1 ///////////////////////////////////////////////////////////////////////////////
3 /// \file lzma2_decoder.c
4 /// \brief LZMA2 decoder
6 // Copyright (C) 1999-2008 Igor Pavlov
7 // Copyright (C) 2008 Lasse Collin
9 // This library is free software; you can redistribute it and/or
10 // modify it under the terms of the GNU Lesser General Public
11 // License as published by the Free Software Foundation; either
12 // version 2.1 of the License, or (at your option) any later version.
14 // This library is distributed in the hope that it will be useful,
15 // but WITHOUT ANY WARRANTY; without even the implied warranty of
16 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 // Lesser General Public License for more details.
19 ///////////////////////////////////////////////////////////////////////////////
21 #include "lzma2_decoder.h"
22 #include "lz_decoder.h"
23 #include "lzma_decoder.h"
38 /// Sequence after the size fields have been decoded.
39 enum sequence next_sequence;
44 /// Uncompressed size of LZMA chunk
45 size_t uncompressed_size;
47 /// Compressed size of the chunk (naturally equals to uncompressed
48 /// size of uncompressed chunk)
49 size_t compressed_size;
51 /// True if properties are needed. This is false before the
55 /// True if dictionary reset is needed. This is false before the
56 /// first chunk (LZMA or uncompressed).
57 bool need_dictionary_reset;
59 lzma_options_lzma options;
64 lzma2_decode(lzma_coder *restrict coder, lzma_dict *restrict dict,
65 const uint8_t *restrict in, size_t *restrict in_pos,
68 // With SEQ_LZMA it is possible that no new input is needed to do
69 // some progress. The rest of the sequences assume that there is
70 // at least one byte of input.
71 while (*in_pos < in_size || coder->sequence == SEQ_LZMA)
72 switch (coder->sequence) {
74 const uint32_t control = in[*in_pos];
77 // Dictionary reset implies that next LZMA chunk has to set
79 if (control >= 0xE0 || control == 1) {
81 coder->need_dictionary_reset = false;
82 coder->need_properties = true;
83 } else if (coder->need_dictionary_reset) {
84 return LZMA_DATA_ERROR;
87 if (control >= 0x80) {
88 // LZMA chunk. The highest five bits of the
89 // uncompressed size are taken from the control byte.
90 coder->uncompressed_size = (control & 0x1F) << 16;
91 coder->sequence = SEQ_UNCOMPRESSED_1;
93 // See if there are new properties or if we need to
95 if (control >= 0xC0) {
96 // When there are new properties, state reset
97 // is done at SEQ_PROPERTIES.
98 coder->need_properties = false;
99 coder->next_sequence = SEQ_PROPERTIES;
101 } else if (coder->need_properties) {
102 return LZMA_DATA_ERROR;
105 coder->next_sequence = SEQ_LZMA;
107 // If only state reset is wanted with old
108 // properties, do the resetting here for
111 coder->lzma.reset(coder->lzma.coder,
117 return LZMA_STREAM_END;
119 // Invalid control values
121 return LZMA_DATA_ERROR;
123 // It's uncompressed chunk
124 coder->sequence = SEQ_COMPRESSED_0;
125 coder->next_sequence = SEQ_COPY;
131 case SEQ_UNCOMPRESSED_1:
132 coder->uncompressed_size += (uint32_t)(in[(*in_pos)++]) << 8;
133 coder->sequence = SEQ_UNCOMPRESSED_2;
136 case SEQ_UNCOMPRESSED_2:
137 coder->uncompressed_size += in[(*in_pos)++] + 1;
138 coder->sequence = SEQ_COMPRESSED_0;
139 coder->lzma.set_uncompressed(coder->lzma.coder,
140 coder->uncompressed_size);
143 case SEQ_COMPRESSED_0:
144 coder->compressed_size = (uint32_t)(in[(*in_pos)++]) << 8;
145 coder->sequence = SEQ_COMPRESSED_1;
148 case SEQ_COMPRESSED_1:
149 coder->compressed_size += in[(*in_pos)++] + 1;
150 coder->sequence = coder->next_sequence;
154 if (lzma_lzma_lclppb_decode(&coder->options, in[(*in_pos)++]))
155 return LZMA_DATA_ERROR;
157 coder->lzma.reset(coder->lzma.coder, &coder->options);
159 coder->sequence = SEQ_LZMA;
163 // Store the start offset so that we can update
164 // coder->compressed_size later.
