isomorfeus-ferret 0.12.7 → 0.13.0

Sign up to get free protection for your applications and to get access to all the features.
Files changed (164) hide show
  1. checksums.yaml +4 -4
  2. data/LICENSE +101 -19
  3. data/README.md +54 -1
  4. data/ext/isomorfeus_ferret_ext/bm_bitvector.c +22 -30
  5. data/ext/isomorfeus_ferret_ext/bm_hash.c +6 -12
  6. data/ext/isomorfeus_ferret_ext/bm_micro_string.c +3 -6
  7. data/ext/isomorfeus_ferret_ext/bm_store.c +11 -22
  8. data/ext/isomorfeus_ferret_ext/brotli_common_dictionary.c +1 -1
  9. data/ext/isomorfeus_ferret_ext/brotli_dec_decode.c +1 -1
  10. data/ext/isomorfeus_ferret_ext/bzip_blocksort.c +1094 -0
  11. data/ext/isomorfeus_ferret_ext/bzip_huffman.c +205 -0
  12. data/ext/isomorfeus_ferret_ext/bzlib.c +1572 -0
  13. data/ext/isomorfeus_ferret_ext/bzlib.h +282 -0
  14. data/ext/isomorfeus_ferret_ext/bzlib_compress.c +672 -0
  15. data/ext/isomorfeus_ferret_ext/bzlib_crctable.c +104 -0
  16. data/ext/isomorfeus_ferret_ext/bzlib_decompress.c +652 -0
  17. data/ext/isomorfeus_ferret_ext/bzlib_private.h +509 -0
  18. data/ext/isomorfeus_ferret_ext/bzlib_randtable.c +84 -0
  19. data/ext/isomorfeus_ferret_ext/fio_tmpfile.h +53 -53
  20. data/ext/isomorfeus_ferret_ext/frb_analysis.c +785 -1192
  21. data/ext/isomorfeus_ferret_ext/frb_index.c +492 -474
  22. data/ext/isomorfeus_ferret_ext/frb_qparser.c +48 -60
  23. data/ext/isomorfeus_ferret_ext/frb_search.c +1520 -1002
  24. data/ext/isomorfeus_ferret_ext/frb_store.c +96 -96
  25. data/ext/isomorfeus_ferret_ext/frb_threading.h +0 -1
  26. data/ext/isomorfeus_ferret_ext/frb_utils.c +147 -196
  27. data/ext/isomorfeus_ferret_ext/frt_analysis.c +695 -1090
  28. data/ext/isomorfeus_ferret_ext/frt_analysis.h +174 -170
  29. data/ext/isomorfeus_ferret_ext/frt_array.c +2 -4
  30. data/ext/isomorfeus_ferret_ext/frt_bitvector.c +9 -16
  31. data/ext/isomorfeus_ferret_ext/frt_bitvector.h +32 -81
  32. data/ext/isomorfeus_ferret_ext/frt_document.c +15 -20
  33. data/ext/isomorfeus_ferret_ext/frt_document.h +10 -10
  34. data/ext/isomorfeus_ferret_ext/frt_except.c +5 -12
  35. data/ext/isomorfeus_ferret_ext/frt_field_index.c +3 -3
  36. data/ext/isomorfeus_ferret_ext/frt_field_index.h +6 -7
  37. data/ext/isomorfeus_ferret_ext/frt_filter.c +35 -46
  38. data/ext/isomorfeus_ferret_ext/frt_fs_store.c +1 -0
  39. data/ext/isomorfeus_ferret_ext/frt_global.c +105 -63
  40. data/ext/isomorfeus_ferret_ext/frt_global.h +7 -3
  41. data/ext/isomorfeus_ferret_ext/frt_hash.c +1 -2
  42. data/ext/isomorfeus_ferret_ext/frt_ind.c +32 -35
  43. data/ext/isomorfeus_ferret_ext/frt_ind.h +9 -9
  44. data/ext/isomorfeus_ferret_ext/frt_index.c +580 -399
  45. data/ext/isomorfeus_ferret_ext/frt_index.h +272 -291
  46. data/ext/isomorfeus_ferret_ext/frt_mempool.c +1 -2
  47. data/ext/isomorfeus_ferret_ext/frt_multimapper.c +4 -7
  48. data/ext/isomorfeus_ferret_ext/frt_q_boolean.c +67 -91
  49. data/ext/isomorfeus_ferret_ext/frt_q_const_score.c +35 -38
  50. data/ext/isomorfeus_ferret_ext/frt_q_filtered_query.c +53 -72
  51. data/ext/isomorfeus_ferret_ext/frt_q_fuzzy.c +25 -32
  52. data/ext/isomorfeus_ferret_ext/frt_q_match_all.c +21 -23
  53. data/ext/isomorfeus_ferret_ext/frt_q_multi_term.c +66 -103
  54. data/ext/isomorfeus_ferret_ext/frt_q_parser.c +207 -195
  55. data/ext/isomorfeus_ferret_ext/frt_q_phrase.c +20 -16
  56. data/ext/isomorfeus_ferret_ext/frt_q_prefix.c +17 -14
  57. data/ext/isomorfeus_ferret_ext/frt_q_range.c +102 -131
  58. data/ext/isomorfeus_ferret_ext/frt_q_span.c +179 -178
  59. data/ext/isomorfeus_ferret_ext/frt_q_term.c +47 -60
  60. data/ext/isomorfeus_ferret_ext/frt_q_wildcard.c +18 -16
  61. data/ext/isomorfeus_ferret_ext/frt_ram_store.c +45 -84
  62. data/ext/isomorfeus_ferret_ext/frt_search.c +105 -146
  63. data/ext/isomorfeus_ferret_ext/frt_search.h +331 -320
  64. data/ext/isomorfeus_ferret_ext/frt_similarity.c +5 -13
  65. data/ext/isomorfeus_ferret_ext/frt_similarity.h +7 -12
  66. data/ext/isomorfeus_ferret_ext/frt_sort.c +105 -149
  67. data/ext/isomorfeus_ferret_ext/frt_store.c +13 -7
  68. data/ext/isomorfeus_ferret_ext/frt_store.h +10 -2
  69. data/ext/isomorfeus_ferret_ext/frt_threading.h +0 -1
  70. data/ext/isomorfeus_ferret_ext/isomorfeus_ferret.c +21 -109
  71. data/ext/isomorfeus_ferret_ext/isomorfeus_ferret.h +2 -32
  72. data/ext/isomorfeus_ferret_ext/lz4.c +2495 -0
  73. data/ext/isomorfeus_ferret_ext/lz4.h +774 -0
  74. data/ext/isomorfeus_ferret_ext/lz4frame.c +1899 -0
  75. data/ext/isomorfeus_ferret_ext/lz4frame.h +623 -0
  76. data/ext/isomorfeus_ferret_ext/lz4hc.c +1615 -0
  77. data/ext/isomorfeus_ferret_ext/lz4hc.h +413 -0
  78. data/ext/isomorfeus_ferret_ext/lz4xxhash.c +1030 -0
  79. data/ext/isomorfeus_ferret_ext/lz4xxhash.h +328 -0
  80. data/ext/isomorfeus_ferret_ext/stem_modules.h +0 -86
  81. data/ext/isomorfeus_ferret_ext/test.c +1 -2
  82. data/ext/isomorfeus_ferret_ext/test_1710.c +11 -12
  83. data/ext/isomorfeus_ferret_ext/test_analysis.c +590 -583
  84. data/ext/isomorfeus_ferret_ext/test_compound_io.c +1 -1
  85. data/ext/isomorfeus_ferret_ext/test_document.c +19 -15
  86. data/ext/isomorfeus_ferret_ext/test_except.c +1 -2
  87. data/ext/isomorfeus_ferret_ext/test_fields.c +59 -60
  88. data/ext/isomorfeus_ferret_ext/test_file_deleter.c +10 -27
  89. data/ext/isomorfeus_ferret_ext/test_filter.c +11 -8
  90. data/ext/isomorfeus_ferret_ext/test_hash.c +2 -2
  91. data/ext/isomorfeus_ferret_ext/test_hashset.c +1 -1
  92. data/ext/isomorfeus_ferret_ext/test_highlighter.c +15 -11
  93. data/ext/isomorfeus_ferret_ext/test_index.c +372 -365
  94. data/ext/isomorfeus_ferret_ext/test_q_const_score.c +5 -3
  95. data/ext/isomorfeus_ferret_ext/test_q_filtered.c +5 -3
  96. data/ext/isomorfeus_ferret_ext/test_q_fuzzy.c +13 -10
  97. data/ext/isomorfeus_ferret_ext/test_q_parser.c +45 -7
  98. data/ext/isomorfeus_ferret_ext/test_q_span.c +15 -12
  99. data/ext/isomorfeus_ferret_ext/test_ram_store.c +3 -3
  100. data/ext/isomorfeus_ferret_ext/test_search.c +60 -62
  101. data/ext/isomorfeus_ferret_ext/test_segments.c +5 -4
  102. data/ext/isomorfeus_ferret_ext/test_sort.c +17 -14
  103. data/ext/isomorfeus_ferret_ext/test_store.c +2 -0
  104. data/ext/isomorfeus_ferret_ext/test_term.c +3 -1
  105. data/ext/isomorfeus_ferret_ext/test_term_vectors.c +9 -10
  106. data/ext/isomorfeus_ferret_ext/test_test.c +1 -2
  107. data/ext/isomorfeus_ferret_ext/test_threading.c +9 -10
  108. data/ext/isomorfeus_ferret_ext/testhelper.c +1 -2
  109. data/lib/isomorfeus/ferret/version.rb +1 -1
  110. metadata +27 -57
  111. data/ext/isomorfeus_ferret_ext/email.rl +0 -21
  112. data/ext/isomorfeus_ferret_ext/frt_scanner.c +0 -900
  113. data/ext/isomorfeus_ferret_ext/frt_scanner.h +0 -28
  114. data/ext/isomorfeus_ferret_ext/frt_scanner_mb.c +0 -6706
  115. data/ext/isomorfeus_ferret_ext/frt_scanner_utf8.c +0 -4420
  116. data/ext/isomorfeus_ferret_ext/scanner.h +0 -28
  117. data/ext/isomorfeus_ferret_ext/scanner.in +0 -43
  118. data/ext/isomorfeus_ferret_ext/scanner.rl +0 -84
  119. data/ext/isomorfeus_ferret_ext/scanner_mb.rl +0 -200
  120. data/ext/isomorfeus_ferret_ext/scanner_utf8.rl +0 -85
  121. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_basque.c +0 -1167
  122. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_basque.h +0 -6
  123. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_catalan.c +0 -1433
  124. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_catalan.h +0 -6
  125. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_danish.c +0 -301
  126. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_danish.h +0 -6
  127. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_dutch.c +0 -590
  128. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_dutch.h +0 -6
  129. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_english.c +0 -1049
  130. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_english.h +0 -6
  131. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_finnish.c +0 -705
  132. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_finnish.h +0 -6
  133. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_french.c +0 -1239
  134. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_french.h +0 -6
  135. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_german.c +0 -477
  136. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_german.h +0 -6
  137. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_hungarian.c +0 -1217
  138. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_hungarian.h +0 -7
  139. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_indonesian.c +0 -394
  140. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_indonesian.h +0 -6
  141. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_irish.c +0 -457
  142. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_irish.h +0 -6
  143. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_italian.c +0 -1009
  144. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_italian.h +0 -6
  145. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_norwegian.c +0 -259
  146. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_norwegian.h +0 -6
  147. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_porter.c +0 -704
  148. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_porter.h +0 -6
  149. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_portuguese.c +0 -948
  150. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_portuguese.h +0 -6
  151. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_spanish.c +0 -1028
  152. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_spanish.h +0 -6
  153. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_swedish.c +0 -275
  154. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_swedish.h +0 -6
  155. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_2_hungarian.c +0 -849
  156. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_2_hungarian.h +0 -6
  157. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_2_romanian.c +0 -952
  158. data/ext/isomorfeus_ferret_ext/stem_ISO_8859_2_romanian.h +0 -6
  159. data/ext/isomorfeus_ferret_ext/stem_KOI8_R_russian.c +0 -669
  160. data/ext/isomorfeus_ferret_ext/stem_KOI8_R_russian.h +0 -6
  161. data/ext/isomorfeus_ferret_ext/stem_modules.txt +0 -63
  162. data/ext/isomorfeus_ferret_ext/uchar-ucs4.rl +0 -1854
  163. data/ext/isomorfeus_ferret_ext/uchar-utf8.rl +0 -1999
  164. data/ext/isomorfeus_ferret_ext/url.rl +0 -27
@@ -0,0 +1,1899 @@
1
+ /*
2
+ * LZ4 auto-framing library
3
+ * Copyright (C) 2011-2016, Yann Collet.
4
+ *
5
+ * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
6
+ *
7
+ * Redistribution and use in source and binary forms, with or without
8
+ * modification, are permitted provided that the following conditions are
9
+ * met:
10
+ *
11
+ * - Redistributions of source code must retain the above copyright
12
+ * notice, this list of conditions and the following disclaimer.
13
+ * - Redistributions in binary form must reproduce the above
14
+ * copyright notice, this list of conditions and the following disclaimer
15
+ * in the documentation and/or other materials provided with the
16
+ * distribution.
17
+ *
18
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
19
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
20
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
21
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
22
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
23
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
24
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
25
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
26
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
27
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
28
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
29
+ *
30
+ * You can contact the author at :
31
+ * - LZ4 homepage : http://www.lz4.org
32
+ * - LZ4 source repository : https://github.com/lz4/lz4
33
+ */
34
+
35
+ /* LZ4F is a stand-alone API to create LZ4-compressed Frames
36
+ * in full conformance with specification v1.6.1 .
37
+ * This library rely upon memory management capabilities (malloc, free)
38
+ * provided either by <stdlib.h>,
39
+ * or redirected towards another library of user's choice
40
+ * (see Memory Routines below).
41
+ */
42
+
43
+
44
+ /*-************************************
45
+ * Compiler Options
46
+ **************************************/
47
+ #ifdef _MSC_VER /* Visual Studio */
48
+ # pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */
49
+ #endif
50
+
51
+
52
+ /*-************************************
53
+ * Tuning parameters
54
+ **************************************/
55
+ /*
56
+ * LZ4F_HEAPMODE :
57
+ * Select how default compression functions will allocate memory for their hash table,
58
+ * in memory stack (0:default, fastest), or in memory heap (1:requires malloc()).
59
+ */
60
+ #ifndef LZ4F_HEAPMODE
61
+ # define LZ4F_HEAPMODE 0
62
+ #endif
63
+
64
+
65
+ /*-************************************
66
+ * Memory routines
67
+ **************************************/
68
+ /*
69
+ * User may redirect invocations of
70
+ * malloc(), calloc() and free()
71
+ * towards another library or solution of their choice
72
+ * by modifying below section.
