Coverage Report

Created: 2025-08-18 06:39

/src/brotli/c/common/transform.c
Line
Count
Source (jump to first uncovered line)
1
/* Copyright 2013 Google Inc. All Rights Reserved.
2
3
   Distributed under MIT license.
4
   See file LICENSE for detail or copy at https://opensource.org/licenses/MIT
5
*/
6
7
#include "platform.h"
8
#include "transform.h"
9
10
#if defined(__cplusplus) || defined(c_plusplus)
11
extern "C" {
12
#endif
13
14
/* RFC 7932 transforms string data */
15
static const BROTLI_MODEL("small") char kPrefixSuffix[217] =
16
      "\1 \2, \10 of the \4 of \2s \1.\5 and \4 "
17
/* 0x  _0 _2  __5        _E    _3  _6 _8     _E */
18
      "in \1\"\4 to \2\">\1\n\2. \1]\5 for \3 a \6 "
19
/* 2x     _3_ _5    _A_  _D_ _F  _2 _4     _A   _E */
20
      "that \1\'\6 with \6 from \4 by \1(\6. T"
21
/* 4x       _5_ _7      _E      _5    _A _C */
22
      "he \4 on \4 as \4 is \4ing \2\n\t\1:\3ed "
23
/* 6x     _3    _8    _D    _2    _7_ _ _A _C */
24
      "\2=\"\4 at \3ly \1,\2=\'\5.com/\7. This \5"
25
/* 8x  _0 _ _3    _8   _C _E _ _1     _7       _F */
26
      " not \3er \3al \4ful \4ive \5less \4es"
27
/* Ax       _5   _9   _D    _2    _7     _D */
28
      "t \4ize \2\xc2\xa0\4ous \5 the \2e "; /* \0 - implicit trailing zero. */
29
/* Cx    _2    _7___ ___ _A    _F     _5        _8 */
30
31
static const BROTLI_MODEL("small") uint16_t kPrefixSuffixMap[50] = {
32
  0x00, 0x02, 0x05, 0x0E, 0x13, 0x16, 0x18, 0x1E, 0x23, 0x25,
33
  0x2A, 0x2D, 0x2F, 0x32, 0x34, 0x3A, 0x3E, 0x45, 0x47, 0x4E,
34
  0x55, 0x5A, 0x5C, 0x63, 0x68, 0x6D, 0x72, 0x77, 0x7A, 0x7C,
35
  0x80, 0x83, 0x88, 0x8C, 0x8E, 0x91, 0x97, 0x9F, 0xA5, 0xA9,
36
  0xAD, 0xB2, 0xB7, 0xBD, 0xC2, 0xC7, 0xCA, 0xCF, 0xD5, 0xD8
37
};
38
39
/* RFC 7932 transforms */
40
static const BROTLI_MODEL("small") uint8_t kTransformsData[] = {
41
  49, BROTLI_TRANSFORM_IDENTITY, 49,
42
  49, BROTLI_TRANSFORM_IDENTITY, 0,
43
   0, BROTLI_TRANSFORM_IDENTITY, 0,
44
  49, BROTLI_TRANSFORM_OMIT_FIRST_1, 49,
45
  49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 0,
46
  49, BROTLI_TRANSFORM_IDENTITY, 47,
47
   0, BROTLI_TRANSFORM_IDENTITY, 49,
48
   4, BROTLI_TRANSFORM_IDENTITY, 0,
49
  49, BROTLI_TRANSFORM_IDENTITY, 3,
50
  49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 49,
51
  49, BROTLI_TRANSFORM_IDENTITY, 6,
52
  49, BROTLI_TRANSFORM_OMIT_FIRST_2, 49,
53
  49, BROTLI_TRANSFORM_OMIT_LAST_1, 49,
54
   1, BROTLI_TRANSFORM_IDENTITY, 0,
55
  49, BROTLI_TRANSFORM_IDENTITY, 1,
56
   0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 0,
57
  49, BROTLI_TRANSFORM_IDENTITY, 7,
58
  49, BROTLI_TRANSFORM_IDENTITY, 9,
59
  48, BROTLI_TRANSFORM_IDENTITY, 0,
60
  49, BROTLI_TRANSFORM_IDENTITY, 8,
61
  49, BROTLI_TRANSFORM_IDENTITY, 5,
62
  49, BROTLI_TRANSFORM_IDENTITY, 10,
63
  49, BROTLI_TRANSFORM_IDENTITY, 11,
64
  49, BROTLI_TRANSFORM_OMIT_LAST_3, 49,
65
  49, BROTLI_TRANSFORM_IDENTITY, 13,
66
  49, BROTLI_TRANSFORM_IDENTITY, 14,
67
  49, BROTLI_TRANSFORM_OMIT_FIRST_3, 49,
68
  49, BROTLI_TRANSFORM_OMIT_LAST_2, 49,
69
  49, BROTLI_TRANSFORM_IDENTITY, 15,
70
  49, BROTLI_TRANSFORM_IDENTITY, 16,
71
   0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 49,
72
