define DICTBUILDER_H_001 + +/*-************************************* +* Version +***************************************/ +#define DiB_VERSION_MAJOR 0 /* for breaking interface changes */ +#define DiB_VERSION_MINOR 0 /* for new (non-breaking) interface capabilities */ +#define DiB_VERSION_RELEASE 1 /* for tweaks, bug-fixes, or development */ +#define DiB_VERSION_NUMBER (DiB_VERSION_MAJOR *100*100 + DiB_VERSION_MINOR *100 + DiB_VERSION_RELEASE) +unsigned DiB_versionNumber (void); + + +/*-************************************* +* Public type +***************************************/ +typedef struct { + unsigned selectivityLevel; /* 0 means default; larger => bigger selection => larger dictionary */ + unsigned compressionLevel; /* 0 means default; target a specific zstd compression level */ +} DiB_params_t; + + +/*-************************************* +* Public functions +***************************************/ +/*! DiB_trainFromBuffer + Train a dictionary from a memory buffer @samplesBuffer + where @nbSamples samples have been stored concatenated. + Each sample size is provided into an orderly table @sampleSizes. + Resulting dictionary will be saved into @dictBuffer. + @parameters is optional and can be provided with 0 values to mean "default". + @result : size of dictionary stored into @dictBuffer (<= @dictBufferSize) + or an error code, which can be tested by DiB_isError(). + note : DiB_trainFromBuffer() will send notifications into stderr if instructed to, using DiB_setNotificationLevel() +*/ +size_t DiB_trainFromBuffer(void* dictBuffer, size_t dictBufferSize, + const void* samplesBuffer, const size_t* sampleSizes, unsigned nbSamples, + DiB_params_t parameters); + + +/*! DiB_trainFromFiles + Train a dictionary from a set of files provided by @fileNamesTable + Resulting dictionary is written into file @dictFileName. + @parameters is optional and can be provided with 0 values. + @result : 0 == ok. Any other : error. +*/ +int DiB_trainFromFiles(const char* dictFileName, unsigned maxDictSize, + const char** fileNamesTable, unsigned nbFiles, + DiB_params_t parameters); + + +/*-************************************* +* Helper functions +***************************************/ +unsigned DiB_isError(size_t errorCode); +const char* DiB_getErrorName(size_t errorCode); + +/*! DiB_setNotificationLevel + Set amount of notification to be displayed on the console. + default initial value : 0 = no console notification. + Note : not thread-safe (use a global constant) +*/ +void DiB_setNotificationLevel(unsigned l); + + +#endif diff --git a/dictBuilder/divsufsort.c b/dictBuilder/divsufsort.c new file mode 100644 index 000000000..312813597 --- /dev/null +++ b/dictBuilder/divsufsort.c @@ -0,0 +1,404 @@ +/* + * divsufsort.c for libdivsufsort + * Copyright (c) 2003-2008 Yuta Mori All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, + * copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following + * conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES + * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +/*- Compiler specifics -*/ +#ifdef __clang__ +#pragma clang diagnostic ignored "-Wshorten-64-to-32" +#endif + +/*- Dependencies -*/ +#include "divsufsort_private.h" +#ifdef _OPENMP +# include +#endif + + +/*- Private Functions -*/ + +/* Sorts suffixes of type B*. */ +static +saidx_t +sort_typeBstar(const sauchar_t *T, saidx_t *SA, + saidx_t *bucket_A, saidx_t *bucket_B, + saidx_t n) { + saidx_t *PAb, *ISAb, *buf; +#ifdef _OPENMP + saidx_t *curbuf; + saidx_t l; +#endif + saidx_t i, j, k, t, m, bufsize; + saint_t c0, c1; +#ifdef _OPENMP + saint_t d0, d1; + int tmp; +#endif + + /* Initialize bucket arrays. */ + for(i = 0; i < BUCKET_A_SIZE; ++i) { bucket_A[i] = 0; } + for(i = 0; i < BUCKET_B_SIZE; ++i) { bucket_B[i] = 0; } + + /* Count the number of occurrences of the first one or two characters of each + type A, B and B* suffix. Moreover, store the beginning position of all + type B* suffixes into the array SA. */ + for(i = n - 1, m = n, c0 = T[n - 1]; 0 <= i;) { + /* type A suffix. */ + do { ++BUCKET_A(c1 = c0); } while((0 <= --i) && ((c0 = T[i]) >= c1)); + if(0 <= i) { + /* type B* suffix. */ + ++BUCKET_BSTAR(c0, c1); + SA[--m] = i; + /* type B suffix. */ + for(--i, c1 = c0; (0 <= i) && ((c0 = T[i]) <= c1); --i, c1 = c0) { + ++BUCKET_B(c0, c1); + } + } + } + m = n - m; +/* +note: + A type B* suffix is lexicographically smaller than a type B suffix that + begins with the same first two characters. +*/ + + /* Calculate the index of start/end point of each bucket. */ + for(c0 = 0, i = 0, j = 0; c0 < ALPHABET_SIZE; ++c0) { + t = i + BUCKET_A(c0); + BUCKET_A(c0) = i + j; /* start point */ + i = t + BUCKET_B(c0, c0); + for(c1 = c0 + 1; c1 < ALPHABET_SIZE; ++c1) { + j += BUCKET_BSTAR(c0, c1); + BUCKET_BSTAR(c0, c1) = j; /* end point */ + i += BUCKET_B(c0, c1); + } + } + + if(0 < m) { + /* Sort the type B* suffixes by their first two characters. */ + PAb = SA + n - m; ISAb = SA + m; + for(i = m - 2; 0 <= i; --i) { + t = PAb[i], c0 = T[t], c1 = T[t + 1]; + SA[--BUCKET_BSTAR(c0, c1)] = i; + } + t = PAb[m - 1], c0 = T[t], c1 = T[t + 1]; + SA[--BUCKET_BSTAR(c0, c1)] = m - 1; + + /* Sort the type B* substrings using sssort. */ +#ifdef _OPENMP + tmp = omp_get_max_threads(); + buf = SA + m, bufsize = (n - (2 * m)) / tmp; + c0 = ALPHABET_SIZE - 2, c1 = ALPHABET_SIZE - 1, j = m; +#pragma omp parallel default(shared) private(curbuf, k, l, d0, d1, tmp) + { + tmp = omp_get_thread_num(); + curbuf = buf + tmp * bufsize; + k = 0; + for(;;) { + #pragma omp critical(sssort_lock) + { + if(0 < (l = j)) { + d0 = c0, d1 = c1; + do { + k = BUCKET_BSTAR(d0, d1); + if(--d1 <= d0) { + d1 = ALPHABET_SIZE - 1; + if(--d0 < 0) { break; } + } + } while(((l - k) <= 1) && (0 < (l = k))); + c0 = d0, c1 = d1, j = k; + } + } + if(l == 0) { break; } + sssort(T, PAb, SA + k, SA + l, + curbuf, bufsize, 2, n, *(SA + k) == (m - 1)); + } + } +#else + buf = SA + m, bufsize = n - (2 * m); + for(c0 = ALPHABET_SIZE - 2, j = m; 0 < j; --c0) { + for(c1 = ALPHABET_SIZE - 1; c0 < c1; j = i, --c1) { + i = BUCKET_BSTAR(c0, c1); + if(1 < (j - i)) { + sssort(T, PAb, SA + i, SA + j, + buf, bufsize, 2, n, *(SA + i) == (m - 1)); + } + } + } +#endif + + /* Compute ranks of type B* substrings. */ + for(i = m - 1; 0 <= i; --i) { + if(0 <= SA[i]) { + j = i; + do { ISAb[SA[i]] = i; } while((0 <= --i) && (0 <= SA[i])); + SA[i + 1] = i - j; + if(i <= 0) { break; } + } + j = i; + do { ISAb[SA[i] = ~SA[i]] = j; } while(SA[--i] < 0); + ISAb[SA[i]] = j; + } + + /* Construct the inverse suffix array of type B* suffixes using trsort. */ + trsort(ISAb, SA, m, 1); + + /* Set the sorted order of tyoe B* suffixes. */ + for(i = n - 1, j = m, c0 = T[n - 1]; 0 <= i;) { + for(--i, c1 = c0; (0 <= i) && ((c0 = T[i]) >= c1); --i, c1 = c0) { } + if(0 <= i) { + t = i; + for(--i, c1 = c0; (0 <= i) && ((c0 = T[i]) <= c1); --i, c1 = c0) { } + SA[ISAb[--j]] = ((t == 0) || (1 < (t - i))) ? t : ~t; + } + } + + /* Calculate the index of start/end point of each bucket. */ + BUCKET_B(ALPHABET_SIZE - 1, ALPHABET_SIZE - 1) = n; /* end point */ + for(c0 = ALPHABET_SIZE - 2, k = m - 1; 0 <= c0; --c0) { + i = BUCKET_A(c0 + 1) - 1; + for(c1 = ALPHABET_SIZE - 1; c0 < c1; --c1) { + t = i - BUCKET_B(c0, c1); + BUCKET_B(c0, c1) = i; /* end point */ + + /* Move all type B* suffixes to the correct position. */ + for(i = t, j = BUCKET_BSTAR(c0, c1); + j <= k; + --i, --k) { SA[i] = SA[k]; } + } + BUCKET_BSTAR(c0, c0 + 1) = i - BUCKET_B(c0, c0) + 1; /* start point */ + BUCKET_B(c0, c0) = i; /* end point */ + } + } + + return m; +} + +/* Constructs the suffix array by using the sorted order of type B* suffixes. */ +static +void +construct_SA(const sauchar_t *T, saidx_t *SA, + saidx_t *bucket_A, saidx_t *bucket_B, + saidx_t n, saidx_t m) { + saidx_t *i, *j, *k; + saidx_t s; + saint_t c0, c1, c2; + + if(0 < m) { + /* Construct the sorted order of type B suffixes by using + the sorted order of type B* suffixes. */ + for(c1 = ALPHABET_SIZE - 2; 0 <= c1; --c1) { + /* Scan the suffix array from right to left. */ + for(i = SA + BUCKET_BSTAR(c1, c1 + 1), + j = SA + BUCKET_A(c1 + 1) - 1, k = NULL, c2 = -1; + i <= j; + --j) { + if(0 < (s = *j)) { + assert(T[s] == c1); + assert(((s + 1) < n) && (T[s] <= T[s + 1])); + assert(T[s - 1] <= T[s]); + *j = ~s; + c0 = T[--s]; + if((0 < s) && (T[s - 1] > c0)) { s = ~s; } + if(c0 != c2) { + if(0 <= c2) { BUCKET_B(c2, c1) = k - SA; } + k = SA + BUCKET_B(c2 = c0, c1); + } + assert(k < j); + *k-- = s; + } else { + assert(((s == 0) && (T[s] == c1)) || (s < 0)); + *j = ~s; + } + } + } + } + + /* Construct the suffix array by using + the sorted order of type B suffixes. */ + k = SA + BUCKET_A(c2 = T[n - 1]); + *k++ = (T[n - 2] < c2) ? ~(n - 1) : (n - 1); + /* Scan the suffix array from left to right. */ + for(i = SA, j = SA + n; i < j; ++i) { + if(0 < (s = *i)) { + assert(T[s - 1] >= T[s]); + c0 = T[--s]; + if((s == 0) || (T[s - 1] < c0)) { s = ~s; } + if(c0 != c2) { + BUCKET_A(c2) = k - SA; + k = SA + BUCKET_A(c2 = c0); + } + assert(i < k); + *k++ = s; + } else { + assert(s < 0); + *i = ~s; + } + } +} + +/* Constructs the burrows-wheeler transformed string directly + by using the sorted order of type B* suffixes. */ +static +saidx_t +construct_BWT(const sauchar_t *T, saidx_t *SA, + saidx_t *bucket_A, saidx_t *bucket_B, + saidx_t n, saidx_t m) { + saidx_t *i, *j, *k, *orig; + saidx_t s; + saint_t c0, c1, c2; + + if(0 < m) { + /* Construct the sorted order of type B suffixes by using + the sorted order of type B* suffixes. */ + for(c1 = ALPHABET_SIZE - 2; 0 <= c1; --c1) { + /* Scan the suffix array from right to left. */ + for(i = SA + BUCKET_BSTAR(c1, c1 + 1), + j = SA + BUCKET_A(c1 + 1) - 1, k = NULL, c2 = -1; + i <= j; + --j) { + if(0 < (s = *j)) { + assert(T[s] == c1); + assert(((s + 1) < n) && (T[s] <= T[s + 1])); + assert(T[s - 1] <= T[s]); + c0 = T[--s]; + *j = ~((saidx_t)c0); + if((0 < s) && (T[s - 1] > c0)) { s = ~s; } + if(c0 != c2) { + if(0 <= c2) { BUCKET_B(c2, c1) = k - SA; } + k = SA + BUCKET_B(c2 = c0, c1); + } + assert(k < j); + *k-- = s; + } else if(s != 0) { + *j = ~s; +#ifndef NDEBUG + } else { + assert(T[s] == c1); +#endif + } + } + } + } + + /* Construct the BWTed string by using + the sorted order of type B suffixes. */ + k = SA + BUCKET_A(c2 = T[n - 1]); + *k++ = (T[n - 2] < c2) ? ~((saidx_t)T[n - 2]) : (n - 1); + /* Scan the suffix array from left to right. */ + for(i = SA, j = SA + n, orig = SA; i < j; ++i) { + if(0 < (s = *i)) { + assert(T[s - 1] >= T[s]); + c0 = T[--s]; + *i = c0; + if((0 < s) && (T[s - 1] < c0)) { s = ~((saidx_t)T[s - 1]); } + if(c0 != c2) { + BUCKET_A(c2) = k - SA; + k = SA + BUCKET_A(c2 = c0); + } + assert(i < k); + *k++ = s; + } else if(s != 0) { + *i = ~s; + } else { + orig = i; + } + } + + return orig - SA; +} + + +/*---------------------------------------------------------------------------*/ + +/*- Function -*/ + +saint_t +divsufsort(const sauchar_t *T, saidx_t *SA, saidx_t n) { + saidx_t *bucket_A, *bucket_B; + saidx_t m; + saint_t err = 0; + + /* Check arguments. */ + if((T == NULL) || (SA == NULL) || (n < 0)) { return -1; } + else if(n == 0) { return 0; } + else if(n == 1) { SA[0] = 0; return 0; } + else if(n == 2) { m = (T[0] < T[1]); SA[m ^ 1] = 0, SA[m] = 1; return 0; } + + bucket_A = (saidx_t *)malloc(BUCKET_A_SIZE * sizeof(saidx_t)); + bucket_B = (saidx_t *)malloc(BUCKET_B_SIZE * sizeof(saidx_t)); + + /* Suffixsort. */ + if((bucket_A != NULL) && (bucket_B != NULL)) { + m = sort_typeBstar(T, SA, bucket_A, bucket_B, n); + construct_SA(T, SA, bucket_A, bucket_B, n, m); + } else { + err = -2; + } + + free(bucket_B); + free(bucket_A); + + return err; +} + +saidx_t +divbwt(const sauchar_t *T, sauchar_t *U, saidx_t *A, saidx_t n) { + saidx_t *B; + saidx_t *bucket_A, *bucket_B; + saidx_t m, pidx, i; + + /* Check arguments. */ + if((T == NULL) || (U == NULL) || (n < 0)) { return -1; } + else if(n <= 1) { if(n == 1) { U[0] = T[0]; } return n; } + + if((B = A) == NULL) { B = (saidx_t *)malloc((size_t)(n + 1) * sizeof(saidx_t)); } + bucket_A = (saidx_t *)malloc(BUCKET_A_SIZE * sizeof(saidx_t)); + bucket_B = (saidx_t *)malloc(BUCKET_B_SIZE * sizeof(saidx_t)); + + /* Burrows-Wheeler Transform. */ + if((B != NULL) && (bucket_A != NULL) && (bucket_B != NULL)) { + m = sort_typeBstar(T, B, bucket_A, bucket_B, n); + pidx = construct_BWT(T, B, bucket_A, bucket_B, n, m); + + /* Copy to output string. */ + U[0] = T[n - 1]; + for(i = 0; i < pidx; ++i) { U[i + 1] = (sauchar_t)B[i]; } + for(i += 1; i < n; ++i) { U[i] = (sauchar_t)B[i]; } + pidx += 1; + } else { + pidx = -2; + } + + free(bucket_B); + free(bucket_A); + if(A == NULL) { free(B); } + + return pidx; +} + +const char * +divsufsort_version(void) { + return PROJECT_VERSION_FULL; +} diff --git a/dictBuilder/divsufsort.h b/dictBuilder/divsufsort.h new file mode 100644 index 000000000..6d3e64870 --- /dev/null +++ b/dictBuilder/divsufsort.h @@ -0,0 +1,180 @@ +/* + * divsufsort.h for libdivsufsort + * Copyright (c) 2003-2008 Yuta Mori All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, + * copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following + * conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES + * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#ifndef _DIVSUFSORT_H +#define _DIVSUFSORT_H 1 + +#ifdef __cplusplus +extern "C" { +#endif /* __cplusplus */ + +#include + +#ifndef DIVSUFSORT_API +# ifdef DIVSUFSORT_BUILD_DLL +# define DIVSUFSORT_API +# else +# define DIVSUFSORT_API +# endif +#endif + +/*- Datatypes -*/ +#ifndef SAUCHAR_T +#define SAUCHAR_T +typedef uint8_t sauchar_t; +#endif /* SAUCHAR_T */ +#ifndef SAINT_T +#define SAINT_T +typedef int32_t saint_t; +#endif /* SAINT_T */ +#ifndef SAIDX_T +#define SAIDX_T +typedef int32_t saidx_t; +#endif /* SAIDX_T */ +#ifndef PRIdSAINT_T +#define PRIdSAINT_T PRId32 +#endif /* PRIdSAINT_T */ +#ifndef PRIdSAIDX_T +#define PRIdSAIDX_T PRId32 +#endif /* PRIdSAIDX_T */ + + +/*- Prototypes -*/ + +/** + * Constructs the suffix array of a given string. + * @param T[0..n-1] The input string. + * @param SA[0..n-1] The output array of suffixes. + * @param n The length of the given string. + * @return 0 if no error occurred, -1 or -2 otherwise. + */ +DIVSUFSORT_API +saint_t +divsufsort(const sauchar_t *T, saidx_t *SA, saidx_t n); + +/** + * Constructs the burrows-wheeler transformed string of a given string. + * @param T[0..n-1] The input string. + * @param U[0..n-1] The output string. (can be T) + * @param A[0..n-1] The temporary array. (can be NULL) + * @param n The length of the given string. + * @return The primary index if no error occurred, -1 or -2 otherwise. + */ +DIVSUFSORT_API +saidx_t +divbwt(const sauchar_t *T, sauchar_t *U, saidx_t *A, saidx_t n); + +/** + * Returns the version of the divsufsort library. + * @return The version number string. + */ +DIVSUFSORT_API +const char * +divsufsort_version(void); + + +/** + * Constructs the burrows-wheeler transformed string of a given string and suffix array. + * @param T[0..n-1] The input string. + * @param U[0..n-1] The output string. (can be T) + * @param SA[0..n-1] The suffix array. (can be NULL) + * @param n The length of the given string. + * @param idx The output primary index. + * @return 0 if no error occurred, -1 or -2 otherwise. + */ +DIVSUFSORT_API +saint_t +bw_transform(const sauchar_t *T, sauchar_t *U, + saidx_t *SA /* can NULL */, + saidx_t n, saidx_t *idx); + +/** + * Inverse BW-transforms a given BWTed string. + * @param T[0..n-1] The input string. + * @param U[0..n-1] The output string. (can be T) + * @param A[0..n-1] The temporary array. (can be NULL) + * @param n The length of the given string. + * @param idx The primary index. + * @return 0 if no error occurred, -1 or -2 otherwise. + */ +DIVSUFSORT_API +saint_t +inverse_bw_transform(const sauchar_t *T, sauchar_t *U, + saidx_t *A /* can NULL */, + saidx_t n, saidx_t idx); + +/** + * Checks the correctness of a given suffix array. + * @param T[0..n-1] The input string. + * @param SA[0..n-1] The input suffix array. + * @param n The length of the given string. + * @param verbose The verbose mode. + * @return 0 if no error occurred. + */ +DIVSUFSORT_API +saint_t +sufcheck(const sauchar_t *T, const saidx_t *SA, saidx_t n, saint_t verbose); + +/** + * Search for the pattern P in the string T. + * @param T[0..Tsize-1] The input string. + * @param Tsize The length of the given string. + * @param P[0..Psize-1] The input pattern string. + * @param Psize The length of the given pattern string. + * @param SA[0..SAsize-1] The input suffix array. + * @param SAsize The length of the given suffix array. + * @param idx The output index. + * @return The count of matches if no error occurred, -1 otherwise. + */ +DIVSUFSORT_API +saidx_t +sa_search(const sauchar_t *T, saidx_t Tsize, + const sauchar_t *P, saidx_t Psize, + const saidx_t *SA, saidx_t SAsize, + saidx_t *left); + +/** + * Search for the character c in the string T. + * @param T[0..Tsize-1] The input string. + * @param Tsize The length of the given string. + * @param SA[0..SAsize-1] The input suffix array. + * @param SAsize The length of the given suffix array. + * @param c The input character. + * @param idx The output index. + * @return The count of matches if no error occurred, -1 otherwise. + */ +DIVSUFSORT_API +saidx_t +sa_simplesearch(const sauchar_t *T, saidx_t Tsize, + const saidx_t *SA, saidx_t SAsize, + saint_t c, saidx_t *left); + + +#ifdef __cplusplus +} /* extern "C" */ +#endif /* __cplusplus */ + +#endif /* _DIVSUFSORT_H */ diff --git a/dictBuilder/divsufsort_private.h b/dictBuilder/divsufsort_private.h new file mode 100644 index 000000000..0a18f6d28 --- /dev/null +++ b/dictBuilder/divsufsort_private.h @@ -0,0 +1,212 @@ +/* + * divsufsort_private.h for libdivsufsort + * Copyright (c) 2003-2008 Yuta Mori All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, + * copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following + * conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES + * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#ifndef _DIVSUFSORT_PRIVATE_H +#define _DIVSUFSORT_PRIVATE_H 1 + +#ifdef __cplusplus +extern "C" { +#endif /* __cplusplus */ + +/* ************************* +* Includes +***************************/ +#include +#include /* unconditional */ +#include +#include "config.h" /* unconditional */ + + +#if HAVE_STRING_H +# include +#endif +#if HAVE_MEMORY_H +# include +#endif +#if HAVE_STDDEF_H +# include +#endif +#if HAVE_STRINGS_H +# ifdef _WIN32 +# include +# else +# include +# endif +#endif +#if HAVE_INTTYPES_H +# include +#else +# if HAVE_STDINT_H +# include +# endif +#endif +#if defined(BUILD_DIVSUFSORT64) +# include "divsufsort64.h" +# ifndef SAIDX_T +# define SAIDX_T +# define saidx_t saidx64_t +# endif /* SAIDX_T */ +# ifndef PRIdSAIDX_T +# define PRIdSAIDX_T PRIdSAIDX64_T +# endif /* PRIdSAIDX_T */ +# define divsufsort divsufsort64 +# define divbwt divbwt64 +# define divsufsort_version divsufsort64_version +# define bw_transform bw_transform64 +# define inverse_bw_transform inverse_bw_transform64 +# define sufcheck sufcheck64 +# define sa_search sa_search64 +# define sa_simplesearch sa_simplesearch64 +# define sssort sssort64 +# define trsort trsort64 +#else +# include "divsufsort.h" +#endif + + +/*- Constants -*/ +#if !defined(UINT8_MAX) +# define UINT8_MAX (255) +#endif /* UINT8_MAX */ +#if defined(ALPHABET_SIZE) && (ALPHABET_SIZE < 1) +# undef ALPHABET_SIZE +#endif +#if !defined(ALPHABET_SIZE) +# define ALPHABET_SIZE (UINT8_MAX + 1) +#endif +/* for divsufsort.c */ +#define BUCKET_A_SIZE (ALPHABET_SIZE) +#define BUCKET_B_SIZE (ALPHABET_SIZE * ALPHABET_SIZE) +/* for sssort.c */ +#if defined(SS_INSERTIONSORT_THRESHOLD) +# if SS_INSERTIONSORT_THRESHOLD < 1 +# undef SS_INSERTIONSORT_THRESHOLD +# define SS_INSERTIONSORT_THRESHOLD (1) +# endif +#else +# define SS_INSERTIONSORT_THRESHOLD (8) +#endif +#if defined(SS_BLOCKSIZE) +# if SS_BLOCKSIZE < 0 +# undef SS_BLOCKSIZE +# define SS_BLOCKSIZE (0) +# elif 32768 <= SS_BLOCKSIZE +# undef SS_BLOCKSIZE +# define SS_BLOCKSIZE (32767) +# endif +#else +# define SS_BLOCKSIZE (1024) +#endif +/* minstacksize = log(SS_BLOCKSIZE) / log(3) * 2 */ +#if SS_BLOCKSIZE == 0 +# if defined(BUILD_DIVSUFSORT64) +# define SS_MISORT_STACKSIZE (96) +# else +# define SS_MISORT_STACKSIZE (64) +# endif +#elif SS_BLOCKSIZE <= 4096 +# define SS_MISORT_STACKSIZE (16) +#else +# define SS_MISORT_STACKSIZE (24) +#endif +#if defined(BUILD_DIVSUFSORT64) +# define SS_SMERGE_STACKSIZE (64) +#else +# define SS_SMERGE_STACKSIZE (32) +#endif +/* for trsort.c */ +#define TR_INSERTIONSORT_THRESHOLD (8) +#if defined(BUILD_DIVSUFSORT64) +# define TR_STACKSIZE (96) +#else +# define TR_STACKSIZE (64) +#endif + + +/*- Macros -*/ +#ifndef SWAP +# define SWAP(_a, _b) do { t = (_a); (_a) = (_b); (_b) = t; } while(0) +#endif /* SWAP */ +#ifndef MIN +# define MIN(_a, _b) (((_a) < (_b)) ? (_a) : (_b)) +#endif /* MIN */ +#ifndef MAX +# define MAX(_a, _b) (((_a) > (_b)) ? (_a) : (_b)) +#endif /* MAX */ +#define STACK_PUSH(_a, _b, _c, _d)\ + do {\ + assert(ssize < STACK_SIZE);\ + stack[ssize].a = (_a), stack[ssize].b = (_b),\ + stack[ssize].c = (_c), stack[ssize++].d = (_d);\ + } while(0) +#define STACK_PUSH5(_a, _b, _c, _d, _e)\ + do {\ + assert(ssize < STACK_SIZE);\ + stack[ssize].a = (_a), stack[ssize].b = (_b),\ + stack[ssize].c = (_c), stack[ssize].d = (_d), stack[ssize++].e = (_e);\ + } while(0) +#define STACK_POP(_a, _b, _c, _d)\ + do {\ + assert(0 <= ssize);\ + if(ssize == 0) { return; }\ + (_a) = stack[--ssize].a, (_b) = stack[ssize].b,\ + (_c) = stack[ssize].c, (_d) = stack[ssize].d;\ + } while(0) +#define STACK_POP5(_a, _b, _c, _d, _e)\ + do {\ + assert(0 <= ssize);\ + if(ssize == 0) { return; }\ + (_a) = stack[--ssize].a, (_b) = stack[ssize].b,\ + (_c) = stack[ssize].c, (_d) = stack[ssize].d, (_e) = stack[ssize].e;\ + } while(0) +/* for divsufsort.c */ +#define BUCKET_A(_c0) bucket_A[(_c0)] +#if ALPHABET_SIZE == 256 +#define BUCKET_B(_c0, _c1) (bucket_B[((_c1) << 8) | (_c0)]) +#define BUCKET_BSTAR(_c0, _c1) (bucket_B[((_c0) << 8) | (_c1)]) +#else +#define BUCKET_B(_c0, _c1) (bucket_B[(_c1) * ALPHABET_SIZE + (_c0)]) +#define BUCKET_BSTAR(_c0, _c1) (bucket_B[(_c0) * ALPHABET_SIZE + (_c1)]) +#endif + + +/*- Private Prototypes -*/ +/* sssort.c */ +void +sssort(const sauchar_t *Td, const saidx_t *PA, + saidx_t *first, saidx_t *last, + saidx_t *buf, saidx_t bufsize, + saidx_t depth, saidx_t n, saint_t lastsuffix); +/* trsort.c */ +void +trsort(saidx_t *ISA, saidx_t *SA, saidx_t n, saidx_t depth); + + +#ifdef __cplusplus +} /* extern "C" */ +#endif /* __cplusplus */ + +#endif /* _DIVSUFSORT_PRIVATE_H */ diff --git a/dictBuilder/lfs.h b/dictBuilder/lfs.h new file mode 100644 index 000000000..7ef88f0b4 --- /dev/null +++ b/dictBuilder/lfs.h @@ -0,0 +1,56 @@ +/* + * lfs.h for libdivsufsort + * Copyright (c) 2003-2008 Yuta Mori All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, + * copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following + * conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES + * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#ifndef _LFS_H +#define _LFS_H 1 + +#ifdef __cplusplus +extern "C" { +#endif /* __cplusplus */ + +#ifndef __STRICT_ANSI__ +# define LFS_OFF_T off_t +# define LFS_FOPEN fopen +# define LFS_FTELL ftello +# define LFS_FSEEK fseeko +# define LFS_PRId PRIdMAX +#else +# define LFS_OFF_T long +# define LFS_FOPEN fopen +# define LFS_FTELL ftell +# define LFS_FSEEK fseek +# define LFS_PRId "ld" +#endif +#ifndef PRIdOFF_T +# define PRIdOFF_T LFS_PRId +#endif + + +#ifdef __cplusplus +} /* extern "C" */ +#endif /* __cplusplus */ + +#endif /* _LFS_H */ diff --git a/dictBuilder/sssort.c b/dictBuilder/sssort.c new file mode 100644 index 000000000..07cb5bd1b --- /dev/null +++ b/dictBuilder/sssort.c @@ -0,0 +1,844 @@ +/* + * sssort.c for libdivsufsort + * Copyright (c) 2003-2008 Yuta Mori All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, + * copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following + * conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES + * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +/*- Compiler specifics -*/ +#ifdef __clang__ +#pragma clang diagnostic ignored "-Wshorten-64-to-32" +#endif + +#if defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) +/* inline is defined */ +#elif defined(_MSC_VER) +# define inline __inline +#else +# define inline /* disable inline */ +#endif + +#ifdef _MSC_VER /* Visual Studio */ +# pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */ +# define FORCE_INLINE static __forceinline +#else +# if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L /* C99 */ +# ifdef __GNUC__ +# define FORCE_INLINE static inline __attribute__((always_inline)) +# else +# define FORCE_INLINE static inline +# endif +# else +# define FORCE_INLINE static +# endif /* __STDC_VERSION__ */ +#endif + +/*- Dependencies -*/ +#include "divsufsort_private.h" + + +/*- Private Functions -*/ + +static const saint_t lg_table[256]= { + -1,0,1,1,2,2,2,2,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4, + 5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, + 6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, + 6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, + 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, + 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, + 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, + 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7 +}; + +#if (SS_BLOCKSIZE == 0) || (SS_INSERTIONSORT_THRESHOLD < SS_BLOCKSIZE) + +static INLINE +saint_t +ss_ilg(saidx_t n) { +#if SS_BLOCKSIZE == 0 +# if defined(BUILD_DIVSUFSORT64) + return (n >> 32) ? + ((n >> 48) ? + ((n >> 56) ? + 56 + lg_table[(n >> 56) & 0xff] : + 48 + lg_table[(n >> 48) & 0xff]) : + ((n >> 40) ? + 40 + lg_table[(n >> 40) & 0xff] : + 32 + lg_table[(n >> 32) & 0xff])) : + ((n & 0xffff0000) ? + ((n & 0xff000000) ? + 24 + lg_table[(n >> 24) & 0xff] : + 16 + lg_table[(n >> 16) & 0xff]) : + ((n & 0x0000ff00) ? + 8 + lg_table[(n >> 8) & 0xff] : + 0 + lg_table[(n >> 0) & 0xff])); +# else + return (n & 0xffff0000) ? + ((n & 0xff000000) ? + 24 + lg_table[(n >> 24) & 0xff] : + 16 + lg_table[(n >> 16) & 0xff]) : + ((n & 0x0000ff00) ? + 8 + lg_table[(n >> 8) & 0xff] : + 0 + lg_table[(n >> 0) & 0xff]); +# endif +#elif SS_BLOCKSIZE < 256 + return lg_table[n]; +#else + return (n & 0xff00) ? + 8 + lg_table[(n >> 8) & 0xff] : + 0 + lg_table[(n >> 0) & 0xff]; +#endif +} + +#endif /* (SS_BLOCKSIZE == 0) || (SS_INSERTIONSORT_THRESHOLD < SS_BLOCKSIZE) */ + +#if SS_BLOCKSIZE != 0 + +static const saint_t sqq_table[256] = { + 0, 16, 22, 27, 32, 35, 39, 42, 45, 48, 50, 53, 55, 57, 59, 61, + 64, 65, 67, 69, 71, 73, 75, 76, 78, 80, 81, 83, 84, 86, 87, 89, + 90, 91, 93, 94, 96, 97, 98, 99, 101, 102, 103, 104, 106, 107, 108, 109, +110, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, +128, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, +143, 144, 144, 145, 146, 147, 148, 149, 150, 150, 151, 152, 153, 154, 155, 155, +156, 157, 158, 159, 160, 160, 161, 162, 163, 163, 164, 165, 166, 167, 167, 168, +169, 170, 170, 171, 172, 173, 173, 174, 175, 176, 176, 177, 178, 178, 179, 180, +181, 181, 182, 183, 183, 184, 185, 185, 186, 187, 187, 188, 189, 189, 190, 191, +192, 192, 193, 193, 194, 195, 195, 196, 197, 197, 198, 199, 199, 200, 201, 201, +202, 203, 203, 204, 204, 205, 206, 206, 207, 208, 208, 209, 209, 210, 211, 211, +212, 212, 213, 214, 214, 215, 215, 216, 217, 217, 218, 218, 219, 219, 220, 221, +221, 222, 222, 223, 224, 224, 225, 225, 226, 226, 227, 227, 228, 229, 229, 230, +230, 231, 231, 232, 232, 233, 234, 234, 235, 235, 236, 236, 237, 237, 238, 238, +239, 240, 240, 241, 241, 242, 242, 243, 243, 244, 244, 245, 245, 246, 246, 247, +247, 248, 248, 249, 249, 250, 250, 251, 251, 252, 252, 253, 253, 