73
+ */
74
+ #ifndef LZ4_SRC_INCLUDED /* avoid redefinition when sources are coalesced */
75
+ # include <stdlib.h> /* malloc, calloc, free */
76
+ # define ALLOC(s) malloc(s)
77
+ # define ALLOC_AND_ZERO(s) calloc(1,(s))
78
+ # define FREEMEM(p) free(p)
79
+ #endif
80
+
81
+ #include <string.h> /* memset, memcpy, memmove */
82
+ #ifndef LZ4_SRC_INCLUDED /* avoid redefinition when sources are coalesced */
83
+ # define MEM_INIT(p,v,s) memset((p),(v),(s))
84
+ #endif
85
+
86
+
87
+ /*-************************************
88
+ * Library declarations
89
+ **************************************/
90
+ #define LZ4F_STATIC_LINKING_ONLY
91
+ #include "lz4frame.h"
92
+ #define LZ4_STATIC_LINKING_ONLY
93
+ #include "lz4.h"
94
+ #define LZ4_HC_STATIC_LINKING_ONLY
95
+ #include "lz4hc.h"
96
+ #define XXH_STATIC_LINKING_ONLY
97
+ #include "lz4xxhash.h"
98
+
99
+
100
+ /*-************************************
101
+ * Debug
102
+ **************************************/
103
+ #if defined(LZ4_DEBUG) && (LZ4_DEBUG>=1)
104
+ # include <assert.h>
105
+ #else
106
+ # ifndef assert
107
+ # define assert(condition) ((void)0)
108
+ # endif
109
+ #endif
110
+
111
+ #define LZ4F_STATIC_ASSERT(c) { enum { LZ4F_static_assert = 1/(int)(!!(c)) }; } /* use only *after* variable declarations */
112
+
113
+ #if defined(LZ4_DEBUG) && (LZ4_DEBUG>=2) && !defined(DEBUGLOG)
114
+ # include <stdio.h>
115
+ static int g_debuglog_enable = 1;
116
+ # define DEBUGLOG(l, ...) { \
117
+ if ((g_debuglog_enable) && (l<=LZ4_DEBUG)) { \
118
+ fprintf(stderr, __FILE__ ": "); \
119
+ fprintf(stderr, __VA_ARGS__); \
120
+ fprintf(stderr, " \n"); \
121
+ } }
122
+ #else
123
+ # define DEBUGLOG(l, ...) {} /* disabled */
124
+ #endif
125
+
126
+
127
+ /*-************************************
128
+ * Basic Types
129
+ **************************************/
130
+ #if !defined (__VMS) && (defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) )
131
+ # include <stdint.h>
132
+ typedef uint8_t BYTE;
133
+ typedef uint16_t U16;
134
+ typedef uint32_t U32;
135
+ typedef int32_t S32;
136
+ typedef uint64_t U64;
137
+ #else
138
+ typedef unsigned char BYTE;
139
+ typedef unsigned short U16;
140
+ typedef unsigned int U32;
141
+ typedef signed int S32;
142
+ typedef unsigned long long U64;
143
+ #endif
144
+
145
+
146
+ /* unoptimized version; solves endianess & alignment issues */
147
+ static U32 LZ4F_readLE32 (const void* src)
148
+ {
149
+ const BYTE* const srcPtr = (const BYTE*)src;
150
+ U32 value32 = srcPtr[0];
151
+ value32 += ((U32)srcPtr[1])<< 8;
152
+ value32 += ((U32)srcPtr[2])<<16;
153
+ value32 += ((U32)srcPtr[3])<<24;
154
+ return value32;
155
+ }
156
+
157
+ static void LZ4F_writeLE32 (void* dst, U32 value32)
158
+ {
159
+ BYTE* const dstPtr = (BYTE*)dst;
160
+ dstPtr[0] = (BYTE)value32;
161
+ dstPtr[1] = (BYTE)(value32 >> 8);
162
+ dstPtr[2] = (BYTE)(value32 >> 16);
163
+ dstPtr[3] = (BYTE)(value32 >> 24);
164
+ }
165
+
166
+ static U64 LZ4F_readLE64 (const void* src)
167
+ {
168
+ const BYTE* const srcPtr = (const BYTE*)src;
169
+ U64 value64 = srcPtr[0];
170
+ value64 += ((U64)srcPtr[1]<<8);
171
+ value64 += ((U64)srcPtr[2]<<16);
172
+ value64 += ((U64)srcPtr[3]<<24);
173
+ value64 += ((U64)srcPtr[4]<<32);
174
+ value64 += ((U64)srcPtr[5]<<40);
175
+ value64 += ((U64)srcPtr[6]<<48);
176
+ value64 += ((U64)srcPtr[7]<<56);
177
+ return value64;
178
+ }
179
+
180
+ static void LZ4F_writeLE64 (void* dst, U64 value64)
181
+ {
182
+ BYTE* const dstPtr = (BYTE*)dst;
183
+ dstPtr[0] = (BYTE)value64;
184
+ dstPtr[1] = (BYTE)(value64 >> 8);
185
+ dstPtr[2] = (BYTE)(value64 >> 16);
186
+ dstPtr[3] = (BYTE)(value64 >> 24);
187
+ dstPtr[4] = (BYTE)(value64 >> 32);
188
+ dstPtr[5] = (BYTE)(value64 >> 40);
189
+ dstPtr[6] = (BYTE)(value64 >> 48);
190
+ dstPtr[7] = (BYTE)(value64 >> 56);
191
+ }
192
+
193
+
194
+ /*-************************************
195
+ * Constants
196
+ **************************************/
197
+ #ifndef LZ4_SRC_INCLUDED /* avoid double definition */
198
+ # define KB *(1<<10)
199
+ # define MB *(1<<20)
200
+ # define GB *(1<<30)
201
+ #endif
202
+
203
+ #define _1BIT 0x01
204
+ #define _2BITS 0x03
205
+ #define _3BITS 0x07
206
+ #define _4BITS 0x0F
207
+ #define _8BITS 0xFF
208
+
209
+ #define LZ4F_MAGIC_SKIPPABLE_START 0x184D2A50U
210
+ #define LZ4F_MAGICNUMBER 0x184D2204U
211
+ #define LZ4F_BLOCKUNCOMPRESSED_FLAG 0x80000000U
212
+ #define LZ4F_BLOCKSIZEID_DEFAULT LZ4F_max64KB
213
+
214
+ static const size_t minFHSize = LZ4F_HEADER_SIZE_MIN; /* 7 */
215
+ static const size_t maxFHSize = LZ4F_HEADER_SIZE_MAX; /* 19 */
216
+ static const size_t BHSize = LZ4F_BLOCK_HEADER_SIZE; /* block header : size, and compress flag */
217
+ static const size_t BFSize = LZ4F_BLOCK_CHECKSUM_SIZE; /* block footer : checksum (optional) */
218
+
219
+
220
+ /*-************************************
221
+ * Structures and local types
222
+ **************************************/
223
+ typedef struct LZ4F_cctx_s
224
+ {
225
+ LZ4F_preferences_t prefs;
226
+ U32 version;
227
+ U32 cStage;
228
+ const LZ4F_CDict* cdict;
229
+ size_t maxBlockSize;
230
+ size_t maxBufferSize;
231
+ BYTE* tmpBuff;
232
+ BYTE* tmpIn;
233
+ size_t tmpInSize;
234
+ U64 totalInSize;
235
+ XXH32_state_t xxh;
236
+ void* lz4CtxPtr;
237
+ U16 lz4CtxAlloc; /* sized for: 0 = none, 1 = lz4 ctx, 2 = lz4hc ctx */
238
+ U16 lz4CtxState; /* in use as: 0 = none, 1 = lz4 ctx, 2 = lz4hc ctx */
239
+ } LZ4F_cctx_t;
240
+
241
+
242
+ /*-************************************
243
+ * Error management
244
+ **************************************/
245
+ #define LZ4F_GENERATE_STRING(STRING) #STRING,
246
+ static const char* LZ4F_errorStrings[] = { LZ4F_LIST_ERRORS(LZ4F_GENERATE_STRING) };
247
+
248
+
249
+ unsigned LZ4F_isError(LZ4F_errorCode_t code)
250
+ {
251
+ return (code > (LZ4F_errorCode_t)(-LZ4F_ERROR_maxCode));
252
+ }
253
+
254
+ const char* LZ4F_getErrorName(LZ4F_errorCode_t code)
255
+ {
256
+ static const char* codeError = "Unspecified error code";
257
+ if (LZ4F_isError(code)) return LZ4F_errorStrings[-(int)(code)];
258
+ return codeError;
259
+ }
260
+
261
+ LZ4F_errorCodes LZ4F_getErrorCode(size_t functionResult)
262
+ {
263
+ if (!LZ4F_isError(functionResult)) return LZ4F_OK_NoError;
264
+ return (LZ4F_errorCodes)(-(ptrdiff_t)functionResult);
265
+ }
266
+
267
+ static LZ4F_errorCode_t err0r(LZ4F_errorCodes code)
268
+ {
269
+ /* A compilation error here means sizeof(ptrdiff_t) is not large enough */
270
+ LZ4F_STATIC_ASSERT(sizeof(ptrdiff_t) >= sizeof(size_t));
271
+ return (LZ4F_errorCode_t)-(ptrdiff_t)code;
272
+ }
273
+
274
+ unsigned LZ4F_getVersion(void) { return LZ4F_VERSION; }
275
+
276
+ int LZ4F_compressionLevel_max(void) { return LZ4HC_CLEVEL_MAX; }
277
+
278
+ size_t LZ4F_getBlockSize(unsigned blockSizeID)
279
+ {
280
+ static const size_t blockSizes[4] = { 64 KB, 256 KB, 1 MB, 4 MB };
281
+
282
+ if (blockSizeID == 0) blockSizeID = LZ4F_BLOCKSIZEID_DEFAULT;
283
+ if (blockSizeID < LZ4F_max64KB || blockSizeID > LZ4F_max4MB)
284
+ return err0r(LZ4F_ERROR_maxBlockSize_invalid);
285
+ blockSizeID -= LZ4F_max64KB;
286
+ return blockSizes[blockSizeID];
287
+ }
288
+
289
+ /*-************************************
290
+ * Private functions
291
+ **************************************/
292
+ #define MIN(a,b) ( (a) < (b) ? (a) : (b) )
293
+
294
+ static BYTE LZ4F_headerChecksum (const void* header, size_t length)
295
+ {
296
+ U32 const xxh = XXH32(header, length, 0);
297
+ return (BYTE)(xxh >> 8);
298
+ }
299
+
300
+
301
+ /*-************************************
302
+ * Simple-pass compression functions
303
+ **************************************/
304
+ static LZ4F_blockSizeID_t LZ4F_optimalBSID(const LZ4F_blockSizeID_t requestedBSID,
305
+ const size_t srcSize)
306
+ {
307
+ LZ4F_blockSizeID_t proposedBSID = LZ4F_max64KB;
308
+ size_t maxBlockSize = 64 KB;
309
+ while (requestedBSID > proposedBSID) {
310
+ if (srcSize <= maxBlockSize)
311
+ return proposedBSID;
312
+ proposedBSID = (LZ4F_blockSizeID_t)((int)proposedBSID + 1);
313
+ maxBlockSize <<= 2;
314
+ }
315
+ return requestedBSID;
316
+ }
317
+
318
+ /*! LZ4F_compressBound_internal() :
319
+ * Provides dstCapacity given a srcSize to guarantee operation success in worst case situations.
320
+ * prefsPtr is optional : if NULL is provided, preferences will be set to cover worst case scenario.
321
+ * @return is always the same for a srcSize and prefsPtr, so it can be relied upon to size reusable buffers.
322
+ * When srcSize==0, LZ4F_compressBound() provides an upper bound for LZ4F_flush() and LZ4F_compressEnd() operations.
323
+ */
324
+ static size_t LZ4F_compressBound_internal(size_t srcSize,
325
+ const LZ4F_preferences_t* preferencesPtr,
326
+ size_t alreadyBuffered)
327
+ {
328
+ LZ4F_preferences_t prefsNull = LZ4F_INIT_PREFERENCES;
329
+ prefsNull.frameInfo.contentChecksumFlag = LZ4F_contentChecksumEnabled; /* worst case */
330
+ prefsNull.frameInfo.blockChecksumFlag = LZ4F_blockChecksumEnabled; /* worst case */
331
+ { const LZ4F_preferences_t* const prefsPtr = (preferencesPtr==NULL) ? &prefsNull : preferencesPtr;
332
+ U32 const flush = prefsPtr->autoFlush | (srcSize==0);
333
+ LZ4F_blockSizeID_t const blockID = prefsPtr->frameInfo.blockSizeID;
334
+ size_t const blockSize = LZ4F_getBlockSize(blockID);
335
+ size_t const maxBuffered = blockSize - 1;
336
+ size_t const bufferedSize = MIN(alreadyBuffered, maxBuffered);
337
+ size_t const maxSrcSize = srcSize + bufferedSize;
338
+ unsigned const nbFullBlocks = (unsigned)(maxSrcSize / blockSize);
339
+ size_t const partialBlockSize = maxSrcSize & (blockSize-1);
340
+ size_t const lastBlockSize = flush ? partialBlockSize : 0;
341
+ unsigned const nbBlocks = nbFullBlocks + (lastBlockSize>0);
342
+
343
+ size_t const blockCRCSize = BFSize * prefsPtr->frameInfo.blockChecksumFlag;
344
+ size_t const frameEnd = BHSize + (prefsPtr->frameInfo.contentChecksumFlag*BFSize);
345
+
346
+ return ((BHSize + blockCRCSize) * nbBlocks) +
347
+ (blockSize * nbFullBlocks) + lastBlockSize + frameEnd;
348
+ }
349
+ }
350
+
351
+ size_t LZ4F_compressFrameBound(size_t srcSize, const LZ4F_preferences_t* preferencesPtr)
352
+ {
353
+ LZ4F_preferences_t prefs;
354
+ size_t const headerSize = maxFHSize; /* max header size, including optional fields */
355
+
356
+ if (preferencesPtr!=NULL) prefs = *preferencesPtr;
357
+ else MEM_INIT(&prefs, 0, sizeof(prefs));
358
+ prefs.autoFlush = 1;
359
+
360
+ return headerSize + LZ4F_compressBound_internal(srcSize, &prefs, 0);;
361
+ }
362
+
363
+
364
+ /*! LZ4F_compressFrame_usingCDict() :
365
+ * Compress srcBuffer using a dictionary, in a single step.
366
+ * cdict can be NULL, in which case, no dictionary is used.
367
+ * dstBuffer MUST be >= LZ4F_compressFrameBound(srcSize, preferencesPtr).
