diff --git a/release/c/wuffs-unsupported-snapshot.c b/release/c/wuffs-unsupported-snapshot.c index fbe67e72..fc2efa17 100644 --- a/release/c/wuffs-unsupported-snapshot.c +++ b/release/c/wuffs-unsupported-snapshot.c @@ -10658,13 +10658,16 @@ struct wuffs_lzma__decoder__struct { // ---------------- Status Codes +extern const char wuffs_lzip__error__bad_checksum[]; +extern const char wuffs_lzip__error__bad_footer[]; +extern const char wuffs_lzip__error__bad_header[]; extern const char wuffs_lzip__error__truncated_input[]; // ---------------- Public Consts #define WUFFS_LZIP__DECODER_DST_HISTORY_RETAIN_LENGTH_MAX_INCL_WORST_CASE 0u -#define WUFFS_LZIP__DECODER_WORKBUF_LEN_MAX_INCL_WORST_CASE 1u +#define WUFFS_LZIP__DECODER_WORKBUF_LEN_MAX_INCL_WORST_CASE 4294967568u // ---------------- Struct Declarations @@ -10777,12 +10780,17 @@ struct wuffs_lzip__decoder__struct { wuffs_base__vtable null_vtable; bool f_ignore_checksum; + uint64_t f_dsize_have; + uint64_t f_ssize_have; uint32_t p_transform_io; uint32_t p_do_transform_io; } private_impl; struct { + wuffs_crc32__ieee_hasher f_crc32; + wuffs_lzma__decoder f_lzma; + struct { uint64_t scratch; } s_do_transform_io; @@ -52546,20 +52554,37 @@ wuffs_lzma__decoder__set_quirk( } uint32_t v_v = 0; + uint32_t v_n = 0; if (a_key == 1348001792u) { if (a_value == 0u) { - } else if (((a_value & 255u) != 2u) || ((a_value >> 8u) > 40u)) { - return wuffs_base__make_status(wuffs_base__error__bad_argument); + self->private_impl.f_format_extension = 0u; + return wuffs_base__make_status(NULL); + } else if ((a_value & 255u) == 1u) { + if ((a_value >> 8u) <= 255u) { + self->private_impl.f_format_extension = ((uint32_t)(a_value)); + v_v = (self->private_impl.f_format_extension >> 8u); + v_n = (((uint32_t)(1u)) << (v_v & 31u)); + wuffs_private_impl__u32__sat_sub_indirect(&v_n, ((v_n >> 4u) * ((v_v >> 5u) & 7u))); + if ((v_n < 4096u) || (536870912u < v_n)) { + return wuffs_base__make_status(wuffs_base__error__bad_argument); + } + self->private_impl.f_dict_size = v_n; + return wuffs_base__make_status(NULL); + } + } else if ((a_value & 255u) == 2u) { + if ((a_value >> 8u) <= 40u) { + self->private_impl.f_format_extension = ((uint32_t)(a_value)); + v_v = (self->private_impl.f_format_extension >> 8u); + if (v_v < 40u) { + self->private_impl.f_dict_size = ((2u | (v_v & 1u)) << ((v_v >> 1u) + 11u)); + } else { + self->private_impl.f_dict_size = 4294967295u; + } + return wuffs_base__make_status(NULL); + } } - self->private_impl.f_format_extension = ((uint32_t)(a_value)); - v_v = (self->private_impl.f_format_extension >> 8u); - if (v_v < 40u) { - self->private_impl.f_dict_size = ((2u | (v_v & 1u)) << ((v_v >> 1u) + 11u)); - } else { - self->private_impl.f_dict_size = 4294967295u; - } - return wuffs_base__make_status(NULL); + return wuffs_base__make_status(wuffs_base__error__bad_argument); } return wuffs_base__make_status(wuffs_base__error__unsupported_option); } @@ -52771,7 +52796,7 @@ wuffs_lzma__decoder__do_transform_io( self->private_impl.f_lzma2_need_properties = true; self->private_impl.f_lzma2_need_dict_reset = true; while (true) { - if (self->private_impl.f_format_extension == 0u) { + if ((self->private_impl.f_format_extension & 255u) == 0u) { { WUFFS_BASE__COROUTINE_SUSPENSION_POINT(1); if (WUFFS_BASE__UNLIKELY(iop_a_src == io2_a_src)) { @@ -52867,6 +52892,12 @@ wuffs_lzma__decoder__do_transform_io( goto exit; } wuffs_lzma__decoder__initialize_probs(self); + } else if ((self->private_impl.f_format_extension & 255u) == 1u) { + self->private_impl.f_lc = 3u; + self->private_impl.f_lp = 0u; + self->private_impl.f_pb = 2u; + self->private_impl.f_decoded_length = 18446744073709551615u; + wuffs_lzma__decoder__initialize_probs(self); } else { while (((uint64_t)(io2_a_src - iop_a_src)) <= 0u) { status = wuffs_base__make_status(wuffs_base__suspension__short_read); @@ -53172,7 +53203,7 @@ wuffs_lzma__decoder__do_transform_io( goto exit; } } - if (self->private_impl.f_format_extension == 0u) { + if ((self->private_impl.f_format_extension & 255u) < 2u) { break; } else if (self->private_impl.f_lzma2_encoded_length_have != self->private_impl.f_lzma2_encoded_length_want) { status = wuffs_base__make_status(wuffs_lzma__error__bad_lzma2_header); @@ -53536,6 +53567,9 @@ wuffs_lzma__decoder__initialize_probs( // ---------------- Status Codes Implementations +const char wuffs_lzip__error__bad_checksum[] = "#lzip: bad checksum"; +const char wuffs_lzip__error__bad_footer[] = "#lzip: bad footer"; +const char wuffs_lzip__error__bad_header[] = "#lzip: bad header"; const char wuffs_lzip__error__truncated_input[] = "#lzip: truncated input"; // ---------------- Private Consts @@ -53610,6 +53644,20 @@ wuffs_lzip__decoder__initialize( } } + { + wuffs_base__status z = wuffs_crc32__ieee_hasher__initialize( + &self->private_data.f_crc32, sizeof(self->private_data.f_crc32), WUFFS_VERSION, options); + if (z.repr) { + return z; + } + } + { + wuffs_base__status z = wuffs_lzma__decoder__initialize( + &self->private_data.f_lzma, sizeof(self->private_data.f_lzma), WUFFS_VERSION, options); + if (z.repr) { + return z; + } + } self->private_impl.magic = WUFFS_BASE__MAGIC; self->private_impl.vtable_for__wuffs_base__io_transformer.vtable_name = wuffs_base__io_transformer__vtable_name; @@ -53700,7 +53748,7 @@ wuffs_lzip__decoder__dst_history_retain_length( return wuffs_base__utility__make_optional_u63(false, 0u); } - return wuffs_base__utility__make_optional_u63(true, 0u); + return wuffs_lzma__decoder__dst_history_retain_length(&self->private_data.f_lzma); } // -------- func lzip.decoder.workbuf_len @@ -53717,7 +53765,7 @@ wuffs_lzip__decoder__workbuf_len( return wuffs_base__utility__empty_range_ii_u64(); } - return wuffs_base__utility__make_range_ii_u64(1u, 1u); + return wuffs_lzma__decoder__workbuf_len(&self->private_data.f_lzma); } // -------- func lzip.decoder.transform_io @@ -53798,6 +53846,16 @@ wuffs_lzip__decoder__do_transform_io( wuffs_base__slice_u8 a_workbuf) { wuffs_base__status status = wuffs_base__make_status(NULL); + uint8_t v_c8 = 0; + uint32_t v_c32 = 0; + uint64_t v_c64 = 0; + uint64_t v_dmark = 0; + uint64_t v_smark = 0; + wuffs_base__status v_status = wuffs_base__make_status(NULL); + uint32_t v_checksum_want = 0; + uint32_t v_checksum_have = 0; + uint64_t v_size_want = 0; + uint8_t* iop_a_dst = NULL; uint8_t* io0_a_dst WUFFS_BASE__POTENTIALLY_UNUSED = NULL; uint8_t* io1_a_dst WUFFS_BASE__POTENTIALLY_UNUSED = NULL; @@ -53811,20 +53869,234 @@ wuffs_lzip__decoder__do_transform_io( io2_a_dst = iop_a_dst; } } + const uint8_t* iop_a_src = NULL; + const uint8_t* io0_a_src WUFFS_BASE__POTENTIALLY_UNUSED = NULL; + const uint8_t* io1_a_src WUFFS_BASE__POTENTIALLY_UNUSED = NULL; + const uint8_t* io2_a_src WUFFS_BASE__POTENTIALLY_UNUSED = NULL; + if (a_src && a_src->data.ptr) { + io0_a_src = a_src->data.ptr; + io1_a_src = io0_a_src + a_src->meta.ri; + iop_a_src = io1_a_src; + io2_a_src = io0_a_src + a_src->meta.wi; + } uint32_t coro_susp_point = self->private_impl.p_do_transform_io; switch (coro_susp_point) { WUFFS_BASE__COROUTINE_SUSPENSION_POINT_0; - self->private_data.s_do_transform_io.scratch = 120u; - WUFFS_BASE__COROUTINE_SUSPENSION_POINT(1); - if (iop_a_dst == io2_a_dst) { - status = wuffs_base__make_status(wuffs_base__suspension__short_write); - goto suspend; + while (true) { + { + WUFFS_BASE__COROUTINE_SUSPENSION_POINT(1); + uint64_t t_0; + if (WUFFS_BASE__LIKELY(io2_a_src - iop_a_src >= 5)) { + t_0 = ((uint64_t)(wuffs_base__peek_u40le__no_bounds_check(iop_a_src))); + iop_a_src += 5; + } else { + self->private_data.s_do_transform_io.scratch = 0; + WUFFS_BASE__COROUTINE_SUSPENSION_POINT(2); + while (true) { + if (WUFFS_BASE__UNLIKELY(iop_a_src == io2_a_src)) { + status = wuffs_base__make_status(wuffs_base__suspension__short_read); + goto suspend; + } + uint64_t* scratch = &self->private_data.s_do_transform_io.scratch; + uint32_t num_bits_0 = ((uint32_t)(*scratch >> 56)); + *scratch <<= 8; + *scratch >>= 8; + *scratch |= ((uint64_t)(*iop_a_src++)) << num_bits_0; + if (num_bits_0 == 32) { + t_0 = ((uint64_t)(*scratch)); + break; + } + num_bits_0 += 8u; + *scratch |= ((uint64_t)(num_bits_0)) << 56; + } + } + v_c64 = t_0; + } + if (v_c64 != 5641951820u) { + status = wuffs_base__make_status(wuffs_lzip__error__bad_header); + goto exit; + } + { + WUFFS_BASE__COROUTINE_SUSPENSION_POINT(3); + if (WUFFS_BASE__UNLIKELY(iop_a_src == io2_a_src)) { + status = wuffs_base__make_status(wuffs_base__suspension__short_read); + goto suspend; + } + uint8_t t_1 = *iop_a_src++; + v_c8 = t_1; + } + v_status = wuffs_lzma__decoder__set_quirk(&self->private_data.f_lzma, 1348001792u, (1u | (((uint64_t)(v_c8)) << 8u))); + if ( ! wuffs_base__status__is_ok(&v_status)) { + if (v_status.repr == wuffs_base__error__bad_argument) { + status = wuffs_base__make_status(wuffs_lzip__error__bad_header); + goto exit; + } + status = v_status; + if (wuffs_base__status__is_error(&status)) { + goto exit; + } else if (wuffs_base__status__is_suspension(&status)) { + status = wuffs_base__make_status(wuffs_base__error__cannot_return_a_suspension); + goto exit; + } + goto ok; + } + self->private_impl.f_ssize_have = 0u; + self->private_impl.f_dsize_have = 0u; + while (true) { + v_dmark = ((uint64_t)(iop_a_dst - io0_a_dst)); + v_smark = ((uint64_t)(iop_a_src - io0_a_src)); + { + if (a_dst) { + a_dst->meta.wi = ((size_t)(iop_a_dst - a_dst->data.ptr)); + } + if (a_src) { + a_src->meta.ri = ((size_t)(iop_a_src - a_src->data.ptr)); + } + wuffs_base__status t_2 = wuffs_lzma__decoder__transform_io(&self->private_data.f_lzma, a_dst, a_src, a_workbuf); + v_status = t_2; + if (a_dst) { + iop_a_dst = a_dst->data.ptr + a_dst->meta.wi; + } + if (a_src) { + iop_a_src = a_src->data.ptr + a_src->meta.ri; + } + } + self->private_impl.f_ssize_have += wuffs_private_impl__io__count_since(v_smark, ((uint64_t)(iop_a_src - io0_a_src))); + self->private_impl.f_dsize_have += wuffs_private_impl__io__count_since(v_dmark, ((uint64_t)(iop_a_dst - io0_a_dst))); + if ( ! self->private_impl.f_ignore_checksum) { + wuffs_crc32__ieee_hasher__update(&self->private_data.f_crc32, wuffs_private_impl__io__since(v_dmark, ((uint64_t)(iop_a_dst - io0_a_dst)), io0_a_dst)); + } + if (wuffs_base__status__is_ok(&v_status)) { + break; + } + status = v_status; + WUFFS_BASE__COROUTINE_SUSPENSION_POINT_MAYBE_SUSPEND(4); + } + { + WUFFS_BASE__COROUTINE_SUSPENSION_POINT(5); + uint32_t t_3; + if (WUFFS_BASE__LIKELY(io2_a_src - iop_a_src >= 4)) { + t_3 = wuffs_base__peek_u32le__no_bounds_check(iop_a_src); + iop_a_src += 4; + } else { + self->private_data.s_do_transform_io.scratch = 0; + WUFFS_BASE__COROUTINE_SUSPENSION_POINT(6); + while (true) { + if (WUFFS_BASE__UNLIKELY(iop_a_src == io2_a_src)) { + status = wuffs_base__make_status(wuffs_base__suspension__short_read); + goto suspend; + } + uint64_t* scratch = &self->private_data.s_do_transform_io.scratch; + uint32_t num_bits_3 = ((uint32_t)(*scratch >> 56)); + *scratch <<= 8; + *scratch >>= 8; + *scratch |= ((uint64_t)(*iop_a_src++)) << num_bits_3; + if (num_bits_3 == 24) { + t_3 = ((uint32_t)(*scratch)); + break; + } + num_bits_3 += 8u; + *scratch |= ((uint64_t)(num_bits_3)) << 56; + } + } + v_checksum_want = t_3; + } + if ( ! self->private_impl.f_ignore_checksum) { + v_checksum_have = wuffs_crc32__ieee_hasher__checksum_u32(&self->private_data.f_crc32); + if (v_checksum_have != v_checksum_want) { + status = wuffs_base__make_status(wuffs_lzip__error__bad_checksum); + goto exit; + } + } + { + WUFFS_BASE__COROUTINE_SUSPENSION_POINT(7); + uint64_t t_4; + if (WUFFS_BASE__LIKELY(io2_a_src - iop_a_src >= 8)) { + t_4 = wuffs_base__peek_u64le__no_bounds_check(iop_a_src); + iop_a_src += 8; + } else { + self->private_data.s_do_transform_io.scratch = 0; + WUFFS_BASE__COROUTINE_SUSPENSION_POINT(8); + while (true) { + if (WUFFS_BASE__UNLIKELY(iop_a_src == io2_a_src)) { + status = wuffs_base__make_status(wuffs_base__suspension__short_read); + goto suspend; + } + uint64_t* scratch = &self->private_data.s_do_transform_io.scratch; + uint32_t num_bits_4 = ((uint32_t)(*scratch >> 56)); + *scratch <<= 8; + *scratch >>= 8; + *scratch |= ((uint64_t)(*iop_a_src++)) << num_bits_4; + if (num_bits_4 == 56) { + t_4 = ((uint64_t)(*scratch)); + break; + } + num_bits_4 += 8u; + *scratch |= ((uint64_t)(num_bits_4)) << 56; + } + } + v_size_want = t_4; + } + if (self->private_impl.f_dsize_have != v_size_want) { + status = wuffs_base__make_status(wuffs_lzip__error__bad_footer); + goto exit; + } + { + WUFFS_BASE__COROUTINE_SUSPENSION_POINT(9); + uint64_t t_5; + if (WUFFS_BASE__LIKELY(io2_a_src - iop_a_src >= 8)) { + t_5 = wuffs_base__peek_u64le__no_bounds_check(iop_a_src); + iop_a_src += 8; + } else { + self->private_data.s_do_transform_io.scratch = 0; + WUFFS_BASE__COROUTINE_SUSPENSION_POINT(10); + while (true) { + if (WUFFS_BASE__UNLIKELY(iop_a_src == io2_a_src)) { + status = wuffs_base__make_status(wuffs_base__suspension__short_read); + goto suspend; + } + uint64_t* scratch = &self->private_data.s_do_transform_io.scratch; + uint32_t num_bits_5 = ((uint32_t)(*scratch >> 56)); + *scratch <<= 8; + *scratch >>= 8; + *scratch |= ((uint64_t)(*iop_a_src++)) << num_bits_5; + if (num_bits_5 == 56) { + t_5 = ((uint64_t)(*scratch)); + break; + } + num_bits_5 += 8u; + *scratch |= ((uint64_t)(num_bits_5)) << 56; + } + } + v_size_want = t_5; + } + if ((v_size_want < 26u) || (2251799813685248u < v_size_want)) { + status = wuffs_base__make_status(wuffs_lzip__error__bad_footer); + goto exit; + } else if (self->private_impl.f_ssize_have != (v_size_want - 26u)) { + status = wuffs_base__make_status(wuffs_lzip__error__bad_footer); + goto exit; + } + while (((uint64_t)(io2_a_src - iop_a_src)) < 4u) { + if (a_src && a_src->meta.closed) { + goto label__outer__break; + } + status = wuffs_base__make_status(wuffs_base__suspension__short_read); + WUFFS_BASE__COROUTINE_SUSPENSION_POINT_MAYBE_SUSPEND(11); + } + v_c32 = wuffs_base__peek_u32le__no_bounds_check(iop_a_src); + if (v_c32 != 1346984524u) { + break; + } + wuffs_private_impl__ignore_status(wuffs_crc32__ieee_hasher__initialize(&self->private_data.f_crc32, + sizeof (wuffs_crc32__ieee_hasher), WUFFS_VERSION, WUFFS_INITIALIZE__LEAVE_INTERNAL_BUFFERS_UNINITIALIZED)); + wuffs_private_impl__ignore_status(wuffs_lzma__decoder__initialize(&self->private_data.f_lzma, + sizeof (wuffs_lzma__decoder), WUFFS_VERSION, WUFFS_INITIALIZE__LEAVE_INTERNAL_BUFFERS_UNINITIALIZED)); } - *iop_a_dst++ = ((uint8_t)(self->private_data.s_do_transform_io.scratch)); + label__outer__break:; - goto ok; ok: self->private_impl.p_do_transform_io = 0; goto exit; @@ -53839,6 +54111,9 @@ wuffs_lzip__decoder__do_transform_io( if (a_dst && a_dst->data.ptr) { a_dst->meta.wi = ((size_t)(iop_a_dst - a_dst->data.ptr)); } + if (a_src && a_src->data.ptr) { + a_src->meta.ri = ((size_t)(iop_a_src - a_src->data.ptr)); + } return status; } diff --git a/std/lzip/decode_lzip.wuffs b/std/lzip/decode_lzip.wuffs index 66d5c230..8c5b4898 100644 --- a/std/lzip/decode_lzip.wuffs +++ b/std/lzip/decode_lzip.wuffs @@ -8,19 +8,32 @@ // // SPDX-License-Identifier: Apache-2.0 OR MIT +// -------- + +// The Lzip file format specification is at +// https://www.nongnu.org/lzip/manual/lzip_manual.html#File-format + use "std/crc32" use "std/lzma" +pub status "#bad checksum" +pub status "#bad footer" +pub status "#bad header" pub status "#truncated input" pub const DECODER_DST_HISTORY_RETAIN_LENGTH_MAX_INCL_WORST_CASE : base.u64 = 0 - -pub const DECODER_WORKBUF_LEN_MAX_INCL_WORST_CASE : base.u64 = 1 +pub const DECODER_WORKBUF_LEN_MAX_INCL_WORST_CASE : base.u64 = 0xFFFF_FFFF + 273 pub struct decoder? implements base.io_transformer( ignore_checksum : base.bool, + dsize_have : base.u64, + ssize_have : base.u64, + util : base.utility, +) + ( + crc32 : crc32.ieee_hasher, + lzma : lzma.decoder, ) pub func decoder.get_quirk(key: base.u32) base.u64 { @@ -39,13 +52,11 @@ pub func decoder.set_quirk!(key: base.u32, value: base.u64) base.status { } pub func decoder.dst_history_retain_length() base.optional_u63 { - return this.util.make_optional_u63(has_value: true, value: 0) + return this.lzma.dst_history_retain_length() } pub func decoder.workbuf_len() base.range_ii_u64 { - return this.util.make_range_ii_u64( - min_incl: DECODER_WORKBUF_LEN_MAX_INCL_WORST_CASE, - max_incl: DECODER_WORKBUF_LEN_MAX_INCL_WORST_CASE) + return this.lzma.workbuf_len() } pub func decoder.transform_io?