254, 254, 255 +}; + +static INLINE +saidx_t +ss_isqrt(saidx_t x) { + saidx_t y, e; + + if(x >= (SS_BLOCKSIZE * SS_BLOCKSIZE)) { return SS_BLOCKSIZE; } + e = (x & 0xffff0000) ? + ((x & 0xff000000) ? + 24 + lg_table[(x >> 24) & 0xff] : + 16 + lg_table[(x >> 16) & 0xff]) : + ((x & 0x0000ff00) ? + 8 + lg_table[(x >> 8) & 0xff] : + 0 + lg_table[(x >> 0) & 0xff]); + + if(e >= 16) { + y = sqq_table[x >> ((e - 6) - (e & 1))] << ((e >> 1) - 7); + if(e >= 24) { y = (y + 1 + x / y) >> 1; } + y = (y + 1 + x / y) >> 1; + } else if(e >= 8) { + y = (sqq_table[x >> ((e - 6) - (e & 1))] >> (7 - (e >> 1))) + 1; + } else { + return sqq_table[x] >> 4; + } + + return (x < (y * y)) ? y - 1 : y; +} + +#endif /* SS_BLOCKSIZE != 0 */ + + +/*---------------------------------------------------------------------------*/ + +/* Compares two suffixes. */ +static INLINE +saint_t +ss_compare(const sauchar_t *T, + const saidx_t *p1, const saidx_t *p2, + saidx_t depth) { + const sauchar_t *U1, *U2, *U1n, *U2n; + + for(U1 = T + depth + *p1, + U2 = T + depth + *p2, + U1n = T + *(p1 + 1) + 2, + U2n = T + *(p2 + 1) + 2; + (U1 < U1n) && (U2 < U2n) && (*U1 == *U2); + ++U1, ++U2) { + } + + return U1 < U1n ? + (U2 < U2n ? *U1 - *U2 : 1) : + (U2 < U2n ? -1 : 0); +} + + +/*---------------------------------------------------------------------------*/ + +#if (SS_BLOCKSIZE != 1) && (SS_INSERTIONSORT_THRESHOLD != 1) + +/* Insertionsort for small size groups */ +static +void +ss_insertionsort(const sauchar_t *T, const saidx_t *PA, + saidx_t *first, saidx_t *last, saidx_t depth) { + saidx_t *i, *j; + saidx_t t; + saint_t r; + + for(i = last - 2; first <= i; --i) { + for(t = *i, j = i + 1; 0 < (r = ss_compare(T, PA + t, PA + *j, depth));) { + do { *(j - 1) = *j; } while((++j < last) && (*j < 0)); + if(last <= j) { break; } + } + if(r == 0) { *j = ~*j; } + *(j - 1) = t; + } +} + +#endif /* (SS_BLOCKSIZE != 1) && (SS_INSERTIONSORT_THRESHOLD != 1) */ + + +/*---------------------------------------------------------------------------*/ + +#if (SS_BLOCKSIZE == 0) || (SS_INSERTIONSORT_THRESHOLD < SS_BLOCKSIZE) + +static INLINE +void +ss_fixdown(const sauchar_t *Td, const saidx_t *PA, + saidx_t *SA, saidx_t i, saidx_t size) { + saidx_t j, k; + saidx_t v; + saint_t c, d, e; + + for(v = SA[i], c = Td[PA[v]]; (j = 2 * i + 1) < size; SA[i] = SA[k], i = k) { + d = Td[PA[SA[k = j++]]]; + if(d < (e = Td[PA[SA[j]]])) { k = j; d = e; } + if(d <= c) { break; } + } + SA[i] = v; +} + +/* Simple top-down heapsort. */ +static +void +ss_heapsort(const sauchar_t *Td, const saidx_t *PA, saidx_t *SA, saidx_t size) { + saidx_t i, m; + saidx_t t; + + m = size; + if((size % 2) == 0) { + m--; + if(Td[PA[SA[m / 2]]] < Td[PA[SA[m]]]) { SWAP(SA[m], SA[m / 2]); } + } + + for(i = m / 2 - 1; 0 <= i; --i) { ss_fixdown(Td, PA, SA, i, m); } + if((size % 2) == 0) { SWAP(SA[0], SA[m]); ss_fixdown(Td, PA, SA, 0, m); } + for(i = m - 1; 0 < i; --i) { + t = SA[0], SA[0] = SA[i]; + ss_fixdown(Td, PA, SA, 0, i); + SA[i] = t; + } +} + + +/*---------------------------------------------------------------------------*/ + +/* Returns the median of three elements. */ +static INLINE +saidx_t * +ss_median3(const sauchar_t *Td, const saidx_t *PA, + saidx_t *v1, saidx_t *v2, saidx_t *v3) { + saidx_t *t; + if(Td[PA[*v1]] > Td[PA[*v2]]) { SWAP(v1, v2); } + if(Td[PA[*v2]] > Td[PA[*v3]]) { + if(Td[PA[*v1]] > Td[PA[*v3]]) { return v1; } + else { return v3; } + } + return v2; +} + +/* Returns the median of five elements. */ +static INLINE +saidx_t * +ss_median5(const sauchar_t *Td, const saidx_t *PA, + saidx_t *v1, saidx_t *v2, saidx_t *v3, saidx_t *v4, saidx_t *v5) { + saidx_t *t; + if(Td[PA[*v2]] > Td[PA[*v3]]) { SWAP(v2, v3); } + if(Td[PA[*v4]] > Td[PA[*v5]]) { SWAP(v4, v5); } + if(Td[PA[*v2]] > Td[PA[*v4]]) { SWAP(v2, v4); SWAP(v3, v5); } + if(Td[PA[*v1]] > Td[PA[*v3]]) { SWAP(v1, v3); } + if(Td[PA[*v1]] > Td[PA[*v4]]) { SWAP(v1, v4); SWAP(v3, v5); } + if(Td[PA[*v3]] > Td[PA[*v4]]) { return v4; } + return v3; +} + +/* Returns the pivot element. */ +static INLINE +saidx_t * +ss_pivot(const sauchar_t *Td, const saidx_t *PA, saidx_t *first, saidx_t *last) { + saidx_t *middle; + saidx_t t; + + t = last - first; + middle = first + t / 2; + + if(t <= 512) { + if(t <= 32) { + return ss_median3(Td, PA, first, middle, last - 1); + } else { + t >>= 2; + return ss_median5(Td, PA, first, first + t, middle, last - 1 - t, last - 1); + } + } + t >>= 3; + first = ss_median3(Td, PA, first, first + t, first + (t << 1)); + middle = ss_median3(Td, PA, middle - t, middle, middle + t); + last = ss_median3(Td, PA, last - 1 - (t << 1), last - 1 - t, last - 1); + return ss_median3(Td, PA, first, middle, last); +} + + +/*---------------------------------------------------------------------------*/ + +/* Binary partition for substrings. */ +static INLINE +saidx_t * +ss_partition(const saidx_t *PA, + saidx_t *first, saidx_t *last, saidx_t depth) { + saidx_t *a, *b; + saidx_t t; + for(a = first - 1, b = last;;) { + for(; (++a < b) && ((PA[*a] + depth) >= (PA[*a + 1] + 1));) { *a = ~*a; } + for(; (a < --b) && ((PA[*b] + depth) < (PA[*b + 1] + 1));) { } + if(b <= a) { break; } + t = ~*b; + *b = *a; + *a = t; + } + if(first < a) { *first = ~*first; } + return a; +} + +/* Multikey introsort for medium size groups. */ +static +void +ss_mintrosort(const sauchar_t *T, const saidx_t *PA, + saidx_t *first, saidx_t *last, + saidx_t depth) { +#define STACK_SIZE SS_MISORT_STACKSIZE + struct { saidx_t *a, *b, c; saint_t d; } stack[STACK_SIZE]; + const sauchar_t *Td; + saidx_t *a, *b, *c, *d, *e, *f; + saidx_t s, t; + saint_t ssize; + saint_t limit; + saint_t v, x = 0; + + for(ssize = 0, limit = ss_ilg(last - first);;) { + + if((last - first) <= SS_INSERTIONSORT_THRESHOLD) { +#if 1 < SS_INSERTIONSORT_THRESHOLD + if(1 < (last - first)) { ss_insertionsort(T, PA, first, last, depth); } +#endif + STACK_POP(first, last, depth, limit); + continue; + } + + Td = T + depth; + if(limit-- == 0) { ss_heapsort(Td, PA, first, last - first); } + if(limit < 0) { + for(a = first + 1, v = Td[PA[*first]]; a < last; ++a) { + if((x = Td[PA[*a]]) != v) { + if(1 < (a - first)) { break; } + v = x; + first = a; + } + } + if(Td[PA[*first] - 1] < v) { + first = ss_partition(PA, first, a, depth); + } + if((a - first) <= (last - a)) { + if(1 < (a - first)) { + STACK_PUSH(a, last, depth, -1); + last = a, depth += 1, limit = ss_ilg(a - first); + } else { + first = a, limit = -1; + } + } else { + if(1 < (last - a)) { + STACK_PUSH(first, a, depth + 1, ss_ilg(a - first)); + first = a, limit = -1; + } else { + last = a, depth += 1, limit = ss_ilg(a - first); + } + } + continue; + } + + /* choose pivot */ + a = ss_pivot(Td, PA, first, last); + v = Td[PA[*a]]; + SWAP(*first, *a); + + /* partition */ + for(b = first; (++b < last) && ((x = Td[PA[*b]]) == v);) { } + if(((a = b) < last) && (x < v)) { + for(; (++b < last) && ((x = Td[PA[*b]]) <= v);) { + if(x == v) { SWAP(*b, *a); ++a; } + } + } + for(c = last; (b < --c) && ((x = Td[PA[*c]]) == v);) { } + if((b < (d = c)) && (x > v)) { + for(; (b < --c) && ((x = Td[PA[*c]]) >= v);) { + if(x == v) { SWAP(*c, *d); --d; } + } + } + for(; b < c;) { + SWAP(*b, *c); + for(; (++b < c) && ((x = Td[PA[*b]]) <= v);) { + if(x == v) { SWAP(*b, *a); ++a; } + } + for(; (b < --c) && ((x = Td[PA[*c]]) >= v);) { + if(x == v) { SWAP(*c, *d); --d; } + } + } + + if(a <= d) { + c = b - 1; + + if((s = a - first) > (t = b - a)) { s = t; } + for(e = first, f = b - s; 0 < s; --s, ++e, ++f) { SWAP(*e, *f); } + if((s = d - c) > (t = last - d - 1)) { s = t; } + for(e = b, f = last - s; 0 < s; --s, ++e, ++f) { SWAP(*e, *f); } + + a = first + (b - a), c = last - (d - c); + b = (v <= Td[PA[*a] - 1]) ? a : ss_partition(PA, a, c, depth); + + if((a - first) <= (last - c)) { + if((last - c) <= (c - b)) { + STACK_PUSH(b, c, depth + 1, ss_ilg(c - b)); + STACK_PUSH(c, last, depth, limit); + last = a; + } else if((a - first) <= (c - b)) { + STACK_PUSH(c, last, depth, limit); + STACK_PUSH(b, c, depth + 1, ss_ilg(c - b)); + last = a; + } else { + STACK_PUSH(c, last, depth, limit); + STACK_PUSH(first, a, depth, limit); + first = b, last = c, depth += 1, limit = ss_ilg(c - b); + } + } else { + if((a - first) <= (c - b)) { + STACK_PUSH(b, c, depth + 1, ss_ilg(c - b)); + STACK_PUSH(first, a, depth, limit); + first = c; + } else if((last - c) <= (c - b)) { + STACK_PUSH(first, a, depth, limit); + STACK_PUSH(b, c, depth + 1, ss_ilg(c - b)); + first = c; + } else { + STACK_PUSH(first, a, depth, limit); + STACK_PUSH(c, last, depth, limit); + first = b, last = c, depth += 1, limit = ss_ilg(c - b); + } + } + } else { + limit += 1; + if(Td[PA[*first] - 1] < v) { + first = ss_partition(PA, first, last, depth); + limit = ss_ilg(last - first); + } + depth += 1; + } + } +#undef STACK_SIZE +} + +#endif /* (SS_BLOCKSIZE == 0) || (SS_INSERTIONSORT_THRESHOLD < SS_BLOCKSIZE) */ + + +/*---------------------------------------------------------------------------*/ + +#if SS_BLOCKSIZE != 0 + +static INLINE +void +ss_blockswap(saidx_t *a, saidx_t *b, saidx_t n) { + saidx_t t; + for(; 0 < n; --n, ++a, ++b) { + t = *a, *a = *b, *b = t; + } +} + +static INLINE +void +ss_rotate(saidx_t *first, saidx_t *middle, saidx_t *last) { + saidx_t *a, *b, t; + saidx_t l, r; + l = middle - first, r = last - middle; + for(; (0 < l) && (0 < r);) { + if(l == r) { ss_blockswap(first, middle, l); break; } + if(l < r) { + a = last - 1, b = middle - 1; + t = *a; + do { + *a-- = *b, *b-- = *a; + if(b < first) { + *a = t; + last = a; + if((r -= l + 1) <= l) { break; } + a -= 1, b = middle - 1; + t = *a; + } + } while(1); + } else { + a = first, b = middle; + t = *a; + do { + *a++ = *b, *b++ = *a; + if(last <= b) { + *a = t; + first = a + 1; + if((l -= r + 1) <= r) { break; } + a += 1, b = middle; + t = *a; + } + } while(1); + } + } +} + + +/*---------------------------------------------------------------------------*/ + +static +void +ss_inplacemerge(const sauchar_t *T, const saidx_t *PA, + saidx_t *first, saidx_t *middle, saidx_t *last, + saidx_t depth) { + const saidx_t *p; + saidx_t *a, *b; + saidx_t len, half; + saint_t q, r; + saint_t x; + + for(;;) { + if(*(last - 1) < 0) { x = 1; p = PA + ~*(last - 1); } + else { x = 0; p = PA + *(last - 1); } + for(a = first, len = middle - first, half = len >> 1, r = -1; + 0 < len; + len = half, half >>= 1) { + b = a + half; + q = ss_compare(T, PA + ((0 <= *b) ? *b : ~*b), p, depth); + if(q < 0) { + a = b + 1; + half -= (len & 1) ^ 1; + } else { + r = q; + } + } + if(a < middle) { + if(r == 0) { *a = ~*a; } + ss_rotate(a, middle, last); + last -= middle - a; + middle = a; + if(first == middle) { break; } + } + --last; + if(x != 0) { while(*--last < 0) { } } + if(middle == last) { break; } + } +} + + +/*---------------------------------------------------------------------------*/ + +/* Merge-forward with internal buffer. */ +static +void +ss_mergeforward(const sauchar_t *T, const saidx_t *PA, + saidx_t *first, saidx_t *middle, saidx_t *last, + saidx_t *buf, saidx_t depth) { + saidx_t *a, *b, *c, *bufend; + saidx_t t; + saint_t r; + + bufend = buf + (middle - first) - 1; + ss_blockswap(buf, first, middle - first); + + for(t = *(a = first), b = buf, c = middle;;) { + r = ss_compare(T, PA + *b, PA + *c, depth); + if(r < 0) { + do { + *a++ = *b; + if(bufend <= b) { *bufend = t; return; } + *b++ = *a; + } while(*b < 0); + } else if(r > 0) { + do { + *a++ = *c, *c++ = *a; + if(last <= c) { + while(b < bufend) { *a++ = *b, *b++ = *a; } + *a = *b, *b = t; + return; + } + } while(*c < 0); + } else { + *c = ~*c; + do { + *a++ = *b; + if(bufend <= b) { *bufend = t; return; } + *b++ = *a; + } while(*b < 0); + + do { + *a++ = *c, *c++ = *a; + if(last <= c) { + while(b < bufend) { *a++ = *b, *b++ = *a; } + *a = *b, *b = t; + return; + } + } while(*c < 0); + } + } +} + +/* Merge-backward with internal buffer. */ +static +void +ss_mergebackward(const sauchar_t *T, const saidx_t *PA, + saidx_t *first, saidx_t *middle, saidx_t *last, + saidx_t *buf, saidx_t depth) { + const saidx_t *p1, *p2; + saidx_t *a, *b, *c, *bufend; + saidx_t t; + saint_t r; + saint_t x; + + bufend = buf + (last - middle) - 1; + ss_blockswap(buf, middle, last - middle); + + x = 0; + if(*bufend < 0) { p1 = PA + ~*bufend; x |= 1; } + else { p1 = PA + *bufend; } + if(*(middle - 1) < 0) { p2 = PA + ~*(middle - 1); x |= 2; } + else { p2 = PA + *(middle - 1); } + for(t = *(a = last - 1), b = bufend, c = middle - 1;;) { + r = ss_compare(T, p1, p2, depth); + if(0 < r) { + if(x & 1) { do { *a-- = *b, *b-- = *a; } while(*b < 0); x ^= 1; } + *a-- = *b; + if(b <= buf) { *buf = t; break; } + *b-- = *a; + if(*b < 0) { p1 = PA + ~*b; x |= 1; } + else { p1 = PA + *b; } + } else if(r < 0) { + if(x & 2) { do { *a-- = *c, *c-- = *a; } while(*c < 0); x ^= 2; } + *a-- = *c, *c-- = *a; + if(c < first) { + while(buf < b) { *a-- = *b, *b-- = *a; } + *a = *b, *b = t; + break; + } + if(*c < 0) { p2 = PA + ~*c; x |= 2; } + else { p2 = PA + *c; } + } else { + if(x & 1) { do { *a-- = *b, *b-- = *a; } while(*b < 0); x ^= 1; } + *a-- = ~*b; + if(b <= buf) { *buf = t; break; } + *b-- = *a; + if(x & 2) { do { *a-- = *c, *c-- = *a; } while(*c < 0); x ^= 2; } + *a-- = *c, *c-- = *a; + if(c < first) { + while(buf < b) { *a-- = *b, *b-- = *a; } + *a = *b, *b = t; + break; + } + if(*b < 0) { p1 = PA + ~*b; x |= 1; } + else { p1 = PA + *b; } + if(*c < 0) { p2 = PA + ~*c; x |= 2; } + else { p2 = PA + *c; } + } + } +} + +/* D&C based merge. */ +static +void +ss_swapmerge(const sauchar_t *T, const saidx_t *PA, + saidx_t *first, saidx_t *middle, saidx_t *last, + saidx_t *buf, saidx_t bufsize, saidx_t depth) { +#define STACK_SIZE SS_SMERGE_STACKSIZE +#define GETIDX(a) ((0 <= (a)) ? (a) : (~(a))) +#define MERGE_CHECK(a, b, c)\ + do {\ + if(((c) & 1) ||\ + (((c) & 2) && (ss_compare(T, PA + GETIDX(*((a) - 1)), PA + *(a), depth) == 0))) {\ + *(a) = ~*(a);\ + }\ + if(((c) & 4) && ((ss_compare(T, PA + GETIDX(*((b) - 1)), PA + *(b), depth) == 0))) {\ + *(b) = ~*(b);\ + }\ + } while(0) + struct { saidx_t *a, *b, *c; saint_t d; } stack[STACK_SIZE]; + saidx_t *l, *r, *lm, *rm; + saidx_t m, len, half; + saint_t ssize; + saint_t check, next; + + for(check = 0, ssize = 0;;) { + if((last - middle) <= bufsize) { + if((first < middle) && (middle < last)) { + ss_mergebackward(T, PA, first, middle, last, buf, depth); + } + MERGE_CHECK(first, last, check); + STACK_POP(first, middle, last, check); + continue; + } + + if((middle - first) <= bufsize) { + if(first < middle) { + ss_mergeforward(T, PA, first, middle, last, buf, depth); + } + MERGE_CHECK(first, last, check); + STACK_POP(first, middle, last, check); + continue; + } + + for(m = 0, len = MIN(middle - first, last - middle), half = len >> 1; + 0 < len; + len = half, half >>= 1) { + if(ss_compare(T, PA + GETIDX(*(middle + m + half)), + PA + GETIDX(*(middle - m - half - 1)), depth) < 0) { + m += half + 1; + half -= (len & 1) ^ 1; + } + } + + if(0 < m) { + lm = middle - m, rm = middle + m; + ss_blockswap(lm, middle, m); + l = r = middle, next = 0; + if(rm < last) { + if(*rm < 0) { + *rm = ~*rm; + if(first < lm) { for(; *--l < 0;) { } next |= 4; } + next |= 1; + } else if(first < lm) { + for(; *r < 0; ++r) { } + next |= 2; + } + } + + if((l - first) <= (last - r)) { + STACK_PUSH(r, rm, last, (next & 3) | (check & 4)); + middle = lm, last = l, check = (check & 3) | (next & 4); + } else { + if((next & 2) && (r == middle)) { next ^= 6; } + STACK_PUSH(first, lm, l, (check & 3) | (next & 4)); + first = r, middle = rm, check = (next & 3) | (check & 4); + } + } else { + if(ss_compare(T, PA + GETIDX(*(middle - 1)), PA + *middle, depth) == 0) { + *middle = ~*middle; + } + MERGE_CHECK(first, last, check); + STACK_POP(first, middle, last, check); + } + } +#undef STACK_SIZE +} + +#endif /* SS_BLOCKSIZE != 0 */ + + +/*---------------------------------------------------------------------------*/ + +/*- Function -*/ + +/* Substring sort */ +void +sssort(const sauchar_t *T, const saidx_t *PA, + saidx_t *first, saidx_t *last, + saidx_t *buf, saidx_t bufsize, + saidx_t depth, saidx_t n, saint_t lastsuffix) { + saidx_t *a; +#if SS_BLOCKSIZE != 0 + saidx_t *b, *middle, *curbuf; + saidx_t j, k, curbufsize, limit; +#endif + saidx_t i; + + if(lastsuffix != 0) { ++first; } + +#if SS_BLOCKSIZE == 0 + ss_mintrosort(T, PA, first, last, depth); +#else + if((bufsize < SS_BLOCKSIZE) && + (bufsize < (last - first)) && + (bufsize < (limit = ss_isqrt(last - first)))) { + if(SS_BLOCKSIZE < limit) { limit = SS_BLOCKSIZE; } + buf = middle = last - limit, bufsize = limit; + } else { + middle = last, limit = 0; + } + for(a = first, i = 0; SS_BLOCKSIZE < (middle - a); a += SS_BLOCKSIZE, ++i) { +#if SS_INSERTIONSORT_THRESHOLD < SS_BLOCKSIZE + ss_mintrosort(T, PA, a, a + SS_BLOCKSIZE, depth); +#elif 1 < SS_BLOCKSIZE + ss_insertionsort(T, PA, a, a + SS_BLOCKSIZE, depth); +#endif + curbufsize = last - (a + SS_BLOCKSIZE); + curbuf = a + SS_BLOCKSIZE; + if(curbufsize <= bufsize) { curbufsize = bufsize, curbuf = buf; } + for(b = a, k = SS_BLOCKSIZE, j = i; j & 1; b -= k, k <<= 1, j >>= 1) { + ss_swapmerge(T, PA, b - k, b, b + k, curbuf, curbufsize, depth); + } + } +#if SS_INSERTIONSORT_THRESHOLD < SS_BLOCKSIZE + ss_mintrosort(T, PA, a, middle, depth); +#elif 1 < SS_BLOCKSIZE + ss_insertionsort(T, PA, a, middle, depth); +#endif + for(k = SS_BLOCKSIZE; i != 0; k <<= 1, i >>= 1) { + if(i & 1) { + ss_swapmerge(T, PA, a - k, a, middle, buf, bufsize, depth); + a -= k; + } + } + if(limit != 0) { +#if SS_INSERTIONSORT_THRESHOLD < SS_BLOCKSIZE + ss_mintrosort(T, PA, middle, last, depth); +#elif 1 < SS_BLOCKSIZE + ss_insertionsort(T, PA, middle, last, depth); +#endif + ss_inplacemerge(T, PA, first, middle, last, depth); + } +#endif + + if(lastsuffix != 0) { + /* Insert last type B* suffix. */ + saidx_t PAi[2]; PAi[0] = PA[*(first - 1)], PAi[1] = n - 2; + for(a = first, i = *(first - 1); + (a < last) && ((*a < 0) || (0 < ss_compare(T, &(PAi[0]), PA + *a, depth))); + ++a) { + *(a - 1) = *a; + } + *(a - 1) = i; + } +} diff --git a/dictBuilder/trsort.c b/dictBuilder/trsort.c new file mode 100644 index 000000000..9e21e68ec --- /dev/null +++ b/dictBuilder/trsort.c @@ -0,0 +1,615 @@ +/* + * trsort.c for libdivsufsort + * Copyright (c) 2003-2008 Yuta Mori All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, + * copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following + * conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES + * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +/*- Compiler specifics -*/ +#ifdef __clang__ +#pragma clang diagnostic ignored "-Wshorten-64-to-32" +#endif + +#if defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) +/* inline is defined */ +#elif defined(_MSC_VER) +# define inline __inline +#else +# define inline /* disable inline */ +#endif + +#ifdef _MSC_VER /* Visual Studio */ +# pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */ +# define FORCE_INLINE static __forceinline +#else +# if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L /* C99 */ +# ifdef __GNUC__ +# define FORCE_INLINE static inline __attribute__((always_inline)) +# else +# define FORCE_INLINE static inline +# endif +# else +# define FORCE_INLINE static +# endif /* __STDC_VERSION__ */ +#endif + +/*- Dependencies -*/ +#include "divsufsort_private.h" + + +/*- Private Functions -*/ + +static const saint_t lg_table[256]= { + -1,0,1,1,2,2,2,2,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4, + 5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, + 6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, + 6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, + 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, + 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, + 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, + 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7 +}; + +static INLINE +saint_t +tr_ilg(saidx_t n) { +#if defined(BUILD_DIVSUFSORT64) + return (n >> 32) ? + ((n >> 48) ? + ((n >> 56) ? + 56 + lg_table[(n >> 56) & 0xff] : + 48 + lg_table[(n >> 48) & 0xff]) : + ((n >> 40) ? + 40 + lg_table[(n >> 40) & 0xff] : + 32 + lg_table[(n >> 32) & 0xff])) : + ((n & 0xffff0000) ? + ((n & 0xff000000) ? + 24 + lg_table[(n >> 24) & 0xff] : + 16 + lg_table[(n >> 16) & 0xff]) : + ((n & 0x0000ff00) ? + 8 + lg_table[(n >> 8) & 0xff] : + 0 + lg_table[(n >> 0) & 0xff])); +#else + return (n & 0xffff0000) ? + ((n & 0xff000000) ? + 24 + lg_table[(n >> 24) & 0xff] : + 16 + lg_table[(n >> 16) & 0xff]) : + ((n & 0x0000ff00) ? + 8 + lg_table[(n >> 8) & 0xff] : + 0 + lg_table[(n >> 0) & 0xff]); +#endif +} + + +/*---------------------------------------------------------------------------*/ + +/* Simple insertionsort for small size groups. */ +static +void +tr_insertionsort(const saidx_t *ISAd, saidx_t *first, saidx_t *last) { + saidx_t *a, *b; + saidx_t t, r; + + for(a = first + 1; a < last; ++a) { + for(t = *a, b = a - 1; 0 > (r = ISAd[t] - ISAd[*b]);) { + do { *(b + 1) = *b; } while((first <= --b) && (*b < 0)); + if(b < first) { break; } + } + if(r == 0) { *b = ~*b; } + *(b + 1) = t; + } +} + + +/*---------------------------------------------------------------------------*/ + +static INLINE +void +tr_fixdown(const saidx_t *ISAd, saidx_t *SA, saidx_t i, saidx_t size) { + saidx_t j, k; + saidx_t v; + saidx_t c, d, e; + + for(v = SA[i], c = ISAd[v]; (j = 2 * i + 1) < size; SA[i] = SA[k], i = k) { + d = ISAd[SA[k = j++]]; + if(d < (e = ISAd[SA[j]])) { k = j; d = e; } + if(d <= c) { break; } + } + SA[i] = v; +} + +/* Simple top-down heapsort. */ +static +void +tr_heapsort(const saidx_t *ISAd, saidx_t *SA, saidx_t size) { + saidx_t i, m; + saidx_t t; + + m = size; + if((size % 2) == 0) { + m--; + if(ISAd[SA[m / 2]] < ISAd[SA[m]]) { SWAP(SA[m], SA[m / 2]); } + } + + for(i = m / 2 - 1; 0 <= i; --i) { tr_fixdown(ISAd, SA, i, m); } + if((size % 2) == 0) { SWAP(SA[0], SA[m]); tr_fixdown(ISAd, SA, 0, m); } + for(i = m - 1; 0 < i; --i) { + t = SA[0], SA[0] = SA[i]; + tr_fixdown(ISAd, SA, 0, i); + SA[i] = t; + } +} + + +/*---------------------------------------------------------------------------*/ + +/* Returns the median of three elements. */ +static INLINE +saidx_t * +tr_median3(const saidx_t *ISAd, saidx_t *v1, saidx_t *v2, saidx_t *v3) { + saidx_t *t; + if(ISAd[*v1] > ISAd[*v2]) { SWAP(v1, v2); } + if(ISAd[*v2] > ISAd[*v3]) { + if(ISAd[*v1] > ISAd[*v3]) { return v1; } + else { return v3; } + } + return v2; +} + +/* Returns the median of five elements. */ +static INLINE +saidx_t * +tr_median5(const saidx_t *ISAd, + saidx_t *v1, saidx_t *v2, saidx_t *v3, saidx_t *v4, saidx_t *v5) { + saidx_t *t; + if(ISAd[*v2] > ISAd[*v3]) { SWAP(v2, v3); } + if(ISAd[*v4] > ISAd[*v5]) { SWAP(v4, v5); } + if(ISAd[*v2] > ISAd[*v4]) { SWAP(v2, v4); SWAP(v3, v5); } + if(ISAd[*v1] > ISAd[*v3]) { SWAP(v1, v3); } + if(ISAd[*v1] > ISAd[*v4]) { SWAP(v1, v4); SWAP(v3, v5); } + if(ISAd[*v3] > ISAd[*v4]) { return v4; } + return v3; +} + +/* Returns the pivot element. */ +static INLINE +saidx_t * +tr_pivot(const saidx_t *ISAd, saidx_t *first, saidx_t *last) { + saidx_t *middle; + saidx_t t; + + t = last - first; + middle = first + t / 2; + + if(t <= 512) { + if(t <= 32) { + return tr_median3(ISAd, first, middle, last - 1); + } else { + t >>= 2; + return tr_median5(ISAd, first, first + t, middle, last - 1 - t, last - 1); + } + } + t >>= 3; + first = tr_median3(ISAd, first, first + t, first + (t << 1)); + middle = tr_median3(ISAd, middle - t, middle, middle + t); + last = tr_median3(ISAd, last - 1 - (t << 1), last - 1 - t, last - 1); + return tr_median3(ISAd, first, middle, last); +} + + +/*---------------------------------------------------------------------------*/ + +typedef struct _trbudget_t trbudget_t; +struct _trbudget_t { + saidx_t chance; + saidx_t remain; + saidx_t incval; + saidx_t count; +}; + +static INLINE +void +trbudget_init(trbudget_t *budget, saidx_t chance, saidx_t incval) { + budget->chance = chance; + budget->remain = budget->incval = incval; +} + +static INLINE +saint_t +trbudget_check(trbudget_t *budget, saidx_t size) { + if(size <= budget->remain) { budget->remain -= size; return 1; } + if(budget->chance == 0) { budget->count += size; return 0; } + budget->remain += budget->incval - size; + budget->chance -= 1; + return 1; +} + + +/*---------------------------------------------------------------------------*/ + +static INLINE +void +tr_partition(const saidx_t *ISAd, + saidx_t *first, saidx_t *middle, saidx_t *last, + saidx_t **pa, saidx_t **pb, saidx_t v) { + saidx_t *a, *b, *c, *d, *e, *f; + saidx_t t, s; + saidx_t x = 0; + + for(b = middle - 1; (++b < last) && ((x = ISAd[*b]) == v);) { } + if(((a = b) < last) && (x < v)) { + for(; (++b < last) && ((x = ISAd[*b]) <= v);) { + if(x == v) { SWAP(*b, *a); ++a; } + } + } + for(c = last; (b < --c) && ((x = ISAd[*c]) == v);) { } + if((b < (d = c)) && (x > v)) { + for(; (b < --c) && ((x = ISAd[*c]) >= v);) { + if(x == v) { SWAP(*c, *d); --d; } + } + } + for(; b < c;) { + SWAP(*b, *c); + for(; (++b < c) && ((x = ISAd[*b]) <= v);) { + if(x == v) { SWAP(*b, *a); ++a; } + } + for(; (b < --c) && ((x = ISAd[*c]) >= v);) { + if(x == v) { SWAP(*c, *d); --d; } + } + } + + if(a <= d) { + c = b - 1; + if((s = a - first) > (t = b - a)) { s = t; } + for(e = first, f = b - s; 0 < s; --s, ++e, ++f) { SWAP(*e, *f); } + if((s = d - c) > (t = last - d - 1)) { s = t; } + for(e = b, f = last - s; 0 < s; --s, ++e, ++f) { SWAP(*e, *f); } + first += (b - a), last -= (d - c); + } + *pa = first, *pb = last; +} + +static +void +tr_copy(saidx_t *ISA, const saidx_t *SA, + saidx_t *first, saidx_t *a, saidx_t *b, saidx_t *last, + saidx_t depth) { + /* sort suffixes of middle partition + by using sorted order of suffixes of left and right partition. */ + saidx_t *c, *d, *e; + saidx_t s, v; + + v = b - SA - 1; + for(c = first, d = a - 1; c <= d; ++c) { + if((0 <= (s = *c - depth)) && (ISA[s] == v)) { + *++d = s; + ISA[s] = d - SA; + } + } + for(c = last - 1, e = d + 1, d = b; e < d; --c) { + if((0 <= (s = *c - depth)) && (ISA[s] == v)) { + *--d = s; + ISA[s] = d - SA; + } + } +} + +static +void +tr_partialcopy(saidx_t *ISA, const saidx_t *SA, + saidx_t *first, saidx_t *a, saidx_t *b, saidx_t *last, + saidx_t depth) { + saidx_t *c, *d, *e; + saidx_t s, v; + saidx_t rank, lastrank, newrank = -1; + + v = b - SA - 1; + lastrank = -1; + for(c = first, d = a - 1; c <= d; ++c) { + if((0 <= (s = *c - depth)) && (ISA[s] == v)) { + *++d = s; + rank = ISA[s + depth]; + if(lastrank != rank) { lastrank = rank; newrank = d - SA; } + ISA[s] = newrank; + } + } + + lastrank = -1; + for(e = d; first <= e; --e) { + rank = ISA[*e]; + if(lastrank != rank) { lastrank = rank; newrank = e - SA; } + if(newrank != rank) { ISA[*e] = newrank; } + } + + lastrank = -1; + for(c = last - 1, e = d + 1, d = b; e < d; --c) { + if((0 <= (s = *c - depth)) && (ISA[s] == v)) { + *--d = s; + rank = ISA[s + depth]; + if(lastrank != rank) { lastrank = rank; newrank = d - SA; } + ISA[s] = newrank; + } + } +} + +static +void +tr_introsort(saidx_t *ISA, const saidx_t *ISAd, + saidx_t *SA, saidx_t *first, saidx_t *last, + trbudget_t *budget) { +#define STACK_SIZE TR_STACKSIZE + struct { const saidx_t *a; saidx_t *b, *c; saint_t d, e; }stack[STACK_SIZE]; + saidx_t *a, *b, *c; + saidx_t t; + saidx_t v, x = 0; + saidx_t incr = ISAd - ISA; + saint_t limit, next; + saint_t ssize, trlink = -1; + + for(ssize = 0, limit = tr_ilg(last - first);;) { + + if(limit < 0) { + if(limit == -1) { + /* tandem repeat partition */ + tr_partition(ISAd - incr, first, first, last, &a, &b, last - SA - 1); + + /* update ranks */ + if(a < last) { + for(c = first, v = a - SA - 1; c < a; ++c) { ISA[*c] = v; } + } + if(b < last) { + for(c = a, v = b - SA - 1; c < b; ++c) { ISA[*c] = v; } + } + + /* push */ + if(1 < (b - a)) { + STACK_PUSH5(NULL, a, b, 0, 0); + STACK_PUSH5(ISAd - incr, first, last, -2, trlink); + trlink = ssize - 2; + } + if((a - first) <= (last - b)) { + if(1 < (a - first)) { + STACK_PUSH5(ISAd, b, last, tr_ilg(last - b), trlink); + last = a, limit = tr_ilg(a - first); + } else if(1 < (last - b)) { + first = b, limit = tr_ilg(last - b); + } else { + STACK_POP5(ISAd, first, last, limit, trlink); + } + } else { + if(1 < (last - b)) { + STACK_PUSH5(ISAd, first, a, tr_ilg(a - first), trlink); + first = b, limit = tr_ilg(last - b); + } else if(1 < (a - first)) { + last = a, limit = tr_ilg(a - first); + } else { + STACK_POP5(ISAd, first, last, limit, trlink); + } + } + } else if(limit == -2) { + /* tandem repeat copy */ + a = stack[--ssize].b, b = stack[ssize].c; + if(stack[ssize].d == 0) { + tr_copy(ISA, SA, first, a, b, last, ISAd - ISA); + } else { + if(0 <= trlink) { stack[trlink].d = -1; } + tr_partialcopy(ISA, SA, first, a, b, last, ISAd - ISA); + } + STACK_POP5(ISAd, first, last, limit, trlink); + } else { + /* sorted partition */ + if(0 <= *first) { + a = first; + do { ISA[*a] = a - SA; } while((++a < last) && (0 <= *a)); + first = a; + } + if(first < last) { + a = first; do { *a = ~*a; } while(*++a < 0); + next = (ISA[*a] != ISAd[*a]) ? tr_ilg(a - first + 1) : -1; + if(++a < last) { for(b = first, v = a - SA - 1; b < a; ++b) { ISA[*b] = v; } } + + /* push */ + if(trbudget_check(budget, a - first)) { + if((a - first) <= (last - a)) { + STACK_PUSH5(ISAd, a, last, -3, trlink); + ISAd += incr, last = a, limit = next; + } else { + if(1 < (last - a)) { + STACK_PUSH5(ISAd + incr, first, a, next, trlink); + first = a, limit = -3; + } else { + ISAd += incr, last = a, limit = next; + } + } + } else { + if(0 <= trlink) { stack[trlink].d = -1; } + if(1 < (last - a)) { + first = a, limit = -3; + } else { + STACK_POP5(ISAd, first, last, limit, trlink); + } + } + } else { + STACK_POP5(ISAd, first, last, limit, trlink); + } + } + continue; + } + + if((last - first) <= TR_INSERTIONSORT_THRESHOLD) { + tr_insertionsort(ISAd, first, last); + limit = -3; + continue; + } + + if(limit-- == 0) { + tr_heapsort(ISAd, first, last - first); + for(a = last - 1; first < a; a = b) { + for(x = ISAd[*a], b = a - 1; (first <= b) && (ISAd[*b] == x); --b) { *b = ~*b; } + } + limit = -3; + continue; + } + + /* choose pivot */ + a = tr_pivot(ISAd, first, last); + SWAP(*first, *a); + v = ISAd[*first]; + + /* partition */ + tr_partition(ISAd, first, first + 1, last, &a, &b, v); + if((last - first) != (b - a)) { + next = (ISA[*a] != v) ? tr_ilg(b - a) : -1; + + /* update ranks */ + for(c = first, v = a - SA - 1; c < a; ++c) { ISA[*c] = v; } + if(b < last) { for(c = a, v = b - SA - 1; c < b; ++c) { ISA[*c] = v; } } + + /* push */ + if((1 < (b - a)) && (trbudget_check(budget, b - a))) { + if((a - first) <= (last - b)) { + if((last - b) <= (b - a)) { + if(1 < (a - first)) { + STACK_PUSH5(ISAd + incr, a, b, next, trlink); + STACK_PUSH5(ISAd, b, last, limit, trlink); + last = a; + } else if(1 < (last - b)) { + STACK_PUSH5(ISAd + incr, a, b, next, trlink); + first = b; + } else { + ISAd += incr, first = a, last = b, limit = next; + } + } else if((a - first) <= (b - a)) { + if(1 < (a - first)) { + STACK_PUSH5(ISAd, b, last, limit, trlink); + STACK_PUSH5(ISAd + incr, a, b, next, trlink); + last = a; + } else { + STACK_PUSH5(ISAd, b, last, limit, trlink); + ISAd += incr, first = a, last = b, limit = next; + } + } else { + STACK_PUSH5(ISAd, b, last, limit, trlink); + STACK_PUSH5(ISAd, first, a, limit, trlink); + ISAd += incr, first = a, last = b, limit = next; + } + } else { + if((a - first) <= (b - a)) { + if(1 < (last - b)) { + STACK_PUSH5(ISAd + incr, a, b, next, trlink); + STACK_PUSH5(ISAd, first, a, limit, trlink); + first = b; + } else if(1 < (a - first)) { + STACK_PUSH5(ISAd + incr, a, b, next, trlink); + last = a; + } else { + ISAd += incr, first = a, last = b, limit = next; + } + } else if((last - b) <= (b - a)) { + if(1 < (last - b)) { + STACK_PUSH5(ISAd, first, a, limit, trlink); + STACK_PUSH5(ISAd + incr, a, b, next, trlink); + first = b; + } else { + STACK_PUSH5(ISAd, first, a, limit, trlink); + ISAd += incr, first = a, last = b, limit = next; + } + } else { + STACK_PUSH5(ISAd, first, a, limit, trlink); + STACK_PUSH5(ISAd, b, last, limit, trlink); + ISAd += incr, first = a, last = b, limit = next; + } + } + } else { + if((1 < (b - a)) && (0 <= trlink)) { stack[trlink].d = -1; } + if((a - first) <= (last - b)) { + if(1 < (a - first)) { + STACK_PUSH5(ISAd, b, last, limit, trlink); + last = a; + } else if(1 < (last - b)) { + first = b; + } else { + STACK_POP5(ISAd, first, last, limit, trlink); + } + } else { + if(1 < (last - b)) { + STACK_PUSH5(ISAd, first, a, limit, trlink); + first = b; + } else if(1 < (a - first)) { + last = a; + } else { + STACK_POP5(ISAd, first, last, limit, trlink); + } + } + } + } else { + if(trbudget_check(budget, last - first)) { + limit = tr_ilg(last - first), ISAd += incr; + } else { + if(0 <= trlink) { stack[trlink].d = -1; } + STACK_POP5(ISAd, first, last, limit, trlink); + } + } + } +#undef STACK_SIZE +} + + + +/*---------------------------------------------------------------------------*/ + +/*- Function -*/ + +/* Tandem repeat sort */ +void +trsort(saidx_t *ISA, saidx_t *SA, saidx_t n, saidx_t depth) { + saidx_t *ISAd; + saidx_t *first, *last; + trbudget_t budget; + saidx_t t, skip, unsorted; + + trbudget_init(&budget, tr_ilg(n) * 2 / 3, n); +/* trbudget_init(&budget, tr_ilg(n) * 3 / 4, n); */ + for(ISAd = ISA + depth; -n < *SA; ISAd += ISAd - ISA) { + first = SA; + skip = 0; + unsorted = 0; + do { + if((t = *first) < 0) { first -= t; skip += t; } + else { + if(skip != 0) { *(first + skip) = skip; skip = 0; } + last = SA + ISA[t] + 1; + if(1 < (last - first)) { + budget.count = 0; + tr_introsort(ISA, ISAd, SA, first, last, &budget); + if(budget.count != 0) { unsorted += budget.count; } + else { skip = first - last; } + } else if((last - first) == 1) { + skip = -1; + } + first = last; + } + } while(first < (SA + n)); + if(skip != 0) { *(first + skip) = skip; } + if(unsorted == 0) { break; } + } +} diff --git a/dictBuilder/utils.c b/dictBuilder/utils.c new file mode 100644 index 000000000..90fb23efa --- /dev/null +++ b/dictBuilder/utils.c @@ -0,0 +1,381 @@ +/* + * utils.c for libdivsufsort + * Copyright (c) 2003-2008 Yuta Mori All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, + * copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following + * conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES + * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include "divsufsort_private.h" + + +/*- Private Function -*/ + +/* Binary search for inverse bwt. */ +static +saidx_t +binarysearch_lower(const saidx_t *A, saidx_t size, saidx_t value) { + saidx_t half, i; + for(i = 0, half = size >> 1; + 0 < size; + size = half, half >>= 1) { + if(A[i + half] < value) { + i += half + 1; + half -= (size & 1) ^ 1; + } + } + return i; +} + + +/*- Functions -*/ + +/* Burrows-Wheeler transform. */ +saint_t +bw_transform(const sauchar_t *T, sauchar_t *U, saidx_t *SA, + saidx_t n, saidx_t *idx) { + saidx_t *A, i, j, p, t; + saint_t c; + + /* Check arguments. */ + if((T == NULL) || (U == NULL) || (n < 0) || (idx == NULL)) { return -1; } + if(n <= 1) { + if(n == 1) { U[0] = T[0]; } + *idx = n; + return 0; + } + + if((A = SA) == NULL) { + i = divbwt(T, U, NULL, n); + if(0 <= i) { *idx = i; i = 0; } + return (saint_t)i; + } + + /* BW transform. */ + if(T == U) { + t = n; + for(i = 0, j = 0; i < n; ++i) { + p = t - 1; + t = A[i]; + if(0 <= p) { + c = T[j]; + U[j] = (j <= p) ? T[p] : (sauchar_t)A[p]; + A[j] = c; + j++; + } else { + *idx = i; + } + } + p = t - 1; + if(0 <= p) { + c = T[j]; + U[j] = (j <= p) ? T[p] : (sauchar_t)A[p]; + A[j] = c; + } else { + *idx = i; + } + } else { + U[0] = T[n - 1]; + for(i = 0; A[i] != 0; ++i) { U[i + 1] = T[A[i] - 1]; } + *idx = i + 1; + for(++i; i < n; ++i) { U[i] = T[A[i] - 1]; } + } + + if(SA == NULL) { + /* Deallocate memory. */ + free(A); + } + + return 0; +} + +/* Inverse Burrows-Wheeler transform. */ +saint_t +inverse_bw_transform(const sauchar_t *T, sauchar_t *U, saidx_t *A, + saidx_t n, saidx_t idx) { + saidx_t C[ALPHABET_SIZE]; + sauchar_t D[ALPHABET_SIZE]; + saidx_t *B; + saidx_t i, p; + saint_t c, d; + + /* Check arguments. */ + if((T == NULL) || (U == NULL) || (n < 0) || (idx < 0) || + (n < idx) || ((0 < n) && (idx == 0))) { + return -1; + } + if(n <= 1) { return 0; } + + if((B = A) == NULL) { + /* Allocate n*sizeof(saidx_t) bytes of memory. */ + if((B = (saidx_t *)malloc((size_t)n * sizeof(saidx_t))) == NULL) { return -2; } + } + + /* Inverse BW transform. */ + for(c = 0; c < ALPHABET_SIZE; ++c) { C[c] = 0; } + for(i = 0; i < n; ++i) { ++C[T[i]]; } + for(c = 0, d = 0, i = 0; c < ALPHABET_SIZE; ++c) { + p = C[c]; + if(0 < p) { + C[c] = i; + D[d++] = (sauchar_t)c; + i += p; + } + } + for(i = 0; i < idx; ++i) { B[C[T[i]]++] = i; } + for( ; i < n; ++i) { B[C[T[i]]++] = i + 1; } + for(c = 0; c < d; ++c) { C[c] = C[D[c]]; } + for(i = 0, p = idx; i < n; ++i) { + U[i] = D[binarysearch_lower(C, d, p)]; + p = B[p - 1]; + } + + if(A == NULL) { + /* Deallocate memory. */ + free(B); + } + + return 0; +} + +/* Checks the suffix array SA of the string T. */ +saint_t +sufcheck(const sauchar_t *T, const saidx_t *SA, + saidx_t n, saint_t verbose) { + saidx_t C[ALPHABET_SIZE]; + saidx_t i, p, q, t; + saint_t c; + + if(verbose) { fprintf(stderr, "sufcheck: "); } + + /* Check arguments. */ + if((T == NULL) || (SA == NULL) || (n < 0)) { + if(verbose) { fprintf(stderr, "Invalid arguments.\n"); } + return -1; + } + if(n == 0) { + if(verbose) { fprintf(stderr, "Done.\n"); } + return 0; + } + + /* check range: [0..n-1] */ + for(i = 0; i < n; ++i) { + if((SA[i] < 0) || (n <= SA[i])) { + if(verbose) { + fprintf(stderr, "Out of the range [0,%" PRIdSAIDX_T "].\n" + " SA[%" PRIdSAIDX_T "]=%" PRIdSAIDX_T "\n", + n - 1, i, SA[i]); + } + return -2; + } + } + + /* check first characters. */ + for(i = 1; i < n; ++i) { + if(T[SA[i - 1]] > T[SA[i]]) { + if(verbose) { + fprintf(stderr, "Suffixes in wrong order.\n" + " T[SA[%" PRIdSAIDX_T "]=%" PRIdSAIDX_T "]=%d" + " > T[SA[%" PRIdSAIDX_T "]=%" PRIdSAIDX_T "]=%d\n", + i - 1, SA[i - 1], T[SA[i - 1]], i, SA[i], T[SA[i]]); + } + return -3; + } + } + + /* check suffixes. */ + for(i = 0; i < ALPHABET_SIZE; ++i) { C[i] = 0; } + for(i = 0; i < n; ++i) { ++C[T[i]]; } + for(i = 0, p = 0; i < ALPHABET_SIZE; ++i) { + t = C[i]; + C[i] = p; + p += t; + } + + q = C[T[n - 1]]; + C[T[n - 1]] += 1; + for(i = 0; i < n; ++i) { + p = SA[i]; + if(0 < p) { + c = T[--p]; + t = C[c]; + } else { + c = T[p = n - 1]; + t = q; + } + if((t < 0) || (p != SA[t])) { + if(verbose) { + fprintf(stderr, "Suffix in wrong position.\n" + " SA[%" PRIdSAIDX_T "]=%" PRIdSAIDX_T " or\n" + " SA[%" PRIdSAIDX_T "]=%" PRIdSAIDX_T "\n", + t, (0 <= t) ? SA[t] : -1, i, SA[i]); + } + return -4; + } + if(t != q) { + ++C[c]; + if((n <= C[c]) || (T[SA[C[c]]] != c)) { C[c] = -1; } + } + } + + if(1 <= verbose) { fprintf(stderr, "Done.\n"); } + return 0; +} + + +static +int +_compare(const sauchar_t *T, saidx_t Tsize, + const sauchar_t *P, saidx_t Psize, + saidx_t suf, saidx_t *match) { + saidx_t i, j; + saint_t r; + for(i = suf + *match, j = *match, r = 0; + (i < Tsize) && (j < Psize) && ((r = T[i] - P[j]) == 0); ++i, ++j) { } + *match = j; + return (r == 0) ? -(j != Psize) : r; +} + +/* Search for the pattern P in the string T. */ +saidx_t +sa_search(const sauchar_t *T, saidx_t Tsize, + const sauchar_t *P, saidx_t Psize, + const saidx_t *SA, saidx_t SAsize, + saidx_t *idx) { + saidx_t size, lsize, rsize, half; + saidx_t match, lmatch, rmatch; + saidx_t llmatch, lrmatch, rlmatch, rrmatch; + saidx_t i, j, k; + saint_t r; + + if(idx != NULL) { *idx = -1; } + if((T == NULL) || (P == NULL) || (SA == NULL) || + (Tsize < 0) || (Psize < 0) || (SAsize < 0)) { return -1; } + if((Tsize == 0) || (SAsize == 0)) { return 0; } + if(Psize == 0) { if(idx != NULL) { *idx = 0; } return SAsize; } + + for(i = j = k = 0, lmatch = rmatch = 0, size = SAsize, half = size >> 1; + 0 < size; + size = half, half >>= 1) { + match = MIN(lmatch, rmatch); + r = _compare(T, Tsize, P, Psize, SA[i + half], &match); + if(r < 0) { + i += half + 1; + half -= (size & 1) ^ 1; + lmatch = match; + } else if(r > 0) { + rmatch = match; + } else { + lsize = half, j = i, rsize = size - half - 1, k = i + half + 1; + + /* left part */ + for(llmatch = lmatch, lrmatch = match, half = lsize >> 1; + 0 < lsize; + lsize = half, half >>= 1) { + lmatch = MIN(llmatch, lrmatch); + r = _compare(T, Tsize, P, Psize, SA[j + half], &lmatch); + if(r < 0) { + j += half + 1; + half -= (lsize & 1) ^ 1; + llmatch = lmatch; + } else { + lrmatch = lmatch; + } + } + + /* right part */ + for(rlmatch = match, rrmatch = rmatch, half = rsize >> 1; + 0 < rsize; + rsize = half, half >>= 1) { + rmatch = MIN(rlmatch, rrmatch); + r = _compare(T, Tsize, P, Psize, SA[k + half], &rmatch); + if(r <= 0) { + k += half + 1; + half -= (rsize & 1) ^ 1; + rlmatch = rmatch; + } else { + rrmatch = rmatch; + } + } + + break; + } + } + + if(idx != NULL) { *idx = (0 < (k - j)) ? j : i; } + return k - j; +} + +/* Search for the character c in the string T. */ +saidx_t +sa_simplesearch(const sauchar_t *T, saidx_t Tsize, + const saidx_t *SA, saidx_t SAsize, + saint_t c, saidx_t *idx) { + saidx_t size, lsize, rsize, half; + saidx_t i, j, k, p; + saint_t r; + + if(idx != NULL) { *idx = -1; } + if((T == NULL) || (SA == NULL) || (Tsize < 0) || (SAsize < 0)) { return -1; } + if((Tsize == 0) || (SAsize == 0)) { return 0; } + + for(i = j = k = 0, size = SAsize, half = size >> 1; + 0 < size; + size = half, half >>= 1) { + p = SA[i + half]; + r = (p < Tsize) ? T[p] - c : -1; + if(r < 0) { + i += half + 1; + half -= (size & 1) ^ 1; + } else if(r == 0) { + lsize = half, j = i, rsize = size - half - 1, k = i + half + 1; + + /* left part */ + for(half = lsize >> 1; + 0 < lsize; + lsize = half, half >>= 1) { + p = SA[j + half]; + r = (p < Tsize) ? T[p] - c : -1; + if(r < 0) { + j += half + 1; + half -= (lsize & 1) ^ 1; + } + } + + /* right part */ + for(half = rsize >> 1; + 0 < rsize; + rsize = half, half >>= 1) { + p = SA[k + half]; + r = (p < Tsize) ? T[p] - c : -1; + if(r <= 0) { + k += half + 1; + half -= (rsize & 1) ^ 1; + } + } + + break; + } + } + + if(idx != NULL) { *idx = (0 < (k - j)) ? j : i; } + return k - j; +} diff --git a/lib/Makefile b/lib/Makefile index a7172b6f8..e0ca0c626 100644 --- a/lib/Makefile +++ b/lib/Makefile @@ -53,7 +53,7 @@ LIBDIR ?= $(PREFIX)/lib INCLUDEDIR=$(PREFIX)/include ZSTD_FILES := zstd_compress.c zstd_decompress.c fse.c huff0.c -ZSTD_LEGACY:= legacy/zstd_v01.c legacy/zstd_v02.c legacy/zstd_v03.c +ZSTD_LEGACY:= legacy/zstd_v01.c legacy/zstd_v02.c legacy/zstd_v03.c legacy/zstd_v04.c ifeq ($(ZSTD_LEGACY_SUPPORT), 0) CPPFLAGS += -DZSTD_LEGACY_SUPPORT=0 diff --git a/lib/bitstream.h b/lib/bitstream.h index fbd0f3f37..e0487e87e 100644 --- a/lib/bitstream.h +++ b/lib/bitstream.h @@ -1,8 +1,8 @@ /* ****************************************************************** bitstream - Part of NewGen Entropy library + Part of FSE library header file (to include) - Copyright (C) 2013-2015, Yann Collet. + Copyright (C) 2013-2016, Yann Collet. BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) @@ -31,7 +31,6 @@ You can contact the author at : - Source repository : https://github.com/Cyan4973/FiniteStateEntropy - - Public forum : https://groups.google.com/forum/#!forum/lz4c ****************************************************************** */ #ifndef BITSTREAM_H_MODULE #define BITSTREAM_H_MODULE @@ -47,17 +46,17 @@ extern "C" { * these functions are defined into a .h to be included. */ -/****************************************** -* Includes +/*-**************************************** +* Dependencies ******************************************/ #include "mem.h" /* unaligned access routines */ #include "error_private.h" /* error codes and messages */ -/******************************************** -* bitStream compression API (write forward) +/*-****************************************** +* bitStream encoding API (write forward) ********************************************/ -/* +/*! * bitStream can mix input from multiple sources. * A critical property of these streams is that they encode and decode in **reverse** direction. * So the first bit sequence you add will be the last to be read, like a LIFO stack. @@ -71,32 +70,32 @@ typedef struct char* endPtr; } BIT_CStream_t; -MEM_STATIC size_t BIT_initCStream(BIT_CStream_t* bitC, void* dstBuffer, size_t maxDstSize); +MEM_STATIC size_t BIT_initCStream(BIT_CStream_t* bitC, void* dstBuffer, size_t dstCapacity); MEM_STATIC void BIT_addBits(BIT_CStream_t* bitC, size_t value, unsigned nbBits); MEM_STATIC void BIT_flushBits(BIT_CStream_t* bitC); MEM_STATIC size_t BIT_closeCStream(BIT_CStream_t* bitC); -/* -* Start by initCStream, providing the maximum size of write buffer to write into. +/*! +* Start by initCStream, providing the size of buffer to write into. * bitStream will never write outside of this buffer. -* buffer must be at least as large as a size_t, otherwise function result will be an error code. +* @dstCapacity must be >= sizeof(size_t), otherwise @return will be an error code. * * bits are first added to a local register. * Local register is size_t, hence 64-bits on 64-bits systems, or 32-bits on 32-bits systems. -* Writing data into memory is a manual operation, performed by the flushBits function. +* Writing data into memory is an explicit operation, performed by the flushBits function. * Hence keep track how many bits are potentially stored into local register to avoid register overflow. * After a flushBits, a maximum of 7 bits might still be stored into local register. * -* Avoid storing elements of more than 25 bits if you want compatibility with 32-bits bitstream readers. +* Avoid storing elements of more than 24 bits if you want compatibility with 32-bits bitstream readers. * * Last operation is to close the bitStream. * The function returns the final size of CStream in bytes. -* If data couldn't fit into dstBuffer, it will return a 0 ( == not storable) +* If data couldn't fit into @dstBuffer, it will return a 0 ( == not storable) */ -/********************************************** -* bitStream decompression API (read backward) +/*-******************************************** +* bitStream decoding API (read backward) **********************************************/ typedef struct { @@ -118,19 +117,19 @@ MEM_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD); MEM_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t* bitD); -/* +/*! * Start by invoking BIT_initDStream(). * A chunk of the bitStream is then stored into a local register. * Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t). * You can then retrieve bitFields stored into the local register, **in reverse order**. -* Local register is manually filled from memory by the BIT_reloadDStream() method. +* Local register is explicitly reloaded from memory by the BIT_reloadDStream() method. * A reload guarantee a minimum of ((8*sizeof(size_t))-7) bits when its result is BIT_DStream_unfinished. * Otherwise, it can be less than that, so proceed accordingly. * Checking if DStream has reached its end can be performed with BIT_endOfDStream() */ -/****************************************** +/*-**************************************** * unsafe API ******************************************/ MEM_STATIC void BIT_addBitsFast(BIT_CStream_t* bitC, size_t value, unsigned nbBits); @@ -144,7 +143,7 @@ MEM_STATIC size_t BIT_readBitsFast(BIT_DStream_t* bitD, unsigned nbBits); -/**************************************************************** +/*-************************************************************** * Helper functions ****************************************************************/ MEM_STATIC unsigned BIT_highbit32 (register U32 val) @@ -170,10 +169,9 @@ MEM_STATIC unsigned BIT_highbit32 (register U32 val) } -/**************************************************************** +/*-************************************************************** * bitStream encoding ****************************************************************/ - MEM_STATIC size_t BIT_initCStream(BIT_CStream_t* bitC, void* startPtr, size_t maxSize) { bitC->bitContainer = 0; @@ -240,10 +238,9 @@ MEM_STATIC size_t BIT_closeCStream(BIT_CStream_t* bitC) } -/********************************************************** +/*-******************************************************** * bitStream decoding **********************************************************/ - /*!BIT_initDStream * Initialize a BIT_DStream_t. * @bitD : a pointer to an already allocated BIT_DStream_t structure @@ -255,8 +252,7 @@ MEM_STATIC size_t BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, si { if (srcSize < 1) { memset(bitD, 0, sizeof(*bitD)); return ERROR(srcSize_wrong); } - if (srcSize >= sizeof(size_t)) /* normal case */ - { + if (srcSize >= sizeof(size_t)) { /* normal case */ U32 contain32; bitD->start = (const char*)srcBuffer; bitD->ptr = (const char*)srcBuffer + srcSize - sizeof(size_t); @@ -264,9 +260,7 @@ MEM_STATIC size_t BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, si contain32 = ((const BYTE*)srcBuffer)[srcSize-1]; if (contain32 == 0) return ERROR(GENERIC); /* endMark not present */ bitD->bitsConsumed = 8 - BIT_highbit32(contain32); - } - else - { + } else { U32 contain32; bitD->start = (const char*)srcBuffer; bitD->ptr = bitD->start; @@ -342,23 +336,20 @@ MEM_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD) if (bitD->bitsConsumed > (sizeof(bitD->bitContainer)*8)) /* should never happen */ return BIT_DStream_overflow; - if (bitD->ptr >= bitD->start + sizeof(bitD->bitContainer)) - { + if (bitD->ptr >= bitD->start + sizeof(bitD->bitContainer)) { bitD->ptr -= bitD->bitsConsumed >> 3; bitD->bitsConsumed &= 7; bitD->bitContainer = MEM_readLEST(bitD->ptr); return BIT_DStream_unfinished; } - if (bitD->ptr == bitD->start) - { + if (bitD->ptr == bitD->start) { if (bitD->bitsConsumed < sizeof(bitD->bitContainer)*8) return BIT_DStream_endOfBuffer; return BIT_DStream_completed; } { U32 nbBytes = bitD->bitsConsumed >> 3; BIT_DStream_status result = BIT_DStream_unfinished; - if (bitD->ptr - nbBytes < bitD->start) - { + if (bitD->ptr - nbBytes < bitD->start) { nbBytes = (U32)(bitD->ptr - bitD->start); /* ptr > start */ result = BIT_DStream_endOfBuffer; } diff --git a/lib/error_private.h b/lib/error_private.h index e56753899..c0c3f4900 100644 --- a/lib/error_private.h +++ b/lib/error_private.h @@ -40,14 +40,14 @@ extern "C" { #endif -/* ***************************************** -* Includes +/* **************************************** +* Dependencies ******************************************/ -#include /* size_t, ptrdiff_t */ +#include /* size_t */ #include "error_public.h" /* enum list */ -/* ***************************************** +/* **************************************** * Compiler-specific ******************************************/ #if defined(__GNUC__) @@ -61,43 +61,52 @@ extern "C" { #endif -/* ***************************************** -* Error Codes +/*-**************************************** +* Customization ******************************************/ +typedef ZSTD_ErrorCode ERR_enum; #define PREFIX(name) ZSTD_error_##name + +/*-**************************************** +* Error codes handling +******************************************/ #ifdef ERROR -# undef ERROR /* reported already defined on VS 2015 by Rich Geldreich */ +# undef ERROR /* reported already defined on VS 2015 (Rich Geldreich) */ #endif #define ERROR(name) (size_t)-PREFIX(name) ERR_STATIC unsigned ERR_isError(size_t code) { return (code > ERROR(maxCode)); } +ERR_STATIC ERR_enum ERR_getError(size_t code) { if (!ERR_isError(code)) return (ERR_enum)0; return (ERR_enum) (0-code); } -/* ***************************************** + +/*-**************************************** * Error Strings ******************************************/ ERR_STATIC const char* ERR_getErrorName(size_t code) { - static const char* codeError = "Unspecified error code"; - switch( (size_t)(0-code) ) + static const char* notErrorCode = "Unspecified error code"; + switch( ERR_getError(code) ) { - case ZSTD_error_No_Error: return "No error detected"; - case ZSTD_error_GENERIC: return "Error (generic)"; - case ZSTD_error_prefix_unknown: return "Unknown frame descriptor"; - case ZSTD_error_frameParameter_unsupported: return "Unsupported frame parameter"; - case ZSTD_error_frameParameter_unsupportedBy32bitsImplementation: return "Frame parameter unsupported in 32-bits mode"; - case ZSTD_error_init_missing: return "Context should be init first"; - case ZSTD_error_memory_allocation: return "Allocation error : not enough memory"; - case ZSTD_error_dstSize_tooSmall: return "Destination buffer is too small"; - case ZSTD_error_srcSize_wrong: return "Src size incorrect"; - case ZSTD_error_corruption_detected: return "Corrupted block detected"; - case ZSTD_error_tableLog_tooLarge: return "tableLog requires too much memory"; - case ZSTD_error_maxSymbolValue_tooLarge: return "Unsupported max possible Symbol Value : too large"; - case ZSTD_error_maxSymbolValue_tooSmall: return "Specified maxSymbolValue is too small"; - case ZSTD_error_maxCode: - default: return codeError; + case PREFIX(no_error): return "No error detected"; + case PREFIX(GENERIC): return "Error (generic)"; + case PREFIX(prefix_unknown): return "Unknown frame descriptor"; + case PREFIX(frameParameter_unsupported): return "Unsupported frame parameter"; + case PREFIX(frameParameter_unsupportedBy32bits): return "Frame parameter unsupported in 32-bits mode"; + case PREFIX(init_missing): return "Context should be init first"; + case PREFIX(memory_allocation): return "Allocation error : not enough memory"; + case PREFIX(stage_wrong): return "Operation not authorized at current processing stage"; + case PREFIX(dstSize_tooSmall): return "Destination buffer is too small"; + case PREFIX(srcSize_wrong): return "Src size incorrect"; + case PREFIX(corruption_detected): return "Corrupted block detected"; + case PREFIX(tableLog_tooLarge): return "tableLog requires too much memory"; + case PREFIX(maxSymbolValue_tooLarge): return "Unsupported max possible Symbol Value : too large"; + case PREFIX(maxSymbolValue_tooSmall): return "Specified maxSymbolValue is too small"; + case PREFIX(dictionary_corrupted): return "Dictionary is corrupted"; + case PREFIX(maxCode): + default: return notErrorCode; /* should be impossible, due to ERR_getError() */ } } diff --git a/lib/error_public.h b/lib/error_public.h index 78b0e80a9..655e28e0c 100644 --- a/lib/error_public.h +++ b/lib/error_public.h @@ -39,14 +39,14 @@ extern "C" { /* **************************************** -* error list +* error codes list ******************************************/ -enum { - ZSTD_error_No_Error, +typedef enum { + ZSTD_error_no_error, ZSTD_error_GENERIC, ZSTD_error_prefix_unknown, ZSTD_error_frameParameter_unsupported, - ZSTD_error_frameParameter_unsupportedBy32bitsImplementation, + ZSTD_error_frameParameter_unsupportedBy32bits, ZSTD_error_init_missing, ZSTD_error_memory_allocation, ZSTD_error_stage_wrong, @@ -56,8 +56,9 @@ enum { ZSTD_error_tableLog_tooLarge, ZSTD_error_maxSymbolValue_tooLarge, ZSTD_error_maxSymbolValue_tooSmall, + ZSTD_error_dictionary_corrupted, ZSTD_error_maxCode -}; +} ZSTD_ErrorCode; /* note : functions provide error codes in reverse negative order, so compare with (size_t)(0-enum) */ diff --git a/lib/fse.c b/lib/fse.c index e74c1e833..986a0da15 100644 --- a/lib/fse.c +++ b/lib/fse.c @@ -141,102 +141,6 @@ typedef U32 DTable_max_t[FSE_DTABLE_SIZE_U32(FSE_MAX_TABLELOG)]; /* Function templates */ -size_t FSE_count_generic(unsigned* count, unsigned* maxSymbolValuePtr, const FSE_FUNCTION_TYPE* source, size_t sourceSize, unsigned safe) -{ - const FSE_FUNCTION_TYPE* ip = source; - const FSE_FUNCTION_TYPE* const iend = ip+sourceSize; - unsigned maxSymbolValue = *maxSymbolValuePtr; - unsigned max=0; - int s; - - U32 Counting1[FSE_MAX_SYMBOL_VALUE+1] = { 0 }; - U32 Counting2[FSE_MAX_SYMBOL_VALUE+1] = { 0 }; - U32 Counting3[FSE_MAX_SYMBOL_VALUE+1] = { 0 }; - U32 Counting4[FSE_MAX_SYMBOL_VALUE+1] = { 0 }; - - /* safety checks */ - if (!sourceSize) - { - memset(count, 0, (maxSymbolValue + 1) * sizeof(FSE_FUNCTION_TYPE)); - *maxSymbolValuePtr = 0; - return 0; - } - if (maxSymbolValue > FSE_MAX_SYMBOL_VALUE) return ERROR(GENERIC); /* maxSymbolValue too large : unsupported */ - if (!maxSymbolValue) maxSymbolValue = FSE_MAX_SYMBOL_VALUE; /* 0 == default */ - - if ((safe) || (sizeof(FSE_FUNCTION_TYPE)>1)) - { - /* check input values, to avoid count table overflow */ - while (ip < iend-3) - { - if (*ip>maxSymbolValue) return ERROR(GENERIC); Counting1[*ip++]++; - if (*ip>maxSymbolValue) return ERROR(GENERIC); Counting2[*ip++]++; - if (*ip>maxSymbolValue) return ERROR(GENERIC); Counting3[*ip++]++; - if (*ip>maxSymbolValue) return ERROR(GENERIC); Counting4[*ip++]++; - } - } - else - { - U32 cached = MEM_read32(ip); ip += 4; - while (ip < iend-15) - { - U32 c = cached; cached = MEM_read32(ip); ip += 4; - Counting1[(BYTE) c ]++; - Counting2[(BYTE)(c>>8) ]++; - Counting3[(BYTE)(c>>16)]++; - Counting4[ c>>24 ]++; - c = cached; cached = MEM_read32(ip); ip += 4; - Counting1[(BYTE) c ]++; - Counting2[(BYTE)(c>>8) ]++; - Counting3[(BYTE)(c>>16)]++; - Counting4[ c>>24 ]++; - c = cached; cached = MEM_read32(ip); ip += 4; - Counting1[(BYTE) c ]++; - Counting2[(BYTE)(c>>8) ]++; - Counting3[(BYTE)(c>>16)]++; - Counting4[ c>>24 ]++; - c = cached; cached = MEM_read32(ip); ip += 4; - Counting1[(BYTE) c ]++; - Counting2[(BYTE)(c>>8) ]++; - Counting3[(BYTE)(c>>16)]++; - Counting4[ c>>24 ]++; - } - ip-=4; - } - - /* finish last symbols */ - while (ipmaxSymbolValue)) return ERROR(GENERIC); Counting1[*ip++]++; } - - for (s=0; s<=(int)maxSymbolValue; s++) - { - count[s] = Counting1[s] + Counting2[s] + Counting3[s] + Counting4[s]; - if (count[s] > max) max = count[s]; - } - - while (!count[maxSymbolValue]) maxSymbolValue--; - *maxSymbolValuePtr = maxSymbolValue; - return (size_t)max; -} - -/* hidden fast variant (unsafe) */ -size_t FSE_FUNCTION_NAME(FSE_countFast, FSE_FUNCTION_EXTENSION) -(unsigned* count, unsigned* maxSymbolValuePtr, const FSE_FUNCTION_TYPE* source, size_t sourceSize) -{ - return FSE_count_generic(count, maxSymbolValuePtr, source, sourceSize, 0); -} - -size_t FSE_FUNCTION_NAME(FSE_count, FSE_FUNCTION_EXTENSION) -(unsigned* count, unsigned* maxSymbolValuePtr, const FSE_FUNCTION_TYPE* source, size_t sourceSize) -{ - if ((sizeof(FSE_FUNCTION_TYPE)==1) && (*maxSymbolValuePtr >= 255)) - { - *maxSymbolValuePtr = 255; - return FSE_count_generic(count, maxSymbolValuePtr, source, sourceSize, 0); - } - return FSE_count_generic(count, maxSymbolValuePtr, source, sourceSize, 1); -} - - static U32 FSE_tableStep(U32 tableSize) { return (tableSize>>1) + (tableSize>>3) + 3; } size_t FSE_buildCTable(FSE_CTable* ct, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog) @@ -264,35 +168,28 @@ size_t FSE_buildCTable(FSE_CTable* ct, const short* normalizedCounter, unsigned /* symbol start positions */ cumul[0] = 0; - for (i=1; i<=maxSymbolValue+1; i++) - { - if (normalizedCounter[i-1]==-1) /* Low proba symbol */ - { + for (i=1; i<=maxSymbolValue+1; i++) { + if (normalizedCounter[i-1]==-1) { /* Low proba symbol */ cumul[i] = cumul[i-1] + 1; tableSymbol[highThreshold--] = (FSE_FUNCTION_TYPE)(i-1); - } - else + } else { cumul[i] = cumul[i-1] + normalizedCounter[i-1]; - } + } } cumul[maxSymbolValue+1] = tableSize+1; /* Spread symbols */ - for (symbol=0; symbol<=maxSymbolValue; symbol++) - { + for (symbol=0; symbol<=maxSymbolValue; symbol++) { int nbOccurences; - for (nbOccurences=0; nbOccurences highThreshold) position = (position + step) & tableMask; /* Low proba area */ - } - } + } } if (position!