  49, BROTLI_TRANSFORM_IDENTITY, 12,
73
   5, BROTLI_TRANSFORM_IDENTITY, 49,
74
   0, BROTLI_TRANSFORM_IDENTITY, 1,
75
  49, BROTLI_TRANSFORM_OMIT_FIRST_4, 49,
76
  49, BROTLI_TRANSFORM_IDENTITY, 18,
77
  49, BROTLI_TRANSFORM_IDENTITY, 17,
78
  49, BROTLI_TRANSFORM_IDENTITY, 19,
79
  49, BROTLI_TRANSFORM_IDENTITY, 20,
80
  49, BROTLI_TRANSFORM_OMIT_FIRST_5, 49,
81
  49, BROTLI_TRANSFORM_OMIT_FIRST_6, 49,
82
  47, BROTLI_TRANSFORM_IDENTITY, 49,
83
  49, BROTLI_TRANSFORM_OMIT_LAST_4, 49,
84
  49, BROTLI_TRANSFORM_IDENTITY, 22,
85
  49, BROTLI_TRANSFORM_UPPERCASE_ALL, 49,
86
  49, BROTLI_TRANSFORM_IDENTITY, 23,
87
  49, BROTLI_TRANSFORM_IDENTITY, 24,
88
  49, BROTLI_TRANSFORM_IDENTITY, 25,
89
  49, BROTLI_TRANSFORM_OMIT_LAST_7, 49,
90
  49, BROTLI_TRANSFORM_OMIT_LAST_1, 26,
91
  49, BROTLI_TRANSFORM_IDENTITY, 27,
92
  49, BROTLI_TRANSFORM_IDENTITY, 28,
93
   0, BROTLI_TRANSFORM_IDENTITY, 12,
94
  49, BROTLI_TRANSFORM_IDENTITY, 29,
95
  49, BROTLI_TRANSFORM_OMIT_FIRST_9, 49,
96
  49, BROTLI_TRANSFORM_OMIT_FIRST_7, 49,
97
  49, BROTLI_TRANSFORM_OMIT_LAST_6, 49,
98
  49, BROTLI_TRANSFORM_IDENTITY, 21,
99
  49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 1,
100
  49, BROTLI_TRANSFORM_OMIT_LAST_8, 49,
101
  49, BROTLI_TRANSFORM_IDENTITY, 31,
102
  49, BROTLI_TRANSFORM_IDENTITY, 32,
103
  47, BROTLI_TRANSFORM_IDENTITY, 3,
104
  49, BROTLI_TRANSFORM_OMIT_LAST_5, 49,
105
  49, BROTLI_TRANSFORM_OMIT_LAST_9, 49,
106
   0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 1,
107
  49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 8,
108
   5, BROTLI_TRANSFORM_IDENTITY, 21,
109
  49, BROTLI_TRANSFORM_UPPERCASE_ALL, 0,
110
  49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 10,
111
  49, BROTLI_TRANSFORM_IDENTITY, 30,
112
   0, BROTLI_TRANSFORM_IDENTITY, 5,
113
  35, BROTLI_TRANSFORM_IDENTITY, 49,
114
  47, BROTLI_TRANSFORM_IDENTITY, 2,
115
  49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 17,
116
  49, BROTLI_TRANSFORM_IDENTITY, 36,
117
  49, BROTLI_TRANSFORM_IDENTITY, 33,
118
   5, BROTLI_TRANSFORM_IDENTITY, 0,
119
  49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 21,
120
  49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 5,
121
  49, BROTLI_TRANSFORM_IDENTITY, 37,
122
   0, BROTLI_TRANSFORM_IDENTITY, 30,
123
  49, BROTLI_TRANSFORM_IDENTITY, 38,
124
   0, BROTLI_TRANSFORM_UPPERCASE_ALL, 0,
125
  49, BROTLI_TRANSFORM_IDENTITY, 39,
126
   0, BROTLI_TRANSFORM_UPPERCASE_ALL, 49,
127
  49, BROTLI_TRANSFORM_IDENTITY, 34,
128
  49, BROTLI_TRANSFORM_UPPERCASE_ALL, 8,
129
  49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 12,
130
   0, BROTLI_TRANSFORM_IDENTITY, 21,
131
  49, BROTLI_TRANSFORM_IDENTITY, 40,
132
   0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 12,
133
  49, BROTLI_TRANSFORM_IDENTITY, 41,
134
  49, BROTLI_TRANSFORM_IDENTITY, 42,
135
  49, BROTLI_TRANSFORM_UPPERCASE_ALL, 17,
136
  49, BROTLI_TRANSFORM_IDENTITY, 43,
137
   0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 5,
138
  49, BROTLI_TRANSFORM_UPPERCASE_ALL, 10,
139
   0, BROTLI_TRANSFORM_IDENTITY, 34,
140
  49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 33,
141
  49, BROTLI_TRANSFORM_IDENTITY, 44,
142
  49, BROTLI_TRANSFORM_UPPERCASE_ALL, 5,