165 const size_t in_start = *in_pos;
167 // Decode from in[] to *dict.
168 const lzma_ret ret = coder->lzma.code(coder->lzma.coder,
169 dict, in, in_pos, in_size);
171 // Validate and update coder->compressed_size.
172 const size_t in_used = *in_pos - in_start;
173 if (in_used > coder->compressed_size)
174 return LZMA_DATA_ERROR;
176 coder->compressed_size -= in_used;
178 // Return if we didn't finish the chunk, or an error occurred.
179 if (ret != LZMA_STREAM_END)
182 // The LZMA decoder must have consumed the whole chunk now.
183 // We don't need to worry about uncompressed size since it
184 // is checked by the LZMA decoder.
185 if (coder->compressed_size != 0)
186 return LZMA_DATA_ERROR;
188 coder->sequence = SEQ_CONTROL;
193 // Copy from input to the dictionary as is.
194 // FIXME Can copy too much?
195 dict_write(dict, in, in_pos, in_size, &coder->compressed_size);
196 if (coder->compressed_size != 0)
199 coder->sequence = SEQ_CONTROL;
205 return LZMA_PROG_ERROR;
213 lzma2_decoder_end(lzma_coder *coder, lzma_allocator *allocator)
215 assert(coder->lzma.end == NULL);
216 lzma_free(coder->lzma.coder, allocator);
218 lzma_free(coder, allocator);
225 lzma2_decoder_init(lzma_lz_decoder *lz, lzma_allocator *allocator,
226 const void *options, size_t *dict_size)
228 if (lz->coder == NULL) {
229 lz->coder = lzma_alloc(sizeof(lzma_coder), allocator);
230 if (lz->coder == NULL)
231 return LZMA_MEM_ERROR;
233 lz->code = &lzma2_decode;
234 lz->end = &lzma2_decoder_end;
236 lz->coder->lzma = LZMA_LZ_DECODER_INIT;
239 lz->coder->sequence = SEQ_CONTROL;
240 lz->coder->need_properties = true;
241 lz->coder->need_dictionary_reset = true;
243 return lzma_lzma_decoder_create(&lz->coder->lzma,
244 allocator, options, dict_size);
249 lzma_lzma2_decoder_init(lzma_next_coder *next, lzma_allocator *allocator,
250 const lzma_filter_info *filters)
252 // LZMA2 can only be the last filter in the chain. This is enforced
253 // by the raw_decoder initialization.
254 assert(filters[1].init == NULL);
256 return lzma_lz_decoder_init(next, allocator, filters,
257 &lzma2_decoder_init);
262 lzma_lzma2_decoder_memusage(const void *options)
264 return sizeof(lzma_coder)
265 + lzma_lzma_decoder_memusage_nocheck(options);
270 lzma_lzma2_props_decode(void **options, lzma_allocator *allocator,
271 const uint8_t *props, size_t props_size)
274 return LZMA_OPTIONS_ERROR;
276 // Check that reserved bits are unset.
278 return LZMA_OPTIONS_ERROR;
280 // Decode the dictionary size.
282 return LZMA_OPTIONS_ERROR;
284 lzma_options_lzma *opt = lzma_alloc(
285 sizeof(lzma_options_lzma), allocator);
287 return LZMA_MEM_ERROR;
289 if (props[0] == 40) {
290 opt->dict_size = UINT32_MAX;
292 opt->dict_size = 2 | (props[0] & 1);
293 opt->dict_size <<= props[0] / 2 + 11;
296 opt->preset_dict = NULL;
297 opt->preset_dict_size = 0;