(dst: base.io_writer, src: base.io_reader, workbuf: slice base.u8) { @@ -61,5 +72,98 @@ pub func decoder.transform_io?(dst: base.io_writer, src: base.io_reader, workbuf } pri func decoder.do_transform_io?(dst: base.io_writer, src: base.io_reader, workbuf: slice base.u8) { - args.dst.write_u8?(a: 'x') + var c8 : base.u8 + var c32 : base.u32 + var c64 : base.u64 + var dmark : base.u64 + var smark : base.u64 + var status : base.status + var checksum_want : base.u32 + var checksum_have : base.u32 + var size_want : base.u64 + + while.outer true { + // Read the magic number and version number. + // + // Some of the test/3pdata/xzsuite/good*v0*.lz files have a 0x00 + // version number (instead of 0x01). We reject those, just like + // /usr/bin/lzip does and as per the Lzip spec. + c64 = args.src.read_u40le_as_u64?() + if c64 <> 'LZIP\x01'le { + return "#bad header" + } + + // Read the dictionary size. + c8 = args.src.read_u8?() + status = this.lzma.set_quirk!( + key: lzma.QUIRK_FORMAT_EXTENSION, + value: 0x01 | ((c8 as base.u64) << 8)) + if not status.is_ok() { + if status == base."#bad argument" { + return "#bad header" + } + return status + } + + // Decode LZMA. + this.ssize_have = 0 + this.dsize_have = 0 + while true { + dmark = args.dst.mark() + smark = args.src.mark() + status =? this.lzma.transform_io?(dst: args.dst, src: args.src, workbuf: args.workbuf) + this.ssize_have ~mod+= args.src.count_since(mark: smark) + this.dsize_have ~mod+= args.dst.count_since(mark: dmark) + + if not this.ignore_checksum { + this.crc32.update!(x: args.dst.since(mark: dmark)) + } + + if status.is_ok() { + break + } + yield? status + } endwhile + + // Check the checksum. + checksum_want = args.src.read_u32le?() + if not this.ignore_checksum { + checksum_have = this.crc32.checksum_u32() + if checksum_have <> checksum_want { + return "#bad checksum" + } + } + + // Check the uncompressed size. + size_want = args.src.read_u64le?() + if this.dsize_have <> size_want { + return "#bad footer" + } + + // Check the compressed size, which includes 6 bytes of header and 20 + // bytes of footer. Per the spec, it's also capped at 2 PiB. + size_want = args.src.read_u64le?() + if (size_want < 26) or (0x8_0000_0000_0000 < size_want) { + return "#bad footer" + } else if this.ssize_have <> (size_want - 26) { + return "#bad footer" + } + + // Reset state and continue the outer loop, if not at EOF and it looks + // like there's another LZIP-encoded chunk. + while args.src.length() < 4, + post args.src.length() >= 4, + { + if args.src.is_closed() { + break.outer + } + yield? base."$short read" + } endwhile + c32 = args.src.peek_u32le() + if c32 <> 'LZIP'le { + break.outer + } + this.crc32.reset!() + this.lzma.reset!() + } endwhile.outer } diff --git a/std/lzma/decode_lzma.wuffs b/std/lzma/decode_lzma.wuffs index 5199bf52..cb085fb5 100644 --- a/std/lzma/decode_lzma.wuffs +++ b/std/lzma/decode_lzma.wuffs @@ -273,22 +273,40 @@ pub func decoder.get_quirk(key: base.u32) base.u64 { pub func decoder.set_quirk!(key: base.u32, value: base.u64) base.status { var v : base.u32 + var n : base.u32 if args.key == QUIRK_FORMAT_EXTENSION { if args.value == 0 { - // No-op. - } else if ((args.value & 0xFF) <> 0x02) or ((args.value >> 8) > 40) { - return base."