368
+ * The LZ4F_preferences_t structure is optional : you may provide NULL as argument,
369
+ * however, it's the only way to provide a dictID, so it's not recommended.
370
+ * @return : number of bytes written into dstBuffer,
371
+ * or an error code if it fails (can be tested using LZ4F_isError())
372
+ */
373
+ size_t LZ4F_compressFrame_usingCDict(LZ4F_cctx* cctx,
374
+ void* dstBuffer, size_t dstCapacity,
375
+ const void* srcBuffer, size_t srcSize,
376
+ const LZ4F_CDict* cdict,
377
+ const LZ4F_preferences_t* preferencesPtr)
378
+ {
379
+ LZ4F_preferences_t prefs;
380
+ LZ4F_compressOptions_t options;
381
+ BYTE* const dstStart = (BYTE*) dstBuffer;
382
+ BYTE* dstPtr = dstStart;
383
+ BYTE* const dstEnd = dstStart + dstCapacity;
384
+
385
+ if (preferencesPtr!=NULL)
386
+ prefs = *preferencesPtr;
387
+ else
388
+ MEM_INIT(&prefs, 0, sizeof(prefs));
389
+ if (prefs.frameInfo.contentSize != 0)
390
+ prefs.frameInfo.contentSize = (U64)srcSize; /* auto-correct content size if selected (!=0) */
391
+
392
+ prefs.frameInfo.blockSizeID = LZ4F_optimalBSID(prefs.frameInfo.blockSizeID, srcSize);
393
+ prefs.autoFlush = 1;
394
+ if (srcSize <= LZ4F_getBlockSize(prefs.frameInfo.blockSizeID))
395
+ prefs.frameInfo.blockMode = LZ4F_blockIndependent; /* only one block => no need for inter-block link */
396
+
397
+ MEM_INIT(&options, 0, sizeof(options));
398
+ options.stableSrc = 1;
399
+
400
+ if (dstCapacity < LZ4F_compressFrameBound(srcSize, &prefs)) /* condition to guarantee success */
401
+ return err0r(LZ4F_ERROR_dstMaxSize_tooSmall);
402
+
403
+ { size_t const headerSize = LZ4F_compressBegin_usingCDict(cctx, dstBuffer, dstCapacity, cdict, &prefs); /* write header */
404
+ if (LZ4F_isError(headerSize)) return headerSize;
405
+ dstPtr += headerSize; /* header size */ }
406
+
407
+ assert(dstEnd >= dstPtr);
408
+ { size_t const cSize = LZ4F_compressUpdate(cctx, dstPtr, (size_t)(dstEnd-dstPtr), srcBuffer, srcSize, &options);
409
+ if (LZ4F_isError(cSize)) return cSize;
410
+ dstPtr += cSize; }
411
+
412
+ assert(dstEnd >= dstPtr);
413
+ { size_t const tailSize = LZ4F_compressEnd(cctx, dstPtr, (size_t)(dstEnd-dstPtr), &options); /* flush last block, and generate suffix */
414
+ if (LZ4F_isError(tailSize)) return tailSize;
415
+ dstPtr += tailSize; }
416
+
417
+ assert(dstEnd >= dstStart);
418
+ return (size_t)(dstPtr - dstStart);
419
+ }
420
+
421
+
422
+ /*! LZ4F_compressFrame() :
423
+ * Compress an entire srcBuffer into a valid LZ4 frame, in a single step.
424
+ * dstBuffer MUST be >= LZ4F_compressFrameBound(srcSize, preferencesPtr).
425
+ * The LZ4F_preferences_t structure is optional : you can provide NULL as argument. All preferences will be set to default.
426
+ * @return : number of bytes written into dstBuffer.
427
+ * or an error code if it fails (can be tested using LZ4F_isError())
428
+ */
429
+ size_t LZ4F_compressFrame(void* dstBuffer, size_t dstCapacity,
430
+ const void* srcBuffer, size_t srcSize,
431
+ const LZ4F_preferences_t* preferencesPtr)
432
+ {
433
+ size_t result;
434
+ #if (LZ4F_HEAPMODE)
435
+ LZ4F_cctx_t *cctxPtr;
436
+ result = LZ4F_createCompressionContext(&cctxPtr, LZ4F_VERSION);
437
+ if (LZ4F_isError(result)) return result;
438
+ #else
439
+ LZ4F_cctx_t cctx;
440
+ LZ4_stream_t lz4ctx;
441
+ LZ4F_cctx_t *cctxPtr = &cctx;
442
+
443
+ DEBUGLOG(4, "LZ4F_compressFrame");
444
+ MEM_INIT(&cctx, 0, sizeof(cctx));
445
+ cctx.version = LZ4F_VERSION;
446
+ cctx.maxBufferSize = 5 MB; /* mess with real buffer size to prevent dynamic allocation; works only because autoflush==1 & stableSrc==1 */
447
+ if (preferencesPtr == NULL ||
448
+ preferencesPtr->compressionLevel < LZ4HC_CLEVEL_MIN)
449
+ {
450
+ LZ4_initStream(&lz4ctx, sizeof(lz4ctx));
451
+ cctxPtr->lz4CtxPtr = &lz4ctx;
452
+ cctxPtr->lz4CtxAlloc = 1;
453
+ cctxPtr->lz4CtxState = 1;
454
+ }
455
+ #endif
456
+
457
+ result = LZ4F_compressFrame_usingCDict(cctxPtr, dstBuffer, dstCapacity,
458
+ srcBuffer, srcSize,
459
+ NULL, preferencesPtr);
460
+
461
+ #if (LZ4F_HEAPMODE)
462
+ LZ4F_freeCompressionContext(cctxPtr);
463
+ #else
464
+ if (preferencesPtr != NULL &&
465
+ preferencesPtr->compressionLevel >= LZ4HC_CLEVEL_MIN)
466
+ {
467
+ FREEMEM(cctxPtr->lz4CtxPtr);
468
+ }
469
+ #endif
470
+ return result;
471
+ }
472
+
473
+
474
+ /*-***************************************************
475
+ * Dictionary compression
476
+ *****************************************************/
477
+
478
+ struct LZ4F_CDict_s {
479
+ void* dictContent;
480
+ LZ4_stream_t* fastCtx;
481
+ LZ4_streamHC_t* HCCtx;
482
+ }; /* typedef'd to LZ4F_CDict within lz4frame_static.h */
483
+
484
+ /*! LZ4F_createCDict() :
485
+ * When compressing multiple messages / blocks with the same dictionary, it's recommended to load it just once.
486
+ * LZ4F_createCDict() will create a digested dictionary, ready to start future compression operations without startup delay.
487
+ * LZ4F_CDict can be created once and shared by multiple threads concurrently, since its usage is read-only.
488
+ * `dictBuffer` can be released after LZ4F_CDict creation, since its content is copied within CDict
489
+ * @return : digested dictionary for compression, or NULL if failed */
490
+ LZ4F_CDict* LZ4F_createCDict(const void* dictBuffer, size_t dictSize)
491
+ {
492
+ const char* dictStart = (const char*)dictBuffer;
493
+ LZ4F_CDict* cdict = (LZ4F_CDict*) ALLOC(sizeof(*cdict));
494
+ DEBUGLOG(4, "LZ4F_createCDict");
495
+ if (!cdict) return NULL;
496
+ if (dictSize > 64 KB) {
497
+ dictStart += dictSize - 64 KB;
498
+ dictSize = 64 KB;
499
+ }
500
+ cdict->dictContent = ALLOC(dictSize);
501
+ cdict->fastCtx = LZ4_createStream();
502
+ cdict->HCCtx = LZ4_createStreamHC();
503
+ if (!cdict->dictContent || !cdict->fastCtx || !cdict->HCCtx) {
504
+ LZ4F_freeCDict(cdict);
505
+ return NULL;
506
+ }
507
+ memcpy(cdict->dictContent, dictStart, dictSize);
508
+ LZ4_loadDict (cdict->fastCtx, (const char*)cdict->dictContent, (int)dictSize);
509
+ LZ4_setCompressionLevel(cdict->HCCtx, LZ4HC_CLEVEL_DEFAULT);
510
+ LZ4_loadDictHC(cdict->HCCtx, (const char*)cdict->dictContent, (int)dictSize);
511
+ return cdict;
512
+ }
513
+
514
+ void LZ4F_freeCDict(LZ4F_CDict* cdict)
515
+ {
516
+ if (cdict==NULL) return; /* support free on NULL */
517
+ FREEMEM(cdict->dictContent);
518
+ LZ4_freeStream(cdict->fastCtx);
519
+ LZ4_freeStreamHC(cdict->HCCtx);
520
+ FREEMEM(cdict);
521
+ }
522
+
523
+
524
+ /*-*********************************
525
+ * Advanced compression functions
526
+ ***********************************/
527
+
528
+ /*! LZ4F_createCompressionContext() :
529
+ * The first thing to do is to create a compressionContext object, which will be used in all compression operations.
530
+ * This is achieved using LZ4F_createCompressionContext(), which takes as argument a version and an LZ4F_preferences_t structure.
531
+ * The version provided MUST be LZ4F_VERSION. It is intended to track potential incompatible differences between different binaries.
532
+ * The function will provide a pointer to an allocated LZ4F_compressionContext_t object.
533
+ * If the result LZ4F_errorCode_t is not OK_NoError, there was an error during context creation.
534
+ * Object can release its memory using LZ4F_freeCompressionContext();
535
+ */
536
+ LZ4F_errorCode_t LZ4F_createCompressionContext(LZ4F_cctx** LZ4F_compressionContextPtr, unsigned version)
537
+ {
538
+ LZ4F_cctx_t* const cctxPtr = (LZ4F_cctx_t*)ALLOC_AND_ZERO(sizeof(LZ4F_cctx_t));
539
+ if (cctxPtr==NULL) return err0r(LZ4F_ERROR_allocation_failed);
540
+
541
+ cctxPtr->version = version;
542
+ cctxPtr->cStage = 0; /* Next stage : init stream */
543
+
544
+ *LZ4F_compressionContextPtr = cctxPtr;
545
+
546
+ return LZ4F_OK_NoError;
547
+ }
548
+
549
+
550
+ LZ4F_errorCode_t LZ4F_freeCompressionContext(LZ4F_cctx* cctxPtr)
551
+ {
552
+ if (cctxPtr != NULL) { /* support free on NULL */
553
+ FREEMEM(cctxPtr->lz4CtxPtr); /* note: LZ4_streamHC_t and LZ4_stream_t are simple POD types */
554
+ FREEMEM(cctxPtr->tmpBuff);
555
+ FREEMEM(cctxPtr);
556
+ }
557
+
558
+ return LZ4F_OK_NoError;
559
+ }
560
+
561
+
562
+ /**
563
+ * This function prepares the internal LZ4(HC) stream for a new compression,
564
+ * resetting the context and attaching the dictionary, if there is one.
565
+ *
566
+ * It needs to be called at the beginning of each independent compression
567
+ * stream (i.e., at the beginning of a frame in blockLinked mode, or at the
568
+ * beginning of each block in blockIndependent mode).
569
+ */
570
+ static void LZ4F_initStream(void* ctx,
571
+ const LZ4F_CDict* cdict,
572
+ int level,
573
+ LZ4F_blockMode_t blockMode) {
574
+ if (level < LZ4HC_CLEVEL_MIN) {
575
+ if (cdict != NULL || blockMode == LZ4F_blockLinked) {
576
+ /* In these cases, we will call LZ4_compress_fast_continue(),
577
+ * which needs an already reset context. Otherwise, we'll call a
578
+ * one-shot API. The non-continued APIs internally perform their own
579
+ * resets at the beginning of their calls, where they know what
580
+ * tableType they need the context to be in. So in that case this
581
+ * would be misguided / wasted work. */
582
+ LZ4_resetStream_fast((LZ4_stream_t*)ctx);
583
+ }
584
+ LZ4_attach_dictionary((LZ4_stream_t *)ctx, cdict ? cdict->fastCtx : NULL);
585
+ } else {
586
+ LZ4_resetStreamHC_fast((LZ4_streamHC_t*)ctx, level);
587
+ LZ4_attach_HC_dictionary((LZ4_streamHC_t *)ctx, cdict ? cdict->HCCtx : NULL);
588
+ }
589
+ }
590
+
591
+
592
+ /*! LZ4F_compressBegin_usingCDict() :
593
+ * init streaming compression and writes frame header into dstBuffer.
594
+ * dstBuffer must be >= LZ4F_HEADER_SIZE_MAX bytes.
595
+ * @return : number of bytes written into dstBuffer for the header
596
+ * or an error code (can be tested using LZ4F_isError())
597
+ */
598
+ size_t LZ4F_compressBegin_usingCDict(LZ4F_cctx* cctxPtr,
599
+ void* dstBuffer, size_t dstCapacity,
600
+ const LZ4F_CDict* cdict,
601
+ const LZ4F_preferences_t* preferencesPtr)
602
+ {
603
+ LZ4F_preferences_t prefNull;
604
+ BYTE* const dstStart = (BYTE*)dstBuffer;
605
+ BYTE* dstPtr = dstStart;
606
+ BYTE* headerStart;
607
+
608
+ if (dstCapacity < maxFHSize) return err0r(LZ4F_ERROR_dstMaxSize_tooSmall);
609
+ MEM_INIT(&prefNull, 0, sizeof(prefNull));
610
+ if (preferencesPtr == NULL) preferencesPtr = &prefNull;
611
+ cctxPtr->prefs = *preferencesPtr;
612
+
613
+ /* Ctx Management */
614
+ { U16 const ctxTypeID = (cctxPtr->prefs.compressionLevel < LZ4HC_CLEVEL_MIN) ? 1 : 2;
615
+ if (cctxPtr->lz4CtxAlloc < ctxTypeID) {
616
+ FREEMEM(cctxPtr->lz4CtxPtr);
617
+ if (cctxPtr->prefs.compressionLevel < LZ4HC_CLEVEL_MIN) {
618
+ cctxPtr->lz4CtxPtr = LZ4_createStream();
619
+ } else {
620
+ cctxPtr->lz4CtxPtr = LZ4_createStreamHC();
621
+ }
622
+ if (cctxPtr->lz4CtxPtr == NULL)
623
+ return err0r(LZ4F_ERROR_allocation_failed);
624
+ cctxPtr->lz4CtxAlloc = ctxTypeID;
625
+ cctxPtr->lz4CtxState = ctxTypeID;
626
+ } else if (cctxPtr->lz4CtxState != ctxTypeID) {
627
+ /* otherwise, a sufficient buffer is allocated, but we need to
628
+ * reset it to the correct context type */
629
+ if (cctxPtr->prefs.compressionLevel < LZ4HC_CLEVEL_MIN) {
630
+ LZ4_initStream((LZ4_stream_t *) cctxPtr->lz4CtxPtr, sizeof (LZ4_stream_t));
631
+ } else {
632
+ LZ4_initStreamHC((LZ4_streamHC_t *) cctxPtr->lz4CtxPtr, sizeof(LZ4_streamHC_t));
633
+ LZ4_setCompressionLevel((LZ4_streamHC_t *) cctxPtr->lz4CtxPtr, cctxPtr->prefs.compressionLevel);
634
+ }
635
+ cctxPtr->lz4CtxState = ctxTypeID;
636
+ }
637
+ }
638
+
639
+ /* Buffer Management */
640
+ if (cctxPtr->prefs.frameInfo.blockSizeID == 0)
641
+ cctxPtr->prefs.frameInfo.blockSizeID = LZ4F_BLOCKSIZEID_DEFAULT;
642
+ cctxPtr->maxBlockSize = LZ4F_getBlockSize(cctxPtr->prefs.frameInfo.blockSizeID);
643
+
644
+ { size_t const requiredBuffSize = preferencesPtr->autoFlush ?