143
  45, BROTLI_TRANSFORM_IDENTITY, 49,
144
   0, BROTLI_TRANSFORM_IDENTITY, 33,
145
  49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 30,
146
  49, BROTLI_TRANSFORM_UPPERCASE_ALL, 30,
147
  49, BROTLI_TRANSFORM_IDENTITY, 46,
148
  49, BROTLI_TRANSFORM_UPPERCASE_ALL, 1,
149
  49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 34,
150
   0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 33,
151
   0, BROTLI_TRANSFORM_UPPERCASE_ALL, 30,
152
   0, BROTLI_TRANSFORM_UPPERCASE_ALL, 1,
153
  49, BROTLI_TRANSFORM_UPPERCASE_ALL, 33,
154
  49, BROTLI_TRANSFORM_UPPERCASE_ALL, 21,
155
  49, BROTLI_TRANSFORM_UPPERCASE_ALL, 12,
156
   0, BROTLI_TRANSFORM_UPPERCASE_ALL, 5,
157
  49, BROTLI_TRANSFORM_UPPERCASE_ALL, 34,
158
   0, BROTLI_TRANSFORM_UPPERCASE_ALL, 12,
159
   0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 30,
160
   0, BROTLI_TRANSFORM_UPPERCASE_ALL, 34,
161
   0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 34,
162
};
163
164
static const BROTLI_MODEL("small")
165
BrotliTransforms kBrotliTransforms = {
166
  sizeof(kPrefixSuffix),
167
  (const uint8_t*)kPrefixSuffix,
168
  kPrefixSuffixMap,
169
  sizeof(kTransformsData) / (3 * sizeof(kTransformsData[0])),
170
  kTransformsData,
171
  NULL,  /* no extra parameters */
172
  {0, 12, 27, 23, 42, 63, 56, 48, 59, 64}
173
};
174
175
5.34k
const BrotliTransforms* BrotliGetTransforms(void) {
176
5.34k
  return &kBrotliTransforms;
177
5.34k
}
178
179
18.0k
static int ToUpperCase(uint8_t* p) {
180
18.0k
  if (p[0] < 0xC0) {
181
13.9k
    if (p[0] >= 'a' && p[0] <= 'z') {
182
9.83k
      p[0] ^= 32;
183
9.83k
    }
184
13.9k
    return 1;
185
13.9k
  }
186
  /* An overly simplified uppercasing model for UTF-8. */
187
4.04k
  if (p[0] < 0xE0) {
188
1.98k
    p[1] ^= 32;
189
1.98k
    return 2;
190
1.98k
  }
191
  /* An arbitrary transform for three byte characters. */
192
2.05k
  p[2] ^= 5;
193
2.05k
  return 3;
194
4.04k
}
195
196
0
static int Shift(uint8_t* word, int word_len, uint16_t parameter) {
197
  /* Limited sign extension: scalar < (1 << 24). */
198
0
  uint32_t scalar =
199
0
      (parameter & 0x7FFFu) + (0x1000000u - (parameter & 0x8000u));
200
0
  if (word[0] < 0x80) {
201
    /* 1-byte rune / 0sssssss / 7 bit scalar (ASCII). */
202
0
    scalar += (uint32_t)word[0];
203
0
    word[0] = (uint8_t)(scalar & 0x7Fu);
204
0
    return 1;
205
0
  } else if (word[0] < 0xC0) {
206
    /* Continuation / 10AAAAAA. */
207
0
    return 1;
208
0
  } else if (word[0] < 0xE0) {
209
    /* 2-byte rune / 110sssss AAssssss / 11 bit scalar. */
210
0
    if (word_len < 2) return 1;
211
0
    scalar += (uint32_t)((word[1] & 0x3Fu) | ((word[0] & 0x1Fu) << 6u));
212
0
    word[0] = (uint8_t)(0xC0 | ((scalar >> 6u) & 0x1F));
213
0
    word[1] = (uint8_t)((word[1] & 0xC0) | (scalar & 0x3F));
214
0
    return 2;
215
0
  } else if (word[0] < 0xF0) {
216
    /* 3-byte rune / 1110ssss AAssssss BBssssss / 16 bit scalar. */
217
0
    if (word_len < 3) return word_len;
218
0
    scalar += (uint32_t)((word[2] & 0x3Fu) | ((word[1] & 0x3Fu) << 6u) |
219
0
        ((word[0] & 0x0Fu) << 12u));
220
0
    word[0] = (uint8_t)(0xE0 | ((scalar >> 12u) & 0x0F));
221
0