=0) return ERROR(GENERIC); /* Must have gone through all positions */ /* Build table */ - for (i=0; i= largeLimit) noLarge=0; symbolNext[s] = normalizedCounter[s]; - } - } + } } /* Spread symbols */ - for (s=0; s<=maxSymbolValue; s++) - { + for (s=0; s<=maxSymbolValue; s++) { int i; - for (i=0; i highThreshold) position = (position + step) & tableMask; /* lowprob area */ - } - } + } } if (position!=0) return ERROR(GENERIC); /* position must reach all cells once, otherwise normalizedCounter is incorrect */ /* Build Decoding table */ { U32 i; - for (i=0; i1) /* stops at 1 */ - { - if (previous0) - { + while (remaining>1) { /* stops at 1 */ + if (previous0) { unsigned start = charnum; while (!normalizedCounter[charnum]) charnum++; - while (charnum >= start+24) - { + while (charnum >= start+24) { start+=24; bitStream += 0xFFFFU << bitCount; if ((!writeIsSafe) && (out > oend-2)) return ERROR(dstSize_tooSmall); /* Buffer overflow */ @@ -473,24 +350,21 @@ static size_t FSE_writeNCount_generic (void* header, size_t headerBufferSize, out+=2; bitStream>>=16; } - while (charnum >= start+3) - { + while (charnum >= start+3) { start+=3; bitStream += 3 << bitCount; bitCount += 2; } bitStream += (charnum-start) << bitCount; bitCount += 2; - if (bitCount>16) - { + if (bitCount>16) { if ((!writeIsSafe) && (out > oend - 2)) return ERROR(dstSize_tooSmall); /* Buffer overflow */ out[0] = (BYTE)bitStream; out[1] = (BYTE)(bitStream>>8); out += 2; bitStream >>= 16; bitCount -= 16; - } - } + } } { short count = normalizedCounter[charnum++]; const short max = (short)((2*threshold-1)-remaining); @@ -504,16 +378,14 @@ static size_t FSE_writeNCount_generic (void* header, size_t headerBufferSize, previous0 = (count==1); while (remaining>=1; } - if (bitCount>16) - { + if (bitCount>16) { if ((!writeIsSafe) && (out > oend - 2)) return ERROR(dstSize_tooSmall); /* Buffer overflow */ out[0] = (BYTE)bitStream; out[1] = (BYTE)(bitStream>>8); out += 2; bitStream >>= 16; bitCount -= 16; - } - } + } } /* flush remaining bitStream */ if ((!writeIsSafe) && (out > oend - 2)) return ERROR(dstSize_tooSmall); /* Buffer overflow */ @@ -564,27 +436,19 @@ size_t FSE_readNCount (short* normalizedCounter, unsigned* maxSVPtr, unsigned* t threshold = 1<1) && (charnum<=*maxSVPtr)) - { - if (previous0) - { + while ((remaining>1) && (charnum<=*maxSVPtr)) { + if (previous0) { unsigned n0 = charnum; - while ((bitStream & 0xFFFF) == 0xFFFF) - { + while ((bitStream & 0xFFFF) == 0xFFFF) { n0+=24; - if (ip < iend-5) - { + if (ip < iend-5) { ip+=2; bitStream = MEM_readLE32(ip) >> bitCount; - } - else - { + } else { bitStream >>= 16; bitCount+=16; - } - } - while ((bitStream & 3) == 3) - { + } } + while ((bitStream & 3) == 3) { n0+=3; bitStream>>=2; bitCount+=2; @@ -593,8 +457,7 @@ size_t FSE_readNCount (short* normalizedCounter, unsigned* maxSVPtr, unsigned* t bitCount += 2; if (n0 > *maxSVPtr) return ERROR(maxSymbolValue_tooSmall); while (charnum < n0) normalizedCounter[charnum++] = 0; - if ((ip <= iend-7) || (ip + (bitCount>>3) <= iend-4)) - { + if ((ip <= iend-7) || (ip + (bitCount>>3) <= iend-4)) { ip += bitCount>>3; bitCount &= 7; bitStream = MEM_readLE32(ip) >> bitCount; @@ -606,13 +469,10 @@ size_t FSE_readNCount (short* normalizedCounter, unsigned* maxSVPtr, unsigned* t const short max = (short)((2*threshold-1)-remaining); short count; - if ((bitStream & (threshold-1)) < (U32)max) - { + if ((bitStream & (threshold-1)) < (U32)max) { count = (short)(bitStream & (threshold-1)); bitCount += nbBits-1; - } - else - { + } else { count = (short)(bitStream & (2*threshold-1)); if (count >= threshold) count -= max; bitCount += nbBits; @@ -622,27 +482,20 @@ size_t FSE_readNCount (short* normalizedCounter, unsigned* maxSVPtr, unsigned* t remaining -= FSE_abs(count); normalizedCounter[charnum++] = count; previous0 = !count; - while (remaining < threshold) - { + while (remaining < threshold) { nbBits--; threshold >>= 1; } - { - if ((ip <= iend-7) || (ip + (bitCount>>3) <= iend-4)) - { - ip += bitCount>>3; - bitCount &= 7; - } - else - { - bitCount -= (int)(8 * (iend - 4 - ip)); - ip = iend - 4; - } - bitStream = MEM_readLE32(ip) >> (bitCount & 31); + if ((ip <= iend-7) || (ip + (bitCount>>3) <= iend-4)) { + ip += bitCount>>3; + bitCount &= 7; + } else { + bitCount -= (int)(8 * (iend - 4 - ip)); + ip = iend - 4; } - } - } + bitStream = MEM_readLE32(ip) >> (bitCount & 31); + } } if (remaining != 1) return ERROR(GENERIC); *maxSVPtr = charnum-1; @@ -652,10 +505,130 @@ size_t FSE_readNCount (short* normalizedCounter, unsigned* maxSVPtr, unsigned* t } -/**************************************************************** +/*-************************************************************** +* Counting histogram +****************************************************************/ +/*! FSE_count_simple + This function just counts byte values within @src, + and store the histogram into @count. + This function is unsafe : it doesn't check that all values within @src can fit into @count. + For this reason, prefer using a table @count with 256 elements. + @return : highest count for a single element +*/ +static size_t FSE_count_simple(unsigned* count, unsigned* maxSymbolValuePtr, + const void* src, size_t srcSize) +{ + const BYTE* ip = (const BYTE*)src; + const BYTE* const end = ip + srcSize; + unsigned maxSymbolValue = *maxSymbolValuePtr; + unsigned max=0; + U32 s; + + memset(count, 0, (maxSymbolValue+1)*sizeof(*count)); + if (srcSize==0) { *maxSymbolValuePtr = 0; return 0; } + + while (ip max) max = count[s]; + + return (size_t)max; +} + + +static size_t FSE_count_parallel(unsigned* count, unsigned* maxSymbolValuePtr, + const void* source, size_t sourceSize, + unsigned checkMax) +{ + const BYTE* ip = (const BYTE*)source; + const BYTE* const iend = ip+sourceSize; + unsigned maxSymbolValue = *maxSymbolValuePtr; + unsigned max=0; + U32 s; + + U32 Counting1[256] = { 0 }; + U32 Counting2[256] = { 0 }; + U32 Counting3[256] = { 0 }; + U32 Counting4[256] = { 0 }; + + /* safety checks */ + if (!sourceSize) { + memset(count, 0, maxSymbolValue + 1); + *maxSymbolValuePtr = 0; + return 0; + } + if (!maxSymbolValue) maxSymbolValue = 255; /* 0 == default */ + + { /* by stripes of 16 bytes */ + U32 cached = MEM_read32(ip); ip += 4; + while (ip < iend-15) { + U32 c = cached; cached = MEM_read32(ip); ip += 4; + Counting1[(BYTE) c ]++; + Counting2[(BYTE)(c>>8) ]++; + Counting3[(BYTE)(c>>16)]++; + Counting4[ c>>24 ]++; + c = cached; cached = MEM_read32(ip); ip += 4; + Counting1[(BYTE) c ]++; + Counting2[(BYTE)(c>>8) ]++; + Counting3[(BYTE)(c>>16)]++; + Counting4[ c>>24 ]++; + c = cached; cached = MEM_read32(ip); ip += 4; + Counting1[(BYTE) c ]++; + Counting2[(BYTE)(c>>8) ]++; + Counting3[(BYTE)(c>>16)]++; + Counting4[ c>>24 ]++; + c = cached; cached = MEM_read32(ip); ip += 4; + Counting1[(BYTE) c ]++; + Counting2[(BYTE)(c>>8) ]++; + Counting3[(BYTE)(c>>16)]++; + Counting4[ c>>24 ]++; + } + ip-=4; + } + + /* finish last symbols */ + while (ipmaxSymbolValue; s--) { + Counting1[s] += Counting2[s] + Counting3[s] + Counting4[s]; + if (Counting1[s]) return ERROR(maxSymbolValue_tooSmall); + } } + + for (s=0; s<=maxSymbolValue; s++) { + count[s] = Counting1[s] + Counting2[s] + Counting3[s] + Counting4[s]; + if (count[s] > max) max = count[s]; + } + + while (!count[maxSymbolValue]) maxSymbolValue--; + *maxSymbolValuePtr = maxSymbolValue; + return (size_t)max; +} + +/* fast variant (unsafe : won't check if src contains values beyond count[] limit) */ +size_t FSE_countFast(unsigned* count, unsigned* maxSymbolValuePtr, + const void* source, size_t sourceSize) +{ + if (sourceSize < 1500) return FSE_count_simple(count, maxSymbolValuePtr, source, sourceSize); + return FSE_count_parallel(count, maxSymbolValuePtr, source, sourceSize, 0); +} + +size_t FSE_count(unsigned* count, unsigned* maxSymbolValuePtr, + const void* source, size_t sourceSize) +{ + if (*maxSymbolValuePtr <255) + return FSE_count_parallel(count, maxSymbolValuePtr, source, sourceSize, 1); + *maxSymbolValuePtr = 255; + return FSE_countFast(count, maxSymbolValuePtr, source, sourceSize); +} + + +/*-************************************************************** * FSE Compression Code ****************************************************************/ -/* +/*! FSE_CTable is a variable size structure which contains : U16 tableLog; U16 maxSymbolValue; @@ -686,7 +659,6 @@ void FSE_freeCTable (FSE_CTable* ct) free(ct); } - /* provides the minimum logSize to safely represent a distribution */ static unsigned FSE_minTableLog(size_t srcSize, unsigned maxSymbolValue) { @@ -723,22 +695,18 @@ static size_t FSE_normalizeM2(short* norm, U32 tableLog, const unsigned* count, U32 lowThreshold = (U32)(total >> tableLog); U32 lowOne = (U32)((total * 3) >> (tableLog + 1)); - for (s=0; s<=maxSymbolValue; s++) - { - if (count[s] == 0) - { + for (s=0; s<=maxSymbolValue; s++) { + if (count[s] == 0) { norm[s]=0; continue; } - if (count[s] <= lowThreshold) - { + if (count[s] <= lowThreshold) { norm[s] = -1; distributed++; total -= count[s]; continue; } - if (count[s] <= lowOne) - { + if (count[s] <= lowOne) { norm[s] = 1; distributed++; total -= count[s]; @@ -748,25 +716,20 @@ static size_t FSE_normalizeM2(short* norm, U32 tableLog, const unsigned* count, } ToDistribute = (1 << tableLog) - distributed; - if ((total / ToDistribute) > lowOne) - { + if ((total / ToDistribute) > lowOne) { /* risk of rounding to zero */ lowOne = (U32)((total * 3) / (ToDistribute * 2)); - for (s=0; s<=maxSymbolValue; s++) - { - if ((norm[s] == -2) && (count[s] <= lowOne)) - { + for (s=0; s<=maxSymbolValue; s++) { + if ((norm[s] == -2) && (count[s] <= lowOne)) { norm[s] = 1; distributed++; total -= count[s]; continue; - } - } + } } ToDistribute = (1 << tableLog) - distributed; } - if (distributed == maxSymbolValue+1) - { + if (distributed == maxSymbolValue+1) { /* all values are pretty poor; probably incompressible data (should have already been detected); find max, then give all remaining points to max */ @@ -782,10 +745,8 @@ static size_t FSE_normalizeM2(short* norm, U32 tableLog, const unsigned* count, U64 const mid = (1ULL << (vStepLog-1)) - 1; U64 const rStep = ((((U64)1<> vStepLog); U32 sEnd = (U32)(end >> vStepLog); @@ -794,9 +755,7 @@ static size_t FSE_normalizeM2(short* norm, U32 tableLog, const unsigned* count, return ERROR(GENERIC); norm[s] = (short)weight; tmpTotal = end; - } - } - } + } } } return 0; } @@ -809,7 +768,7 @@ size_t FSE_normalizeCount (short* normalizedCounter, unsigned tableLog, /* Sanity checks */ if (tableLog==0) tableLog = FSE_DEFAULT_TABLELOG; if (tableLog < FSE_MIN_TABLELOG) return ERROR(GENERIC); /* Unsupported size */ - if (tableLog > FSE_MAX_TABLELOG) return ERROR(GENERIC); /* Unsupported size */ + if (tableLog > FSE_MAX_TABLELOG) return ERROR(tableLog_tooLarge); /* Unsupported size */ if (tableLog < FSE_minTableLog(total, maxSymbolValue)) return ERROR(GENERIC); /* Too small tableLog, compression potentially impossible */ { @@ -823,38 +782,23 @@ size_t FSE_normalizeCount (short* normalizedCounter, unsigned tableLog, short largestP=0; U32 lowThreshold = (U32)(total >> tableLog); - for (s=0; s<=maxSymbolValue; s++) - { - if (count[s] == total) return 0; - if (count[s] == 0) - { - normalizedCounter[s]=0; - continue; - } - if (count[s] <= lowThreshold) - { + for (s=0; s<=maxSymbolValue; s++) { + if (count[s] == total) return 0; /* rle special case */ + if (count[s] == 0) { normalizedCounter[s]=0; continue; } + if (count[s] <= lowThreshold) { normalizedCounter[s] = -1; stillToDistribute--; - } - else - { + } else { short proba = (short)((count[s]*step) >> scale); - if (proba<8) - { + if (proba<8) { U64 restToBeat = vStep * rtbTable[proba]; proba += (count[s]*step) - ((U64)proba< restToBeat; } - if (proba > largestP) - { - largestP=proba; - largest=s; - } + if (proba > largestP) largestP=proba, largest=s; normalizedCounter[s] = proba; stillToDistribute -= proba; - } - } - if (-stillToDistribute >= (normalizedCounter[largest] >> 1)) - { + } } + if (-stillToDistribute >= (normalizedCounter[largest] >> 1)) { /* corner case, need another normalization method */ size_t errorCode = FSE_normalizeM2(normalizedCounter, tableLog, count, total, maxSymbolValue); if (FSE_isError(errorCode)) return errorCode; @@ -904,10 +848,12 @@ size_t FSE_buildCTable_raw (FSE_CTable* ct, unsigned nbBits) tableU16[s] = (U16)(tableSize + s); /* Build Symbol Transformation Table */ - for (s=0; s<=maxSymbolValue; s++) { - symbolTT[s].deltaNbBits = nbBits << 16; - symbolTT[s].deltaFindState = s-1; + const U32 deltaNbBits = (nbBits << 16) - (1 << nbBits); + for (s=0; s<=maxSymbolValue; s++) { + symbolTT[s].deltaNbBits = deltaNbBits; + symbolTT[s].deltaFindState = s-1; + } } return 0; @@ -930,10 +876,8 @@ size_t FSE_buildCTable_rle (FSE_CTable* ct, BYTE symbolValue) tableU16[1] = 0; /* just in case */ /* Build Symbol Transformation Table */ - { - symbolTT[symbolValue].deltaNbBits = 0; - symbolTT[symbolValue].deltaFindState = 0; - } + symbolTT[symbolValue].deltaNbBits = 0; + symbolTT[symbolValue].deltaFindState = 0; return 0; } @@ -963,15 +907,13 @@ static size_t FSE_compress_usingCTable_generic (void* dst, size_t dstSize, #define FSE_FLUSHBITS(s) (fast ? BIT_flushBitsFast(s) : BIT_flushBits(s)) /* join to even */ - if (srcSize & 1) - { + if (srcSize & 1) { FSE_encodeSymbol(&bitC, &CState1, *--ip); FSE_FLUSHBITS(&bitC); } /* join to mod 4 */ - if ((sizeof(bitC.bitContainer)*8 > FSE_MAX_TABLELOG*4+7 ) && (srcSize & 2)) /* test bit 2 */ - { + if ((sizeof(bitC.bitContainer)*8 > FSE_MAX_TABLELOG*4+7 ) && (srcSize & 2)) { /* test bit 2 */ FSE_encodeSymbol(&bitC, &CState2, *--ip); FSE_encodeSymbol(&bitC, &CState1, *--ip); FSE_FLUSHBITS(&bitC); @@ -987,8 +929,7 @@ static size_t FSE_compress_usingCTable_generic (void* dst, size_t dstSize, FSE_encodeSymbol(&bitC, &CState1, *--ip); - if (sizeof(bitC.bitContainer)*8 > FSE_MAX_TABLELOG*4+7 ) /* this test must be static */ - { + if (sizeof(bitC.bitContainer)*8 > FSE_MAX_TABLELOG*4+7 ) { /* this test must be static */ FSE_encodeSymbol(&bitC, &CState2, *--ip); FSE_encodeSymbol(&bitC, &CState1, *--ip); } @@ -1071,7 +1012,7 @@ size_t FSE_compress (void* dst, size_t dstSize, const void* src, size_t srcSize) } -/********************************************************* +/*-******************************************************* * Decompression (Byte symbols) *********************************************************/ size_t FSE_buildDTable_rle (FSE_DTable* dt, BYTE symbolValue) @@ -1109,8 +1050,7 @@ size_t FSE_buildDTable_raw (FSE_DTable* dt, unsigned nbBits) /* Build Decoding Table */ DTableH->tableLog = (U16)nbBits; DTableH->fastMode = 1; - for (s=0; s<=maxSymbolValue; s++) - { + for (s=0; s<=maxSymbolValue; s++) { dinfo[s].newState = 0; dinfo[s].symbol = (BYTE)s; dinfo[s].nbBits = (BYTE)nbBits; @@ -1144,8 +1084,7 @@ FORCE_INLINE size_t FSE_decompress_usingDTable_generic( #define FSE_GETSYMBOL(statePtr) fast ? FSE_decodeSymbolFast(statePtr, &bitD) : FSE_decodeSymbol(statePtr, &bitD) /* 4 symbols per loop */ - for ( ; (BIT_reloadDStream(&bitD)==BIT_DStream_unfinished) && (op sizeof(bitD.bitContainer)*8) /* This test must be static */ @@ -1166,8 +1105,7 @@ FORCE_INLINE size_t FSE_decompress_usingDTable_generic( /* tail */ /* note : BIT_reloadDStream(&bitD) >= FSE_DStream_partiallyFilled; Ends at exactly BIT_DStream_completed */ - while (1) - { + while (1) { if ( (BIT_reloadDStream(&bitD)>BIT_DStream_completed) || (op==omax) || (BIT_endOfDStream(&bitD) && (fast || FSE_endOfDState(&state1))) ) break; diff --git a/lib/fse.h b/lib/fse.h index dd1190f86..db6f49cfa 100644 --- a/lib/fse.h +++ b/lib/fse.h @@ -46,7 +46,7 @@ extern "C" { #include /* size_t, ptrdiff_t */ -/* ***************************************** +/*-**************************************** * FSE simple functions ******************************************/ size_t FSE_compress(void* dst, size_t maxDstSize, @@ -124,13 +124,13 @@ or to save and provide normalized distribution using external method. /*! FSE_count(): - Provides the precise count of each symbol within a table 'count' - 'count' is a table of unsigned int, of minimum size (maxSymbolValuePtr[0]+1). - maxSymbolValuePtr[0] will be updated if detected smaller than initially expected - return : the count of the most frequent symbol (which is not identified) - if return == srcSize, there is only one symbol. - if FSE_isError(return), it's an error code. */ -size_t FSE_count(unsigned* count, unsigned* maxSymbolValuePtr, const unsigned char* src, size_t srcSize); + Provides the precise count of each byte within a table 'count' + 'count' is a table of unsigned int, of minimum size (*maxSymbolValuePtr+1). + *maxSymbolValuePtr will be updated if detected smaller than initial value. + @return : the count of the most frequent symbol (which is not identified) + if return == srcSize, there is only one symbol. + Can also return an error code, which can be tested with FSE_isError() */ +size_t FSE_count(unsigned* count, unsigned* maxSymbolValuePtr, const void* src, size_t srcSize); /*! FSE_optimalTableLog(): @@ -170,18 +170,18 @@ void FSE_freeCTable (FSE_CTable* ct); /*! FSE_buildCTable(): - Builds 'ct', which must be already allocated, using FSE_createCTable() + Builds @ct, which must be already allocated, using FSE_createCTable() return : 0 or an errorCode, which can be tested using FSE_isError() */ size_t FSE_buildCTable(FSE_CTable* ct, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog); /*! FSE_compress_usingCTable(): - Compress 'src' using 'ct' into 'dst' which must be already allocated - return : size of compressed data (<= maxDstSize) - or 0 if compressed data could not fit into 'dst' + Compress @src using @ct into @dst which must be already allocated + return : size of compressed data (<= @dstCapacity) + or 0 if compressed data could not fit into @dst or an errorCode, which can be tested using FSE_isError() */ -size_t FSE_compress_usingCTable (void* dst, size_t maxDstSize, const void* src, size_t srcSize, const FSE_CTable* ct); +size_t FSE_compress_usingCTable (void* dst, size_t dstCapacity, const void* src, size_t srcSize, const FSE_CTable* ct); /*! Tutorial : @@ -221,7 +221,7 @@ If there is an error, both functions will return an ErrorCode (which can be test 'CTable' can then be used to compress 'src', with FSE_compress_usingCTable(). Similar to FSE_count(), the convention is that 'src' is assumed to be a table of char of size 'srcSize' -The function returns the size of compressed data (without header), necessarily <= maxDstSize. +The function returns the size of compressed data (without header), necessarily <= @dstCapacity. If it returns '0', compressed data could not fit into 'dst'. If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError()). */ @@ -253,11 +253,11 @@ size_t FSE_buildDTable (FSE_DTable* dt, const short* normalizedCounter, unsigned /*! FSE_decompress_usingDTable(): - Decompress compressed source 'cSrc' of size 'cSrcSize' using 'dt' - into 'dst' which must be already allocated. - return : size of regenerated data (necessarily <= maxDstSize) + Decompress compressed source @cSrc of size @cSrcSize using @dt + into @dst which must be already allocated. + return : size of regenerated data (necessarily <= @dstCapacity) or an errorCode, which can be tested using FSE_isError() */ -size_t FSE_decompress_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const FSE_DTable* dt); +size_t FSE_decompress_usingDTable(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, const FSE_DTable* dt); /*! Tutorial : diff --git a/lib/fse_static.h b/lib/fse_static.h index a881e351f..ca303db84 100644 --- a/lib/fse_static.h +++ b/lib/fse_static.h @@ -63,8 +63,8 @@ extern "C" { /* ***************************************** * FSE advanced API *******************************************/ -size_t FSE_countFast(unsigned* count, unsigned* maxSymbolValuePtr, const unsigned char* src, size_t srcSize); -/* same as FSE_count(), but blindly trust that all values within src are <= *maxSymbolValuePtr */ +size_t FSE_countFast(unsigned* count, unsigned* maxSymbolValuePtr, const void* src, size_t srcSize); +/* same as FSE_count(), but blindly trusts that all byte values within src are <= *maxSymbolValuePtr */ size_t FSE_buildCTable_raw (FSE_CTable* ct, unsigned nbBits); /* build a fake FSE_CTable, designed to not compress an input, where each symbol uses nbBits */ @@ -223,8 +223,7 @@ static unsigned char FSE_decodeSymbolFast(FSE_DState_t* DStatePtr, BIT_DStream_t /* ***************************************** * Implementation of inlined functions *******************************************/ -typedef struct -{ +typedef struct { int deltaFindState; U32 deltaNbBits; } FSE_symbolCompressionTransform; /* total 8 bytes */ @@ -240,6 +239,19 @@ MEM_STATIC void FSE_initCState(FSE_CState_t* statePtr, const FSE_CTable* ct) statePtr->stateLog = tableLog; } +MEM_STATIC void FSE_initCState2(FSE_CState_t* statePtr, const FSE_CTable* ct, U32 symbol) +{ + FSE_initCState(statePtr, ct); + { + const FSE_symbolCompressionTransform symbolTT = ((const FSE_symbolCompressionTransform*)(statePtr->symbolTT))[symbol]; + const U16* stateTable = (const U16*)(statePtr->stateTable); + U32 nbBitsOut = (U32)((symbolTT.deltaNbBits + (1<<15)) >> 16); + statePtr->value = (nbBitsOut << 16) - symbolTT.deltaNbBits; + statePtr->value = stateTable[(statePtr->value >> nbBitsOut) + symbolTT.deltaFindState]; + + } +} + MEM_STATIC void FSE_encodeSymbol(BIT_CStream_t* bitC, FSE_CState_t* statePtr, U32 symbol) { const FSE_symbolCompressionTransform symbolTT = ((const FSE_symbolCompressionTransform*)(statePtr->symbolTT))[symbol]; @@ -278,6 +290,17 @@ MEM_STATIC void FSE_initDState(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD, con DStatePtr->table = dt + 1; } +MEM_STATIC size_t FSE_getStateValue(FSE_DState_t* DStatePtr) +{ + return DStatePtr->state; +} + +MEM_STATIC BYTE FSE_peakSymbol(FSE_DState_t* DStatePtr) +{ + const FSE_decode_t DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state]; + return DInfo.symbol; +} + MEM_STATIC BYTE FSE_decodeSymbol(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD) { const FSE_decode_t DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state]; diff --git a/lib/huff0.c b/lib/huff0.c index 26a7639ff..929bc87b1 100644 --- a/lib/huff0.c +++ b/lib/huff0.c @@ -49,7 +49,6 @@ # pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */ #else # ifdef __GNUC__ -# define GCC_VERSION (__GNUC__ * 100 + __GNUC_MINOR__) # define FORCE_INLINE static inline __attribute__((always_inline)) # else # define FORCE_INLINE static inline @@ -107,7 +106,8 @@ typedef struct nodeElt_s { @dst : destination buffer @CTable : huffman tree to save, using huff0 representation @return : size of saved CTable */ -size_t HUF_writeCTable (void* dst, size_t maxDstSize, const HUF_CElt* CTable, U32 maxSymbolValue, U32 huffLog) +size_t HUF_writeCTable (void* dst, size_t maxDstSize, + const HUF_CElt* CTable, U32 maxSymbolValue, U32 huffLog) { BYTE bitsToWeight[HUF_MAX_TABLELOG + 1]; BYTE huffWeight[HUF_MAX_SYMBOL_VALUE + 1]; @@ -129,10 +129,8 @@ size_t HUF_writeCTable (void* dst, size_t maxDstSize, const HUF_CElt* CTable, U3 size = FSE_compress(op+1, maxDstSize-1, huffWeight, maxSymbolValue); /* don't need last symbol stat : implied */ if (HUF_isError(size)) return size; if (size >= 128) return ERROR(GENERIC); /* should never happen, since maxSymbolValue <= 255 */ - if ((size <= 1) || (size >= maxSymbolValue/2)) - { - if (size==1) /* RLE */ - { + if ((size <= 1) || (size >= maxSymbolValue/2)) { + if (size==1) { /* RLE */ /* only possible case : serie of 1 (because there are at least 2) */ /* can only be 2^n or (2^n-1), otherwise not an huffman tree */ BYTE code; @@ -173,6 +171,66 @@ size_t HUF_writeCTable (void* dst, size_t maxDstSize, const HUF_CElt* CTable, U3 } +static size_t HUF_readStats(BYTE* huffWeight, size_t