#bad argument" - } - this.format_extension = (args.value & 0xFFFF_FFFF) as base.u32 + this.format_extension = 0 + return ok - v = this.format_extension >> 8 - if v < 40 { - this.dict_size = (2 | (v & 1)) << ((v >> 1) + 11) - } else { - this.dict_size = 0xFFFF_FFFF + } else if (args.value & 0xFF) == 1 { + if (args.value >> 8) <= 0xFF { + this.format_extension = (args.value & 0xFFFF_FFFF) as base.u32 + v = this.format_extension >> 8 + n = (1 as base.u32) << (v & 0x1F) + n ~sat-= (n >> 4) * ((v >> 5) & 7) + if (n < (1 << 12)) or ((1 << 29) < n) { + return base."#bad argument" + } + this.dict_size = n + return ok + } + + } else if (args.value & 0xFF) == 2 { + if (args.value >> 8) <= 40 { + this.format_extension = (args.value & 0xFFFF_FFFF) as base.u32 + v = this.format_extension >> 8 + if v < 40 { + this.dict_size = (2 | (v & 1)) << ((v >> 1) + 11) + } else { + this.dict_size = 0xFFFF_FFFF + } + return ok + } } - return ok + + return base."#bad argument" } return base."#unsupported option" } @@ -352,7 +370,7 @@ pri func decoder.do_transform_io?(dst: base.io_writer, src: base.io_reader, work // header, see // https://github.com/jljusten/LZMA-SDK/blob/781863cdf592da3e97420f50de5dac056ad352a5/C/Lzma2Dec.c#L17-L23 while true { - if this.format_extension == 0 { // LZMA1. + if (this.format_extension & 0xFF) == 0 { // LZMA1. prop_byte = args.src.read_u8?() if prop_byte >= 225 { return "#bad header" @@ -379,6 +397,13 @@ pri func decoder.do_transform_io?(dst: base.io_writer, src: base.io_reader, work this.initialize_probs!() + } else if (this.format_extension & 0xFF) == 1 { // Lzip. + this.lc = 3 + this.lp = 0 + this.pb = 2 + this.decoded_length = 0xFFFF_FFFF_FFFF_FFFF + this.initialize_probs!() + } else { // LZMA2. while args.src.length() <= 0, post args.src.length() > 0, @@ -514,7 +539,7 @@ pri func decoder.do_transform_io?(dst: base.io_writer, src: base.io_reader, work } } - if this.format_extension == 0 { + if (this.format_extension & 0xFF) < 2 { break } else if this.lzma2_encoded_length_have <> this.lzma2_encoded_length_want { return "#bad LZMA2 header" diff --git a/std/lzma/decode_quirks.wuffs b/std/lzma/decode_quirks.wuffs index 1a10a9cf..2c7a1105 100644 --- a/std/lzma/decode_quirks.wuffs +++ b/std/lzma/decode_quirks.wuffs @@ -22,7 +22,13 @@ pri const QUIRKS_BASE : base.u32 = 0x5058_E000 // format. Zero means to use the default LZMA format. // // When positive, the low byte of the u64 value indicates the extension. -// Currently, the only valid extension is 0x02, indicating LZMA2. +// Currently, the only valid extensions are: +// - 0x01 indicates Lzip. +// - 0x02 indicates LZMA2. +// +// For Lzip, the high 7 bytes (shifted right by 8) must be less than or equal +// to 0xFF and indicates the dictionary size. See section "6 File format" of +// https://www.nongnu.org/lzip/manual/lzip_manual.html // // For LZMA2, the high 7 bytes (shifted right by 8) must be less than or equal // to 40 and indicates the dictionary size. See section "5.3.1. LZMA2" of diff --git a/test/3pdata/mzcat-checksums-of-xzsuite.txt b/test/3pdata/mzcat-checksums-of-xzsuite.txt index f06c99fe..449cf17b 100644 --- a/test/3pdata/mzcat-checksums-of-xzsuite.txt +++ b/test/3pdata/mzcat-checksums-of-xzsuite.txt @@ -17,14 +17,10 @@ BAD 7759aecd test/3pdata/xzsuite/bad-1-lzma2-8.xz BAD 3b80e9ac test/3pdata/xzsuite/bad-1-lzma2-9.xz