645
+ ((cctxPtr->prefs.frameInfo.blockMode == LZ4F_blockLinked) ? 64 KB : 0) : /* only needs past data up to window size */
646
+ cctxPtr->maxBlockSize + ((cctxPtr->prefs.frameInfo.blockMode == LZ4F_blockLinked) ? 128 KB : 0);
647
+
648
+ if (cctxPtr->maxBufferSize < requiredBuffSize) {
649
+ cctxPtr->maxBufferSize = 0;
650
+ FREEMEM(cctxPtr->tmpBuff);
651
+ cctxPtr->tmpBuff = (BYTE*)ALLOC_AND_ZERO(requiredBuffSize);
652
+ if (cctxPtr->tmpBuff == NULL) return err0r(LZ4F_ERROR_allocation_failed);
653
+ cctxPtr->maxBufferSize = requiredBuffSize;
654
+ } }
655
+ cctxPtr->tmpIn = cctxPtr->tmpBuff;
656
+ cctxPtr->tmpInSize = 0;
657
+ (void)XXH32_reset(&(cctxPtr->xxh), 0);
658
+
659
+ /* context init */
660
+ cctxPtr->cdict = cdict;
661
+ if (cctxPtr->prefs.frameInfo.blockMode == LZ4F_blockLinked) {
662
+ /* frame init only for blockLinked : blockIndependent will be init at each block */
663
+ LZ4F_initStream(cctxPtr->lz4CtxPtr, cdict, cctxPtr->prefs.compressionLevel, LZ4F_blockLinked);
664
+ }
665
+ if (preferencesPtr->compressionLevel >= LZ4HC_CLEVEL_MIN) {
666
+ LZ4_favorDecompressionSpeed((LZ4_streamHC_t*)cctxPtr->lz4CtxPtr, (int)preferencesPtr->favorDecSpeed);
667
+ }
668
+
669
+ /* Magic Number */
670
+ LZ4F_writeLE32(dstPtr, LZ4F_MAGICNUMBER);
671
+ dstPtr += 4;
672
+ headerStart = dstPtr;
673
+
674
+ /* FLG Byte */
675
+ *dstPtr++ = (BYTE)(((1 & _2BITS) << 6) /* Version('01') */
676
+ + ((cctxPtr->prefs.frameInfo.blockMode & _1BIT ) << 5)
677
+ + ((cctxPtr->prefs.frameInfo.blockChecksumFlag & _1BIT ) << 4)
678
+ + ((unsigned)(cctxPtr->prefs.frameInfo.contentSize > 0) << 3)
679
+ + ((cctxPtr->prefs.frameInfo.contentChecksumFlag & _1BIT ) << 2)
680
+ + (cctxPtr->prefs.frameInfo.dictID > 0) );
681
+ /* BD Byte */
682
+ *dstPtr++ = (BYTE)((cctxPtr->prefs.frameInfo.blockSizeID & _3BITS) << 4);
683
+ /* Optional Frame content size field */
684
+ if (cctxPtr->prefs.frameInfo.contentSize) {
685
+ LZ4F_writeLE64(dstPtr, cctxPtr->prefs.frameInfo.contentSize);
686
+ dstPtr += 8;
687
+ cctxPtr->totalInSize = 0;
688
+ }
689
+ /* Optional dictionary ID field */
690
+ if (cctxPtr->prefs.frameInfo.dictID) {
691
+ LZ4F_writeLE32(dstPtr, cctxPtr->prefs.frameInfo.dictID);
692
+ dstPtr += 4;
693
+ }
694
+ /* Header CRC Byte */
695
+ *dstPtr = LZ4F_headerChecksum(headerStart, (size_t)(dstPtr - headerStart));
696
+ dstPtr++;
697
+
698
+ cctxPtr->cStage = 1; /* header written, now request input data block */
699
+ return (size_t)(dstPtr - dstStart);
700
+ }
701
+
702
+
703
+ /*! LZ4F_compressBegin() :
704
+ * init streaming compression and writes frame header into dstBuffer.
705
+ * dstBuffer must be >= LZ4F_HEADER_SIZE_MAX bytes.
706
+ * preferencesPtr can be NULL, in which case default parameters are selected.
707
+ * @return : number of bytes written into dstBuffer for the header
708
+ * or an error code (can be tested using LZ4F_isError())
709
+ */
710
+ size_t LZ4F_compressBegin(LZ4F_cctx* cctxPtr,
711
+ void* dstBuffer, size_t dstCapacity,
712
+ const LZ4F_preferences_t* preferencesPtr)
713
+ {
714
+ return LZ4F_compressBegin_usingCDict(cctxPtr, dstBuffer, dstCapacity,
715
+ NULL, preferencesPtr);
716
+ }
717
+
718
+
719
+ /* LZ4F_compressBound() :
720
+ * @return minimum capacity of dstBuffer for a given srcSize to handle worst case scenario.
721
+ * LZ4F_preferences_t structure is optional : if NULL, preferences will be set to cover worst case scenario.
722
+ * This function cannot fail.
723
+ */
724
+ size_t LZ4F_compressBound(size_t srcSize, const LZ4F_preferences_t* preferencesPtr)
725
+ {
726
+ if (preferencesPtr && preferencesPtr->autoFlush) {
727
+ return LZ4F_compressBound_internal(srcSize, preferencesPtr, 0);
728
+ }
729
+ return LZ4F_compressBound_internal(srcSize, preferencesPtr, (size_t)-1);
730
+ }
731
+
732
+
733
+ typedef int (*compressFunc_t)(void* ctx, const char* src, char* dst, int srcSize, int dstSize, int level, const LZ4F_CDict* cdict);
734
+
735
+
736
+ /*! LZ4F_makeBlock():
737
+ * compress a single block, add header and optional checksum.
738
+ * assumption : dst buffer capacity is >= BHSize + srcSize + crcSize
739
+ */
740
+ static size_t LZ4F_makeBlock(void* dst,
741
+ const void* src, size_t srcSize,
742
+ compressFunc_t compress, void* lz4ctx, int level,
743
+ const LZ4F_CDict* cdict,
744
+ LZ4F_blockChecksum_t crcFlag)
745
+ {
746
+ BYTE* const cSizePtr = (BYTE*)dst;
747
+ U32 cSize = (U32)compress(lz4ctx, (const char*)src, (char*)(cSizePtr+BHSize),
748
+ (int)(srcSize), (int)(srcSize-1),
749
+ level, cdict);
750
+ if (cSize == 0) { /* compression failed */
751
+ DEBUGLOG(5, "LZ4F_makeBlock: compression failed, creating a raw block (size %u)", (U32)srcSize);
752
+ cSize = (U32)srcSize;
753
+ LZ4F_writeLE32(cSizePtr, cSize | LZ4F_BLOCKUNCOMPRESSED_FLAG);
754
+ memcpy(cSizePtr+BHSize, src, srcSize);
755
+ } else {
756
+ LZ4F_writeLE32(cSizePtr, cSize);
757
+ }
758
+ if (crcFlag) {
759
+ U32 const crc32 = XXH32(cSizePtr+BHSize, cSize, 0); /* checksum of compressed data */
760
+ LZ4F_writeLE32(cSizePtr+BHSize+cSize, crc32);
761
+ }
762
+ return BHSize + cSize + ((U32)crcFlag)*BFSize;
763
+ }
764
+
765
+
766
+ static int LZ4F_compressBlock(void* ctx, const char* src, char* dst, int srcSize, int dstCapacity, int level, const LZ4F_CDict* cdict)
767
+ {
768
+ int const acceleration = (level < 0) ? -level + 1 : 1;
769
+ LZ4F_initStream(ctx, cdict, level, LZ4F_blockIndependent);
770
+ if (cdict) {
771
+ return LZ4_compress_fast_continue((LZ4_stream_t*)ctx, src, dst, srcSize, dstCapacity, acceleration);
772
+ } else {
773
+ return LZ4_compress_fast_extState_fastReset(ctx, src, dst, srcSize, dstCapacity, acceleration);
774
+ }
775
+ }
776
+
777
+ static int LZ4F_compressBlock_continue(void* ctx, const char* src, char* dst, int srcSize, int dstCapacity, int level, const LZ4F_CDict* cdict)
778
+ {
779
+ int const acceleration = (level < 0) ? -level + 1 : 1;
780
+ (void)cdict; /* init once at beginning of frame */
781
+ return LZ4_compress_fast_continue((LZ4_stream_t*)ctx, src, dst, srcSize, dstCapacity, acceleration);
782
+ }
783
+
784
+ static int LZ4F_compressBlockHC(void* ctx, const char* src, char* dst, int srcSize, int dstCapacity, int level, const LZ4F_CDict* cdict)
785
+ {
786
+ LZ4F_initStream(ctx, cdict, level, LZ4F_blockIndependent);
787
+ if (cdict) {
788
+ return LZ4_compress_HC_continue((LZ4_streamHC_t*)ctx, src, dst, srcSize, dstCapacity);
789
+ }
790
+ return LZ4_compress_HC_extStateHC_fastReset(ctx, src, dst, srcSize, dstCapacity, level);
791
+ }
792
+
793
+ static int LZ4F_compressBlockHC_continue(void* ctx, const char* src, char* dst, int srcSize, int dstCapacity, int level, const LZ4F_CDict* cdict)
794
+ {
795
+ (void)level; (void)cdict; /* init once at beginning of frame */
796
+ return LZ4_compress_HC_continue((LZ4_streamHC_t*)ctx, src, dst, srcSize, dstCapacity);
797
+ }
798
+
799
+ static compressFunc_t LZ4F_selectCompression(LZ4F_blockMode_t blockMode, int level)
800
+ {
801
+ if (level < LZ4HC_CLEVEL_MIN) {
802
+ if (blockMode == LZ4F_blockIndependent) return LZ4F_compressBlock;
803
+ return LZ4F_compressBlock_continue;
804
+ }
805
+ if (blockMode == LZ4F_blockIndependent) return LZ4F_compressBlockHC;
806
+ return LZ4F_compressBlockHC_continue;
807
+ }
808
+
809
+ static int LZ4F_localSaveDict(LZ4F_cctx_t* cctxPtr)
810
+ {
811
+ if (cctxPtr->prefs.compressionLevel < LZ4HC_CLEVEL_MIN)
812
+ return LZ4_saveDict ((LZ4_stream_t*)(cctxPtr->lz4CtxPtr), (char*)(cctxPtr->tmpBuff), 64 KB);
813
+ return LZ4_saveDictHC ((LZ4_streamHC_t*)(cctxPtr->lz4CtxPtr), (char*)(cctxPtr->tmpBuff), 64 KB);
814
+ }
815
+
816
+ typedef enum { notDone, fromTmpBuffer, fromSrcBuffer } LZ4F_lastBlockStatus;
817
+
818
+ /*! LZ4F_compressUpdate() :
819
+ * LZ4F_compressUpdate() can be called repetitively to compress as much data as necessary.
820
+ * dstBuffer MUST be >= LZ4F_compressBound(srcSize, preferencesPtr).
821
+ * LZ4F_compressOptions_t structure is optional : you can provide NULL as argument.
822
+ * @return : the number of bytes written into dstBuffer. It can be zero, meaning input data was just buffered.