hwSize, U32* rankStats, + U32* nbSymbolsPtr, U32* tableLogPtr, + const void* src, size_t srcSize); + + +size_t HUF_readCTable (HUF_CElt* CTable, U32 maxSymbolValue, const void* src, size_t srcSize) +{ + BYTE huffWeight[HUF_MAX_SYMBOL_VALUE + 1]; + U32 rankVal[HUF_ABSOLUTEMAX_TABLELOG + 1]; /* large enough for values from 0 to 16 */ + U32 tableLog = 0; + size_t iSize; + U32 nbSymbols = 0; + U32 n; + U32 nextRankStart; + //memset(huffWeight, 0, sizeof(huffWeight)); /* is not necessary, even though some analyzer complain ... */ + + /* get symbol weights */ + iSize = HUF_readStats(huffWeight, HUF_MAX_SYMBOL_VALUE+1, rankVal, &nbSymbols, &tableLog, src, srcSize); + if (HUF_isError(iSize)) return iSize; + + /* check result */ + if (tableLog > HUF_MAX_TABLELOG) return ERROR(tableLog_tooLarge); + if (nbSymbols > maxSymbolValue+1) return ERROR(maxSymbolValue_tooSmall); + + /* Prepare base value per rank */ + nextRankStart = 0; + for (n=1; n<=tableLog; n++) { + U32 current = nextRankStart; + nextRankStart += (rankVal[n] << (n-1)); + rankVal[n] = current; + } + + /* fill nbBits */ + for (n=0; n0; n--) { + valPerRank[n] = min; /* get starting value within each rank */ + min += nbPerRank[n]; + min >>= 1; + } } + for (n=0; n<=maxSymbolValue; n++) + CTable[n].val = valPerRank[CTable[n].nbBits]++; /* assign value within rank, symbol order */ + } + + return iSize; +} + + static U32 HUF_setMaxHeight(nodeElt* huffNode, U32 lastNonNull, U32 maxNbBits) { int totalCost = 0; @@ -186,8 +244,7 @@ static U32 HUF_setMaxHeight(nodeElt* huffNode, U32 lastNonNull, U32 maxNbBits) const U32 baseCost = 1 << (largestBits - maxNbBits); U32 n = lastNonNull; - while (huffNode[n].nbBits > maxNbBits) - { + while (huffNode[n].nbBits > maxNbBits) { totalCost += baseCost - (1 << (largestBits - huffNode[n].nbBits)); huffNode[n].nbBits = (BYTE)maxNbBits; n --; @@ -206,18 +263,15 @@ static U32 HUF_setMaxHeight(nodeElt* huffNode, U32 lastNonNull, U32 maxNbBits) /* Get pos of last (smallest) symbol per rank */ memset(rankLast, 0xF0, sizeof(rankLast)); - for (pos=n ; pos >= 0; pos--) - { + for (pos=n ; pos >= 0; pos--) { if (huffNode[pos].nbBits >= currentNbBits) continue; currentNbBits = huffNode[pos].nbBits; /* < maxNbBits */ rankLast[maxNbBits-currentNbBits] = pos; } - while (totalCost > 0) - { + while (totalCost > 0) { U32 nBitsToDecrease = BIT_highbit32(totalCost) + 1; - for ( ; nBitsToDecrease > 1; nBitsToDecrease--) - { + for ( ; nBitsToDecrease > 1; nBitsToDecrease--) { U32 highPos = rankLast[nBitsToDecrease]; U32 lowPos = rankLast[nBitsToDecrease-1]; if (highPos == noSymbol) continue; @@ -226,8 +280,7 @@ static U32 HUF_setMaxHeight(nodeElt* huffNode, U32 lastNonNull, U32 maxNbBits) U32 highTotal = huffNode[highPos].count; U32 lowTotal = 2 * huffNode[lowPos].count; if (highTotal <= lowTotal) break; - } - } + } } /* only triggered when no more rank 1 symbol left => find closest one (note : there is necessarily at least one !) */ while ((nBitsToDecrease<=HUF_MAX_TABLELOG) && (rankLast[nBitsToDecrease] == noSymbol)) /* HUF_MAX_TABLELOG test just to please gcc 5+; but it should not be necessary */ nBitsToDecrease ++; @@ -237,18 +290,14 @@ static U32 HUF_setMaxHeight(nodeElt* huffNode, U32 lastNonNull, U32 maxNbBits) huffNode[rankLast[nBitsToDecrease]].nbBits ++; if (rankLast[nBitsToDecrease] == 0) /* special case, reached largest symbol */ rankLast[nBitsToDecrease] = noSymbol; - else - { + else { rankLast[nBitsToDecrease]--; if (huffNode[rankLast[nBitsToDecrease]].nbBits != maxNbBits-nBitsToDecrease) rankLast[nBitsToDecrease] = noSymbol; /* this rank is now empty */ - } - } + } } - while (totalCost < 0) /* Sometimes, cost correction overshoot */ - { - if (rankLast[1] == noSymbol) /* special case : no rank 1 symbol (using maxNbBits-1); let's create one from largest rank 0 (using maxNbBits) */ - { + while (totalCost < 0) { /* Sometimes, cost correction overshoot */ + if (rankLast[1] == noSymbol) { /* special case : no rank 1 symbol (using maxNbBits-1); let's create one from largest rank 0 (using maxNbBits) */ while (huffNode[n].nbBits == maxNbBits) n--; huffNode[n+1].nbBits--; rankLast[1] = n+1; @@ -258,9 +307,7 @@ static U32 HUF_setMaxHeight(nodeElt* huffNode, U32 lastNonNull, U32 maxNbBits) huffNode[ rankLast[1] + 1 ].nbBits--; rankLast[1]++; totalCost ++; - } - } - } + } } } return maxNbBits; } @@ -277,15 +324,13 @@ static void HUF_sort(nodeElt* huffNode, const U32* count, U32 maxSymbolValue) U32 n; memset(rank, 0, sizeof(rank)); - for (n=0; n<=maxSymbolValue; n++) - { + for (n=0; n<=maxSymbolValue; n++) { U32 r = BIT_highbit32(count[n] + 1); rank[r].base ++; } for (n=30; n>0; n--) rank[n-1].base += rank[n].base; for (n=0; n<32; n++) rank[n].current = rank[n].base; - for (n=0; n<=maxSymbolValue; n++) - { + for (n=0; n<=maxSymbolValue; n++) { U32 c = count[n]; U32 r = BIT_highbit32(c+1) + 1; U32 pos = rank[r].current++; @@ -325,8 +370,7 @@ size_t HUF_buildCTable (HUF_CElt* tree, const U32* count, U32 maxSymbolValue, U3 huffNode0[0].count = (U32)(1U<<31); /* create parents */ - while (nodeNb <= nodeRoot) - { + while (nodeNb <= nodeRoot) { U32 n1 = (huffNode[lowS].count < huffNode[lowN].count) ? lowS-- : lowN++; U32 n2 = (huffNode[lowS].count < huffNode[lowN].count) ? lowS-- : lowN++; huffNode[nodeNb].count = huffNode[n1].count + huffNode[n2].count; @@ -354,8 +398,7 @@ size_t HUF_buildCTable (HUF_CElt* tree, const U32* count, U32 maxSymbolValue, U3 { /* determine stating value per rank */ U16 min = 0; - for (n=maxNbBits; n>0; n--) - { + for (n=maxNbBits; n>0; n--) { valPerRank[n] = min; /* get starting value within each rank */ min += nbPerRank[n]; min >>= 1; @@ -385,7 +428,7 @@ size_t HUF_compressBound(size_t size) { return HUF_COMPRESSBOUND(size); } #define HUF_FLUSHBITS_2(stream) \ if (sizeof((stream)->bitContainer)*8 < HUF_MAX_TABLELOG*4+7) HUF_FLUSHBITS(stream) -size_t HUF_compress_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable) +size_t HUF_compress1X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable) { const BYTE* ip = (const BYTE*) src; BYTE* const ostart = (BYTE*)dst; @@ -414,8 +457,7 @@ size_t HUF_compress_usingCTable(void* dst, size_t dstSize, const void* src, size default: ; } - for (; n>0; n-=4) /* note : n&3==0 at this stage */ - { + for (; n>0; n-=4) { /* note : n&3==0 at this stage */ HUF_encodeSymbol(&bitC, ip[n- 1], CTable); HUF_FLUSHBITS_1(&bitC); HUF_encodeSymbol(&bitC, ip[n- 2], CTable); @@ -430,7 +472,7 @@ size_t HUF_compress_usingCTable(void* dst, size_t dstSize, const void* src, size } -static size_t HUF_compress_into4Segments(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable) +size_t HUF_compress4X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable) { size_t segmentSize = (srcSize+3)/4; /* first 3 segments */ size_t errorCode; @@ -444,28 +486,28 @@ static size_t HUF_compress_into4Segments(void* dst, size_t dstSize, const void* if (srcSize < 12) return 0; /* no saving possible : too small input */ op += 6; /* jumpTable */ - errorCode = HUF_compress_usingCTable(op, oend-op, ip, segmentSize, CTable); + errorCode = HUF_compress1X_usingCTable(op, oend-op, ip, segmentSize, CTable); if (HUF_isError(errorCode)) return errorCode; if (errorCode==0) return 0; MEM_writeLE16(ostart, (U16)errorCode); ip += segmentSize; op += errorCode; - errorCode = HUF_compress_usingCTable(op, oend-op, ip, segmentSize, CTable); + errorCode = HUF_compress1X_usingCTable(op, oend-op, ip, segmentSize, CTable); if (HUF_isError(errorCode)) return errorCode; if (errorCode==0) return 0; MEM_writeLE16(ostart+2, (U16)errorCode); ip += segmentSize; op += errorCode; - errorCode = HUF_compress_usingCTable(op, oend-op, ip, segmentSize, CTable); + errorCode = HUF_compress1X_usingCTable(op, oend-op, ip, segmentSize, CTable); if (HUF_isError(errorCode)) return errorCode; if (errorCode==0) return 0; MEM_writeLE16(ostart+4, (U16)errorCode); ip += segmentSize; op += errorCode; - errorCode = HUF_compress_usingCTable(op, oend-op, ip, iend-ip, CTable); + errorCode = HUF_compress1X_usingCTable(op, oend-op, ip, iend-ip, CTable); if (HUF_isError(errorCode)) return errorCode; if (errorCode==0) return 0; @@ -474,9 +516,11 @@ static size_t HUF_compress_into4Segments(void* dst, size_t dstSize, const void* } -size_t HUF_compress2 (void* dst, size_t dstSize, +static size_t HUF_compress_internal ( + void* dst, size_t dstSize, const void* src, size_t srcSize, - unsigned maxSymbolValue, unsigned huffLog) + unsigned maxSymbolValue, unsigned huffLog, + unsigned singleStream) { BYTE* const ostart = (BYTE*)dst; BYTE* op = ostart; @@ -487,7 +531,7 @@ size_t HUF_compress2 (void* dst, size_t dstSize, size_t errorCode; /* checks & inits */ - if (srcSize < 1) return 0; /* Uncompressed */ + if (srcSize < 1) return 0; /* Uncompressed - note : 1 means rle, so first byte must be correct */ if (dstSize < 1) return 0; /* not compressible within dst budget */ if (srcSize > 128 * 1024) return ERROR(srcSize_wrong); /* current block size limit */ if (huffLog > HUF_MAX_TABLELOG) return ERROR(tableLog_tooLarge); @@ -512,8 +556,10 @@ size_t HUF_compress2 (void* dst, size_t dstSize, op += errorCode; /* Compress */ - //errorCode = HUF_compress_usingCTable(op, oend - op, src, srcSize, CTable); /* single segment */ - errorCode = HUF_compress_into4Segments(op, oend - op, src, srcSize, CTable); + if (singleStream) + errorCode = HUF_compress1X_usingCTable(op, oend - op, src, srcSize, CTable); /* single segment */ + else + errorCode = HUF_compress4X_usingCTable(op, oend - op, src, srcSize, CTable); if (HUF_isError(errorCode)) return errorCode; if (errorCode==0) return 0; op += errorCode; @@ -525,13 +571,29 @@ size_t HUF_compress2 (void* dst, size_t dstSize, return op-ostart; } + +size_t HUF_compress1X (void* dst, size_t dstSize, + const void* src, size_t srcSize, + unsigned maxSymbolValue, unsigned huffLog) +{ + return HUF_compress_internal(dst, dstSize, src, srcSize, maxSymbolValue, huffLog, 1); +} + +size_t HUF_compress2 (void* dst, size_t dstSize, + const void* src, size_t srcSize, + unsigned maxSymbolValue, unsigned huffLog) +{ + return HUF_compress_internal(dst, dstSize, src, srcSize, maxSymbolValue, huffLog, 0); +} + + size_t HUF_compress (void* dst, size_t maxDstSize, const void* src, size_t srcSize) { return HUF_compress2(dst, maxDstSize, src, (U32)srcSize, 255, HUF_DEFAULT_TABLELOG); } -/********************************************************* +/* ******************************************************* * Huff0 : Huffman block decompression *********************************************************/ typedef struct { BYTE byte; BYTE nbBits; } HUF_DEltX2; /* single-symbol decoding */ @@ -558,31 +620,24 @@ static size_t HUF_readStats(BYTE* huffWeight, size_t hwSize, U32* rankStats, //memset(huffWeight, 0, hwSize); /* is not necessary, even though some analyzer complain ... */ - if (iSize >= 128) /* special header */ - { - if (iSize >= (242)) /* RLE */ - { + if (iSize >= 128) { /* special header */ + if (iSize >= (242)) { /* RLE */ static int l[14] = { 1, 2, 3, 4, 7, 8, 15, 16, 31, 32, 63, 64, 127, 128 }; oSize = l[iSize-242]; memset(huffWeight, 1, hwSize); iSize = 0; } - else /* Incompressible */ - { + else { /* Incompressible */ oSize = iSize - 127; iSize = ((oSize+1)/2); if (iSize+1 > srcSize) return ERROR(srcSize_wrong); if (oSize >= hwSize) return ERROR(corruption_detected); ip += 1; - for (n=0; n> 4; huffWeight[n+1] = ip[n/2] & 15; - } - } - } - else /* header compressed with FSE (normal case) */ - { + } } } + else { /* header compressed with FSE (normal case) */ if (iSize+1 > srcSize) return ERROR(srcSize_wrong); oSize = FSE_decompress(huffWeight, hwSize-1, ip+1, iSize); /* max (hwSize-1) values decoded, as last one is implied */ if (FSE_isError(oSize)) return oSize; @@ -591,8 +646,7 @@ static size_t HUF_readStats(BYTE* huffWeight, size_t hwSize, U32* rankStats, /* collect weight stats */ memset(rankStats, 0, (HUF_ABSOLUTEMAX_TABLELOG + 1) * sizeof(U32)); weightTotal = 0; - for (n=0; n= HUF_ABSOLUTEMAX_TABLELOG) return ERROR(corruption_detected); rankStats[huffWeight[n]]++; weightTotal += (1 << huffWeight[n]) >> 1; @@ -601,7 +655,7 @@ static size_t HUF_readStats(BYTE* huffWeight, size_t hwSize, U32* rankStats, /* get last non-null symbol weight (implied, total must be 2^n) */ tableLog = BIT_highbit32(weightTotal) + 1; if (tableLog > HUF_ABSOLUTEMAX_TABLELOG) return ERROR(corruption_detected); - { + { /* determine last weight */ U32 total = 1 << tableLog; U32 rest = total - weightTotal; U32 verif = 1 << BIT_highbit32(rest); @@ -621,9 +675,9 @@ static size_t HUF_readStats(BYTE* huffWeight, size_t hwSize, U32* rankStats, } -/**************************/ -/* single-symbol decoding */ -/**************************/ +/*-***************************/ +/* single-symbol decoding */ +/*-***************************/ size_t HUF_readDTableX2 (U16* DTable, const void* src, size_t srcSize) { @@ -645,20 +699,18 @@ size_t HUF_readDTableX2 (U16* DTable, const void* src, size_t srcSize) /* check result */ if (tableLog > DTable[0]) return ERROR(tableLog_tooLarge); /* DTable is too small */ - DTable[0] = (U16)tableLog; /* maybe should separate sizeof DTable, as allocated, from used size of DTable, in case of DTable re-use */ + DTable[0] = (U16)tableLog; /* maybe should separate sizeof allocated DTable, from used size of DTable, in case of re-use */ /* Prepare ranks */ nextRankStart = 0; - for (n=1; n<=tableLog; n++) - { + for (n=1; n<=tableLog; n++) { U32 current = nextRankStart; nextRankStart += (rankVal[n] << (n-1)); rankVal[n] = current; } /* fill DTable */ - for (n=0; n> 1; U32 i; @@ -696,8 +748,7 @@ static inline size_t HUF_decodeStreamX2(BYTE* p, BIT_DStream_t* const bitDPtr, B BYTE* const pStart = p; /* up to 4 symbols at a time */ - while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p <= pEnd-4)) - { + while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p <= pEnd-4)) { HUF_DECODE_SYMBOLX2_2(p, bitDPtr); HUF_DECODE_SYMBOLX2_1(p, bitDPtr); HUF_DECODE_SYMBOLX2_2(p, bitDPtr); @@ -759,94 +810,91 @@ size_t HUF_decompress4X2_usingDTable( const void* cSrc, size_t cSrcSize, const U16* DTable) { + const BYTE* const istart = (const BYTE*) cSrc; + BYTE* const ostart = (BYTE*) dst; + BYTE* const oend = ostart + dstSize; + const void* const dtPtr = DTable; + const HUF_DEltX2* const dt = ((const HUF_DEltX2*)dtPtr) +1; + const U32 dtLog = DTable[0]; + size_t errorCode; + + /* Check */ if (cSrcSize < 10) return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */ - { - const BYTE* const istart = (const BYTE*) cSrc; - BYTE* const ostart = (BYTE*) dst; - BYTE* const oend = ostart + dstSize; - const void* const dtPtr = DTable; - const HUF_DEltX2* const dt = ((const HUF_DEltX2*)dtPtr) +1; - const U32 dtLog = DTable[0]; - size_t errorCode; + /* Init */ + BIT_DStream_t bitD1; + BIT_DStream_t bitD2; + BIT_DStream_t bitD3; + BIT_DStream_t bitD4; + const size_t length1 = MEM_readLE16(istart); + const size_t length2 = MEM_readLE16(istart+2); + const size_t length3 = MEM_readLE16(istart+4); + size_t length4; + const BYTE* const istart1 = istart + 6; /* jumpTable */ + const BYTE* const istart2 = istart1 + length1; + const BYTE* const istart3 = istart2 + length2; + const BYTE* const istart4 = istart3 + length3; + const size_t segmentSize = (dstSize+3) / 4; + BYTE* const opStart2 = ostart + segmentSize; + BYTE* const opStart3 = opStart2 + segmentSize; + BYTE* const opStart4 = opStart3 + segmentSize; + BYTE* op1 = ostart; + BYTE* op2 = opStart2; + BYTE* op3 = opStart3; + BYTE* op4 = opStart4; + U32 endSignal; - /* Init */ - BIT_DStream_t bitD1; - BIT_DStream_t bitD2; - BIT_DStream_t bitD3; - BIT_DStream_t bitD4; - const size_t length1 = MEM_readLE16(istart); - const size_t length2 = MEM_readLE16(istart+2); - const size_t length3 = MEM_readLE16(istart+4); - size_t length4; - const BYTE* const istart1 = istart + 6; /* jumpTable */ - const BYTE* const istart2 = istart1 + length1; - const BYTE* const istart3 = istart2 + length2; - const BYTE* const istart4 = istart3 + length3; - const size_t segmentSize = (dstSize+3) / 4; - BYTE* const opStart2 = ostart + segmentSize; - BYTE* const opStart3 = opStart2 + segmentSize; - BYTE* const opStart4 = opStart3 + segmentSize; - BYTE* op1 = ostart; - BYTE* op2 = opStart2; - BYTE* op3 = opStart3; - BYTE* op4 = opStart4; - U32 endSignal; + length4 = cSrcSize - (length1 + length2 + length3 + 6); + if (length4 > cSrcSize) return ERROR(corruption_detected); /* overflow */ + errorCode = BIT_initDStream(&bitD1, istart1, length1); + if (HUF_isError(errorCode)) return errorCode; + errorCode = BIT_initDStream(&bitD2, istart2, length2); + if (HUF_isError(errorCode)) return errorCode; + errorCode = BIT_initDStream(&bitD3, istart3, length3); + if (HUF_isError(errorCode)) return errorCode; + errorCode = BIT_initDStream(&bitD4, istart4, length4); + if (HUF_isError(errorCode)) return errorCode; - length4 = cSrcSize - (length1 + length2 + length3 + 6); - if (length4 > cSrcSize) return ERROR(corruption_detected); /* overflow */ - errorCode = BIT_initDStream(&bitD1, istart1, length1); - if (HUF_isError(errorCode)) return errorCode; - errorCode = BIT_initDStream(&bitD2, istart2, length2); - if (HUF_isError(errorCode)) return errorCode; - errorCode = BIT_initDStream(&bitD3, istart3, length3); - if (HUF_isError(errorCode)) return errorCode; - errorCode = BIT_initDStream(&bitD4, istart4, length4); - if (HUF_isError(errorCode)) return errorCode; - - /* 16-32 symbols per loop (4-8 symbols per stream) */ + /* 16-32 symbols per loop (4-8 symbols per stream) */ + endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4); + for ( ; (endSignal==BIT_DStream_unfinished) && (op4<(oend-7)) ; ) { + HUF_DECODE_SYMBOLX2_2(op1, &bitD1); + HUF_DECODE_SYMBOLX2_2(op2, &bitD2); + HUF_DECODE_SYMBOLX2_2(op3, &bitD3); + HUF_DECODE_SYMBOLX2_2(op4, &bitD4); + HUF_DECODE_SYMBOLX2_1(op1, &bitD1); + HUF_DECODE_SYMBOLX2_1(op2, &bitD2); + HUF_DECODE_SYMBOLX2_1(op3, &bitD3); + HUF_DECODE_SYMBOLX2_1(op4, &bitD4); + HUF_DECODE_SYMBOLX2_2(op1, &bitD1); + HUF_DECODE_SYMBOLX2_2(op2, &bitD2); + HUF_DECODE_SYMBOLX2_2(op3, &bitD3); + HUF_DECODE_SYMBOLX2_2(op4, &bitD4); + HUF_DECODE_SYMBOLX2_0(op1, &bitD1); + HUF_DECODE_SYMBOLX2_0(op2, &bitD2); + HUF_DECODE_SYMBOLX2_0(op3, &bitD3); + HUF_DECODE_SYMBOLX2_0(op4, &bitD4); endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4); - for ( ; (endSignal==BIT_DStream_unfinished) && (op4<(oend-7)) ; ) - { - HUF_DECODE_SYMBOLX2_2(op1, &bitD1); - HUF_DECODE_SYMBOLX2_2(op2, &bitD2); - HUF_DECODE_SYMBOLX2_2(op3, &bitD3); - HUF_DECODE_SYMBOLX2_2(op4, &bitD4); - HUF_DECODE_SYMBOLX2_1(op1, &bitD1); - HUF_DECODE_SYMBOLX2_1(op2, &bitD2); - HUF_DECODE_SYMBOLX2_1(op3, &bitD3); - HUF_DECODE_SYMBOLX2_1(op4, &bitD4); - HUF_DECODE_SYMBOLX2_2(op1, &bitD1); - HUF_DECODE_SYMBOLX2_2(op2, &bitD2); - HUF_DECODE_SYMBOLX2_2(op3, &bitD3); - HUF_DECODE_SYMBOLX2_2(op4, &bitD4); - HUF_DECODE_SYMBOLX2_0(op1, &bitD1); - HUF_DECODE_SYMBOLX2_0(op2, &bitD2); - HUF_DECODE_SYMBOLX2_0(op3, &bitD3); - HUF_DECODE_SYMBOLX2_0(op4, &bitD4); - - endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4); - } - - /* check corruption */ - if (op1 > opStart2) return ERROR(corruption_detected); - if (op2 > opStart3) return ERROR(corruption_detected); - if (op3 > opStart4) return ERROR(corruption_detected); - /* note : op4 supposed already verified within main loop */ - - /* finish bitStreams one by one */ - HUF_decodeStreamX2(op1, &bitD1, opStart2, dt, dtLog); - HUF_decodeStreamX2(op2, &bitD2, opStart3, dt, dtLog); - HUF_decodeStreamX2(op3, &bitD3, opStart4, dt, dtLog); - HUF_decodeStreamX2(op4, &bitD4, oend, dt, dtLog); - - /* check */ - endSignal = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4); - if (!endSignal) return ERROR(corruption_detected); - - /* decoded size */ - return dstSize; } + + /* check corruption */ + if (op1 > opStart2) return ERROR(corruption_detected); + if (op2 > opStart3) return ERROR(corruption_detected); + if (op3 > opStart4) return ERROR(corruption_detected); + /* note : op4 supposed already verified within main loop */ + + /* finish bitStreams one by one */ + HUF_decodeStreamX2(op1, &bitD1, opStart2, dt, dtLog); + HUF_decodeStreamX2(op2, &bitD2, opStart3, dt, dtLog); + HUF_decodeStreamX2(op3, &bitD3, opStart4, dt, dtLog); + HUF_decodeStreamX2(op4, &bitD4, oend, dt, dtLog); + + /* check */ + endSignal = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4); + if (!endSignal) return ERROR(corruption_detected); + + /* decoded size */ + return dstSize; } @@ -866,9 +914,9 @@ size_t HUF_decompress4X2 (void* dst, size_t dstSize, const void* cSrc, size_t cS } -/***************************/ +/* *************************/ /* double-symbols decoding */ -/***************************/ +/* *************************/ static void HUF_fillDTableX4Level2(HUF_DEltX4* DTable, U32 sizeLog, const U32 consumed, const U32* rankValOrigin, const int minWeight, @@ -883,8 +931,7 @@ static void HUF_fillDTableX4Level2(HUF_DEltX4* DTable, U32 sizeLog, const U32 co memcpy(rankVal, rankValOrigin, sizeof(rankVal)); /* fill skipped values */ - if (minWeight>1) - { + if (minWeight>1) { U32 i, skipSize = rankVal[minWeight]; MEM_writeLE16(&(DElt.sequence), baseSeq); DElt.nbBits = (BYTE)(consumed); @@ -894,8 +941,7 @@ static void HUF_fillDTableX4Level2(HUF_DEltX4* DTable, U32 sizeLog, const U32 co } /* fill DTable */ - for (s=0; s= minBits) /* enough room for a second symbol */ - { + if (targetLog-nbBits >= minBits) { /* enough room for a second symbol */ U32 sortedRank; int minWeight = nbBits + scaleLog; if (minWeight < 1) minWeight = 1; @@ -946,9 +990,7 @@ static void HUF_fillDTableX4(HUF_DEltX4* DTable, const U32 targetLog, rankValOrigin[nbBits], minWeight, sortedList+sortedRank, sortedListSize-sortedRank, nbBitsBaseline, symbol); - } - else - { + } else { U32 i; const U32 end = start + length; HUF_DEltX4 DElt; @@ -993,8 +1035,7 @@ size_t HUF_readDTableX4 (U32* DTable, const void* src, size_t srcSize) /* Get start index of each weight */ { U32 w, nextRankStart = 0; - for (w=1; w<=maxW; w++) - { + for (w=1; w<=maxW; w++) { U32 current = nextRankStart; nextRankStart += rankStats[w]; rankStart[w] = current; @@ -1006,8 +1047,7 @@ size_t HUF_readDTableX4 (U32* DTable, const void* src, size_t srcSize) /* sort symbols by weight */ { U32 s; - for (s=0; s> consumed; - } - } - } + } } } HUF_fillDTableX4(dt, memLog, sortedSymbol, sizeOfSort, @@ -1061,15 +1096,12 @@ static U32 HUF_decodeLastSymbolX4(void* op, BIT_DStream_t* DStream, const HUF_DE const size_t val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */ memcpy(op, dt+val, 1); if (dt[val].length==1) BIT_skipBits(DStream, dt[val].nbBits); - else - { - if (DStream->bitsConsumed < (sizeof(DStream->bitContainer)*8)) - { + else { + if (DStream->bitsConsumed < (sizeof(DStream->bitContainer)*8)) { BIT_skipBits(DStream, dt[val].nbBits); if (DStream->bitsConsumed > (sizeof(DStream->bitContainer)*8)) DStream->bitsConsumed = (sizeof(DStream->bitContainer)*8); /* ugly hack; works only because it's the last symbol. Note : can't easily extract nbBits from just this symbol */ - } - } + } } return 1; } @@ -1090,8 +1122,7 @@ static inline size_t HUF_decodeStreamX4(BYTE* p, BIT_DStream_t* bitDPtr, BYTE* c BYTE* const pStart = p; /* up to 8 symbols at a time */ - while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p < pEnd-7)) - { + while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p < pEnd-7)) { HUF_DECODE_SYMBOLX4_2(p, bitDPtr); HUF_DECODE_SYMBOLX4_1(p, bitDPtr); HUF_DECODE_SYMBOLX4_2(p, bitDPtr); @@ -1207,8 +1238,7 @@ size_t HUF_decompress4X4_usingDTable( /* 16-32 symbols per loop (4-8 symbols per stream) */ endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4); - for ( ; (endSignal==BIT_DStream_unfinished) && (op4<(oend-7)) ; ) - { + for ( ; (endSignal==BIT_DStream_unfinished) && (op4<(oend-7)) ; ) { HUF_DECODE_SYMBOLX4_2(op1, &bitD1); HUF_DECODE_SYMBOLX4_2(op2, &bitD2); HUF_DECODE_SYMBOLX4_2(op3, &bitD3); @@ -1266,9 +1296,9 @@ size_t HUF_decompress4X4 (void* dst, size_t dstSize, const void* cSrc, size_t cS } -/**********************************/ +/* ********************************/ /* quad-symbol decoding */ -/**********************************/ +/* ********************************/ typedef struct { BYTE nbBits; BYTE nbBytes; } HUF_DDescX6; typedef union { BYTE byte[4]; U32 sequence; } HUF_DSeqX6; @@ -1288,22 +1318,18 @@ static void HUF_fillDTableX6LevelN(HUF_DDescX6* DDescription, HUF_DSeqX6* DSeque memcpy(rankVal, rankValOrigin[consumed], sizeof(rankVal)); /* fill skipped values */ - if (minWeight>1) - { + if (minWeight>1) { U32 i; const U32 skipSize = rankVal[minWeight]; - for (i = 0; i < skipSize; i++) - { + for (i = 0; i < skipSize; i++) { DSequence[i] = baseSeq; DDescription[i] = DDesc; - } - } + } } /* fill DTable */ DDesc.nbBytes++; symbolStartPos = rankStart[minWeight]; - for (s=symbolStartPos; s= 1 (sorted) */ const int nbBits = nbBitsBaseline - weight; /* >= 1 (by construction) */ @@ -1313,25 +1339,20 @@ static void HUF_fillDTableX6LevelN(HUF_DDescX6* DDescription, HUF_DSeqX6* DSeque baseSeq.byte[level] = symbol; DDesc.nbBits = (BYTE)totalBits; - if ((level<3) && (sizeLog-totalBits >= minBits)) /* enough room for another symbol */ - { + if ((level<3) && (sizeLog-totalBits >= minBits)) { /* enough room for another symbol */ int nextMinWeight = totalBits + scaleLog; if (nextMinWeight < 1) nextMinWeight = 1; HUF_fillDTableX6LevelN(DDescription+start, DSequence+start, sizeLog-nbBits, rankValOrigin, totalBits, nextMinWeight, maxWeight, sortedSymbols, sortedListSize, rankStart, nbBitsBaseline, baseSeq, DDesc); /* recursive (max : level 3) */ - } - else - { + } else { U32 i; const U32 end = start + length; - for (i = start; i < end; i++) - { + for (i = start; i < end; i++) { DDescription[i] = DDesc; DSequence[i] = baseSeq; - } - } + } } rankVal[weight] += length; } } @@ -1365,8 +1386,7 @@ size_t HUF_readDTableX6 (U32* DTable, const void* src, size_t srcSize) /* Get start index of each weight */ { U32 w, nextRankStart = 0; - for (w=1; w<=maxW; w++) - { + for (w=1; w<=maxW; w++) { U32 current = nextRankStart; nextRankStart += rankStats[w]; rankStart[w] = current; @@ -1378,8 +1398,7 @@ size_t HUF_readDTableX6 (U32* DTable, const void* src, size_t srcSize) /* sort symbols by weight */ { U32 s; - for (s=0; s> consumed; - } - } - } + } } } /* fill tables */ { @@ -1445,15 +1459,13 @@ static U32 HUF_decodeLastSymbolsX6(void* op, const U32 maxL, BIT_DStream_t* DStr { const size_t val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */ U32 length = dd[val].nbBytes; - if (length <= maxL) - { + if (length <= maxL) { memcpy(op, ds+val, length); BIT_skipBits(DStream, dd[val].nbBits); return length; } memcpy(op, ds+val, maxL); - if (DStream->bitsConsumed < (sizeof(DStream->bitContainer)*8)) - { + if (DStream->bitsConsumed < (sizeof(DStream->bitContainer)*8)) { BIT_skipBits(DStream, dd[val].nbBits); if (DStream->bitsConsumed > (sizeof(DStream->bitContainer)*8)) DStream->bitsConsumed = (sizeof(DStream->bitContainer)*8); /* ugly hack; works only because it's the last symbol. Note : can't easily extract nbBits from just this symbol */ @@ -1482,8 +1494,7 @@ static inline size_t HUF_decodeStreamX6(BYTE* p, BIT_DStream_t* bitDPtr, BYTE* c BYTE* const pStart = p; /* up to 16 symbols at a time */ - while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p <= pEnd-16)) - { + while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p <= pEnd-16)) { HUF_DECODE_SYMBOLX6_2(p, bitDPtr); HUF_DECODE_SYMBOLX6_1(p, bitDPtr); HUF_DECODE_SYMBOLX6_2(p, bitDPtr); @@ -1548,97 +1559,95 @@ size_t HUF_decompress4X6_usingDTable( const void* cSrc, size_t cSrcSize, const U32* DTable) { + const BYTE* const istart = (const BYTE*) cSrc; + BYTE* const ostart = (BYTE*) dst; + BYTE* const oend = ostart + dstSize; + + const U32 dtLog = DTable[0]; + const void* const