/src/h2o/deps/brotli/c/enc/static_dict.c
Line | Count | Source |
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 "./static_dict.h" |
8 | | |
9 | | #include "../common/dictionary.h" |
10 | | #include "./find_match_length.h" |
11 | | #include "./port.h" |
12 | | #include "./static_dict_lut.h" |
13 | | |
14 | | #if defined(__cplusplus) || defined(c_plusplus) |
15 | | extern "C" { |
16 | | #endif |
17 | | |
18 | | static const uint8_t kUppercaseFirst = 10; |
19 | | static const uint8_t kOmitLastNTransforms[10] = { |
20 | | 0, 12, 27, 23, 42, 63, 56, 48, 59, 64, |
21 | | }; |
22 | | |
23 | 0 | static BROTLI_INLINE uint32_t Hash(const uint8_t *data) { |
24 | 0 | uint32_t h = BROTLI_UNALIGNED_LOAD32(data) * kDictHashMul32; |
25 | | /* The higher bits contain more mixture from the multiplication, |
26 | | so we take our results from there. */ |
27 | 0 | return h >> (32 - kDictNumBits); |
28 | 0 | } |
29 | | |
30 | | static BROTLI_INLINE void AddMatch(size_t distance, size_t len, size_t len_code, |
31 | 0 | uint32_t* matches) { |
32 | 0 | uint32_t match = (uint32_t)((distance << 5) + len_code); |
33 | 0 | matches[len] = BROTLI_MIN(uint32_t, matches[len], match); |
34 | 0 | } |
35 | | |
36 | | static BROTLI_INLINE size_t DictMatchLength(const BrotliDictionary* dictionary, |
37 | | const uint8_t* data, |
38 | | size_t id, |
39 | | size_t len, |
40 | 0 | size_t maxlen) { |
41 | 0 | const size_t offset = dictionary->offsets_by_length[len] + len * id; |
42 | 0 | return FindMatchLengthWithLimit(&dictionary->data[offset], data, |
43 | 0 | BROTLI_MIN(size_t, len, maxlen)); |
44 | 0 | } |
45 | | |
46 | | static BROTLI_INLINE BROTLI_BOOL IsMatch(const BrotliDictionary* dictionary, |
47 | 0 | DictWord w, const uint8_t* data, size_t max_length) { |
48 | 0 | if (w.len > max_length) { |
49 | 0 | return BROTLI_FALSE; |
50 | 0 | } else { |
51 | 0 | const size_t offset = dictionary->offsets_by_length[w.len] + |
52 | 0 | (size_t)w.len * (size_t)w.idx; |
53 | 0 | const uint8_t* dict = &dictionary->data[offset]; |
54 | 0 | if (w.transform == 0) { |
55 | | /* Match against base dictionary word. */ |
56 | 0 | return |
57 | 0 | TO_BROTLI_BOOL(FindMatchLengthWithLimit(dict, data, w.len) == w.len); |
58 | 0 | } else if (w.transform == 10) { |
59 | | /* Match against uppercase first transform. |
60 | | Note that there are only ASCII uppercase words in the lookup table. */ |
61 | 0 | return TO_BROTLI_BOOL(dict[0] >= 'a' && dict[0] <= 'z' && |
62 | 0 | (dict[0] ^ 32) == data[0] && |
63 | 0 | FindMatchLengthWithLimit(&dict[1], &data[1], w.len - 1u) == |
64 | 0 | w.len - 1u); |
65 | 0 | } else { |
66 | | /* Match against uppercase all transform. |
67 | | Note that there are only ASCII uppercase words in the lookup table. */ |
68 | 0 | size_t i; |
69 | 0 | for (i = 0; i < w.len; ++i) { |
70 | 0 | if (dict[i] >= 'a' && dict[i] <= 'z') { |
71 | 0 | if ((dict[i] ^ 32) != data[i]) return BROTLI_FALSE; |
72 | 0 | } else { |
73 | 0 | if (dict[i] != data[i]) return BROTLI_FALSE; |
74 | 0 | } |
75 | 0 | } |
76 | 0 | return BROTLI_TRUE; |
77 | 0 | } |
78 | 0 | } |
79 | 0 | } |