823
+ * or an error code if it fails (which can be tested using LZ4F_isError())
824
+ */
825
+ size_t LZ4F_compressUpdate(LZ4F_cctx* cctxPtr,
826
+ void* dstBuffer, size_t dstCapacity,
827
+ const void* srcBuffer, size_t srcSize,
828
+ const LZ4F_compressOptions_t* compressOptionsPtr)
829
+ {
830
+ LZ4F_compressOptions_t cOptionsNull;
831
+ size_t const blockSize = cctxPtr->maxBlockSize;
832
+ const BYTE* srcPtr = (const BYTE*)srcBuffer;
833
+ const BYTE* const srcEnd = srcPtr + srcSize;
834
+ BYTE* const dstStart = (BYTE*)dstBuffer;
835
+ BYTE* dstPtr = dstStart;
836
+ LZ4F_lastBlockStatus lastBlockCompressed = notDone;
837
+ compressFunc_t const compress = LZ4F_selectCompression(cctxPtr->prefs.frameInfo.blockMode, cctxPtr->prefs.compressionLevel);
838
+
839
+ DEBUGLOG(4, "LZ4F_compressUpdate (srcSize=%zu)", srcSize);
840
+
841
+ if (cctxPtr->cStage != 1) return err0r(LZ4F_ERROR_GENERIC);
842
+ if (dstCapacity < LZ4F_compressBound_internal(srcSize, &(cctxPtr->prefs), cctxPtr->tmpInSize))
843
+ return err0r(LZ4F_ERROR_dstMaxSize_tooSmall);
844
+ MEM_INIT(&cOptionsNull, 0, sizeof(cOptionsNull));
845
+ if (compressOptionsPtr == NULL) compressOptionsPtr = &cOptionsNull;
846
+
847
+ /* complete tmp buffer */
848
+ if (cctxPtr->tmpInSize > 0) { /* some data already within tmp buffer */
849
+ size_t const sizeToCopy = blockSize - cctxPtr->tmpInSize;
850
+ if (sizeToCopy > srcSize) {
851
+ /* add src to tmpIn buffer */
852
+ memcpy(cctxPtr->tmpIn + cctxPtr->tmpInSize, srcBuffer, srcSize);
853
+ srcPtr = srcEnd;
854
+ cctxPtr->tmpInSize += srcSize;
855
+ /* still needs some CRC */
856
+ } else {
857
+ /* complete tmpIn block and then compress it */
858
+ lastBlockCompressed = fromTmpBuffer;
859
+ memcpy(cctxPtr->tmpIn + cctxPtr->tmpInSize, srcBuffer, sizeToCopy);
860
+ srcPtr += sizeToCopy;
861
+
862
+ dstPtr += LZ4F_makeBlock(dstPtr,
863
+ cctxPtr->tmpIn, blockSize,
864
+ compress, cctxPtr->lz4CtxPtr, cctxPtr->prefs.compressionLevel,
865
+ cctxPtr->cdict,
866
+ cctxPtr->prefs.frameInfo.blockChecksumFlag);
867
+
868
+ if (cctxPtr->prefs.frameInfo.blockMode==LZ4F_blockLinked) cctxPtr->tmpIn += blockSize;
869
+ cctxPtr->tmpInSize = 0;
870
+ }
871
+ }
872
+
873
+ while ((size_t)(srcEnd - srcPtr) >= blockSize) {
874
+ /* compress full blocks */
875
+ lastBlockCompressed = fromSrcBuffer;
876
+ dstPtr += LZ4F_makeBlock(dstPtr,
877
+ srcPtr, blockSize,
878
+ compress, cctxPtr->lz4CtxPtr, cctxPtr->prefs.compressionLevel,
879
+ cctxPtr->cdict,
880
+ cctxPtr->prefs.frameInfo.blockChecksumFlag);
881
+ srcPtr += blockSize;
882
+ }
883
+
884
+ if ((cctxPtr->prefs.autoFlush) && (srcPtr < srcEnd)) {
885
+ /* compress remaining input < blockSize */
886
+ lastBlockCompressed = fromSrcBuffer;
887
+ dstPtr += LZ4F_makeBlock(dstPtr,
888
+ srcPtr, (size_t)(srcEnd - srcPtr),
889
+ compress, cctxPtr->lz4CtxPtr, cctxPtr->prefs.compressionLevel,
890
+ cctxPtr->cdict,
891
+ cctxPtr->prefs.frameInfo.blockChecksumFlag);
892
+ srcPtr = srcEnd;
893
+ }
894
+
895
+ /* preserve dictionary if necessary */
896
+ if ((cctxPtr->prefs.frameInfo.blockMode==LZ4F_blockLinked) && (lastBlockCompressed==fromSrcBuffer)) {
897
+ if (compressOptionsPtr->stableSrc) {
898
+ cctxPtr->tmpIn = cctxPtr->tmpBuff;
899
+ } else {
900
+ int const realDictSize = LZ4F_localSaveDict(cctxPtr);
901
+ if (realDictSize==0) return err0r(LZ4F_ERROR_GENERIC);
902
+ cctxPtr->tmpIn = cctxPtr->tmpBuff + realDictSize;
903
+ }
904
+ }
905
+
906
+ /* keep tmpIn within limits */
907
+ if ((cctxPtr->tmpIn + blockSize) > (cctxPtr->tmpBuff + cctxPtr->maxBufferSize) /* necessarily LZ4F_blockLinked && lastBlockCompressed==fromTmpBuffer */
908
+ && !(cctxPtr->prefs.autoFlush))
909
+ {
910
+ int const realDictSize = LZ4F_localSaveDict(cctxPtr);
911
+ cctxPtr->tmpIn = cctxPtr->tmpBuff + realDictSize;
912
+ }
913
+
914
+ /* some input data left, necessarily < blockSize */
915
+ if (srcPtr < srcEnd) {
916
+ /* fill tmp buffer */
917
+ size_t const sizeToCopy = (size_t)(srcEnd - srcPtr);
918
+ memcpy(cctxPtr->tmpIn, srcPtr, sizeToCopy);
919
+ cctxPtr->tmpInSize = sizeToCopy;
920
+ }
921
+
922
+ if (cctxPtr->prefs.frameInfo.contentChecksumFlag == LZ4F_contentChecksumEnabled)
923
+ (void)XXH32_update(&(cctxPtr->xxh), srcBuffer, srcSize);
924
+
925
+ cctxPtr->totalInSize += srcSize;
926
+ return (size_t)(dstPtr - dstStart);
927
+ }
928
+
929
+
930
+ /*! LZ4F_flush() :
931
+ * When compressed data must be sent immediately, without waiting for a block to be filled,
932
+ * invoke LZ4_flush(), which will immediately compress any remaining data stored within LZ4F_cctx.
933
+ * The result of the function is the number of bytes written into dstBuffer.
934
+ * It can be zero, this means there was no data left within LZ4F_cctx.
935
+ * The function outputs an error code if it fails (can be tested using LZ4F_isError())
936
+ * LZ4F_compressOptions_t* is optional. NULL is a valid argument.
937
+ */
938
+ size_t LZ4F_flush(LZ4F_cctx* cctxPtr,
939
+ void* dstBuffer, size_t dstCapacity,
940
+ const LZ4F_compressOptions_t* compressOptionsPtr)
941
+ {
942
+ BYTE* const dstStart = (BYTE*)dstBuffer;
943
+ BYTE* dstPtr = dstStart;
944
+ compressFunc_t compress;
945
+
946
+ if (cctxPtr->tmpInSize == 0) return 0; /* nothing to flush */
947
+ if (cctxPtr->cStage != 1) return err0r(LZ4F_ERROR_GENERIC);
948
+ if (dstCapacity < (cctxPtr->tmpInSize + BHSize + BFSize))
949
+ return err0r(LZ4F_ERROR_dstMaxSize_tooSmall);
950
+ (void)compressOptionsPtr; /* not yet useful */
951
+
952
+ /* select compression function */
953
+ compress = LZ4F_selectCompression(cctxPtr->prefs.frameInfo.blockMode, cctxPtr->prefs.compressionLevel);
954
+
955
+ /* compress tmp buffer */
956
+ dstPtr += LZ4F_makeBlock(dstPtr,
957
+ cctxPtr->tmpIn, cctxPtr->tmpInSize,
958
+ compress, cctxPtr->lz4CtxPtr, cctxPtr->prefs.compressionLevel,
959
+ cctxPtr->cdict,
960
+ cctxPtr->prefs.frameInfo.blockChecksumFlag);
961
+ assert(((void)"flush overflows dstBuffer!", (size_t)(dstPtr - dstStart) <= dstCapacity));
962
+
963
+ if (cctxPtr->prefs.frameInfo.blockMode == LZ4F_blockLinked)
964
+ cctxPtr->tmpIn += cctxPtr->tmpInSize;
965
+ cctxPtr->tmpInSize = 0;
966
+
967
+ /* keep tmpIn within limits */
968
+ if ((cctxPtr->tmpIn + cctxPtr->maxBlockSize) > (cctxPtr->tmpBuff + cctxPtr->maxBufferSize)) { /* necessarily LZ4F_blockLinked */
969
+ int const realDictSize = LZ4F_localSaveDict(cctxPtr);
970
+ cctxPtr->tmpIn = cctxPtr->tmpBuff + realDictSize;
971
+ }
972
+
973
+ return (size_t)(dstPtr - dstStart);
974
+ }
975
+
976
+
977
+ /*! LZ4F_compressEnd() :
978
+ * When you want to properly finish the compressed frame, just call LZ4F_compressEnd().
979
+ * It will flush whatever data remained within compressionContext (like LZ4_flush())
980
+ * but also properly finalize the frame, with an endMark and an (optional) checksum.
981
+ * LZ4F_compressOptions_t structure is optional : you can provide NULL as argument.
982
+ * @return: the number of bytes written into dstBuffer (necessarily >= 4 (endMark size))
983
+ * or an error code if it fails (can be tested using LZ4F_isError())
984
+ * The context can then be used again to compress a new frame, starting with LZ4F_compressBegin().
985
+ */
986
+ size_t LZ4F_compressEnd(LZ4F_cctx* cctxPtr,
987
+ void* dstBuffer, size_t dstCapacity,
988
+ const LZ4F_compressOptions_t* compressOptionsPtr)
989
+ {
990
+ BYTE* const dstStart = (BYTE*)dstBuffer;
991
+ BYTE* dstPtr = dstStart;
992
+
993
+ size_t const flushSize = LZ4F_flush(cctxPtr, dstBuffer, dstCapacity, compressOptionsPtr);
994
+ DEBUGLOG(5,"LZ4F_compressEnd: dstCapacity=%u", (unsigned)dstCapacity);
995
+ if (LZ4F_isError(flushSize)) return flushSize;
996
+ dstPtr += flushSize;
997
+
998
+ assert(flushSize <= dstCapacity);
999
+ dstCapacity -= flushSize;
1000
+
1001
+ if (dstCapacity < 4) return err0r(LZ4F_ERROR_dstMaxSize_tooSmall);
1002
+ LZ4F_writeLE32(dstPtr, 0);
1003
+ dstPtr += 4; /* endMark */
1004
+
1005
+ if (cctxPtr->prefs.frameInfo.contentChecksumFlag == LZ4F_contentChecksumEnabled) {
1006
+ U32 const xxh = XXH32_digest(&(cctxPtr->xxh));
1007
+ if (dstCapacity < 8) return err0r(LZ4F_ERROR_dstMaxSize_tooSmall);
1008
+ DEBUGLOG(5,"Writing 32-bit content checksum");
1009
+ LZ4F_writeLE32(dstPtr, xxh);
1010
+ dstPtr+=4; /* content Checksum */
1011
+ }
1012
+
1013
+ cctxPtr->cStage = 0; /* state is now re-usable (with identical preferences) */
1014
+ cctxPtr->maxBufferSize = 0; /* reuse HC context */
1015
+
1016
+ if (cctxPtr->prefs.frameInfo.contentSize) {
1017
+ if (cctxPtr->prefs.frameInfo.contentSize != cctxPtr->totalInSize)
1018
+ return err0r(LZ4F_ERROR_frameSize_wrong);
1019
+ }
1020
+
1021
+ return (size_t)(dstPtr - dstStart);
1022
+ }
1023
+
1024
+
1025
+ /*-***************************************************
1026
+ * Frame Decompression
1027
+ *****************************************************/
1028
+
1029
+ typedef enum {
1030
+ dstage_getFrameHeader=0, dstage_storeFrameHeader,
1031
+ dstage_init,
1032
+ dstage_getBlockHeader, dstage_storeBlockHeader,
1033
+ dstage_copyDirect, dstage_getBlockChecksum,
1034
+ dstage_getCBlock, dstage_storeCBlock,
1035
+ dstage_flushOut,
1036
+ dstage_getSuffix, dstage_storeSuffix,
1037
+ dstage_getSFrameSize, dstage_storeSFrameSize,
1038
+ dstage_skipSkippable
1039
+ } dStage_t;
1040
+
1041
+ struct LZ4F_dctx_s {
1042
+ LZ4F_frameInfo_t frameInfo;
1043
+ U32 version;
1044
+ dStage_t dStage;
1045
+ U64 frameRemainingSize;
1046
+ size_t maxBlockSize;
1047
+ size_t maxBufferSize;
1048
+ BYTE* tmpIn;
1049
+ size_t tmpInSize;
1050
+ size_t tmpInTarget;
1051
+ BYTE* tmpOutBuffer;
1052
+ const BYTE* dict;
1053
+ size_t dictSize;
1054
+ BYTE* tmpOut;
1055
+ size_t tmpOutSize;
1056
+ size_t tmpOutStart;
1057
+ XXH32_state_t xxh;
1058
+ XXH32_state_t blockChecksum;
1059
+ BYTE header[LZ4F_HEADER_SIZE_MAX];
1060
+ }; /* typedef'd to LZ4F_dctx in lz4frame.h */
1061
+
1062
+
1063
+ /*! LZ4F_createDecompressionContext() :
1064
+ * Create a decompressionContext object, which will track all decompression operations.
1065
+ * Provides a pointer to a fully allocated and initialized LZ4F_decompressionContext object.
1066
+ * Object can later be released using LZ4F_freeDecompressionContext().
1067
+ * @return : if != 0, there was an error during context creation.
1068
+ */
1069
+ LZ4F_errorCode_t LZ4F_createDecompressionContext(LZ4F_dctx** LZ4F_decompressionContextPtr, unsigned versionNumber)
1070
+ {
1071
+ LZ4F_dctx* const dctx = (LZ4F_dctx*)ALLOC_AND_ZERO(sizeof(LZ4F_dctx));
1072
+ if (dctx == NULL) { /* failed allocation */
1073
+ *LZ4F_decompressionContextPtr = NULL;
1074
+ return err0r(LZ4F_ERROR_allocation_failed);
1075
+ }
1076
+
1077
+ dctx->version = versionNumber;
1078
+ *LZ4F_decompressionContextPtr = dctx;
1079
+ return LZ4F_OK_NoError;
1080
+ }
1081
+
1082
+ LZ4F_errorCode_t LZ4F_freeDecompressionContext(LZ4F_dctx* dctx)
1083
+ {
1084
+ LZ4F_errorCode_t result = LZ4F_OK_NoError;
1085
+ if (dctx != NULL) { /* can accept NULL input, like free() */
1086
+ result = (LZ4F_errorCode_t)dctx->dStage;
1087
+ FREEMEM(dctx->tmpIn);
1088
+ FREEMEM(dctx->tmpOutBuffer);
1089
+ FREEMEM(dctx);
1090
+ }
1091
+ return result;
1092
+ }
1093
+
1094
+
1095
+ /*==--- Streaming Decompression operations ---==*/
1096
+
1097
+ void LZ4F_resetDecompressionContext(LZ4F_dctx* dctx)
1098
+ {
1099
+ dctx->dStage = dstage_getFrameHeader;
1100
+ dctx->dict = NULL;
1101
+ dctx->dictSize = 0;
1102
+ }
1103
+
1104
+
1105
+ /*! LZ4F_decodeHeader() :
1106
+ * input : `src` points at the **beginning of the frame**
1107
+ * output : set internal values of dctx, such as
1108
+ * dctx->frameInfo and dctx->dStage.
1109
+ * Also allocates internal buffers.