ddPtr = DTable+1; + const HUF_DDescX6* dd = (const HUF_DDescX6*)ddPtr; + const void* const dsPtr = DTable + 1 + ((size_t)1<<(dtLog-1)); + const HUF_DSeqX6* ds = (const HUF_DSeqX6*)dsPtr; + size_t errorCode; + + /* Check */ if (cSrcSize < 10) return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */ - { - const BYTE* const istart = (const BYTE*) cSrc; - BYTE* const ostart = (BYTE*) dst; - BYTE* const oend = ostart + dstSize; + /* Init */ + BIT_DStream_t bitD1; + BIT_DStream_t bitD2; + BIT_DStream_t bitD3; + BIT_DStream_t bitD4; + const size_t length1 = MEM_readLE16(istart); + const size_t length2 = MEM_readLE16(istart+2); + const size_t length3 = MEM_readLE16(istart+4); + size_t length4; + const BYTE* const istart1 = istart + 6; /* jumpTable */ + const BYTE* const istart2 = istart1 + length1; + const BYTE* const istart3 = istart2 + length2; + const BYTE* const istart4 = istart3 + length3; + const size_t segmentSize = (dstSize+3) / 4; + BYTE* const opStart2 = ostart + segmentSize; + BYTE* const opStart3 = opStart2 + segmentSize; + BYTE* const opStart4 = opStart3 + segmentSize; + BYTE* op1 = ostart; + BYTE* op2 = opStart2; + BYTE* op3 = opStart3; + BYTE* op4 = opStart4; + U32 endSignal; - const U32 dtLog = DTable[0]; - const void* const ddPtr = DTable+1; - const HUF_DDescX6* dd = (const HUF_DDescX6*)ddPtr; - const void* const dsPtr = DTable + 1 + ((size_t)1<<(dtLog-1)); - const HUF_DSeqX6* ds = (const HUF_DSeqX6*)dsPtr; - size_t errorCode; + length4 = cSrcSize - (length1 + length2 + length3 + 6); + if (length4 > cSrcSize) return ERROR(corruption_detected); /* overflow */ + errorCode = BIT_initDStream(&bitD1, istart1, length1); + if (HUF_isError(errorCode)) return errorCode; + errorCode = BIT_initDStream(&bitD2, istart2, length2); + if (HUF_isError(errorCode)) return errorCode; + errorCode = BIT_initDStream(&bitD3, istart3, length3); + if (HUF_isError(errorCode)) return errorCode; + errorCode = BIT_initDStream(&bitD4, istart4, length4); + if (HUF_isError(errorCode)) return errorCode; - /* Init */ - BIT_DStream_t bitD1; - BIT_DStream_t bitD2; - BIT_DStream_t bitD3; - BIT_DStream_t bitD4; - const size_t length1 = MEM_readLE16(istart); - const size_t length2 = MEM_readLE16(istart+2); - const size_t length3 = MEM_readLE16(istart+4); - size_t length4; - const BYTE* const istart1 = istart + 6; /* jumpTable */ - const BYTE* const istart2 = istart1 + length1; - const BYTE* const istart3 = istart2 + length2; - const BYTE* const istart4 = istart3 + length3; - const size_t segmentSize = (dstSize+3) / 4; - BYTE* const opStart2 = ostart + segmentSize; - BYTE* const opStart3 = opStart2 + segmentSize; - BYTE* const opStart4 = opStart3 + segmentSize; - BYTE* op1 = ostart; - BYTE* op2 = opStart2; - BYTE* op3 = opStart3; - BYTE* op4 = opStart4; - U32 endSignal; + /* 16-64 symbols per loop (4-16 symbols per stream) */ + endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4); + for ( ; (op3 <= opStart4) && (endSignal==BIT_DStream_unfinished) && (op4<=(oend-16)) ; ) { + HUF_DECODE_SYMBOLX6_2(op1, &bitD1); + HUF_DECODE_SYMBOLX6_2(op2, &bitD2); + HUF_DECODE_SYMBOLX6_2(op3, &bitD3); + HUF_DECODE_SYMBOLX6_2(op4, &bitD4); + HUF_DECODE_SYMBOLX6_1(op1, &bitD1); + HUF_DECODE_SYMBOLX6_1(op2, &bitD2); + HUF_DECODE_SYMBOLX6_1(op3, &bitD3); + HUF_DECODE_SYMBOLX6_1(op4, &bitD4); + HUF_DECODE_SYMBOLX6_2(op1, &bitD1); + HUF_DECODE_SYMBOLX6_2(op2, &bitD2); + HUF_DECODE_SYMBOLX6_2(op3, &bitD3); + HUF_DECODE_SYMBOLX6_2(op4, &bitD4); + HUF_DECODE_SYMBOLX6_0(op1, &bitD1); + HUF_DECODE_SYMBOLX6_0(op2, &bitD2); + HUF_DECODE_SYMBOLX6_0(op3, &bitD3); + HUF_DECODE_SYMBOLX6_0(op4, &bitD4); - length4 = cSrcSize - (length1 + length2 + length3 + 6); - if (length4 > cSrcSize) return ERROR(corruption_detected); /* overflow */ - errorCode = BIT_initDStream(&bitD1, istart1, length1); - if (HUF_isError(errorCode)) return errorCode; - errorCode = BIT_initDStream(&bitD2, istart2, length2); - if (HUF_isError(errorCode)) return errorCode; - errorCode = BIT_initDStream(&bitD3, istart3, length3); - if (HUF_isError(errorCode)) return errorCode; - errorCode = BIT_initDStream(&bitD4, istart4, length4); - if (HUF_isError(errorCode)) return errorCode; - - /* 16-64 symbols per loop (4-16 symbols per stream) */ endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4); - for ( ; (op3 <= opStart4) && (endSignal==BIT_DStream_unfinished) && (op4<=(oend-16)) ; ) - { - HUF_DECODE_SYMBOLX6_2(op1, &bitD1); - HUF_DECODE_SYMBOLX6_2(op2, &bitD2); - HUF_DECODE_SYMBOLX6_2(op3, &bitD3); - HUF_DECODE_SYMBOLX6_2(op4, &bitD4); - HUF_DECODE_SYMBOLX6_1(op1, &bitD1); - HUF_DECODE_SYMBOLX6_1(op2, &bitD2); - HUF_DECODE_SYMBOLX6_1(op3, &bitD3); - HUF_DECODE_SYMBOLX6_1(op4, &bitD4); - HUF_DECODE_SYMBOLX6_2(op1, &bitD1); - HUF_DECODE_SYMBOLX6_2(op2, &bitD2); - HUF_DECODE_SYMBOLX6_2(op3, &bitD3); - HUF_DECODE_SYMBOLX6_2(op4, &bitD4); - HUF_DECODE_SYMBOLX6_0(op1, &bitD1); - HUF_DECODE_SYMBOLX6_0(op2, &bitD2); - HUF_DECODE_SYMBOLX6_0(op3, &bitD3); - HUF_DECODE_SYMBOLX6_0(op4, &bitD4); - - endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4); - } - - /* check corruption */ - if (op1 > opStart2) return ERROR(corruption_detected); - if (op2 > opStart3) return ERROR(corruption_detected); - if (op3 > opStart4) return ERROR(corruption_detected); - /* note : op4 supposed already verified within main loop */ - - /* finish bitStreams one by one */ - HUF_decodeStreamX6(op1, &bitD1, opStart2, DTable, dtLog); - HUF_decodeStreamX6(op2, &bitD2, opStart3, DTable, dtLog); - HUF_decodeStreamX6(op3, &bitD3, opStart4, DTable, dtLog); - HUF_decodeStreamX6(op4, &bitD4, oend, DTable, dtLog); - - /* check */ - endSignal = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4); - if (!endSignal) return ERROR(corruption_detected); - - /* decoded size */ - return dstSize; } + + /* check corruption */ + if (op1 > opStart2) return ERROR(corruption_detected); + if (op2 > opStart3) return ERROR(corruption_detected); + if (op3 > opStart4) return ERROR(corruption_detected); + /* note : op4 supposed already verified within main loop */ + + /* finish bitStreams one by one */ + HUF_decodeStreamX6(op1, &bitD1, opStart2, DTable, dtLog); + HUF_decodeStreamX6(op2, &bitD2, opStart3, DTable, dtLog); + HUF_decodeStreamX6(op3, &bitD3, opStart4, DTable, dtLog); + HUF_decodeStreamX6(op4, &bitD4, oend, DTable, dtLog); + + /* check */ + endSignal = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4); + if (!endSignal) return ERROR(corruption_detected); + + /* decoded size */ + return dstSize; } @@ -1657,9 +1666,9 @@ size_t HUF_decompress4X6 (void* dst, size_t dstSize, const void* cSrc, size_t cS } -/**********************************/ +/* ********************************/ /* Generic decompression selector */ -/**********************************/ +/* ********************************/ typedef struct { U32 tableTime; U32 decode256Time; } algo_time_t; static const algo_time_t algoTime[16 /* Quantization */][3 /* single, double, quad */] = diff --git a/lib/huff0.h b/lib/huff0.h index 2ebd5cfbd..fe28d7bea 100644 --- a/lib/huff0.h +++ b/lib/huff0.h @@ -1,7 +1,7 @@ /* ****************************************************************** Huff0 : Huffman coder, part of New Generation Entropy library header file - Copyright (C) 2013-2015, Yann Collet. + Copyright (C) 2013-2016, Yann Collet. BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) @@ -30,7 +30,6 @@ You can contact the author at : - Source repository : https://github.com/Cyan4973/FiniteStateEntropy - - Public forum : https://groups.google.com/forum/#!forum/lz4c ****************************************************************** */ #ifndef HUFF0_H #define HUFF0_H @@ -66,8 +65,10 @@ HUF_compress(): HUF_decompress(): Decompress Huff0 data from buffer 'cSrc', of size 'cSrcSize', into already allocated destination buffer 'dst', of size 'dstSize'. - 'dstSize' must be the exact size of original (uncompressed) data. - Note : in contrast with FSE, HUF_decompress can regenerate RLE (cSrcSize==1) and uncompressed (cSrcSize==dstSize) data, because it knows size to regenerate. + @dstSize : must be the **exact** size of original (uncompressed) data. + Note : in contrast with FSE, HUF_decompress can regenerate + RLE (cSrcSize==1) and uncompressed (cSrcSize==dstSize) data, + because it knows size to regenerate. @return : size of regenerated data (== dstSize) or an error code, which can be tested using HUF_isError() */ diff --git a/lib/huff0_static.h b/lib/huff0_static.h index 5df0727d7..84033964a 100644 --- a/lib/huff0_static.h +++ b/lib/huff0_static.h @@ -1,7 +1,7 @@ /* ****************************************************************** - Huff0 : Huffman coder, part of New Generation Entropy library - header file for static linking (only) - Copyright (C) 2013-2015, Yann Collet + Huff0 : Huffman codec, part of New Generation Entropy library + header file, for static linking only + Copyright (C) 2013-2016, Yann Collet BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) @@ -30,7 +30,6 @@ You can contact the author at : - Source repository : https://github.com/Cyan4973/FiniteStateEntropy - - Public forum : https://groups.google.com/forum/#!forum/lz4c ****************************************************************** */ #ifndef HUFF0_STATIC_H #define HUFF0_STATIC_H @@ -47,15 +46,21 @@ extern "C" { /* **************************************** -* Static allocation macros +* Static allocation ******************************************/ /* Huff0 buffer bounds */ #define HUF_CTABLEBOUND 129 #define HUF_BLOCKBOUND(size) (size + (size>>8) + 8) /* only true if incompressible pre-filtered with fast heuristic */ #define HUF_COMPRESSBOUND(size) (HUF_CTABLEBOUND + HUF_BLOCKBOUND(size)) /* Macro version, useful for static allocation */ +/* static allocation of Huff0's Compression Table */ +#define HUF_CREATE_STATIC_CTABLE(name, maxSymbolValue) \ + U32 name##hb[maxSymbolValue+1]; \ + void* name##hv = &(name##hb); \ + HUF_CElt* name = (HUF_CElt*)(name##hv) /* no final ; */ + /* static allocation of Huff0's DTable */ -#define HUF_DTABLE_SIZE(maxTableLog) (1 + (1< /* size_t, ptrdiff_t */ +#include /* memcpy */ + + +/****************************************** +* Compiler-specific +******************************************/ +#if defined(__GNUC__) +# define MEM_STATIC static __attribute__((unused)) +#elif defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) +# define MEM_STATIC static inline +#elif defined(_MSC_VER) +# define MEM_STATIC static __inline +#else +# define MEM_STATIC static /* this version may generate warnings for unused static functions; disable the relevant warning */ +#endif + + +/**************************************************************** +* Basic Types +*****************************************************************/ +#if defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) +# include + typedef uint8_t BYTE; + typedef uint16_t U16; + typedef int16_t S16; + typedef uint32_t U32; + typedef int32_t S32; + typedef uint64_t U64; + typedef int64_t S64; +#else + typedef unsigned char BYTE; + typedef unsigned short U16; + typedef signed short S16; + typedef unsigned int U32; + typedef signed int S32; + typedef unsigned long long U64; + typedef signed long long S64; +#endif + + +/**************************************************************** +* Memory I/O +*****************************************************************/ +/* MEM_FORCE_MEMORY_ACCESS + * By default, access to unaligned memory is controlled by `memcpy()`, which is safe and portable. + * Unfortunately, on some target/compiler combinations, the generated assembly is sub-optimal. + * The below switch allow to select different access method for improved performance. + * Method 0 (default) : use `memcpy()`. Safe and portable. + * Method 1 : `__packed` statement. It depends on compiler extension (ie, not portable). + * This method is safe if your compiler supports it, and *generally* as fast or faster than `memcpy`. + * Method 2 : direct access. This method is portable but violate C standard. + * It can generate buggy code on targets generating assembly depending on alignment. + * But in some circumstances, it's the only known way to get the most performance (ie GCC + ARMv6) + * See http://fastcompression.blogspot.fr/2015/08/accessing-unaligned-memory.html for details. + * Prefer these methods in priority order (0 > 1 > 2) + */ +#ifndef MEM_FORCE_MEMORY_ACCESS /* can be defined externally, on command line for example */ +# if defined(__GNUC__) && ( defined(__ARM_ARCH_6__) || defined(__ARM_ARCH_6J__) || defined(__ARM_ARCH_6K__) || defined(__ARM_ARCH_6Z__) || defined(__ARM_ARCH_6ZK__) || defined(__ARM_ARCH_6T2__) ) +# define MEM_FORCE_MEMORY_ACCESS 2 +# elif defined(__INTEL_COMPILER) || \ + (defined(__GNUC__) && ( defined(__ARM_ARCH_7__) || defined(__ARM_ARCH_7A__) || defined(__ARM_ARCH_7R__) || defined(__ARM_ARCH_7M__) || defined(__ARM_ARCH_7S__) )) +# define MEM_FORCE_MEMORY_ACCESS 1 +# endif +#endif + +MEM_STATIC unsigned MEM_32bits(void) { return sizeof(void*)==4; } +MEM_STATIC unsigned MEM_64bits(void) { return sizeof(void*)==8; } + +MEM_STATIC unsigned MEM_isLittleEndian(void) +{ + const union { U32 u; BYTE c[4]; } one = { 1 }; /* don't use static : performance detrimental */ + return one.c[0]; +} + +#if defined(MEM_FORCE_MEMORY_ACCESS) && (MEM_FORCE_MEMORY_ACCESS==2) + +/* violates C standard on structure alignment. +Only use if no other choice to achieve best performance on target platform */ +MEM_STATIC U16 MEM_read16(const void* memPtr) { return *(const U16*) memPtr; } +MEM_STATIC U32 MEM_read32(const void* memPtr) { return *(const U32*) memPtr; } +MEM_STATIC U64 MEM_read64(const void* memPtr) { return *(const U64*) memPtr; } + +MEM_STATIC void MEM_write16(void* memPtr, U16 value) { *(U16*)memPtr = value; } +MEM_STATIC void MEM_write32(void* memPtr, U32 value) { *(U32*)memPtr = value; } +MEM_STATIC void MEM_write64(void* memPtr, U64 value) { *(U64*)memPtr = value; } + +#elif defined(MEM_FORCE_MEMORY_ACCESS) && (MEM_FORCE_MEMORY_ACCESS==1) + +/* __pack instructions are safer, but compiler specific, hence potentially problematic for some compilers */ +/* currently only defined for gcc and icc */ +typedef union { U16 u16; U32 u32; U64 u64; } __attribute__((packed)) unalign; + +MEM_STATIC U16 MEM_read16(const void* ptr) { return ((const unalign*)ptr)->u16; } +MEM_STATIC U32 MEM_read32(const void* ptr) { return ((const unalign*)ptr)->u32; } +MEM_STATIC U64 MEM_read64(const void* ptr) { return ((const unalign*)ptr)->u64; } + +MEM_STATIC void MEM_write16(void* memPtr, U16 value) { ((unalign*)memPtr)->u16 = value; } +MEM_STATIC void MEM_write32(void* memPtr, U32 value) { ((unalign*)memPtr)->u32 = value; } +MEM_STATIC void MEM_write64(void* memPtr, U64 value) { ((unalign*)memPtr)->u64 = value; } + +#else + +/* default method, safe and standard. + can sometimes prove slower */ + +MEM_STATIC U16 MEM_read16(const void* memPtr) +{ + U16 val; memcpy(&val, memPtr, sizeof(val)); return val; +} + +MEM_STATIC U32 MEM_read32(const void* memPtr) +{ + U32 val; memcpy(&val, memPtr, sizeof(val)); return val; +} + +MEM_STATIC U64 MEM_read64(const void* memPtr) +{ + U64 val; memcpy(&val, memPtr, sizeof(val)); return val; +} + +MEM_STATIC void MEM_write16(void* memPtr, U16 value) +{ + memcpy(memPtr, &value, sizeof(value)); +} + +MEM_STATIC void MEM_write32(void* memPtr, U32 value) +{ + memcpy(memPtr, &value, sizeof(value)); +} + +MEM_STATIC void MEM_write64(void* memPtr, U64 value) +{ + memcpy(memPtr, &value, sizeof(value)); +} + +#endif // MEM_FORCE_MEMORY_ACCESS + + +MEM_STATIC U16 MEM_readLE16(const void* memPtr) +{ + if (MEM_isLittleEndian()) + return MEM_read16(memPtr); + else + { + const BYTE* p = (const BYTE*)memPtr; + return (U16)(p[0] + (p[1]<<8)); + } +} + +MEM_STATIC void MEM_writeLE16(void* memPtr, U16 val) +{ + if (MEM_isLittleEndian()) + { + MEM_write16(memPtr, val); + } + else + { + BYTE* p = (BYTE*)memPtr; + p[0] = (BYTE)val; + p[1] = (BYTE)(val>>8); + } +} + +MEM_STATIC U32 MEM_readLE32(const void* memPtr) +{ + if (MEM_isLittleEndian()) + return MEM_read32(memPtr); + else + { + const BYTE* p = (const BYTE*)memPtr; + return (U32)((U32)p[0] + ((U32)p[1]<<8) + ((U32)p[2]<<16) + ((U32)p[3]<<24)); + } +} + +MEM_STATIC void MEM_writeLE32(void* memPtr, U32 val32) +{ + if (MEM_isLittleEndian()) + { + MEM_write32(memPtr, val32); + } + else + { + BYTE* p = (BYTE*)memPtr; + p[0] = (BYTE)val32; + p[1] = (BYTE)(val32>>8); + p[2] = (BYTE)(val32>>16); + p[3] = (BYTE)(val32>>24); + } +} + +MEM_STATIC U64 MEM_readLE64(const void* memPtr) +{ + if (MEM_isLittleEndian()) + return MEM_read64(memPtr); + else + { + const BYTE* p = (const BYTE*)memPtr; + return (U64)((U64)p[0] + ((U64)p[1]<<8) + ((U64)p[2]<<16) + ((U64)p[3]<<24) + + ((U64)p[4]<<32) + ((U64)p[5]<<40) + ((U64)p[6]<<48) + ((U64)p[7]<<56)); + } +} + +MEM_STATIC void MEM_writeLE64(void* memPtr, U64 val64) +{ + if (MEM_isLittleEndian()) + { + MEM_write64(memPtr, val64); + } + else + { + BYTE* p = (BYTE*)memPtr; + p[0] = (BYTE)val64; + p[1] = (BYTE)(val64>>8); + p[2] = (BYTE)(val64>>16); + p[3] = (BYTE)(val64>>24); + p[4] = (BYTE)(val64>>32); + p[5] = (BYTE)(val64>>40); + p[6] = (BYTE)(val64>>48); + p[7] = (BYTE)(val64>>56); + } +} + +MEM_STATIC size_t MEM_readLEST(const void* memPtr) +{ + if (MEM_32bits()) + return (size_t)MEM_readLE32(memPtr); + else + return (size_t)MEM_readLE64(memPtr); +} + +MEM_STATIC void MEM_writeLEST(void* memPtr, size_t val) +{ + if (MEM_32bits()) + MEM_writeLE32(memPtr, (U32)val); + else + MEM_writeLE64(memPtr, (U64)val); +} + +#if defined (__cplusplus) +} +#endif + +#endif /* MEM_H_MODULE */ + +/* ****************************************************************** + Error codes list + Copyright (C) 2016, Yann Collet + + BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are + met: + + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above + copyright notice, this list of conditions and the following disclaimer + in the documentation and/or other materials provided with the + distribution. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + You can contact the author at : + - Source repository : https://github.com/Cyan4973/zstd +****************************************************************** */ +#ifndef ERROR_PUBLIC_H_MODULE +#define ERROR_PUBLIC_H_MODULE + +#if defined (__cplusplus) +extern "C" { +#endif + + +/* **************************************** +* error list +******************************************/ +enum { + ZSTD_error_No_Error, + ZSTD_error_GENERIC, + ZSTD_error_prefix_unknown, + ZSTD_error_frameParameter_unsupported, + ZSTD_error_frameParameter_unsupportedBy32bitsImplementation, + ZSTD_error_init_missing, + ZSTD_error_memory_allocation, + ZSTD_error_stage_wrong, + ZSTD_error_dstSize_tooSmall, + ZSTD_error_srcSize_wrong, + ZSTD_error_corruption_detected, + ZSTD_error_tableLog_tooLarge, + ZSTD_error_maxSymbolValue_tooLarge, + ZSTD_error_maxSymbolValue_tooSmall, + ZSTD_error_maxCode +}; + +/* note : functions provide error codes in reverse negative order, + so compare with (size_t)(0-enum) */ + + +#if defined (__cplusplus) +} +#endif + +#endif /* ERROR_PUBLIC_H_MODULE */ + + + +/* + zstd - standard compression library + Header File for static linking only + Copyright (C) 2014-2015, Yann Collet. + + BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are + met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above + copyright notice, this list of conditions and the following disclaimer + in the documentation and/or other materials provided with the + distribution. + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + You can contact the author at : + - zstd source repository : https://github.com/Cyan4973/zstd + - ztsd public forum : https://groups.google.com/forum/#!forum/lz4c +*/ +#ifndef ZSTD_STATIC_H +#define ZSTD_STATIC_H + +/* The objects defined into this file shall be considered experimental. + * They are not considered stable, as their prototype may change in the future. + * You can use them for tests, provide feedback, or if you can endure risks of future changes. + */ + +#if defined (__cplusplus) +extern "C" { +#endif + +/* ************************************* +* Types +***************************************/ +#define ZSTD_WINDOWLOG_MAX 26 +#define ZSTD_WINDOWLOG_MIN 18 +#define ZSTD_WINDOWLOG_ABSOLUTEMIN 11 +#define ZSTD_CONTENTLOG_MAX (ZSTD_WINDOWLOG_MAX+1) +#define ZSTD_CONTENTLOG_MIN 4 +#define ZSTD_HASHLOG_MAX 28 +#define ZSTD_HASHLOG_MIN 4 +#define ZSTD_SEARCHLOG_MAX (ZSTD_CONTENTLOG_MAX-1) +#define ZSTD_SEARCHLOG_MIN 1 +#define ZSTD_SEARCHLENGTH_MAX 7 +#define ZSTD_SEARCHLENGTH_MIN 4 + +/** from faster to stronger */ +typedef enum { ZSTD_fast, ZSTD_greedy, ZSTD_lazy, ZSTD_lazy2, ZSTD_btlazy2 } ZSTD_strategy; + +typedef struct +{ + U64 srcSize; /* optional : tells how much bytes are present in the frame. Use 0 if not known. */ + U32 windowLog; /* largest match distance : larger == more compression, more memory needed during decompression */ + U32 contentLog; /* full search segment : larger == more compression, slower, more memory (useless for fast) */ + U32 hashLog; /* dispatch table : larger == more memory, faster */ + U32 searchLog; /* nb of searches : larger == more compression, slower */ + U32 searchLength; /* size of matches : larger == faster decompression, sometimes less compression */ + ZSTD_strategy strategy; +} ZSTD_parameters; + +typedef ZSTDv04_Dctx ZSTD_DCtx; + +/* ************************************* +* Advanced functions +***************************************/ +/** ZSTD_decompress_usingDict +* Same as ZSTD_decompressDCtx, using a Dictionary content as prefix +* Note : dict can be NULL, in which case, it's equivalent to ZSTD_decompressDCtx() */ +static size_t ZSTD_decompress_usingDict(ZSTD_DCtx* ctx, + void* dst, size_t maxDstSize, + const void* src, size_t srcSize, + const void* dict,size_t dictSize); + + +/* ************************************** +* Streaming functions (direct mode) +****************************************/ +static size_t ZSTD_resetDCtx(ZSTD_DCtx* dctx); +static size_t ZSTD_getFrameParams(ZSTD_parameters* params, const void* src, size_t srcSize); +static void ZSTD_decompress_insertDictionary(ZSTD_DCtx* ctx, const void* src, size_t srcSize); + +static size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx* dctx); +static size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t maxDstSize, const void* src, size_t srcSize); + +/** + Streaming decompression, bufferless mode + + A ZSTD_DCtx object is required to track streaming operations. + Use ZSTD_createDCtx() / ZSTD_freeDCtx() to manage it. + A ZSTD_DCtx object can be re-used multiple times. Use ZSTD_resetDCtx() to return to fresh status. + + First operation is to retrieve frame parameters, using ZSTD_getFrameParams(). + This function doesn't consume its input. It needs enough input data to properly decode the frame header. + Objective is to retrieve *params.windowlog, to know minimum amount of memory required during decoding. + Result : 0 when successful, it means the ZSTD_parameters structure has been filled. + >0 : means there is not enough data into src. Provides the expected size to successfully decode header. + errorCode, which can be tested using ZSTD_isError() (For example, if it's not a ZSTD header) + + Then, you can optionally insert a dictionary. + This operation must mimic the compressor behavior, otherwise decompression will fail or be corrupted. + + Then it's possible to start decompression. + Use ZSTD_nextSrcSizeToDecompress() and ZSTD_decompressContinue() alternatively. + ZSTD_nextSrcSizeToDecompress() tells how much bytes to provide as 'srcSize' to ZSTD_decompressContinue(). + ZSTD_decompressContinue() requires this exact amount of bytes, or it will fail. + ZSTD_decompressContinue() needs previous data blocks during decompression, up to (1 << windowlog). + They should preferably be located contiguously, prior to current block. Alternatively, a round buffer is also possible. + + @result of ZSTD_decompressContinue() is the number of bytes regenerated within 'dst'. + It can be zero, which is not an error; it just means ZSTD_decompressContinue() has decoded some header. + + A frame is fully decoded when ZSTD_nextSrcSizeToDecompress() returns zero. + Context can then be reset to start a new decompression. +*/ + + +#if defined (__cplusplus) +} +#endif + +/* ****************************************************************** + Error codes and messages + Copyright (C) 2013-2016, Yann Collet + + BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are + met: + + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above + copyright notice, this list of conditions and the following disclaimer + in the documentation and/or other materials provided with the + distribution. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + You can contact the author at : + - Source repository : https://github.com/Cyan4973/zstd +****************************************************************** */ +/* Note : this module is expected to remain private, do not expose it */ + +#ifndef ERROR_H_MODULE +#define ERROR_H_MODULE + +#if defined (__cplusplus) +extern "C" { +#endif + + +/* ***************************************** +* Includes +******************************************/ +#include /* size_t, ptrdiff_t */ + + +/* ***************************************** +* Compiler-specific +******************************************/ +#if defined(__GNUC__) +# define ERR_STATIC static __attribute__((unused)) +#elif defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) +# define ERR_STATIC static inline +#elif defined(_MSC_VER) +# define ERR_STATIC static __inline +#else +# define ERR_STATIC static /* this version may generate warnings for unused static functions; disable the relevant warning */ +#endif + + +/* ***************************************** +* Error Codes +******************************************/ +#define PREFIX(name) ZSTD_error_##name + +#ifdef ERROR +# undef ERROR /* reported already defined on VS 2015 by Rich Geldreich */ +#endif +#define ERROR(name) (size_t)-PREFIX(name) + +ERR_STATIC unsigned ERR_isError(size_t code) { return (code > ERROR(maxCode)); } + + +/* ***************************************** +* Error Strings +******************************************/ + +ERR_STATIC const char* ERR_getErrorName(size_t code) +{ + static const char* codeError = "Unspecified error code"; + switch( (size_t)(0-code) ) + { + case ZSTD_error_No_Error: return "No error detected"; + case ZSTD_error_GENERIC: return "Error (generic)"; + case ZSTD_error_prefix_unknown: return "Unknown frame descriptor"; + case ZSTD_error_frameParameter_unsupported: return "Unsupported frame parameter"; + case ZSTD_error_frameParameter_unsupportedBy32bitsImplementation: return "Frame parameter unsupported in 32-bits mode"; + case ZSTD_error_init_missing: return "Context should be init first"; + case ZSTD_error_memory_allocation: return "Allocation error : not enough memory"; + case ZSTD_error_dstSize_tooSmall: return "Destination buffer is too small"; + case ZSTD_error_srcSize_wrong: return "Src size incorrect"; + case ZSTD_error_corruption_detected: return "Corrupted block detected"; + case ZSTD_error_tableLog_tooLarge: return "tableLog requires too much memory"; + case ZSTD_error_maxSymbolValue_tooLarge: return "Unsupported max possible Symbol Value : too large"; + case ZSTD_error_maxSymbolValue_tooSmall: return "Specified maxSymbolValue is too small"; + case ZSTD_error_maxCode: + default: return codeError; + } +} + + +#if defined (__cplusplus) +} +#endif + +#endif /* ERROR_H_MODULE */ + + +#endif /* ZSTD_STATIC_H */ + + +/* + zstd_internal - common functions to include + Header File for include + Copyright (C) 2014-2015, Yann Collet. + + BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are + met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above + copyright notice, this list of conditions and the following disclaimer + in the documentation and/or other materials provided with the + distribution. + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + You can contact the author at : + - zstd source repository : https://github.com/Cyan4973/zstd + - ztsd public forum : https://groups.google.com/forum/#!forum/lz4c +*/ +#ifndef ZSTD_CCOMMON_H_MODULE +#define ZSTD_CCOMMON_H_MODULE + +#if defined (__cplusplus) +extern "C" { +#endif + +/* ************************************* +* Common macros +***************************************/ +#define MIN(a,b) ((a)<(b) ? (a) : (b)) +#define MAX(a,b) ((a)>(b) ? (a) : (b)) + + +/* ************************************* +* Common constants +***************************************/ +#define ZSTD_MAGICNUMBER 0xFD2FB524 /* v0.4 */ + +#define KB *(1 <<10) +#define MB *(1 <<20) +#define GB *(1U<<30) + +#define BLOCKSIZE (128 KB) /* define, for static allocation */ + +static const size_t ZSTD_blockHeaderSize = 3; +static const size_t ZSTD_frameHeaderSize_min = 5; +#define ZSTD_frameHeaderSize_max 5 /* define, for static allocation */ + +#define BIT7 128 +#define BIT6 64 +#define BIT5 32 +#define BIT4 16 +#define BIT1 2 +#define BIT0 1 + +#define IS_RAW BIT0 +#define IS_RLE BIT1 + +#define MINMATCH 4 +#define REPCODE_STARTVALUE 4 + +#define MLbits 7 +#define LLbits 6 +#define Offbits 5 +#define MaxML ((1< /* size_t, ptrdiff_t */ + + +/* ***************************************** +* FSE simple functions +******************************************/ +static size_t FSE_decompress(void* dst, size_t maxDstSize, + const void* cSrc, size_t cSrcSize); +/*! +FSE_decompress(): + Decompress FSE data from buffer 'cSrc', of size 'cSrcSize', + into already allocated destination buffer 'dst', of size 'maxDstSize'. + return : size of regenerated data (<= maxDstSize) + or an error code, which can be tested using FSE_isError() + + ** Important ** : FSE_decompress() doesn't decompress non-compressible nor RLE data !!! + Why ? : making this distinction requires a header. + Header management is intentionally delegated to the user layer, which can better manage special cases. +*/ + + +/* ***************************************** +* Tool functions +******************************************/ +/* Error Management */ +static unsigned FSE_isError(size_t code); /* tells if a return value is an error code */ + + + +/* ***************************************** +* FSE detailed API +******************************************/ +/*! +FSE_compress() does the following: +1. count symbol occurrence from source[] into table count[] +2. normalize counters so that sum(count[]) == Power_of_2 (2^tableLog) +3. save normalized counters to memory buffer using writeNCount() +4. build encoding table 'CTable' from normalized counters +5. encode the data stream using encoding table 'CTable' + +FSE_decompress() does the following: +1. read normalized counters with readNCount() +2. build decoding table 'DTable' from normalized counters +3. decode the data stream using decoding table 'DTable' + +The following API allows targeting specific sub-functions for advanced tasks. +For example, it's possible to compress several blocks using the same 'CTable', +or to save and provide normalized distribution using external method. +*/ + + +/* *** DECOMPRESSION *** */ + +/*! +FSE_readNCount(): + Read compactly saved 'normalizedCounter' from 'rBuffer'. + return : size read from 'rBuffer' + or an errorCode, which can be tested using FSE_isError() + maxSymbolValuePtr[0] and tableLogPtr[0] will also be updated with their respective values */ +static size_t FSE_readNCount (short* normalizedCounter, unsigned* maxSymbolValuePtr, unsigned* tableLogPtr, const void* rBuffer, size_t rBuffSize); + +/*! +Constructor and Destructor of type FSE_DTable + Note that its size depends on 'tableLog' */ +typedef unsigned FSE_DTable; /* don't allocate that. It's just a way to be more restrictive than void* */ + +/*! +FSE_buildDTable(): + Builds 'dt', which must be already allocated, using FSE_createDTable() + return : 0, + or an errorCode, which can be tested using FSE_isError() */ +static size_t FSE_buildDTable ( FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog); + +/*! +FSE_decompress_usingDTable(): + Decompress compressed source 'cSrc' of size 'cSrcSize' using 'dt' + into 'dst' which must be already allocated. + return : size of regenerated data (necessarily <= maxDstSize) + or an errorCode, which can be tested using FSE_isError() */ +static size_t FSE_decompress_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const FSE_DTable* dt); + +/*! +Tutorial : +---------- +(Note : these functions only decompress FSE-compressed blocks. + If block is uncompressed, use memcpy() instead + If block is a single repeated byte, use memset() instead ) + +The first step is to obtain the normalized frequencies of symbols. +This can be performed by FSE_readNCount() if it was saved using FSE_writeNCount(). +'normalizedCounter' must be already allocated, and have at least 'maxSymbolValuePtr[0]+1' cells of signed short. +In practice, that means it's necessary to know 'maxSymbolValue' beforehand, +or size the table to handle worst case situations (typically 256). +FSE_readNCount() will provide 'tableLog' and 'maxSymbolValue'. +The result of FSE_readNCount() is the number of bytes read from 'rBuffer'. +Note that 'rBufferSize' must be at least 4 bytes, even if useful information is less than that. +If there is an error, the function will return an error code, which can be tested using FSE_isError(). + +The next step is to build the decompression tables 'FSE_DTable' from 'normalizedCounter'. +This is performed by the function FSE_buildDTable(). +The space required by 'FSE_DTable' must be already allocated using FSE_createDTable(). +If there is an error, the function will return an error code, which can be tested using FSE_isError(). + +'FSE_DTable' can then be used to decompress 'cSrc', with FSE_decompress_usingDTable(). +'cSrcSize' must be strictly correct, otherwise decompression will fail. +FSE_decompress_usingDTable() result will tell how many bytes were regenerated (<=maxDstSize). +If there is an error, the function will return an error code, which can be tested using FSE_isError(). (ex: dst buffer too small) +*/ + + +#if defined (__cplusplus) +} +#endif + +#endif /* FSE_H */ + + +/* ****************************************************************** + bitstream + Part of NewGen Entropy library + header file (to include) + Copyright (C) 2013-2015, Yann Collet. + + BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are + met: + + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above + copyright notice, this list of conditions and the following disclaimer + in the documentation and/or other materials provided with the + distribution. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + You can contact the author at : + - Source repository : https://github.com/Cyan4973/FiniteStateEntropy + - Public forum : https://groups.google.com/forum/#!forum/lz4c +****************************************************************** */ +#ifndef BITSTREAM_H_MODULE +#define BITSTREAM_H_MODULE + +#if defined (__cplusplus) +extern "C" { +#endif + + +/* +* This API consists of small unitary functions, which highly benefit from being inlined. +* Since link-time-optimization is not available for all compilers, +* these functions are defined into a .h to be included. +*/ + +/********************************************** +* bitStream decompression API (read backward) +**********************************************/ +typedef struct +{ + size_t bitContainer; + unsigned bitsConsumed; + const char* ptr; + const char* start; +} BIT_DStream_t; + +typedef enum { BIT_DStream_unfinished = 0, + BIT_DStream_endOfBuffer = 1, + BIT_DStream_completed = 2, + BIT_DStream_overflow = 3 } BIT_DStream_status; /* result of BIT_reloadDStream() */ + /* 1,2,4,8 would be better for bitmap combinations, but slows down performance a bit ... :( */ + +MEM_STATIC size_t BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, size_t srcSize); +MEM_STATIC size_t BIT_readBits(BIT_DStream_t* bitD, unsigned nbBits); +MEM_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD); +MEM_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t* bitD); + + +/* +* Start by invoking BIT_initDStream(). +* A chunk of the bitStream is then stored into a local register. +* Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t). +* You can then retrieve bitFields stored into the local register, **in reverse order**. +* Local register is manually filled from memory by the BIT_reloadDStream() method. +* A reload guarantee a minimum of ((8*sizeof(size_t))-7) bits when its result is BIT_DStream_unfinished. +* Otherwise, it can be less than that, so proceed accordingly. +* Checking if DStream has reached its end can be performed with BIT_endOfDStream() +*/ + + +/****************************************** +* unsafe API +******************************************/ +MEM_STATIC size_t BIT_readBitsFast(BIT_DStream_t* bitD, unsigned nbBits); +/* faster, but works only if nbBits >= 1 */ + + + +/**************************************************************** +* Helper functions +****************************************************************/ +MEM_STATIC unsigned BIT_highbit32 (register U32 val) +{ +# if defined(_MSC_VER) /* Visual */ + unsigned long r=0; + _BitScanReverse ( &r, val ); + return (unsigned) r; +# elif defined(__GNUC__) && (__GNUC__ >= 3) /* Use GCC Intrinsic */ + return 31 - __builtin_clz (val); +# else /* Software version */ + static const unsigned DeBruijnClz[32] = { 0, 9, 1, 10, 13, 21, 2, 29, 11, 14, 16, 18, 22, 25, 3, 30, 8, 12, 20, 28, 15, 17, 24, 7, 19, 27, 23, 6, 26, 5, 4, 31 }; + U32 v = val; + unsigned r; + v |= v >> 1; + v |= v >> 2; + v |= v >> 4; + v |= v >> 8; + v |= v >> 16; + r = DeBruijnClz[ (U32) (v * 0x07C4ACDDU) >> 27]; + return r; +# endif +} + + +/********************************************************** +* bitStream decoding +**********************************************************/ + +/*!BIT_initDStream +* Initialize a BIT_DStream_t. +* @bitD : a pointer to an already allocated BIT_DStream_t structure +* @srcBuffer must point at the beginning of a bitStream +* @srcSize must be the exact size of the bitStream +* @result : size of stream (== srcSize) or an errorCode if a problem is detected +*/ +MEM_STATIC size_t BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, size_t srcSize) +{ + if (srcSize < 1) { memset(bitD, 0, sizeof(*bitD)); return ERROR(srcSize_wrong); } + + if (srcSize >= sizeof(size_t)) /* normal case */ + { + U32 contain32; + bitD->start = (const char*)srcBuffer; + bitD->ptr = (const char*)srcBuffer + srcSize - sizeof(size_t); + bitD->bitContainer = MEM_readLEST(bitD->ptr); + contain32 = ((const BYTE*)srcBuffer)[srcSize-1]; + if (contain32 == 0) return ERROR(GENERIC); /* endMark not present */ + bitD->bitsConsumed = 8 - BIT_highbit32(contain32); + } + else + { + U32 contain32; + bitD->start = (const char*)srcBuffer; + bitD->ptr = bitD->start; + bitD->bitContainer = *(const BYTE*)(bitD->start); + switch(srcSize) + { + case 7: bitD->bitContainer += (size_t)(((const BYTE*)(bitD->start))[6]) << (sizeof(size_t)*8 - 16); + case 6: bitD->bitContainer += (size_t)(((const BYTE*)(bitD->start))[5]) << (sizeof(size_t)*8 - 24); + case 5: bitD->bitContainer += (size_t)(((const BYTE*)(bitD->start))[4]) << (sizeof(size_t)*8 - 32); + case 4: bitD->bitContainer += (size_t)(((const BYTE*)(bitD->start))[3]) << 24; + case 3: bitD->bitContainer += (size_t)(((const BYTE*)(bitD->start))[2]) << 16; + case 2: bitD->bitContainer += (size_t)(((const BYTE*)(bitD->start))[1]) << 8; + default:; + } + contain32 = ((const BYTE*)srcBuffer)[srcSize-1]; + if (contain32 == 0) return ERROR(GENERIC); /* endMark not present */ + bitD->bitsConsumed = 8 - BIT_highbit32(contain32); + bitD->bitsConsumed += (U32)(sizeof(size_t) - srcSize)*8; + } + + return srcSize; +} + +/*!BIT_lookBits + * Provides next n bits from local register + * local register is not modified (bits are still present for next read/look) + * On 32-bits, maxNbBits==25 + * On 64-bits, maxNbBits==57 + * @return : value extracted + */ +MEM_STATIC size_t BIT_lookBits(BIT_DStream_t* bitD, U32 nbBits) +{ + const U32 bitMask = sizeof(bitD->bitContainer)*8 - 1; + return ((bitD->bitContainer << (bitD->bitsConsumed & bitMask)) >> 1) >> ((bitMask-nbBits) & bitMask); +} + +/*! BIT_lookBitsFast : +* unsafe version; only works only if nbBits >= 1 */ +MEM_STATIC size_t BIT_lookBitsFast(BIT_DStream_t* bitD, U32 nbBits) +{ + const U32 bitMask = sizeof(bitD->bitContainer)*8 - 1; + return (bitD->bitContainer << (bitD->bitsConsumed & bitMask)) >> (((bitMask+1)-nbBits) & bitMask); +} + +MEM_STATIC void BIT_skipBits(BIT_DStream_t* bitD, U32 nbBits) +{ + bitD->bitsConsumed += nbBits; +} + +/*!BIT_readBits + * Read next n bits from local register. + * pay attention to not read more than nbBits contained into local register. + * @return : extracted value. + */ +MEM_STATIC size_t BIT_readBits(BIT_DStream_t* bitD, U32 nbBits) +{ + size_t value = BIT_lookBits(bitD, nbBits); + BIT_skipBits(bitD, nbBits); + return value; +} + +/*!BIT_readBitsFast : +* unsafe version; only works only if nbBits >= 1 */ +MEM_STATIC size_t BIT_readBitsFast(BIT_DStream_t* bitD, U32 nbBits) +{ + size_t value = BIT_lookBitsFast(bitD, nbBits); + BIT_skipBits(bitD, nbBits); + return value; +} + +MEM_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD) +{ + if (bitD->bitsConsumed > (sizeof(bitD->bitContainer)*8)) /* should never happen */ + return BIT_DStream_overflow; + + if (bitD->ptr >= bitD->start + sizeof(bitD->bitContainer)) + { + bitD->ptr -= bitD->bitsConsumed >> 3; + bitD->bitsConsumed &= 7; + bitD->bitContainer = MEM_readLEST(bitD->ptr); + return BIT_DStream_unfinished; + } + if (bitD->ptr == bitD->start) + { + if (bitD->bitsConsumed < sizeof(bitD->bitContainer)*8) return BIT_DStream_endOfBuffer; + return BIT_DStream_completed; + } + { + U32 nbBytes = bitD->bitsConsumed >> 3; + BIT_DStream_status result = BIT_DStream_unfinished; + if (bitD->ptr - nbBytes < bitD->start) + { + nbBytes = (U32)(bitD->ptr - bitD->start); /* ptr > start */ + result = BIT_DStream_endOfBuffer; + } + bitD->ptr -= nbBytes; + bitD->bitsConsumed -= nbBytes*8; + bitD->bitContainer = MEM_readLEST(bitD->ptr); /* reminder : srcSize > sizeof(bitD) */ + return result; + } +} + +/*! BIT_endOfDStream +* @return Tells if DStream has reached its exact end +*/ +MEM_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t* DStream) +{ + return ((DStream->ptr == DStream->start) && (DStream->bitsConsumed == sizeof(DStream->bitContainer)*8)); +} + +#if defined (__cplusplus) +} +#endif + +#endif /* BITSTREAM_H_MODULE */ + + + +/* ****************************************************************** + FSE : Finite State Entropy coder + header file for static linking (only) + Copyright (C) 2013-2015, Yann Collet + + BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are + met: + + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above + copyright notice, this list of conditions and the following disclaimer + in the documentation and/or other materials provided with the + distribution. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + You can contact the author at : + - Source repository : https://github.com/Cyan4973/FiniteStateEntropy + - Public forum : https://groups.google.com/forum/#!forum/lz4c +****************************************************************** */ +#ifndef FSE_STATIC_H +#define FSE_STATIC_H + +#if defined (__cplusplus) +extern "C" { +#endif + + +/* ***************************************** +* Static allocation +*******************************************/ +/* FSE buffer bounds */ +#define FSE_NCOUNTBOUND 512 +#define FSE_BLOCKBOUND(size) (size + (size>>7)) +#define FSE_COMPRESSBOUND(size) (FSE_NCOUNTBOUND + FSE_BLOCKBOUND(size)) /* Macro version, useful for static allocation */ + +/* It is possible to statically allocate FSE CTable/DTable as a table of unsigned using below macros */ +#define FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue) (1 + (1<<(maxTableLog-1)) + ((maxSymbolValue+1)*2)) +#define FSE_DTABLE_SIZE_U32(maxTableLog) (1 + (1<= BIT_DStream_completed + +When it's done, verify decompression is fully completed, by checking both DStream and the relevant states. +Checking if DStream has reached its end is performed by : + BIT_endOfDStream(&DStream); +Check also the states. There might be some symbols left there, if some high probability ones (>50%) are possible. + FSE_endOfDState(&DState); +*/ + + +/* ***************************************** +* FSE unsafe API +*******************************************/ +static unsigned char FSE_decodeSymbolFast(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD); +/* faster, but works only if nbBits is always >= 1 (otherwise, result will be corrupted) */ + + +/* ***************************************** +* Implementation of inlined functions +*******************************************/ +/* decompression */ + +typedef struct { + U16 tableLog; + U16 fastMode; +} FSE_DTableHeader; /* sizeof U32 */ + +typedef struct +{ + unsigned short newState; + unsigned char symbol; + unsigned char nbBits; +} FSE_decode_t; /* size == U32 */ + +MEM_STATIC void FSE_initDState(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD, const FSE_DTable* dt) +{ + const void* ptr = dt; + const FSE_DTableHeader* const DTableH = (const FSE_DTableHeader*)ptr; + DStatePtr->state = BIT_readBits(bitD, DTableH->tableLog); + BIT_reloadDStream(bitD); + DStatePtr->table = dt + 1; +} + +MEM_STATIC BYTE FSE_decodeSymbol(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD) +{ + const FSE_decode_t DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state]; + const U32 nbBits = DInfo.nbBits; + BYTE symbol = DInfo.symbol; + size_t lowBits = BIT_readBits(bitD, nbBits); + + DStatePtr->state = DInfo.newState + lowBits; + return symbol; +} + +MEM_STATIC BYTE FSE_decodeSymbolFast(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD) +{ + const FSE_decode_t DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state]; + const U32 nbBits = DInfo.nbBits; + BYTE symbol = DInfo.symbol; + size_t lowBits = BIT_readBitsFast(bitD, nbBits); + + DStatePtr->state = DInfo.newState + lowBits; + return symbol; +} + +MEM_STATIC unsigned FSE_endOfDState(const FSE_DState_t* DStatePtr) +{ + return DStatePtr->state == 0; +} + + +#if defined (__cplusplus) +} +#endif + +#endif /* FSE_STATIC_H */ + +/* ****************************************************************** + FSE : Finite State Entropy coder + Copyright (C) 2013-2015, Yann Collet. + + BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are + met: + + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above + copyright notice, this list of conditions and the following disclaimer + in the documentation and/or other materials provided with the + distribution. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + You can contact the author at : + - FSE source repository : https://github.com/Cyan4973/FiniteStateEntropy + - Public forum : https://groups.google.com/forum/#!forum/lz4c +****************************************************************** */ + +#ifndef FSE_COMMONDEFS_ONLY + +/* ************************************************************** +* Tuning parameters +****************************************************************/ +/*!MEMORY_USAGE : +* Memory usage formula : N->2^N Bytes (examples : 10 -> 1KB; 12 -> 4KB ; 16 -> 64KB; 20 -> 1MB; etc.) +* Increasing memory usage improves compression ratio +* Reduced memory usage can improve speed, due to cache effect +* Recommended max value is 14, for 16KB, which nicely fits into Intel x86 L1 cache */ +#define FSE_MAX_MEMORY_USAGE 14 +#define FSE_DEFAULT_MEMORY_USAGE 13 + +/*!FSE_MAX_SYMBOL_VALUE : +* Maximum symbol value authorized. +* Required for proper stack allocation */ +#define FSE_MAX_SYMBOL_VALUE 255 + + +/* ************************************************************** +* template functions type & suffix +****************************************************************/ +#define FSE_FUNCTION_TYPE BYTE +#define FSE_FUNCTION_EXTENSION +#define FSE_DECODE_TYPE FSE_decode_t + + +#endif /* !FSE_COMMONDEFS_ONLY */ + +/* ************************************************************** +* Compiler specifics +****************************************************************/ +#ifdef _MSC_VER /* Visual Studio */ +# define FORCE_INLINE static __forceinline +# include /* For Visual 2005 */ +# pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */ +# pragma warning(disable : 4214) /* disable: C4214: non-int bitfields */ +#else +# ifdef __GNUC__ +# define GCC_VERSION (__GNUC__ * 100 + __GNUC_MINOR__) +# define FORCE_INLINE static inline __attribute__((always_inline)) +# else +# define FORCE_INLINE static inline +# endif +#endif + + +/* ************************************************************** +* Includes +****************************************************************/ +#include /* malloc, free, qsort */ +#include /* memcpy, memset */ +#include /* printf (debug) */ + + +/* *************************************************************** +* Constants +*****************************************************************/ +#define FSE_MAX_TABLELOG (FSE_MAX_MEMORY_USAGE-2) +#define FSE_MAX_TABLESIZE (1U< FSE_TABLELOG_ABSOLUTE_MAX +#error "FSE_MAX_TABLELOG > FSE_TABLELOG_ABSOLUTE_MAX is not supported" +#endif + + +/* ************************************************************** +* Error Management +****************************************************************/ +#define FSE_STATIC_ASSERT(c) { enum { FSE_static_assert = 1/(int)(!!(c)) }; } /* use only *after* variable declarations */ + + +/* ************************************************************** +* Complex types +****************************************************************/ +typedef U32 DTable_max_t[FSE_DTABLE_SIZE_U32(FSE_MAX_TABLELOG)]; + + +/* ************************************************************** +* Templates +****************************************************************/ +/* + designed to be included + for type-specific functions (template emulation in C) + Objective is to write these functions only once, for improved maintenance +*/ + +/* safety checks */ +#ifndef FSE_FUNCTION_EXTENSION +# error "FSE_FUNCTION_EXTENSION must be defined" +#endif +#ifndef FSE_FUNCTION_TYPE +# error "FSE_FUNCTION_TYPE must be defined" +#endif + +/* Function names */ +#define FSE_CAT(X,Y) X##Y +#define FSE_FUNCTION_NAME(X,Y) FSE_CAT(X,Y) +#define FSE_TYPE_NAME(X,Y) FSE_CAT(X,Y) + +static U32 FSE_tableStep(U32 tableSize) { return (tableSize>>1) + (tableSize>>3) + 3; } + + +static size_t FSE_buildDTable(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog) +{ + FSE_DTableHeader DTableH; + void* const tdPtr = dt+1; /* because dt is unsigned, 32-bits aligned on 32-bits */ + FSE_DECODE_TYPE* const tableDecode = (FSE_DECODE_TYPE*) (tdPtr); + const U32 tableSize = 1 << tableLog; + const U32 tableMask = tableSize-1; + const U32 step = FSE_tableStep(tableSize); + U16 symbolNext[FSE_MAX_SYMBOL_VALUE+1]; + U32 position = 0; + U32 highThreshold = tableSize-1; + const S16 largeLimit= (S16)(1 << (tableLog-1)); + U32 noLarge = 1; + U32 s; + + /* Sanity Checks */ + if (maxSymbolValue > FSE_MAX_SYMBOL_VALUE) return ERROR(maxSymbolValue_tooLarge); + if (tableLog > FSE_MAX_TABLELOG) return ERROR(tableLog_tooLarge); + + /* Init, lay down lowprob symbols */ + DTableH.tableLog = (U16)tableLog; + for (s=0; s<=maxSymbolValue; s++) + { + if (normalizedCounter[s]==-1) + { + tableDecode[highThreshold--].symbol = (FSE_FUNCTION_TYPE)s; + symbolNext[s] = 1; + } + else + { + if (normalizedCounter[s] >= largeLimit) noLarge=0; + symbolNext[s] = normalizedCounter[s]; + } + } + + /* Spread symbols */ + for (s=0; s<=maxSymbolValue; s++) + { + int i; + for (i=0; i highThreshold) position = (position + step) & tableMask; /* lowprob area */ + } + } + + if (position!=0) return ERROR(GENERIC); /* position must reach all cells once, otherwise normalizedCounter is incorrect */ + + /* Build Decoding table */ + { + U32 i; + for (i=0; i FSE_TABLELOG_ABSOLUTE_MAX) return ERROR(tableLog_tooLarge); + bitStream >>= 4; + bitCount = 4; + *tableLogPtr = nbBits; + remaining = (1<1) && (charnum<=*maxSVPtr)) + { + if (previous0) + { + unsigned n0 = charnum; + while ((bitStream & 0xFFFF) == 0xFFFF) + { + n0+=24; + if (ip < iend-5) + { + ip+=2; + bitStream = MEM_readLE32(ip) >> bitCount; + } + else + { + bitStream >>= 16; + bitCount+=16; + } + } + while ((bitStream & 3) == 3) + { + n0+=3; + bitStream>>=2; + bitCount+=2; + } + n0 += bitStream & 3; + bitCount += 2; + if (n0 > *maxSVPtr) return ERROR(maxSymbolValue_tooSmall); + while (charnum < n0) normalizedCounter[charnum++] = 0; + if ((ip <= iend-7) || (ip + (bitCount>>3) <= iend-4)) + { + ip += bitCount>>3; + bitCount &= 7; + bitStream = MEM_readLE32(ip) >> bitCount; + } + else + bitStream >>= 2; + } + { + const short max = (short)((2*threshold-1)-remaining); + short count; + + if ((bitStream & (threshold-1)) < (U32)max) + { + count = (short)(bitStream & (threshold-1)); + bitCount += nbBits-1; + } + else + { + count = (short)(bitStream & (2*threshold-1)); + if (count >= threshold) count -= max; + bitCount += nbBits; + } + + count--; /* extra accuracy */ + remaining -= FSE_abs(count); + normalizedCounter[charnum++] = count; + previous0 = !count; + while (remaining < threshold) + { + nbBits--; + threshold >>= 1; + } + + { + if ((ip <= iend-7) || (ip + (bitCount>>3) <= iend-4)) + { + ip += bitCount>>3; + bitCount &= 7; + } + else + { + bitCount -= (int)(8 * (iend - 4 - ip)); + ip = iend - 4; + } + bitStream = MEM_readLE32(ip) >> (bitCount & 31); + } + } + } + if (remaining != 1) return ERROR(GENERIC); + *maxSVPtr = charnum-1; + + ip += (bitCount+7)>>3; + if ((size_t)(ip-istart) > hbSize) return ERROR(srcSize_wrong); + return ip-istart; +} + + +/********************************************************* +* Decompression (Byte symbols) +*********************************************************/ +static size_t FSE_buildDTable_rle (FSE_DTable* dt, BYTE symbolValue) +{ + void* ptr = dt; + FSE_DTableHeader* const DTableH = (FSE_DTableHeader*)ptr; + void* dPtr = dt + 1; + FSE_decode_t* const cell = (FSE_decode_t*)dPtr; + + DTableH->tableLog = 0; + DTableH->fastMode = 0; + + cell->newState = 0; + cell->symbol = symbolValue; + cell->nbBits = 0; + + return 0; +} + + +static size_t FSE_buildDTable_raw (FSE_DTable* dt, unsigned nbBits) +{ + void* ptr = dt; + FSE_DTableHeader* const DTableH = (FSE_DTableHeader*)ptr; + void* dPtr = dt + 1; + FSE_decode_t* const dinfo = (FSE_decode_t*)dPtr; + const unsigned tableSize = 1 << nbBits; + const unsigned tableMask = tableSize - 1; + const unsigned maxSymbolValue = tableMask; + unsigned s; + + /* Sanity checks */ + if (nbBits < 1) return ERROR(GENERIC); /* min size */ + + /* Build Decoding Table */ + DTableH->tableLog = (U16)nbBits; + DTableH->fastMode = 1; + for (s=0; s<=maxSymbolValue; s++) + { + dinfo[s].newState = 0; + dinfo[s].symbol = (BYTE)s; + dinfo[s].nbBits = (BYTE)nbBits; + } + + return 0; +} + +FORCE_INLINE size_t FSE_decompress_usingDTable_generic( + void* dst, size_t maxDstSize, + const void* cSrc, size_t cSrcSize, + const FSE_DTable* dt, const unsigned fast) +{ + BYTE* const ostart = (BYTE*) dst; + BYTE* op = ostart; + BYTE* const omax = op + maxDstSize; + BYTE* const olimit = omax-3; + + BIT_DStream_t bitD; + FSE_DState_t state1; + FSE_DState_t state2; + size_t errorCode; + + /* Init */ + errorCode = BIT_initDStream(&bitD, cSrc, cSrcSize); /* replaced last arg by maxCompressed Size */ + if (FSE_isError(errorCode)) return errorCode; + + FSE_initDState(&state1, &bitD, dt); + FSE_initDState(&state2, &bitD, dt); + +#define FSE_GETSYMBOL(statePtr) fast ? FSE_decodeSymbolFast(statePtr, &bitD) : FSE_decodeSymbol(statePtr, &bitD) + + /* 4 symbols per loop */ + for ( ; (BIT_reloadDStream(&bitD)==BIT_DStream_unfinished) && (op sizeof(bitD.bitContainer)*8) /* This test must be static */ + BIT_reloadDStream(&bitD); + + op[1] = FSE_GETSYMBOL(&state2); + + if (FSE_MAX_TABLELOG*4+7 > sizeof(bitD.bitContainer)*8) /* This test must be static */ + { if (BIT_reloadDStream(&bitD) > BIT_DStream_unfinished) { op+=2; break; } } + + op[2] = FSE_GETSYMBOL(&state1); + + if (FSE_MAX_TABLELOG*2+7 > sizeof(bitD.bitContainer)*8) /* This test must be static */ + BIT_reloadDStream(&bitD); + + op[3] = FSE_GETSYMBOL(&state2); + } + + /* tail */ + /* note : BIT_reloadDStream(&bitD) >= FSE_DStream_partiallyFilled; Ends at exactly BIT_DStream_completed */ + while (1) + { + if ( (BIT_reloadDStream(&bitD)>BIT_DStream_completed) || (op==omax) || (BIT_endOfDStream(&bitD) && (fast || FSE_endOfDState(&state1))) ) + break; + + *op++ = FSE_GETSYMBOL(&state1); + + if ( (BIT_reloadDStream(&bitD)>BIT_DStream_completed) || (op==omax) || (BIT_endOfDStream(&bitD) && (fast || FSE_endOfDState(&state2))) ) + break; + + *op++ = FSE_GETSYMBOL(&state2); + } + + /* end ? */ + if (BIT_endOfDStream(&bitD) && FSE_endOfDState(&state1) && FSE_endOfDState(&state2)) + return op-ostart; + + if (op==omax) return ERROR(dstSize_tooSmall); /* dst buffer is full, but cSrc unfinished */ + + return ERROR(corruption_detected); +} + + +static size_t FSE_decompress_usingDTable(void* dst, size_t originalSize, + const void* cSrc, size_t cSrcSize, + const FSE_DTable* dt) +{ + const void* ptr = dt; + const FSE_DTableHeader* DTableH = (const FSE_DTableHeader*)ptr; + const U32 fastMode = DTableH->fastMode; + + /* select fast mode (static) */ + if (fastMode) return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 1); + return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 0); +} + + +static size_t FSE_decompress(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize) +{ + const BYTE* const istart = (const BYTE*)cSrc; + const BYTE* ip = istart; + short counting[FSE_MAX_SYMBOL_VALUE+1]; + DTable_max_t dt; /* Static analyzer seems unable to understand this table will be properly initialized later */ + unsigned tableLog; + unsigned maxSymbolValue = FSE_MAX_SYMBOL_VALUE; + size_t errorCode; + + if (cSrcSize<2) return ERROR(srcSize_wrong); /* too small input size */ + + /* normal FSE decoding mode */ + errorCode = FSE_readNCount (counting, &maxSymbolValue, &tableLog, istart, cSrcSize); + if (FSE_isError(errorCode)) return errorCode; + if (errorCode >= cSrcSize) return ERROR(srcSize_wrong); /* too small input size */ + ip += errorCode; + cSrcSize -= errorCode; + + errorCode = FSE_buildDTable (dt, counting, maxSymbolValue, tableLog); + if (FSE_isError(errorCode)) return errorCode; + + /* always return, even if it is an error code */ + return FSE_decompress_usingDTable (dst, maxDstSize, ip, cSrcSize, dt); +} + + + +#endif /* FSE_COMMONDEFS_ONLY */ + + +/* ****************************************************************** + Huff0 : Huffman coder, part of New Generation Entropy library + header file + Copyright (C) 2013-2015, Yann Collet. + + BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are + met: + + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above + copyright notice, this list of conditions and the following disclaimer + in the documentation and/or other materials provided with the + distribution. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + You can contact the author at : + - Source repository : https://github.com/Cyan4973/FiniteStateEntropy + - Public forum : https://groups.google.com/forum/#!forum/lz4c +****************************************************************** */ +#ifndef HUFF0_H +#define HUFF0_H + +#if defined (__cplusplus) +extern "C" { +#endif + + +/* **************************************** +* Dependency +******************************************/ +#include /* size_t */ + + +/* **************************************** +* Huff0 simple functions +******************************************/ +static size_t HUF_decompress(void* dst, size_t dstSize, + const void* cSrc, size_t cSrcSize); +/*! +HUF_decompress(): + Decompress Huff0 data from buffer 'cSrc', of size 'cSrcSize', + into already allocated destination buffer 'dst', of size 'dstSize'. + 'dstSize' must be the exact size of original (uncompressed) data. + Note : in contrast with FSE, HUF_decompress can regenerate RLE (cSrcSize==1) and uncompressed (cSrcSize==dstSize) data, because it knows size to regenerate. + @return : size of regenerated data (== dstSize) + or an error code, which can be tested using HUF_isError() +*/ + + +/* **************************************** +* Tool functions +******************************************/ +/* Error Management */ +static unsigned HUF_isError(size_t code); /* tells if a return value is an error code */ + + +#if defined (__cplusplus) +} +#endif + +#endif /* HUFF0_H */ + + +/* ****************************************************************** + Huff0 : Huffman coder, part of New Generation Entropy library + header file for static linking (only) + Copyright (C) 2013-2015, Yann Collet + + BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions are + met: + + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above + copyright notice, this list of conditions and the following disclaimer + in the documentation and/or other materials provided with the + distribution. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + You can contact the author at : + - Source repository : https://github.com/Cyan4973/FiniteStateEntropy + - Public forum : https://groups.google.com/forum/#!forum/lz4c +****************************************************************** */ +#ifndef HUFF0_STATIC_H +#define HUFF0_STATIC_H + +#if defined (__cplusplus) +extern "C" { +#endif + + +/* **************************************** +* Dependency +******************************************/ +#include "huff0.h" + + +/* **************************************** +* Static allocation macros +******************************************/ +/* static allocation of Huff0's DTable */ +#define HUF_DTABLE_SIZE(maxTableLog) (1 + (1<= 199901L) /* C99 */) +/* inline is defined */ +#elif defined(_MSC_VER) +# define inline __inline +#else +# define inline /* disable inline */ +#endif + + +#ifdef _MSC_VER /* Visual Studio */ +# define FORCE_INLINE static __forceinline +# pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */ +#else +# ifdef __GNUC__ +# define GCC_VERSION (__GNUC__ * 100 + __GNUC_MINOR__) +# define FORCE_INLINE static inline __attribute__((always_inline)) +# else +# define FORCE_INLINE static inline +# endif +#endif + + +/* ************************************************************** +* Includes +****************************************************************/ +#include /* malloc, free, qsort */ +#include /* memcpy, memset */ +#include /* printf (debug) */ + + +/* ************************************************************** +* Constants +****************************************************************/ +#define HUF_ABSOLUTEMAX_TABLELOG 16 /* absolute limit of HUF_MAX_TABLELOG. Beyond that value, code does not work */ +#define HUF_MAX_TABLELOG 12 /* max configured tableLog (for static allocation); can be modified up to HUF_ABSOLUTEMAX_TABLELOG */ +#define HUF_DEFAULT_TABLELOG HUF_MAX_TABLELOG /* tableLog by default, when not specified */ +#define HUF_MAX_SYMBOL_VALUE 255 +#if (HUF_MAX_TABLELOG > HUF_ABSOLUTEMAX_TABLELOG) +# error "HUF_MAX_TABLELOG is too large !" +#endif + + +/* ************************************************************** +* Error Management +****************************************************************/ +static unsigned HUF_isError(size_t code) { return ERR_isError(code); } +#define HUF_STATIC_ASSERT(c) { enum { HUF_static_assert = 1/(int)(!!(c)) }; } /* use only *after* variable declarations */ + + + +/*-******************************************************* +* Huff0 : Huffman block decompression +*********************************************************/ +typedef struct { BYTE byte; BYTE nbBits; } HUF_DEltX2; /* single-symbol decoding */ + +typedef struct { U16 sequence; BYTE nbBits; BYTE length; } HUF_DEltX4; /* double-symbols decoding */ + +typedef struct { BYTE symbol; BYTE weight; } sortedSymbol_t; + +/*! HUF_readStats + Read compact Huffman tree, saved by HUF_writeCTable + @huffWeight : destination buffer + @return : size read from `src` +*/ +static size_t HUF_readStats(BYTE* huffWeight, size_t hwSize, U32* rankStats, + U32* nbSymbolsPtr, U32* tableLogPtr, + const void* src, size_t srcSize) +{ + U32 weightTotal; + U32 tableLog; + const BYTE* ip = (const BYTE*) src; + size_t iSize = ip[0]; + size_t oSize; + U32 n; + + //memset(huffWeight, 0, hwSize); /* is not necessary, even though some analyzer complain ... */ + + if (iSize >= 128) /* special header */ + { + if (iSize >= (242)) /* RLE */ + { + static int l[14] = { 1, 2, 3, 4, 7, 8, 15, 16, 31, 32, 63, 64, 127, 128 }; + oSize = l[iSize-242]; + memset(huffWeight, 1, hwSize); + iSize = 0; + } + else /* Incompressible */ + { + oSize = iSize - 127; + iSize = ((oSize+1)/2); + if (iSize+1 > srcSize) return ERROR(srcSize_wrong); + if (oSize >= hwSize) return ERROR(corruption_detected); + ip += 1; + for (n=0; n> 4; + huffWeight[n+1] = ip[n/2] & 15; + } + } + } + else /* header compressed with FSE (normal case) */ + { + if (iSize+1 > srcSize) return ERROR(srcSize_wrong); + oSize = FSE_decompress(huffWeight, hwSize-1, ip+1, iSize); /* max (hwSize-1) values decoded, as last one is implied */ + if (FSE_isError(oSize)) return oSize; + } + + /* collect weight stats */ + memset(rankStats, 0, (HUF_ABSOLUTEMAX_TABLELOG + 1) * sizeof(U32)); + weightTotal = 0; + for (n=0; n= HUF_ABSOLUTEMAX_TABLELOG) return ERROR(corruption_detected); + rankStats[huffWeight[n]]++; + weightTotal += (1 << huffWeight[n]) >> 1; + } + + /* get last non-null symbol weight (implied, total must be 2^n) */ + tableLog = BIT_highbit32(weightTotal) + 1; + if (tableLog > HUF_ABSOLUTEMAX_TABLELOG) return ERROR(corruption_detected); + { + U32 total = 1 << tableLog; + U32 rest = total - weightTotal; + U32 verif = 1 << BIT_highbit32(rest); + U32 lastWeight = BIT_highbit32(rest) + 1; + if (verif != rest) return ERROR(corruption_detected); /* last value must be a clean power of 2 */ + huffWeight[oSize] = (BYTE)lastWeight; + rankStats[lastWeight]++; + } + + /* check tree construction validity */ + if ((rankStats[1] < 2) || (rankStats[1] & 1)) return ERROR(corruption_detected); /* by construction : at least 2 elts of rank 1, must be even */ + + /* results */ + *nbSymbolsPtr = (U32)(oSize+1); + *tableLogPtr = tableLog; + return iSize+1; +} + + +/**************************/ +/* single-symbol decoding */ +/**************************/ + +static size_t HUF_readDTableX2 (U16* DTable, const void* src, size_t srcSize) +{ + BYTE huffWeight[HUF_MAX_SYMBOL_VALUE + 1]; + U32 rankVal[HUF_ABSOLUTEMAX_TABLELOG + 1]; /* large enough for values from 0 to 16 */ + U32 tableLog = 0; + size_t iSize; + U32 nbSymbols = 0; + U32 n; + U32 nextRankStart; + void* const dtPtr = DTable + 1; + HUF_DEltX2* const dt = (HUF_DEltX2*)dtPtr; + + HUF_STATIC_ASSERT(sizeof(HUF_DEltX2) == sizeof(U16)); /* if compilation fails here, assertion is false */ + //memset(huffWeight, 0, sizeof(huffWeight)); /* is not necessary, even though some analyzer complain ... */ + + iSize = HUF_readStats(huffWeight, HUF_MAX_SYMBOL_VALUE + 1, rankVal, &nbSymbols, &tableLog, src, srcSize); + if (HUF_isError(iSize)) return iSize; + + /* check result */ + if (tableLog > DTable[0]) return ERROR(tableLog_tooLarge); /* DTable is too small */ + DTable[0] = (U16)tableLog; /* maybe should separate sizeof DTable, as allocated, from used size of DTable, in case of DTable re-use */ + + /* Prepare ranks */ + nextRankStart = 0; + for (n=1; n<=tableLog; n++) + { + U32 current = nextRankStart; + nextRankStart += (rankVal[n] << (n-1)); + rankVal[n] = current; + } + + /* fill DTable */ + for (n=0; n> 1; + U32 i; + HUF_DEltX2 D; + D.byte = (BYTE)n; D.nbBits = (BYTE)(tableLog + 1 - w); + for (i = rankVal[w]; i < rankVal[w] + length; i++) + dt[i] = D; + rankVal[w] += length; + } + + return iSize; +} + +static BYTE HUF_decodeSymbolX2(BIT_DStream_t* Dstream, const HUF_DEltX2* dt, const U32 dtLog) +{ + const size_t val = BIT_lookBitsFast(Dstream, dtLog); /* note : dtLog >= 1 */ + const BYTE c = dt[val].byte; + BIT_skipBits(Dstream, dt[val].nbBits); + return c; +} + +#define HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) \ + *ptr++ = HUF_decodeSymbolX2(DStreamPtr, dt, dtLog) + +#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr) \ + if (MEM_64bits() || (HUF_MAX_TABLELOG<=12)) \ + HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) + +#define HUF_DECODE_SYMBOLX2_2(ptr, DStreamPtr) \ + if (MEM_64bits()) \ + HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) + +static inline size_t HUF_decodeStreamX2(BYTE* p, BIT_DStream_t* const bitDPtr, BYTE* const pEnd, const HUF_DEltX2* const dt, const U32 dtLog) +{ + BYTE* const pStart = p; + + /* up to 4 symbols at a time */ + while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p <= pEnd-4)) + { + HUF_DECODE_SYMBOLX2_2(p, bitDPtr); + HUF_DECODE_SYMBOLX2_1(p, bitDPtr); + HUF_DECODE_SYMBOLX2_2(p, bitDPtr); + HUF_DECODE_SYMBOLX2_0(p, bitDPtr); + } + + /* closer to the end */ + while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p < pEnd)) + HUF_DECODE_SYMBOLX2_0(p, bitDPtr); + + /* no more data to retrieve from bitstream, hence no need to reload */ + while (p < pEnd) + HUF_DECODE_SYMBOLX2_0(p, bitDPtr); + + return pEnd-pStart; +} + + +static size_t HUF_decompress4X2_usingDTable( + void* dst, size_t dstSize, + const void* cSrc, size_t cSrcSize, + const U16* DTable) +{ + if (cSrcSize < 10) return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */ + + { + const BYTE* const istart = (const BYTE*) cSrc; + BYTE* const ostart = (BYTE*) dst; + BYTE* const oend = ostart + dstSize; + const void* const dtPtr = DTable; + const HUF_DEltX2* const dt = ((const HUF_DEltX2*)dtPtr) +1; + const U32 dtLog = DTable[0]; + size_t errorCode; + + /* Init */ + BIT_DStream_t bitD1; + BIT_DStream_t bitD2; + BIT_DStream_t bitD3; + BIT_DStream_t bitD4; + const size_t length1 = MEM_readLE16(istart); + const size_t length2 = MEM_readLE16(istart+2); + const size_t length3 = MEM_readLE16(istart+4); + size_t length4; + const BYTE* const istart1 = istart + 6; /* jumpTable */ + const BYTE* const istart2 = istart1 + length1; + const BYTE* const istart3 = istart2 + length2; + const BYTE* const istart4 = istart3 + length3; + const size_t segmentSize = (dstSize+3) / 4; + BYTE* const opStart2 = ostart + segmentSize; + BYTE* const opStart3 = opStart2 + segmentSize; + BYTE* const opStart4 = opStart3 + segmentSize; + BYTE* op1 = ostart; + BYTE* op2 = opStart2; + BYTE* op3 = opStart3; + BYTE* op4 = opStart4; + U32 endSignal; + + length4 = cSrcSize - (length1 + length2 + length3 + 6); + if (length4 > cSrcSize) return ERROR(corruption_detected); /* overflow */ + errorCode = BIT_initDStream(&bitD1, istart1, length1); + if (HUF_isError(errorCode)) return errorCode; + errorCode = BIT_initDStream(&bitD2, istart2, length2); + if (HUF_isError(errorCode)) return errorCode; + errorCode = BIT_initDStream(&bitD3, istart3, length3); + if (HUF_isError(errorCode)) return errorCode; + errorCode = BIT_initDStream(&bitD4, istart4, length4); + if (HUF_isError(errorCode)) return errorCode; + + /* 16-32 symbols per loop (4-8 symbols per stream) */ + endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4); + for ( ; (endSignal==BIT_DStream_unfinished) && (op4<(oend-7)) ; ) + { + HUF_DECODE_SYMBOLX2_2(op1, &bitD1); + HUF_DECODE_SYMBOLX2_2(op2, &bitD2); + HUF_DECODE_SYMBOLX2_2(op3, &bitD3); + HUF_DECODE_SYMBOLX2_2(op4, &bitD4); + HUF_DECODE_SYMBOLX2_1(op1, &bitD1); + HUF_DECODE_SYMBOLX2_1(op2, &bitD2); + HUF_DECODE_SYMBOLX2_1(op3, &bitD3); + HUF_DECODE_SYMBOLX2_1(op4, &bitD4); + HUF_DECODE_SYMBOLX2_2(op1, &bitD1); + HUF_DECODE_SYMBOLX2_2(op2, &bitD2); + HUF_DECODE_SYMBOLX2_2(op3, &bitD3); + HUF_DECODE_SYMBOLX2_2(op4, &bitD4); + HUF_DECODE_SYMBOLX2_0(op1, &bitD1); + HUF_DECODE_SYMBOLX2_0(op2, &bitD2); + HUF_DECODE_SYMBOLX2_0(op3, &bitD3); + HUF_DECODE_SYMBOLX2_0(op4, &bitD4); + + endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4); + } + + /* check corruption */ + if (op1 > opStart2) return ERROR(corruption_detected); + if (op2 > opStart3) return ERROR(corruption_detected); + if (op3 > opStart4) return ERROR(corruption_detected); + /* note : op4 supposed already verified within main loop */ + + /* finish bitStreams one by one */ + HUF_decodeStreamX2(op1, &bitD1, opStart2, dt, dtLog); + HUF_decodeStreamX2(op2, &bitD2, opStart3, dt, dtLog); + HUF_decodeStreamX2(op3, &bitD3, opStart4, dt, dtLog); + HUF_decodeStreamX2(op4, &bitD4, oend, dt, dtLog); + + /* check */ + endSignal = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4); + if (!endSignal) return ERROR(corruption_detected); + + /* decoded size */ + return dstSize; + } +} + + +static size_t HUF_decompress4X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize) +{ + HUF_CREATE_STATIC_DTABLEX2(DTable, HUF_MAX_TABLELOG); + const BYTE* ip = (const BYTE*) cSrc; + size_t errorCode; + + errorCode = HUF_readDTableX2 (DTable, cSrc, cSrcSize); + if (HUF_isError(errorCode)) return errorCode; + if (errorCode >= cSrcSize) return ERROR(srcSize_wrong); + ip += errorCode; + cSrcSize -= errorCode; + + return HUF_decompress4X2_usingDTable (dst, dstSize, ip, cSrcSize, DTable); +} + + +/***************************/ +/* double-symbols decoding */ +/***************************/ + +static void HUF_fillDTableX4Level2(HUF_DEltX4* DTable, U32 sizeLog, const U32 consumed, + const U32* rankValOrigin, const int minWeight, + const sortedSymbol_t* sortedSymbols, const U32 sortedListSize, + U32 nbBitsBaseline, U16 baseSeq) +{ + HUF_DEltX4 DElt; + U32 rankVal[HUF_ABSOLUTEMAX_TABLELOG + 1]; + U32 s; + + /* get pre-calculated rankVal */ + memcpy(rankVal, rankValOrigin, sizeof(rankVal)); + + /* fill skipped values */ + if (minWeight>1) + { + U32 i, skipSize = rankVal[minWeight]; + MEM_writeLE16(&(DElt.sequence), baseSeq); + DElt.nbBits = (BYTE)(consumed); + DElt.length = 1; + for (i = 0; i < skipSize; i++) + DTable[i] = DElt; + } + + /* fill DTable */ + for (s=0; s= 1 */ + + rankVal[weight] += length; + } +} + +typedef U32 rankVal_t[HUF_ABSOLUTEMAX_TABLELOG][HUF_ABSOLUTEMAX_TABLELOG + 1]; + +static void HUF_fillDTableX4(HUF_DEltX4* DTable, const U32 targetLog, + const sortedSymbol_t* sortedList, const U32 sortedListSize, + const U32* rankStart, rankVal_t rankValOrigin, const U32 maxWeight, + const U32 nbBitsBaseline) +{ + U32 rankVal[HUF_ABSOLUTEMAX_TABLELOG + 1]; + const int scaleLog = nbBitsBaseline - targetLog; /* note : targetLog >= srcLog, hence scaleLog <= 1 */ + const U32 minBits = nbBitsBaseline - maxWeight; + U32 s; + + memcpy(rankVal, rankValOrigin, sizeof(rankVal)); + + /* fill DTable */ + for (s=0; s= minBits) /* enough room for a second symbol */ + { + U32 sortedRank; + int minWeight = nbBits + scaleLog; + if (minWeight < 1) minWeight = 1; + sortedRank = rankStart[minWeight]; + HUF_fillDTableX4Level2(DTable+start, targetLog-nbBits, nbBits, + rankValOrigin[nbBits], minWeight, + sortedList+sortedRank, sortedListSize-sortedRank, + nbBitsBaseline, symbol); + } + else + { + U32 i; + const U32 end = start + length; + HUF_DEltX4 DElt; + + MEM_writeLE16(&(DElt.sequence), symbol); + DElt.nbBits = (BYTE)(nbBits); + DElt.length = 1; + for (i = start; i < end; i++) + DTable[i] = DElt; + } + rankVal[weight] += length; + } +} + +static size_t HUF_readDTableX4 (U32* DTable, const void* src, size_t srcSize) +{ + BYTE weightList[HUF_MAX_SYMBOL_VALUE + 1]; + sortedSymbol_t sortedSymbol[HUF_MAX_SYMBOL_VALUE + 1]; + U32 rankStats[HUF_ABSOLUTEMAX_TABLELOG + 1] = { 0 }; + U32 rankStart0[HUF_ABSOLUTEMAX_TABLELOG + 2] = { 0 }; + U32* const rankStart = rankStart0+1; + rankVal_t rankVal; + U32 tableLog, maxW, sizeOfSort, nbSymbols; + const U32 memLog = DTable[0]; + size_t iSize; + void* dtPtr = DTable; + HUF_DEltX4* const dt = ((HUF_DEltX4*)dtPtr) + 1; + + HUF_STATIC_ASSERT(sizeof(HUF_DEltX4) == sizeof(U32)); /* if compilation fails here, assertion is false */ + if (memLog > HUF_ABSOLUTEMAX_TABLELOG) return ERROR(tableLog_tooLarge); + //memset(weightList, 0, sizeof(weightList)); /* is not necessary, even though some analyzer complain ... */ + + iSize = HUF_readStats(weightList, HUF_MAX_SYMBOL_VALUE + 1, rankStats, &nbSymbols, &tableLog, src, srcSize); + if (HUF_isError(iSize)) return iSize; + + /* check result */ + if (tableLog > memLog) return ERROR(tableLog_tooLarge); /* DTable can't fit code depth */ + + /* find maxWeight */ + for (maxW = tableLog; rankStats[maxW]==0; maxW--) {} /* necessarily finds a solution before 0 */ + + /* Get start index of each weight */ + { + U32 w, nextRankStart = 0; + for (w=1; w<=maxW; w++) + { + U32 current = nextRankStart; + nextRankStart += rankStats[w]; + rankStart[w] = current; + } + rankStart[0] = nextRankStart; /* put all 0w symbols at the end of sorted list*/ + sizeOfSort = nextRankStart; + } + + /* sort symbols by weight */ + { + U32 s; + for (s=0; s> consumed; + } + } + } + + HUF_fillDTableX4(dt, memLog, + sortedSymbol, sizeOfSort, + rankStart0, rankVal, maxW, + tableLog+1); + + return iSize; +} + + +static U32 HUF_decodeSymbolX4(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, const U32 dtLog) +{ + const size_t val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */ + memcpy(op, dt+val, 2); + BIT_skipBits(DStream, dt[val].nbBits); + return dt[val].length; +} + +static U32 HUF_decodeLastSymbolX4(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, const U32 dtLog) +{ + const size_t val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */ + memcpy(op, dt+val, 1); + if (dt[val].length==1) BIT_skipBits(DStream, dt[val].nbBits); + else + { + if (DStream->bitsConsumed < (sizeof(DStream->bitContainer)*8)) + { + BIT_skipBits(DStream, dt[val].nbBits); + if (DStream->bitsConsumed > (sizeof(DStream->bitContainer)*8)) + DStream->bitsConsumed = (sizeof(DStream->bitContainer)*8); /* ugly hack; works only because it's the last symbol. Note : can't easily extract nbBits from just this symbol */ + } + } + return 1; +} + + +#define HUF_DECODE_SYMBOLX4_0(ptr, DStreamPtr) \ + ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog) + +#define HUF_DECODE_SYMBOLX4_1(ptr, DStreamPtr) \ + if (MEM_64bits() || (HUF_MAX_TABLELOG<=12)) \ + ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog) + +#define HUF_DECODE_SYMBOLX4_2(ptr, DStreamPtr) \ + if (MEM_64bits()) \ + ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog) + +static inline size_t HUF_decodeStreamX4(BYTE* p, BIT_DStream_t* bitDPtr, BYTE* const pEnd, const HUF_DEltX4* const dt, const U32 dtLog) +{ + BYTE* const pStart = p; + + /* up to 8 symbols at a time */ + while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p < pEnd-7)) + { + HUF_DECODE_SYMBOLX4_2(p, bitDPtr); + HUF_DECODE_SYMBOLX4_1(p, bitDPtr); + HUF_DECODE_SYMBOLX4_2(p, bitDPtr); + HUF_DECODE_SYMBOLX4_0(p, bitDPtr); + } + + /* closer to the end */ + while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p <= pEnd-2)) + HUF_DECODE_SYMBOLX4_0(p, bitDPtr); + + while (p <= pEnd-2) + HUF_DECODE_SYMBOLX4_0(p, bitDPtr); /* no need to reload : reached the end of DStream */ + + if (p < pEnd) + p += HUF_decodeLastSymbolX4(p, bitDPtr, dt, dtLog); + + return p-pStart; +} + +static size_t HUF_decompress4X4_usingDTable( + void* dst, size_t dstSize, + const void* cSrc, size_t cSrcSize, + const U32* DTable) +{ + if (cSrcSize < 10) return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */ + + { + const BYTE* const istart = (const BYTE*) cSrc; + BYTE* const ostart = (BYTE*) dst; + BYTE* const oend = ostart + dstSize; + const void* const dtPtr = DTable; + const HUF_DEltX4* const dt = ((const HUF_DEltX4*)dtPtr) +1; + const U32 dtLog = DTable[0]; + size_t errorCode; + + /* Init */ + BIT_DStream_t bitD1; + BIT_DStream_t bitD2; + BIT_DStream_t bitD3; + BIT_DStream_t bitD4; + const size_t length1 = MEM_readLE16(istart); + const size_t length2 = MEM_readLE16(istart+2); + const size_t length3 = MEM_readLE16(istart+4); + size_t length4; + const BYTE* const istart1 = istart + 6; /* jumpTable */ + const BYTE* const istart2 = istart1 + length1; + const BYTE* const istart3 = istart2 + length2; + const BYTE* const istart4 = istart3 + length3; + const size_t segmentSize = (dstSize+3) / 4; + BYTE* const opStart2 = ostart + segmentSize; + BYTE* const opStart3 = opStart2 + segmentSize; + BYTE* const opStart4 = opStart3 + segmentSize; + BYTE* op1 = ostart; + BYTE* op2 = opStart2; + BYTE* op3 = opStart3; + BYTE* op4 = opStart4; + U32 endSignal; + + length4 = cSrcSize - (length1 + length2 + length3 + 6); + if (length4 > cSrcSize) return ERROR(corruption_detected); /* overflow */ + errorCode = BIT_initDStream(&bitD1, istart1, length1); + if (HUF_isError(errorCode)) return errorCode; + errorCode = BIT_initDStream(&bitD2, istart2, length2); + if (HUF_isError(errorCode)) return errorCode; + errorCode = BIT_initDStream(&bitD3, istart3, length3); + if (HUF_isError(errorCode)) return errorCode; + errorCode = BIT_initDStream(&bitD4, istart4, length4); + if (HUF_isError(errorCode)) return errorCode; + + /* 16-32 symbols per loop (4-8 symbols per stream) */ + endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4); + for ( ; (endSignal==BIT_DStream_unfinished) && (op4<(oend-7)) ; ) + { + HUF_DECODE_SYMBOLX4_2(op1, &bitD1); + HUF_DECODE_SYMBOLX4_2(op2, &bitD2); + HUF_DECODE_SYMBOLX4_2(op3, &bitD3); + HUF_DECODE_SYMBOLX4_2(op4, &bitD4); + HUF_DECODE_SYMBOLX4_1(op1, &bitD1); + HUF_DECODE_SYMBOLX4_1(op2, &bitD2); + HUF_DECODE_SYMBOLX4_1(op3, &bitD3); + HUF_DECODE_SYMBOLX4_1(op4, &bitD4); + HUF_DECODE_SYMBOLX4_2(op1, &bitD1); + HUF_DECODE_SYMBOLX4_2(op2, &bitD2); + HUF_DECODE_SYMBOLX4_2(op3, &bitD3); + HUF_DECODE_SYMBOLX4_2(op4, &bitD4); + HUF_DECODE_SYMBOLX4_0(op1, &bitD1); + HUF_DECODE_SYMBOLX4_0(op2, &bitD2); + HUF_DECODE_SYMBOLX4_0(op3, &bitD3); + HUF_DECODE_SYMBOLX4_0(op4, &bitD4); + + endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4); + } + + /* check corruption */ + if (op1 > opStart2) return ERROR(corruption_detected); + if (op2 > opStart3) return ERROR(corruption_detected); + if (op3 > opStart4) return ERROR(corruption_detected); + /* note : op4 supposed already verified within main loop */ + + /* finish bitStreams one by one */ + HUF_decodeStreamX4(op1, &bitD1, opStart2, dt, dtLog); + HUF_decodeStreamX4(op2, &bitD2, opStart3, dt, dtLog); + HUF_decodeStreamX4(op3, &bitD3, opStart4, dt, dtLog); + HUF_decodeStreamX4(op4, &bitD4, oend, dt, dtLog); + + /* check */ + endSignal = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4); + if (!endSignal) return ERROR(corruption_detected); + + /* decoded size */ + return dstSize; + } +} + + +static size_t HUF_decompress4X4 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize) +{ + HUF_CREATE_STATIC_DTABLEX4(DTable, HUF_MAX_TABLELOG); + const BYTE* ip = (const BYTE*) cSrc; + + size_t hSize = HUF_readDTableX4 (DTable, cSrc, cSrcSize); + if (HUF_isError(hSize)) return hSize; + if (hSize >= cSrcSize) return ERROR(srcSize_wrong); + ip += hSize; + cSrcSize -= hSize; + + return HUF_decompress4X4_usingDTable (dst, dstSize, ip, cSrcSize, DTable); +} + + +/**********************************/ +/* quad-symbol decoding */ +/**********************************/ +typedef struct { BYTE nbBits; BYTE nbBytes; } HUF_DDescX6; +typedef union { BYTE byte[4]; U32 sequence; } HUF_DSeqX6; + +/* recursive, up to level 3; may benefit from