80 | | |
81 | | BROTLI_BOOL BrotliFindAllStaticDictionaryMatches( |
82 | | const BrotliDictionary* dictionary, const uint8_t* data, size_t min_length, |
83 | 0 | size_t max_length, uint32_t* matches) { |
84 | 0 | BROTLI_BOOL has_found_match = BROTLI_FALSE; |
85 | 0 | { |
86 | 0 | size_t offset = kStaticDictionaryBuckets[Hash(data)]; |
87 | 0 | BROTLI_BOOL end = !offset; |
88 | 0 | while (!end) { |
89 | 0 | DictWord w = kStaticDictionaryWords[offset++]; |
90 | 0 | const size_t l = w.len & 0x1F; |
91 | 0 | const size_t n = (size_t)1 << dictionary->size_bits_by_length[l]; |
92 | 0 | const size_t id = w.idx; |
93 | 0 | end = !!(w.len & 0x80); |
94 | 0 | w.len = (uint8_t)l; |
95 | 0 | if (w.transform == 0) { |
96 | 0 | const size_t matchlen = |
97 | 0 | DictMatchLength(dictionary, data, id, l, max_length); |
98 | 0 | const uint8_t* s; |
99 | 0 | size_t minlen; |
100 | 0 | size_t maxlen; |
101 | 0 | size_t len; |
102 | | /* Transform "" + kIdentity + "" */ |
103 | 0 | if (matchlen == l) { |
104 | 0 | AddMatch(id, l, l, matches); |
105 | 0 | has_found_match = BROTLI_TRUE; |
106 | 0 | } |
107 | | /* Transforms "" + kOmitLast1 + "" and "" + kOmitLast1 + "ing " */ |
108 | 0 | if (matchlen >= l - 1) { |
109 | 0 | AddMatch(id + 12 * n, l - 1, l, matches); |
110 | 0 | if (l + 2 < max_length && |
111 | 0 | data[l - 1] == 'i' && data[l] == 'n' && data[l + 1] == 'g' && |
112 | 0 | data[l + 2] == ' ') { |
113 | 0 | AddMatch(id + 49 * n, l + 3, l, matches); |
114 | 0 | } |
115 | 0 | has_found_match = BROTLI_TRUE; |
116 | 0 | } |
117 | | /* Transform "" + kOmitLastN + "" (N = 2 .. 9) */ |
118 | 0 | minlen = min_length; |
119 | 0 | if (l > 9) minlen = BROTLI_MAX(size_t, minlen, l - 9); |
120 | 0 | maxlen = BROTLI_MIN(size_t, matchlen, l - 2); |
121 | 0 | for (len = minlen; len <= maxlen; ++len) { |
122 | 0 | AddMatch(id + kOmitLastNTransforms[l - len] * n, len, l, matches); |
123 | 0 | has_found_match = BROTLI_TRUE; |
124 | 0 | } |
125 | 0 | if (matchlen < l || l + 6 >= max_length) { |
126 | 0 | continue; |
127 | 0 | } |
128 | 0 | s = &data[l]; |
129 | | /* Transforms "" + kIdentity + <suffix> */ |
130 | 0 | if (s[0] == ' ') { |
131 | 0 | AddMatch(id + n, l + 1, l, matches); |
132 | 0 | if (s[1] == 'a') { |
133 | 0 | if (s[2] == ' ') { |
134 | 0 | AddMatch(id + 28 * n, l + 3, l, matches); |
135 | 0 | } else if (s[2] == 's') { |
136 | 0 | if (s[3] == ' ') AddMatch(id + 46 * n, l + 4, l, matches); |
137 | 0 | } else if (s[2] == 't') { |
138 | 0 | if (s[3] == ' ') AddMatch(id + 60 * n, l + 4, l, matches); |
139 | 0 | } else if (s[2] == 'n') { |
140 | 0 | if (s[3] == 'd' && s[4] == ' ') { |
141 | 0 | AddMatch(id + 10 * n, l + 5, l, matches); |
142 | 0 | } |
143 | 0 | } |
144 | 0 | } else if (s[1] == 'b') { |
145 | 0 | if (s[2] == 'y' && s[3] == ' ') { |
146 | 0 | AddMatch(id + 38 * n, l + 4, l, matches); |
147 | 0 | } |
148 | 0 | } else if (s[1] == 'i') { |
149 | 0 | if (s[2] == 'n') { |
150 | 0 | if (s[3] == ' ') AddMatch(id + 16 * n, l + 4, l, matches); |
151 | 0 | } else if (s[2] == 's') { |
152 | 0 | if (s[3] == ' ') AddMatch(id + 47 * n, l + 4, l, matches); |