1110
+ * @return : nb Bytes read from src (necessarily <= srcSize)
1111
+ * or an error code (testable with LZ4F_isError())
1112
+ */
1113
+ static size_t LZ4F_decodeHeader(LZ4F_dctx* dctx, const void* src, size_t srcSize)
1114
+ {
1115
+ unsigned blockMode, blockChecksumFlag, contentSizeFlag, contentChecksumFlag, dictIDFlag, blockSizeID;
1116
+ size_t frameHeaderSize;
1117
+ const BYTE* srcPtr = (const BYTE*)src;
1118
+
1119
+ DEBUGLOG(5, "LZ4F_decodeHeader");
1120
+ /* need to decode header to get frameInfo */
1121
+ if (srcSize < minFHSize) return err0r(LZ4F_ERROR_frameHeader_incomplete); /* minimal frame header size */
1122
+ MEM_INIT(&(dctx->frameInfo), 0, sizeof(dctx->frameInfo));
1123
+
1124
+ /* special case : skippable frames */
1125
+ if ((LZ4F_readLE32(srcPtr) & 0xFFFFFFF0U) == LZ4F_MAGIC_SKIPPABLE_START) {
1126
+ dctx->frameInfo.frameType = LZ4F_skippableFrame;
1127
+ if (src == (void*)(dctx->header)) {
1128
+ dctx->tmpInSize = srcSize;
1129
+ dctx->tmpInTarget = 8;
1130
+ dctx->dStage = dstage_storeSFrameSize;
1131
+ return srcSize;
1132
+ } else {
1133
+ dctx->dStage = dstage_getSFrameSize;
1134
+ return 4;
1135
+ }
1136
+ }
1137
+
1138
+ /* control magic number */
1139
+ #ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
1140
+ if (LZ4F_readLE32(srcPtr) != LZ4F_MAGICNUMBER) {
1141
+ DEBUGLOG(4, "frame header error : unknown magic number");
1142
+ return err0r(LZ4F_ERROR_frameType_unknown);
1143
+ }
1144
+ #endif
1145
+ dctx->frameInfo.frameType = LZ4F_frame;
1146
+
1147
+ /* Flags */
1148
+ { U32 const FLG = srcPtr[4];
1149
+ U32 const version = (FLG>>6) & _2BITS;
1150
+ blockChecksumFlag = (FLG>>4) & _1BIT;
1151
+ blockMode = (FLG>>5) & _1BIT;
1152
+ contentSizeFlag = (FLG>>3) & _1BIT;
1153
+ contentChecksumFlag = (FLG>>2) & _1BIT;
1154
+ dictIDFlag = FLG & _1BIT;
1155
+ /* validate */
1156
+ if (((FLG>>1)&_1BIT) != 0) return err0r(LZ4F_ERROR_reservedFlag_set); /* Reserved bit */
1157
+ if (version != 1) return err0r(LZ4F_ERROR_headerVersion_wrong); /* Version Number, only supported value */
1158
+ }
1159
+
1160
+ /* Frame Header Size */
1161
+ frameHeaderSize = minFHSize + (contentSizeFlag?8:0) + (dictIDFlag?4:0);
1162
+
1163
+ if (srcSize < frameHeaderSize) {
1164
+ /* not enough input to fully decode frame header */
1165
+ if (srcPtr != dctx->header)
1166
+ memcpy(dctx->header, srcPtr, srcSize);
1167
+ dctx->tmpInSize = srcSize;
1168
+ dctx->tmpInTarget = frameHeaderSize;
1169
+ dctx->dStage = dstage_storeFrameHeader;
1170
+ return srcSize;
1171
+ }
1172
+
1173
+ { U32 const BD = srcPtr[5];
1174
+ blockSizeID = (BD>>4) & _3BITS;
1175
+ /* validate */
1176
+ if (((BD>>7)&_1BIT) != 0) return err0r(LZ4F_ERROR_reservedFlag_set); /* Reserved bit */
1177
+ if (blockSizeID < 4) return err0r(LZ4F_ERROR_maxBlockSize_invalid); /* 4-7 only supported values for the time being */
1178
+ if (((BD>>0)&_4BITS) != 0) return err0r(LZ4F_ERROR_reservedFlag_set); /* Reserved bits */
1179
+ }
1180
+
1181
+ /* check header */
1182
+ assert(frameHeaderSize > 5);
1183
+ #ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
1184
+ { BYTE const HC = LZ4F_headerChecksum(srcPtr+4, frameHeaderSize-5);
1185
+ if (HC != srcPtr[frameHeaderSize-1])
1186
+ return err0r(LZ4F_ERROR_headerChecksum_invalid);
1187
+ }
1188
+ #endif
1189
+
1190
+ /* save */
1191
+ dctx->frameInfo.blockMode = (LZ4F_blockMode_t)blockMode;
1192
+ dctx->frameInfo.blockChecksumFlag = (LZ4F_blockChecksum_t)blockChecksumFlag;
1193
+ dctx->frameInfo.contentChecksumFlag = (LZ4F_contentChecksum_t)contentChecksumFlag;
1194
+ dctx->frameInfo.blockSizeID = (LZ4F_blockSizeID_t)blockSizeID;
1195
+ dctx->maxBlockSize = LZ4F_getBlockSize(blockSizeID);
1196
+ if (contentSizeFlag)
1197
+ dctx->frameRemainingSize =
1198
+ dctx->frameInfo.contentSize = LZ4F_readLE64(srcPtr+6);
1199
+ if (dictIDFlag)
1200
+ dctx->frameInfo.dictID = LZ4F_readLE32(srcPtr + frameHeaderSize - 5);
1201
+
1202
+ dctx->dStage = dstage_init;
1203
+
1204
+ return frameHeaderSize;
1205
+ }
1206
+
1207
+
1208
+ /*! LZ4F_headerSize() :
1209
+ * @return : size of frame header
1210
+ * or an error code, which can be tested using LZ4F_isError()
1211
+ */
1212
+ size_t LZ4F_headerSize(const void* src, size_t srcSize)
1213
+ {
1214
+ if (src == NULL) return err0r(LZ4F_ERROR_srcPtr_wrong);
1215
+
1216
+ /* minimal srcSize to determine header size */
1217
+ if (srcSize < LZ4F_MIN_SIZE_TO_KNOW_HEADER_LENGTH)
1218
+ return err0r(LZ4F_ERROR_frameHeader_incomplete);
1219
+
1220
+ /* special case : skippable frames */
1221
+ if ((LZ4F_readLE32(src) & 0xFFFFFFF0U) == LZ4F_MAGIC_SKIPPABLE_START)
1222
+ return 8;
1223
+
1224
+ /* control magic number */
1225
+ #ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
1226
+ if (LZ4F_readLE32(src) != LZ4F_MAGICNUMBER)
1227
+ return err0r(LZ4F_ERROR_frameType_unknown);
1228
+ #endif
1229
+
1230
+ /* Frame Header Size */
1231
+ { BYTE const FLG = ((const BYTE*)src)[4];
1232
+ U32 const contentSizeFlag = (FLG>>3) & _1BIT;
1233
+ U32 const dictIDFlag = FLG & _1BIT;
1234
+ return minFHSize + (contentSizeFlag?8:0) + (dictIDFlag?4:0);
1235
+ }
1236
+ }
1237
+
1238
+ /*! LZ4F_getFrameInfo() :
1239
+ * This function extracts frame parameters (max blockSize, frame checksum, etc.).
1240
+ * Usage is optional. Objective is to provide relevant information for allocation purposes.
1241
+ * This function works in 2 situations :
1242
+ * - At the beginning of a new frame, in which case it will decode this information from `srcBuffer`, and start the decoding process.
1243
+ * Amount of input data provided must be large enough to successfully decode the frame header.
1244
+ * A header size is variable, but is guaranteed to be <= LZ4F_HEADER_SIZE_MAX bytes. It's possible to provide more input data than this minimum.
1245
+ * - After decoding has been started. In which case, no input is read, frame parameters are extracted from dctx.
1246
+ * The number of bytes consumed from srcBuffer will be updated within *srcSizePtr (necessarily <= original value).
1247
+ * Decompression must resume from (srcBuffer + *srcSizePtr).
1248
+ * @return : an hint about how many srcSize bytes LZ4F_decompress() expects for next call,
1249
+ * or an error code which can be tested using LZ4F_isError()
1250
+ * note 1 : in case of error, dctx is not modified. Decoding operations can resume from where they stopped.
1251
+ * note 2 : frame parameters are *copied into* an already allocated LZ4F_frameInfo_t structure.
1252
+ */
1253
+ LZ4F_errorCode_t LZ4F_getFrameInfo(LZ4F_dctx* dctx,
1254
+ LZ4F_frameInfo_t* frameInfoPtr,
1255
+ const void* srcBuffer, size_t* srcSizePtr)
1256
+ {
1257
+ LZ4F_STATIC_ASSERT(dstage_getFrameHeader < dstage_storeFrameHeader);
1258
+ if (dctx->dStage > dstage_storeFrameHeader) {
1259
+ /* frameInfo already decoded */
1260
+ size_t o=0, i=0;
1261
+ *srcSizePtr = 0;
1262
+ *frameInfoPtr = dctx->frameInfo;
1263
+ /* returns : recommended nb of bytes for LZ4F_decompress() */
1264
+ return LZ4F_decompress(dctx, NULL, &o, NULL, &i, NULL);
1265
+ } else {
1266
+ if (dctx->dStage == dstage_storeFrameHeader) {
1267
+ /* frame decoding already started, in the middle of header => automatic fail */
1268
+ *srcSizePtr = 0;
1269
+ return err0r(LZ4F_ERROR_frameDecoding_alreadyStarted);
1270
+ } else {
1271
+ size_t const hSize = LZ4F_headerSize(srcBuffer, *srcSizePtr);
1272
+ if (LZ4F_isError(hSize)) { *srcSizePtr=0; return hSize; }
1273
+ if (*srcSizePtr < hSize) {
1274
+ *srcSizePtr=0;
1275
+ return err0r(LZ4F_ERROR_frameHeader_incomplete);
1276
+ }
1277
+
1278
+ { size_t decodeResult = LZ4F_decodeHeader(dctx, srcBuffer, hSize);
1279
+ if (LZ4F_isError(decodeResult)) {
1280
+ *srcSizePtr = 0;
1281
+ } else {
1282
+ *srcSizePtr = decodeResult;
1283
+ decodeResult = BHSize; /* block header size */
1284
+ }
1285
+ *frameInfoPtr = dctx->frameInfo;
1286
+ return decodeResult;
1287
+ } } }
1288
+ }
1289
+
1290
+
1291
+ /* LZ4F_updateDict() :
1292
+ * only used for LZ4F_blockLinked mode
1293
+ * Condition : dstPtr != NULL
1294
+ */
1295
+ static void LZ4F_updateDict(LZ4F_dctx* dctx,
1296
+ const BYTE* dstPtr, size_t dstSize, const BYTE* dstBufferStart,
1297
+ unsigned withinTmp)
1298
+ {
1299
+ assert(dstPtr != NULL);
1300
+ if (dctx->dictSize==0) {
1301
+ dctx->dict = (const BYTE*)dstPtr; /* priority to prefix mode */
1302
+ }
1303
+ assert(dctx->dict != NULL);
1304
+
1305
+ if (dctx->dict + dctx->dictSize == dstPtr) { /* prefix mode, everything within dstBuffer */
1306
+ dctx->dictSize += dstSize;
1307
+ return;
1308
+ }
1309
+
1310
+ assert(dstPtr >= dstBufferStart);
1311
+ if ((size_t)(dstPtr - dstBufferStart) + dstSize >= 64 KB) { /* history in dstBuffer becomes large enough to become dictionary */
1312
+ dctx->dict = (const BYTE*)dstBufferStart;
1313
+ dctx->dictSize = (size_t)(dstPtr - dstBufferStart) + dstSize;
1314
+ return;
1315
+ }
1316
+
1317
+ assert(dstSize < 64 KB); /* if dstSize >= 64 KB, dictionary would be set into dstBuffer directly */
1318
+
1319
+ /* dstBuffer does not contain whole useful history (64 KB), so it must be saved within tmpOutBuffer */
1320
+ assert(dctx->tmpOutBuffer != NULL);
1321
+
1322
+ if (withinTmp && (dctx->dict == dctx->tmpOutBuffer)) { /* continue history within tmpOutBuffer */
1323
+ /* withinTmp expectation : content of [dstPtr,dstSize] is same as [dict+dictSize,dstSize], so we just extend it */
1324
+ assert(dctx->dict + dctx->dictSize == dctx->tmpOut + dctx->tmpOutStart);
1325
+ dctx->dictSize += dstSize;
1326
+ return;
1327
+ }
1328
+
1329
+ if (withinTmp) { /* copy relevant dict portion in front of tmpOut within tmpOutBuffer */
1330
+ size_t const preserveSize = (size_t)(dctx->tmpOut - dctx->tmpOutBuffer);
1331
+ size_t copySize = 64 KB - dctx->tmpOutSize;
1332
+ const BYTE* const oldDictEnd = dctx->dict + dctx->dictSize - dctx->tmpOutStart;
1333
+ if (dctx->tmpOutSize > 64 KB) copySize = 0;
1334
+ if (copySize > preserveSize) copySize = preserveSize;
1335
+
1336
+ memcpy(dctx->tmpOutBuffer + preserveSize - copySize, oldDictEnd - copySize, copySize);
1337
+
1338
+ dctx->dict = dctx->tmpOutBuffer;
1339
+ dctx->dictSize = preserveSize + dctx->tmpOutStart + dstSize;
1340
+ return;
1341
+ }
1342
+
1343
+ if (dctx->dict == dctx->tmpOutBuffer) { /* copy dst into tmp to complete dict */
1344
+ if (dctx->dictSize + dstSize > dctx->maxBufferSize) { /* tmp buffer not large enough */
1345
+ size_t const preserveSize = 64 KB - dstSize;
1346
+ memcpy(dctx->tmpOutBuffer, dctx->dict + dctx->dictSize - preserveSize, preserveSize);
1347
+ dctx->dictSize = preserveSize;
1348
+ }
1349
+ memcpy(dctx->tmpOutBuffer + dctx->dictSize, dstPtr, dstSize);
1350
+ dctx->dictSize += dstSize;
1351
+ return;
1352
+ }
1353
+
1354
+ /* join dict & dest into tmp */
1355
+ { size_t preserveSize = 64 KB - dstSize;
1356
+ if (preserveSize > dctx->dictSize) preserveSize = dctx->dictSize;
1357
+ memcpy(dctx->tmpOutBuffer, dctx->dict + dctx->dictSize - preserveSize, preserveSize);
1358
+ memcpy(dctx->tmpOutBuffer + preserveSize, dstPtr, dstSize);
1359
+ dctx->dict = dctx->tmpOutBuffer;
1360
+ dctx->dictSize = preserveSize + dstSize;
1361
+ }
1362
+ }
1363
+
1364
+
1365
+
1366
+ /*! LZ4F_decompress() :
1367
+ * Call this function repetitively to regenerate compressed data in srcBuffer.
1368
+ * The function will attempt to decode up to *srcSizePtr bytes from srcBuffer
1369
+ * into dstBuffer of capacity *dstSizePtr.
1370
+ *
1371
+ * The number of bytes regenerated into dstBuffer will be provided within *dstSizePtr (necessarily <= original value).
1372
+ *
1373
+ * The number of bytes effectively read from srcBuffer will be provided within *srcSizePtr (necessarily <= original value).
1374
+ * If number of bytes read is < number of bytes provided, then decompression operation is not complete.
1375
+ * Remaining data will have to be presented again in a subsequent invocation.
1376
+ *
1377
+ * The function result is an hint of the better srcSize to use for next call to LZ4F_decompress.
