std/lzma: accept optional End of Stream code

This commit is contained in:
Nigel Tao committed 2024-02-29 22:01:04 +11:00
1 parent 22b4ff4842
commit 1c0b07f1aa
5 files changed
+172 -26

No files matched your search

+1 -1
View File
@@ -160,7 +160,7 @@ conceptually like this:
AlgOve23 slot = decodeSlot(min(len-2, 3))
AlgOve24 distBiasedBy1 = decodeDistBiasedBy1(slot)
AlgOve25 if distBiasedBy1 == 0xFFFF_FFFF {
AlgOve26 break
AlgOve26 break // End of Stream.
AlgOve27 }
AlgOve28 mrud = (1 + distBiasedBy1, mrud[0], mrud[1], mrud[2])
AlgOve29 goto labelDoTheLZCopy
+2 -7
View File
@@ -90,13 +90,13 @@ pri func decoder.decode_bitstream_slow?(dst: base.io_writer, src: base.io_reader
rep2 = this.stashed_rep2
rep3 = this.stashed_rep3
pos = this.stashed_pos
pos_end = this.stashed_pos_end
lc = this.lc
lp_mask = ((1 as base.u64) << this.lp) - 1
pb_mask = ((1 as base.u64) << this.pb) - 1
// "outer" is the core loop discussed in README.md's "Algorithm Overview".
pos_end = pos ~sat+ this.decoded_length
while.outer pos < pos_end {
// AlgOve00 if decodeTheNextBym() == 0
index_ao00 = (state << 4) | ((pos & pb_mask) as base.u32)
@@ -904,12 +904,6 @@ pri func decoder.decode_bitstream_slow?(dst: base.io_writer, src: base.io_reader
prev_byte = (match_cusp & 0xFF) as base.u8
} endwhile.outer
if bits <> 0 {
return "#bad bitstream trailer"
} else if (this.decoded_length <> 0xFFFF_FFFF_FFFF_FFFF) and (pos <> pos_end) {
return "#bad decoded length"
}
this.stashed_bytes[0] = prev_byte
this.stashed_bytes[1] = match_byte
this.stashed_bits = bits
@@ -920,4 +914,5 @@ pri func decoder.decode_bitstream_slow?(dst: base.io_writer, src: base.io_reader
this.stashed_rep2 = rep2
this.stashed_rep3 = rep3
this.stashed_pos = pos
this.stashed_pos_end = pos_end
}
+54 -9
View File
@@ -60,15 +60,16 @@ pub struct decoder? implements base.io_transformer(
prev_lzma2_chunk_was_uncompressed : base.bool,
stashed_bytes : array[2] base.u8, // 0 is prev_byte, 1 is match_byte.
stashed_bits : base.u32,
stashed_range : base.u32,
stashed_state : base.u32[..= 11],
stashed_rep0 : base.u32[..= 0xFFFF_FFFE],
stashed_rep1 : base.u32[..= 0xFFFF_FFFE],
stashed_rep2 : base.u32[..= 0xFFFF_FFFE],
stashed_rep3 : base.u32[..= 0xFFFF_FFFE],
stashed_pos : base.u64,
stashed_bytes : array[2] base.u8, // 0 is prev_byte, 1 is match_byte.
stashed_bits : base.u32,
stashed_range : base.u32,
stashed_state : base.u32[..= 11],
stashed_rep0 : base.u32[..= 0xFFFF_FFFE],
stashed_rep1 : base.u32[..= 0xFFFF_FFFE],
stashed_rep2 : base.u32[..= 0xFFFF_FFFE],
stashed_rep3 : base.u32[..= 0xFFFF_FFFE],
stashed_pos : base.u64,
stashed_pos_end : base.u64,
util : base.utility,
) + (
@@ -475,6 +476,11 @@ pri func decoder.do_transform_io?(dst: base.io_writer, src: base.io_reader, work
return "#bad code"
}
this.stashed_range = 0xFFFF_FFFF
this.stashed_pos_end = this.stashed_pos ~sat+ this.decoded_length
if (this.stashed_pos_end == 0xFFFF_FFFF_FFFF_FFFF) and
(this.decoded_length <> 0xFFFF_FFFF_FFFF_FFFF) {
return "#unsupported decoded length"
}
this.lzma2_encoded_length_have = 5
while args.workbuf.length() < ((this.dict_size as base.u64) + 273) {
@@ -491,6 +497,23 @@ pri func decoder.do_transform_io?(dst: base.io_writer, src: base.io_reader, work
yield? status
} endwhile
if this.decoded_length == 0xFFFF_FFFF_FFFF_FFFF {
if this.stashed_bits <> 0 {
return "#bad bitstream trailer"
}
} else if this.stashed_pos <> this.stashed_pos_end {
return "#bad decoded length"
} else if this.stashed_bits <> 0 {
// Almost all LZMA chunks have either an explicit decoded_length or
// an End of Stream marker. It is legitimate, though, for a chunk
// to have both. We cater for that here.
this.decode_optional_end_of_stream?(src: args.src, workbuf: args.workbuf)
if this.stashed_bits <> 0 {
return "#bad bitstream trailer"
}
}
if this.format_extension == 0 {
break
} else if this.lzma2_encoded_length_have <> this.lzma2_encoded_length_want {
@@ -543,6 +566,28 @@ pri func decoder.update_stashed_bytes?(dst: base.io_writer, workbuf: roslice bas
} endwhile
}
// Preconditions:
// - this.stashed_pos == this.stashed_pos_end
// - this.stashed_pos_end < 0xFFFF_FFFF_FFFF_FFFF
pri func decoder.decode_optional_end_of_stream?(src: base.io_reader, workbuf: roslice base.u8) {
var w : base.io_writer
var status : base.status
this.stashed_pos_end = 0xFFFF_FFFF_FFFF_FFFF
while true {
io_bind (io: w, data: this.util.empty_slice_u8(), history_position: 0) {
status =? this.decode_bitstream_slow?(dst: w, src: args.src, workbuf: args.workbuf)
}
if status.is_ok() {
break
} else if status == base."$short write" {
return "#bad bitstream trailer"
}
yield? status
} endwhile
this.stashed_pos_end = this.stashed_pos
}
pri func decoder.initialize_dict!() {
this.dict_workbuf_index = 0
this.dict_seen = 0