    word[1] = (uint8_t)((word[1] & 0xC0) | ((scalar >> 6u) & 0x3F));
222
0
    word[2] = (uint8_t)((word[2] & 0xC0) | (scalar & 0x3F));
223
0
    return 3;
224
0
  } else if (word[0] < 0xF8) {
225
    /* 4-byte rune / 11110sss AAssssss BBssssss CCssssss / 21 bit scalar. */
226
0
    if (word_len < 4) return word_len;
227
0
    scalar += (uint32_t)((word[3] & 0x3Fu) | ((word[2] & 0x3Fu) << 6u) |
228
0
        ((word[1] & 0x3Fu) << 12u) | ((word[0] & 0x07u) << 18u));
229
0
    word[0] = (uint8_t)(0xF0 | ((scalar >> 18u) & 0x07));
230
0
    word[1] = (uint8_t)((word[1] & 0xC0) | ((scalar >> 12u) & 0x3F));
231
0
    word[2] = (uint8_t)((word[2] & 0xC0) | ((scalar >> 6u) & 0x3F));
232
0
    word[3] = (uint8_t)((word[3] & 0xC0) | (scalar & 0x3F));
233
0
    return 4;
234
0
  }
235
0
  return 1;
236
0
}
237
238
int BrotliTransformDictionaryWord(uint8_t* dst, const uint8_t* word, int len,
239
31.0k
    const BrotliTransforms* transforms, int transform_idx) {
240
31.0k
  int idx = 0;
241
31.0k
  const uint8_t* prefix = BROTLI_TRANSFORM_PREFIX(transforms, transform_idx);
242
31.0k
  uint8_t type = BROTLI_TRANSFORM_TYPE(transforms, transform_idx);
243
31.0k
  const uint8_t* suffix = BROTLI_TRANSFORM_SUFFIX(transforms, transform_idx);
244
31.0k
  {
245
31.0k
    int prefix_len = *prefix++;
246
48.5k
    while (prefix_len--) { dst[idx++] = *prefix++; }
247
31.0k
  }
248
31.0k
  {
249
31.0k
    const int t = type;
250
31.0k
    int i = 0;
251
31.0k
    if (t <= BROTLI_TRANSFORM_OMIT_LAST_9) {
252
21.2k
      len -= t;
253
21.2k
    } else if (t >= BROTLI_TRANSFORM_OMIT_FIRST_1
254
9.84k
        && t <= BROTLI_TRANSFORM_OMIT_FIRST_9) {
255
2.66k
      int skip = t - (BROTLI_TRANSFORM_OMIT_FIRST_1 - 1);
256
2.66k
      word += skip;
257
2.66k
      len -= skip;
258
2.66k
    }
259
200k
    while (i < len) { dst[idx++] = word[i++]; }
260
31.0k
    if (t == BROTLI_TRANSFORM_UPPERCASE_FIRST) {
261
4.84k
      ToUpperCase(&dst[idx - len]);
262
26.2k
    } else if (t == BROTLI_TRANSFORM_UPPERCASE_ALL) {
263
2.33k
      uint8_t* uppercase = &dst[idx - len];
264
15.4k
      while (len > 0) {
265
13.1k
        int step = ToUpperCase(uppercase);
266
13.1k
        uppercase += step;
267
13.1k
        len -= step;
268
13.1k
      }
269
23.9k
    } else if (t == BROTLI_TRANSFORM_SHIFT_FIRST) {
270
0
      uint16_t param = (uint16_t)(transforms->params[transform_idx * 2]
271
0
          + (transforms->params[transform_idx * 2 + 1] << 8u));
272
0
      Shift(&dst[idx - len], len, param);
273
23.9k
    } else if (t == BROTLI_TRANSFORM_SHIFT_ALL) {
274
0
      uint16_t param = (uint16_t)(transforms->params[transform_idx * 2]
275
0
          + (transforms->params[transform_idx * 2 + 1] << 8u));
276
0
      uint8_t* shift = &dst[idx - len];
277
0
      while (len > 0) {
278
0
        int step = Shift(shift, len, param);
279
0
        shift += step;
280
0
        len -= step;
281
0
      }
282
0
    }
283
31.0k
  }
284
31.0k
  {
285
31.0k
    int suffix_len = *suffix++;
286
87.6k
    while (suffix_len--) { dst[idx++] = *suffix++; }
287
31.0k
    return idx;
288
31.0k
  }
289
31.0k
}
290
291
#if defined(__cplusplus) || defined(c_plusplus)
292
}  /* extern "C" */
293
#endif