1378
+ * Schematically, it's the size of the current (or remaining) compressed block + header of next block.
1379
+ * Respecting the hint provides a small boost to performance, since it allows less buffer shuffling.
1380
+ * Note that this is just a hint, and it's always possible to any srcSize value.
1381
+ * When a frame is fully decoded, @return will be 0.
1382
+ * If decompression failed, @return is an error code which can be tested using LZ4F_isError().
1383
+ */
1384
+ size_t LZ4F_decompress(LZ4F_dctx* dctx,
1385
+ void* dstBuffer, size_t* dstSizePtr,
1386
+ const void* srcBuffer, size_t* srcSizePtr,
1387
+ const LZ4F_decompressOptions_t* decompressOptionsPtr)
1388
+ {
1389
+ LZ4F_decompressOptions_t optionsNull;
1390
+ const BYTE* const srcStart = (const BYTE*)srcBuffer;
1391
+ const BYTE* const srcEnd = srcStart + *srcSizePtr;
1392
+ const BYTE* srcPtr = srcStart;
1393
+ BYTE* const dstStart = (BYTE*)dstBuffer;
1394
+ BYTE* const dstEnd = dstStart ? dstStart + *dstSizePtr : NULL;
1395
+ BYTE* dstPtr = dstStart;
1396
+ const BYTE* selectedIn = NULL;
1397
+ unsigned doAnotherStage = 1;
1398
+ size_t nextSrcSizeHint = 1;
1399
+
1400
+
1401
+ DEBUGLOG(5, "LZ4F_decompress : %p,%u => %p,%u",
1402
+ srcBuffer, (unsigned)*srcSizePtr, dstBuffer, (unsigned)*dstSizePtr);
1403
+ if (dstBuffer == NULL) assert(*dstSizePtr == 0);
1404
+ MEM_INIT(&optionsNull, 0, sizeof(optionsNull));
1405
+ if (decompressOptionsPtr==NULL) decompressOptionsPtr = &optionsNull;
1406
+ *srcSizePtr = 0;
1407
+ *dstSizePtr = 0;
1408
+ assert(dctx != NULL);
1409
+
1410
+ /* behaves as a state machine */
1411
+
1412
+ while (doAnotherStage) {
1413
+
1414
+ switch(dctx->dStage)
1415
+ {
1416
+
1417
+ case dstage_getFrameHeader:
1418
+ DEBUGLOG(6, "dstage_getFrameHeader");
1419
+ if ((size_t)(srcEnd-srcPtr) >= maxFHSize) { /* enough to decode - shortcut */
1420
+ size_t const hSize = LZ4F_decodeHeader(dctx, srcPtr, (size_t)(srcEnd-srcPtr)); /* will update dStage appropriately */
1421
+ if (LZ4F_isError(hSize)) return hSize;
1422
+ srcPtr += hSize;
1423
+ break;
1424
+ }
1425
+ dctx->tmpInSize = 0;
1426
+ if (srcEnd-srcPtr == 0) return minFHSize; /* 0-size input */
1427
+ dctx->tmpInTarget = minFHSize; /* minimum size to decode header */
1428
+ dctx->dStage = dstage_storeFrameHeader;
1429
+ /* fall-through */
1430
+
1431
+ case dstage_storeFrameHeader:
1432
+ DEBUGLOG(6, "dstage_storeFrameHeader");
1433
+ { size_t const sizeToCopy = MIN(dctx->tmpInTarget - dctx->tmpInSize, (size_t)(srcEnd - srcPtr));
1434
+ memcpy(dctx->header + dctx->tmpInSize, srcPtr, sizeToCopy);
1435
+ dctx->tmpInSize += sizeToCopy;
1436
+ srcPtr += sizeToCopy;
1437
+ }
1438
+ if (dctx->tmpInSize < dctx->tmpInTarget) {
1439
+ nextSrcSizeHint = (dctx->tmpInTarget - dctx->tmpInSize) + BHSize; /* rest of header + nextBlockHeader */
1440
+ doAnotherStage = 0; /* not enough src data, ask for some more */
1441
+ break;
1442
+ }
1443
+ { size_t const hSize = LZ4F_decodeHeader(dctx, dctx->header, dctx->tmpInTarget); /* will update dStage appropriately */
1444
+ if (LZ4F_isError(hSize)) return hSize;
1445
+ }
1446
+ break;
1447
+
1448
+ case dstage_init:
1449
+ DEBUGLOG(6, "dstage_init");
1450
+ if (dctx->frameInfo.contentChecksumFlag) (void)XXH32_reset(&(dctx->xxh), 0);
1451
+ /* internal buffers allocation */
1452
+ { size_t const bufferNeeded = dctx->maxBlockSize
1453
+ + ((dctx->frameInfo.blockMode==LZ4F_blockLinked) ? 128 KB : 0);
1454
+ if (bufferNeeded > dctx->maxBufferSize) { /* tmp buffers too small */
1455
+ dctx->maxBufferSize = 0; /* ensure allocation will be re-attempted on next entry*/
1456
+ FREEMEM(dctx->tmpIn);
1457
+ dctx->tmpIn = (BYTE*)ALLOC(dctx->maxBlockSize + BFSize /* block checksum */);
1458
+ if (dctx->tmpIn == NULL)
1459
+ return err0r(LZ4F_ERROR_allocation_failed);
1460
+ FREEMEM(dctx->tmpOutBuffer);
1461
+ dctx->tmpOutBuffer= (BYTE*)ALLOC(bufferNeeded);
1462
+ if (dctx->tmpOutBuffer== NULL)
1463
+ return err0r(LZ4F_ERROR_allocation_failed);
1464
+ dctx->maxBufferSize = bufferNeeded;
1465
+ } }
1466
+ dctx->tmpInSize = 0;
1467
+ dctx->tmpInTarget = 0;
1468
+ dctx->tmpOut = dctx->tmpOutBuffer;
1469
+ dctx->tmpOutStart = 0;
1470
+ dctx->tmpOutSize = 0;
1471
+
1472
+ dctx->dStage = dstage_getBlockHeader;
1473
+ /* fall-through */
1474
+
1475
+ case dstage_getBlockHeader:
1476
+ if ((size_t)(srcEnd - srcPtr) >= BHSize) {
1477
+ selectedIn = srcPtr;
1478
+ srcPtr += BHSize;
1479
+ } else {
1480
+ /* not enough input to read cBlockSize field */
1481
+ dctx->tmpInSize = 0;
1482
+ dctx->dStage = dstage_storeBlockHeader;
1483
+ }
1484
+
1485
+ if (dctx->dStage == dstage_storeBlockHeader) /* can be skipped */
1486
+ case dstage_storeBlockHeader:
1487
+ { size_t const remainingInput = (size_t)(srcEnd - srcPtr);
1488
+ size_t const wantedData = BHSize - dctx->tmpInSize;
1489
+ size_t const sizeToCopy = MIN(wantedData, remainingInput);
1490
+ memcpy(dctx->tmpIn + dctx->tmpInSize, srcPtr, sizeToCopy);
1491
+ srcPtr += sizeToCopy;
1492
+ dctx->tmpInSize += sizeToCopy;
1493
+
1494
+ if (dctx->tmpInSize < BHSize) { /* not enough input for cBlockSize */
1495
+ nextSrcSizeHint = BHSize - dctx->tmpInSize;
1496
+ doAnotherStage = 0;
1497
+ break;
1498
+ }
1499
+ selectedIn = dctx->tmpIn;
1500
+ } /* if (dctx->dStage == dstage_storeBlockHeader) */
1501
+
1502
+ /* decode block header */
1503
+ { U32 const blockHeader = LZ4F_readLE32(selectedIn);
1504
+ size_t const nextCBlockSize = blockHeader & 0x7FFFFFFFU;
1505
+ size_t const crcSize = dctx->frameInfo.blockChecksumFlag * BFSize;
1506
+ if (blockHeader==0) { /* frameEnd signal, no more block */
1507
+ DEBUGLOG(5, "end of frame");
1508
+ dctx->dStage = dstage_getSuffix;
1509
+ break;
1510
+ }
1511
+ if (nextCBlockSize > dctx->maxBlockSize) {
1512
+ return err0r(LZ4F_ERROR_maxBlockSize_invalid);
1513
+ }
1514
+ if (blockHeader & LZ4F_BLOCKUNCOMPRESSED_FLAG) {
1515
+ /* next block is uncompressed */
1516
+ dctx->tmpInTarget = nextCBlockSize;
1517
+ DEBUGLOG(5, "next block is uncompressed (size %u)", (U32)nextCBlockSize);
1518
+ if (dctx->frameInfo.blockChecksumFlag) {
1519
+ (void)XXH32_reset(&dctx->blockChecksum, 0);
1520
+ }
1521
+ dctx->dStage = dstage_copyDirect;
1522
+ break;
1523
+ }
1524
+ /* next block is a compressed block */
1525
+ dctx->tmpInTarget = nextCBlockSize + crcSize;
1526
+ dctx->dStage = dstage_getCBlock;
1527
+ if (dstPtr==dstEnd || srcPtr==srcEnd) {
1528
+ nextSrcSizeHint = BHSize + nextCBlockSize + crcSize;
1529
+ doAnotherStage = 0;
1530
+ }
1531
+ break;
1532
+ }
1533
+
1534
+ case dstage_copyDirect: /* uncompressed block */
1535
+ DEBUGLOG(6, "dstage_copyDirect");
1536
+ { size_t sizeToCopy;
1537
+ if (dstPtr == NULL) {
1538
+ sizeToCopy = 0;
1539
+ } else {
1540
+ size_t const minBuffSize = MIN((size_t)(srcEnd-srcPtr), (size_t)(dstEnd-dstPtr));
1541
+ sizeToCopy = MIN(dctx->tmpInTarget, minBuffSize);
1542
+ memcpy(dstPtr, srcPtr, sizeToCopy);
1543
+ if (dctx->frameInfo.blockChecksumFlag) {
1544
+ (void)XXH32_update(&dctx->blockChecksum, srcPtr, sizeToCopy);
1545
+ }
1546
+ if (dctx->frameInfo.contentChecksumFlag)
1547
+ (void)XXH32_update(&dctx->xxh, srcPtr, sizeToCopy);
1548
+ if (dctx->frameInfo.contentSize)
1549
+ dctx->frameRemainingSize -= sizeToCopy;
1550
+
1551
+ /* history management (linked blocks only)*/
1552
+ if (dctx->frameInfo.blockMode == LZ4F_blockLinked) {
1553
+ LZ4F_updateDict(dctx, dstPtr, sizeToCopy, dstStart, 0);
1554
+ } }
1555
+
1556
+ srcPtr += sizeToCopy;
1557
+ dstPtr += sizeToCopy;
1558
+ if (sizeToCopy == dctx->tmpInTarget) { /* all done */
1559
+ if (dctx->frameInfo.blockChecksumFlag) {
1560
+ dctx->tmpInSize = 0;
1561
+ dctx->dStage = dstage_getBlockChecksum;
1562
+ } else
1563
+ dctx->dStage = dstage_getBlockHeader; /* new block */
1564
+ break;
1565
+ }
1566
+ dctx->tmpInTarget -= sizeToCopy; /* need to copy more */
1567
+ }
1568
+ nextSrcSizeHint = dctx->tmpInTarget +
1569
+ +(dctx->frameInfo.blockChecksumFlag ? BFSize : 0)
1570
+ + BHSize /* next header size */;
1571
+ doAnotherStage = 0;
1572
+ break;
1573
+
1574
+ /* check block checksum for recently transferred uncompressed block */
1575
+ case dstage_getBlockChecksum:
1576
+ DEBUGLOG(6, "dstage_getBlockChecksum");
1577
+ { const void* crcSrc;
1578
+ if ((srcEnd-srcPtr >= 4) && (dctx->tmpInSize==0)) {
1579
+ crcSrc = srcPtr;
1580
+ srcPtr += 4;
1581
+ } else {
1582
+ size_t const stillToCopy = 4 - dctx->tmpInSize;
1583
+ size_t const sizeToCopy = MIN(stillToCopy, (size_t)(srcEnd-srcPtr));
1584
+ memcpy(dctx->header + dctx->tmpInSize, srcPtr, sizeToCopy);
1585
+ dctx->tmpInSize += sizeToCopy;
1586
+ srcPtr += sizeToCopy;
1587
+ if (dctx->tmpInSize < 4) { /* all input consumed */
1588
+ doAnotherStage = 0;
1589
+ break;
1590
+ }
1591
+ crcSrc = dctx->header;
1592
+ }
1593
+ { U32 const readCRC = LZ4F_readLE32(crcSrc);
1594
+ U32 const calcCRC = XXH32_digest(&dctx->blockChecksum);
1595
+ #ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
1596
+ DEBUGLOG(6, "compare block checksum");
1597
+ if (readCRC != calcCRC) {
1598
+ DEBUGLOG(4, "incorrect block checksum: %08X != %08X",
1599
+ readCRC, calcCRC);
1600
+ return err0r(LZ4F_ERROR_blockChecksum_invalid);
1601
+ }
1602
+ #else
1603
+ (void)readCRC;
1604
+ (void)calcCRC;
1605
+ #endif
1606
+ } }
1607
+ dctx->dStage = dstage_getBlockHeader; /* new block */
1608
+ break;
1609
+
1610
+ case dstage_getCBlock:
1611
+ DEBUGLOG(6, "dstage_getCBlock");
1612
+ if ((size_t)(srcEnd-srcPtr) < dctx->tmpInTarget) {
1613
+ dctx->tmpInSize = 0;
1614
+ dctx->dStage = dstage_storeCBlock;
1615
+ break;
1616
+ }
1617
+ /* input large enough to read full block directly */
1618
+ selectedIn = srcPtr;
1619
+ srcPtr += dctx->tmpInTarget;
1620
+
1621
+ if (0) /* always jump over next block */
1622
+ case dstage_storeCBlock:
1623
+ { size_t const wantedData = dctx->tmpInTarget - dctx->tmpInSize;
1624
+ size_t const inputLeft = (size_t)(srcEnd-srcPtr);
1625
+ size_t const sizeToCopy = MIN(wantedData, inputLeft);
1626
+ memcpy(dctx->tmpIn + dctx->tmpInSize, srcPtr, sizeToCopy);
1627
+ dctx->tmpInSize += sizeToCopy;
1628
+ srcPtr += sizeToCopy;
1629
+ if (dctx->tmpInSize < dctx->tmpInTarget) { /* need more input */
1630
+ nextSrcSizeHint = (dctx->tmpInTarget - dctx->tmpInSize)
1631
+ + (dctx->frameInfo.blockChecksumFlag ? BFSize : 0)
1632
+ + BHSize /* next header size */;
1633
+ doAnotherStage = 0;
1634
+ break;
1635
+ }
1636
+ selectedIn = dctx->tmpIn;
1637
+ }
1638
+
1639
+ /* At this stage, input is large enough to decode a block */
1640