BAD 15a2a343 test/3pdata/xzsuite/bad-1-stream_flags-1.xz BAD 15a2a343 test/3pdata/xzsuite/bad-1-stream_flags-3.xz -OK. 8cdc1683 test/3pdata/xzsuite/bad-1-v0-uncomp-size.lz -OK. 8cdc1683 test/3pdata/xzsuite/bad-1-v1-crc32.lz -OK. 8cdc1683 test/3pdata/xzsuite/bad-1-v1-dict-1.lz -OK. 8cdc1683 test/3pdata/xzsuite/bad-1-v1-dict-2.lz -OK. 8cdc1683 test/3pdata/xzsuite/bad-1-v1-magic-2.lz -OK. 8cdc1683 test/3pdata/xzsuite/bad-1-v1-member-size.lz -OK. 8cdc1683 test/3pdata/xzsuite/bad-1-v1-trailing-magic.lz -OK. 8cdc1683 test/3pdata/xzsuite/bad-1-v1-uncomp-size.lz +BAD 15a2a343 test/3pdata/xzsuite/bad-1-v1-crc32.lz +BAD 15a2a343 test/3pdata/xzsuite/bad-1-v1-member-size.lz +BAD 15a2a343 test/3pdata/xzsuite/bad-1-v1-trailing-magic.lz +BAD 15a2a343 test/3pdata/xzsuite/bad-1-v1-uncomp-size.lz BAD 15a2a343 test/3pdata/xzsuite/bad-1-vli-2.xz BAD 31963516 test/3pdata/xzsuite/bad-2-compressed_data_padding.xz BAD 15a2a343 test/3pdata/xzsuite/bad-2-index-1.xz @@ -62,17 +58,13 @@ OK. 00000000 test/3pdata/xzsuite/good-1-lzma2-5.xz OK. 9d269cdc test/3pdata/xzsuite/good-1-riscv-lzma2-1.xz OK. 8d115715 test/3pdata/xzsuite/good-1-riscv-lzma2-2.xz OK. cc270da1 test/3pdata/xzsuite/good-1-sparc-lzma2.xz -OK. 8cdc1683 test/3pdata/xzsuite/good-1-v0-trailing-1.lz -OK. 8cdc1683 test/3pdata/xzsuite/good-1-v0.lz -OK. 8cdc1683 test/3pdata/xzsuite/good-1-v1-trailing-1.lz -OK. 8cdc1683 test/3pdata/xzsuite/good-1-v1-trailing-2.lz -OK. 8cdc1683 test/3pdata/xzsuite/good-1-v1.lz +OK. 15a2a343 test/3pdata/xzsuite/good-1-v1-trailing-1.lz +OK. 15a2a343 test/3pdata/xzsuite/good-1-v1-trailing-2.lz +OK. 15a2a343 test/3pdata/xzsuite/good-1-v1.lz OK. a3679091 test/3pdata/xzsuite/good-1-x86-lzma2.xz OK. 15a2a343 test/3pdata/xzsuite/good-2-lzma2.xz -OK. 8cdc1683 test/3pdata/xzsuite/good-2-v0-v1.lz -OK. 8cdc1683 test/3pdata/xzsuite/good-2-v1-v0.lz -OK. 8cdc1683 test/3pdata/xzsuite/good-2-v1-v1.lz +BAD 31963516 test/3pdata/xzsuite/good-2-v1-v0.lz +OK. 15a2a343 test/3pdata/xzsuite/good-2-v1-v1.lz OK. 15a2a343 test/3pdata/xzsuite/good-known_size-with_eopm.lzma OK. 15a2a343 test/3pdata/xzsuite/good-known_size-without_eopm.lzma OK. 15a2a343 test/3pdata/xzsuite/good-unknown_size-with_eopm.lzma -OK. 8cdc1683 test/3pdata/xzsuite/unsupported-1-v234.lz diff --git a/test/c/std/lzip.c b/test/c/std/lzip.c index 81016a67..148fe349 100644 --- a/test/c/std/lzip.c +++ b/test/c/std/lzip.c @@ -77,7 +77,7 @@ test_wuffs_lzip_decode_interface() { WUFFS_INITIALIZE__LEAVE_INTERNAL_BUFFERS_UNINITIALIZED)); return do_test__wuffs_base__io_transformer( wuffs_lzip__decoder__upcast_as__wuffs_base__io_transformer(&dec), - "test/data/romeo.txt.lz", 0, SIZE_MAX, 1, 0x78); + "test/data/romeo.txt.lz", 0, SIZE_MAX, 942, 0x0A); } // ---------------- Mimic Tests diff --git a/test/mzcat-checksums-of-data.txt b/test/mzcat-checksums-of-data.txt index 92e8efc4..04c8d207 100644 --- a/test/mzcat-checksums-of-data.txt +++ b/test/mzcat-checksums-of-data.txt @@ -29,7 +29,7 @@ OK. abe507ef test/data/romeo.txt.bz2 OK. abe507ef test/data/romeo.txt.delta1.xz OK. abe507ef test/data/romeo.txt.gz OK. abe507ef test/data/romeo.txt.litonlylzma -OK. 8cdc1683 test/data/romeo.txt.lz +OK. abe507ef test/data/romeo.txt.lz OK. abe507ef test/data/romeo.txt.lzma OK. abe507ef test/data/romeo.txt.xz OK. abe507ef test/data/romeo.txt.zlib