+ if (dctx->frameInfo.blockChecksumFlag) {
1641
+ dctx->tmpInTarget -= 4;
1642
+ assert(selectedIn != NULL); /* selectedIn is defined at this stage (either srcPtr, or dctx->tmpIn) */
1643
+ { U32 const readBlockCrc = LZ4F_readLE32(selectedIn + dctx->tmpInTarget);
1644
+ U32 const calcBlockCrc = XXH32(selectedIn, dctx->tmpInTarget, 0);
1645
+ #ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
1646
+ if (readBlockCrc != calcBlockCrc)
1647
+ return err0r(LZ4F_ERROR_blockChecksum_invalid);
1648
+ #else
1649
+ (void)readBlockCrc;
1650
+ (void)calcBlockCrc;
1651
+ #endif
1652
+ } }
1653
+
1654
+ if ((size_t)(dstEnd-dstPtr) >= dctx->maxBlockSize) {
1655
+ const char* dict = (const char*)dctx->dict;
1656
+ size_t dictSize = dctx->dictSize;
1657
+ int decodedSize;
1658
+ assert(dstPtr != NULL);
1659
+ if (dict && dictSize > 1 GB) {
1660
+ /* the dictSize param is an int, avoid truncation / sign issues */
1661
+ dict += dictSize - 64 KB;
1662
+ dictSize = 64 KB;
1663
+ }
1664
+ /* enough capacity in `dst` to decompress directly there */
1665
+ decodedSize = LZ4_decompress_safe_usingDict(
1666
+ (const char*)selectedIn, (char*)dstPtr,
1667
+ (int)dctx->tmpInTarget, (int)dctx->maxBlockSize,
1668
+ dict, (int)dictSize);
1669
+ if (decodedSize < 0) return err0r(LZ4F_ERROR_GENERIC); /* decompression failed */
1670
+ if (dctx->frameInfo.contentChecksumFlag)
1671
+ XXH32_update(&(dctx->xxh), dstPtr, (size_t)decodedSize);
1672
+ if (dctx->frameInfo.contentSize)
1673
+ dctx->frameRemainingSize -= (size_t)decodedSize;
1674
+
1675
+ /* dictionary management */
1676
+ if (dctx->frameInfo.blockMode==LZ4F_blockLinked) {
1677
+ LZ4F_updateDict(dctx, dstPtr, (size_t)decodedSize, dstStart, 0);
1678
+ }
1679
+
1680
+ dstPtr += decodedSize;
1681
+ dctx->dStage = dstage_getBlockHeader;
1682
+ break;
1683
+ }
1684
+
1685
+ /* not enough place into dst : decode into tmpOut */
1686
+ /* ensure enough place for tmpOut */
1687
+ if (dctx->frameInfo.blockMode == LZ4F_blockLinked) {
1688
+ if (dctx->dict == dctx->tmpOutBuffer) {
1689
+ if (dctx->dictSize > 128 KB) {
1690
+ memcpy(dctx->tmpOutBuffer, dctx->dict + dctx->dictSize - 64 KB, 64 KB);
1691
+ dctx->dictSize = 64 KB;
1692
+ }
1693
+ dctx->tmpOut = dctx->tmpOutBuffer + dctx->dictSize;
1694
+ } else { /* dict not within tmp */
1695
+ size_t const reservedDictSpace = MIN(dctx->dictSize, 64 KB);
1696
+ dctx->tmpOut = dctx->tmpOutBuffer + reservedDictSpace;
1697
+ } }
1698
+
1699
+ /* Decode block */
1700
+ { const char* dict = (const char*)dctx->dict;
1701
+ size_t dictSize = dctx->dictSize;
1702
+ int decodedSize;
1703
+ if (dict && dictSize > 1 GB) {
1704
+ /* the dictSize param is an int, avoid truncation / sign issues */
1705
+ dict += dictSize - 64 KB;
1706
+ dictSize = 64 KB;
1707
+ }
1708
+ decodedSize = LZ4_decompress_safe_usingDict(
1709
+ (const char*)selectedIn, (char*)dctx->tmpOut,
1710
+ (int)dctx->tmpInTarget, (int)dctx->maxBlockSize,
1711
+ dict, (int)dictSize);
1712
+ if (decodedSize < 0) /* decompression failed */
1713
+ return err0r(LZ4F_ERROR_decompressionFailed);
1714
+ if (dctx->frameInfo.contentChecksumFlag)
1715
+ XXH32_update(&(dctx->xxh), dctx->tmpOut, (size_t)decodedSize);
1716
+ if (dctx->frameInfo.contentSize)
1717
+ dctx->frameRemainingSize -= (size_t)decodedSize;
1718
+ dctx->tmpOutSize = (size_t)decodedSize;
1719
+ dctx->tmpOutStart = 0;
1720
+ dctx->dStage = dstage_flushOut;
1721
+ }
1722
+ /* fall-through */
1723
+
1724
+ case dstage_flushOut: /* flush decoded data from tmpOut to dstBuffer */
1725
+ DEBUGLOG(6, "dstage_flushOut");
1726
+ if (dstPtr != NULL) {
1727
+ size_t const sizeToCopy = MIN(dctx->tmpOutSize - dctx->tmpOutStart, (size_t)(dstEnd-dstPtr));
1728
+ memcpy(dstPtr, dctx->tmpOut + dctx->tmpOutStart, sizeToCopy);
1729
+
1730
+ /* dictionary management */
1731
+ if (dctx->frameInfo.blockMode == LZ4F_blockLinked)
1732
+ LZ4F_updateDict(dctx, dstPtr, sizeToCopy, dstStart, 1 /*withinTmp*/);
1733
+
1734
+ dctx->tmpOutStart += sizeToCopy;
1735
+ dstPtr += sizeToCopy;
1736
+ }
1737
+ if (dctx->tmpOutStart == dctx->tmpOutSize) { /* all flushed */
1738
+ dctx->dStage = dstage_getBlockHeader; /* get next block */
1739
+ break;
1740
+ }
1741
+ /* could not flush everything : stop there, just request a block header */
1742
+ doAnotherStage = 0;
1743
+ nextSrcSizeHint = BHSize;
1744
+ break;
1745
+
1746
+ case dstage_getSuffix:
1747
+ if (dctx->frameRemainingSize)
1748
+ return err0r(LZ4F_ERROR_frameSize_wrong); /* incorrect frame size decoded */
1749
+ if (!dctx->frameInfo.contentChecksumFlag) { /* no checksum, frame is completed */
1750
+ nextSrcSizeHint = 0;
1751
+ LZ4F_resetDecompressionContext(dctx);
1752
+ doAnotherStage = 0;
1753
+ break;
1754
+ }
1755
+ if ((srcEnd - srcPtr) < 4) { /* not enough size for entire CRC */
1756
+ dctx->tmpInSize = 0;
1757
+ dctx->dStage = dstage_storeSuffix;
1758
+ } else {
1759
+ selectedIn = srcPtr;
1760
+ srcPtr += 4;
1761
+ }
1762
+
1763
+ if (dctx->dStage == dstage_storeSuffix) /* can be skipped */
1764
+ case dstage_storeSuffix:
1765
+ { size_t const remainingInput = (size_t)(srcEnd - srcPtr);
1766
+ size_t const wantedData = 4 - dctx->tmpInSize;
1767
+ size_t const sizeToCopy = MIN(wantedData, remainingInput);
1768
+ memcpy(dctx->tmpIn + dctx->tmpInSize, srcPtr, sizeToCopy);
1769
+ srcPtr += sizeToCopy;
1770
+ dctx->tmpInSize += sizeToCopy;
1771
+ if (dctx->tmpInSize < 4) { /* not enough input to read complete suffix */
1772
+ nextSrcSizeHint = 4 - dctx->tmpInSize;
1773
+ doAnotherStage=0;
1774
+ break;
1775
+ }
1776
+ selectedIn = dctx->tmpIn;
1777
+ } /* if (dctx->dStage == dstage_storeSuffix) */
1778
+
1779
+ /* case dstage_checkSuffix: */ /* no direct entry, avoid initialization risks */
1780
+ { U32 const readCRC = LZ4F_readLE32(selectedIn);
1781
+ U32 const resultCRC = XXH32_digest(&(dctx->xxh));
1782
+ #ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
1783
+ if (readCRC != resultCRC)
1784
+ return err0r(LZ4F_ERROR_contentChecksum_invalid);
1785
+ #else
1786
+ (void)readCRC;
1787
+ (void)resultCRC;
1788
+ #endif
1789
+ nextSrcSizeHint = 0;
1790
+ LZ4F_resetDecompressionContext(dctx);
1791
+ doAnotherStage = 0;
1792
+ break;
1793
+ }
1794
+
1795
+ case dstage_getSFrameSize:
1796
+ if ((srcEnd - srcPtr) >= 4) {
1797
+ selectedIn = srcPtr;
1798
+ srcPtr += 4;
1799
+ } else {
1800
+ /* not enough input to read cBlockSize field */
1801
+ dctx->tmpInSize = 4;
1802
+ dctx->tmpInTarget = 8;
1803
+ dctx->dStage = dstage_storeSFrameSize;
1804
+ }
1805
+
1806
+ if (dctx->dStage == dstage_storeSFrameSize)
1807
+ case dstage_storeSFrameSize:
1808
+ { size_t const sizeToCopy = MIN(dctx->tmpInTarget - dctx->tmpInSize,
1809
+ (size_t)(srcEnd - srcPtr) );
1810
+ memcpy(dctx->header + dctx->tmpInSize, srcPtr, sizeToCopy);
1811
+ srcPtr += sizeToCopy;
1812
+ dctx->tmpInSize += sizeToCopy;
1813
+ if (dctx->tmpInSize < dctx->tmpInTarget) {
1814
+ /* not enough input to get full sBlockSize; wait for more */
1815
+ nextSrcSizeHint = dctx->tmpInTarget - dctx->tmpInSize;
1816
+ doAnotherStage = 0;
1817
+ break;
1818
+ }
1819
+ selectedIn = dctx->header + 4;
1820
+ } /* if (dctx->dStage == dstage_storeSFrameSize) */
1821
+
1822
+ /* case dstage_decodeSFrameSize: */ /* no direct entry */
1823
+ { size_t const SFrameSize = LZ4F_readLE32(selectedIn);
1824
+ dctx->frameInfo.contentSize = SFrameSize;
1825
+ dctx->tmpInTarget = SFrameSize;
1826
+ dctx->dStage = dstage_skipSkippable;
1827
+ break;
1828
+ }
1829
+
1830
+ case dstage_skipSkippable:
1831
+ { size_t const skipSize = MIN(dctx->tmpInTarget, (size_t)(srcEnd-srcPtr));
1832
+ srcPtr += skipSize;
1833
+ dctx->tmpInTarget -= skipSize;
1834
+ doAnotherStage = 0;
1835
+ nextSrcSizeHint = dctx->tmpInTarget;
1836
+ if (nextSrcSizeHint) break; /* still more to skip */
1837
+ /* frame fully skipped : prepare context for a new frame */
1838
+ LZ4F_resetDecompressionContext(dctx);
1839
+ break;
1840
+ }
1841
+ } /* switch (dctx->dStage) */
1842
+ } /* while (doAnotherStage) */
1843
+
1844
+ /* preserve history within tmp whenever necessary */
1845
+ LZ4F_STATIC_ASSERT((unsigned)dstage_init == 2);
1846
+ if ( (dctx->frameInfo.blockMode==LZ4F_blockLinked) /* next block will use up to 64KB from previous ones */
1847
+ && (dctx->dict != dctx->tmpOutBuffer) /* dictionary is not already within tmp */
1848
+ && (dctx->dict != NULL) /* dictionary exists */
1849
+ && (!decompressOptionsPtr->stableDst) /* cannot rely on dst data to remain there for next call */
1850
+ && ((unsigned)(dctx->dStage)-2 < (unsigned)(dstage_getSuffix)-2) ) /* valid stages : [init ... getSuffix[ */
1851
+ {
1852
+ if (dctx->dStage == dstage_flushOut) {
1853
+ size_t const preserveSize = (size_t)(dctx->tmpOut - dctx->tmpOutBuffer);
1854
+ size_t copySize = 64 KB - dctx->tmpOutSize;
1855
+ const BYTE* oldDictEnd = dctx->dict + dctx->dictSize - dctx->tmpOutStart;
1856
+ if (dctx->tmpOutSize > 64 KB) copySize = 0;
1857
+ if (copySize > preserveSize) copySize = preserveSize;
1858
+ assert(dctx->tmpOutBuffer != NULL);
1859
+
1860
+ memcpy(dctx->tmpOutBuffer + preserveSize - copySize, oldDictEnd - copySize, copySize);
1861
+
1862
+ dctx->dict = dctx->tmpOutBuffer;
1863
+ dctx->dictSize = preserveSize + dctx->tmpOutStart;
1864
+ } else {
1865
+ const BYTE* const oldDictEnd = dctx->dict + dctx->dictSize;
1866
+ size_t const newDictSize = MIN(dctx->dictSize, 64 KB);
1867
+
1868
+ memcpy(dctx->tmpOutBuffer, oldDictEnd - newDictSize, newDictSize);
1869
+
1870
+ dctx->dict = dctx->tmpOutBuffer;
1871
+ dctx->dictSize = newDictSize;
1872
+ dctx->tmpOut = dctx->tmpOutBuffer + newDictSize;
1873
+ }
1874
+ }
1875
+
1876
+ *srcSizePtr = (size_t)(srcPtr - srcStart);
1877
+ *dstSizePtr = (size_t)(dstPtr - dstStart);
1878
+ return nextSrcSizeHint;
1879
+ }
1880
+
1881
+ /*! LZ4F_decompress_usingDict() :
1882
+ * Same as LZ4F_decompress(), using a predefined dictionary.
1883
+ * Dictionary is used "in place", without any preprocessing.
1884
+ * It must remain accessible throughout the entire frame decoding.
1885
+ */
1886
+ size_t LZ4F_decompress_usingDict(LZ4F_dctx* dctx,
1887
+ void* dstBuffer, size_t* dstSizePtr,
1888
+ const void* srcBuffer, size_t* srcSizePtr,
1889
+ const void* dict, size_t dictSize,
1890
+ const LZ4F_decompressOptions_t* decompressOptionsPtr)
1891
+ {
1892
+ if (dctx->dStage <= dstage_init) {
1893
+ dctx->dict = (const BYTE*)dict;
1894
+ dctx->dictSize = dictSize;
1895
+ }
1896
+ return LZ4F_decompress(dctx, dstBuffer, dstSizePtr,
1897
+ srcBuffer, srcSizePtr,
1898
+ decompressOptionsPtr);
1899
+ }