153 | 0 | } |
154 | 0 | } else if (s[1] == 'f') { |
155 | 0 | if (s[2] == 'o') { |
156 | 0 | if (s[3] == 'r' && s[4] == ' ') { |
157 | 0 | AddMatch(id + 25 * n, l + 5, l, matches); |
158 | 0 | } |
159 | 0 | } else if (s[2] == 'r') { |
160 | 0 | if (s[3] == 'o' && s[4] == 'm' && s[5] == ' ') { |
161 | 0 | AddMatch(id + 37 * n, l + 6, l, matches); |
162 | 0 | } |
163 | 0 | } |
164 | 0 | } else if (s[1] == 'o') { |
165 | 0 | if (s[2] == 'f') { |
166 | 0 | if (s[3] == ' ') AddMatch(id + 8 * n, l + 4, l, matches); |
167 | 0 | } else if (s[2] == 'n') { |
168 | 0 | if (s[3] == ' ') AddMatch(id + 45 * n, l + 4, l, matches); |
169 | 0 | } |
170 | 0 | } else if (s[1] == 'n') { |
171 | 0 | if (s[2] == 'o' && s[3] == 't' && s[4] == ' ') { |
172 | 0 | AddMatch(id + 80 * n, l + 5, l, matches); |
173 | 0 | } |
174 | 0 | } else if (s[1] == 't') { |
175 | 0 | if (s[2] == 'h') { |
176 | 0 | if (s[3] == 'e') { |
177 | 0 | if (s[4] == ' ') AddMatch(id + 5 * n, l + 5, l, matches); |
178 | 0 | } else if (s[3] == 'a') { |
179 | 0 | if (s[4] == 't' && s[5] == ' ') { |
180 | 0 | AddMatch(id + 29 * n, l + 6, l, matches); |
181 | 0 | } |
182 | 0 | } |
183 | 0 | } else if (s[2] == 'o') { |
184 | 0 | if (s[3] == ' ') AddMatch(id + 17 * n, l + 4, l, matches); |
185 | 0 | } |
186 | 0 | } else if (s[1] == 'w') { |
187 | 0 | if (s[2] == 'i' && s[3] == 't' && s[4] == 'h' && s[5] == ' ') { |
188 | 0 | AddMatch(id + 35 * n, l + 6, l, matches); |
189 | 0 | } |
190 | 0 | } |
191 | 0 | } else if (s[0] == '"') { |
192 | 0 | AddMatch(id + 19 * n, l + 1, l, matches); |
193 | 0 | if (s[1] == '>') { |
194 | 0 | AddMatch(id + 21 * n, l + 2, l, matches); |
195 | 0 | } |
196 | 0 | } else if (s[0] == '.') { |
197 | 0 | AddMatch(id + 20 * n, l + 1, l, matches); |
198 | 0 | if (s[1] == ' ') { |
199 | 0 | AddMatch(id + 31 * n, l + 2, l, matches); |
200 | 0 | if (s[2] == 'T' && s[3] == 'h') { |
201 | 0 | if (s[4] == 'e') { |
202 | 0 | if (s[5] == ' ') AddMatch(id + 43 * n, l + 6, l, matches); |
203 | 0 | } else if (s[4] == 'i') { |
204 | 0 | if (s[5] == 's' && s[6] == ' ') { |
205 | 0 | AddMatch(id + 75 * n, l + 7, l, matches); |
206 | 0 | } |
207 | 0 | } |
208 | 0 | } |
209 | 0 | } |
210 | 0 | } else if (s[0] == ',') { |
211 | 0 | AddMatch(id + 76 * n, l + 1, l, matches); |
212 | 0 | if (s[1] == ' ') { |
213 | 0 | AddMatch(id + 14 * n, l + 2, l, matches); |
214 | 0 | } |
215 | 0 | } else if (s[0] == '\n') { |
216 | 0 | AddMatch(id + 22 * n, l + 1, l, matches); |
217 | 0 | if (s[1] == '\t') { |
218 | 0 | AddMatch(id + 50 * n, l + 2, l, matches); |
219 | 0 | } |
220 | 0 | } else if (s[0] == ']') { |
221 | 0 | AddMatch(id + 24 * n, l + 1, l, matches); |
222 | 0 | } else if (s[0] == '\'') { |
223 | 0 | AddMatch(id + 36 * n, l + 1, l, matches); |
224 | 0 | } else if (s[0] == ':') { |
225 | 0 | AddMatch(id + 51 * n, l + 1, l, matches); |
226 | 0 | } else if (s[0] == '(') { |
227 | 0 | AddMatch(id + 57 * n, l + 1, l, matches); |
228 | 0 | } else if (s[0] == '=') { |
229 | 0 | if (s[1] == '"') { |
230 | 0 | AddMatch(id + 70 * n, l + 2, l, matches); |
231 | 0 | } else if (s[1] == '\'') { |
232 | 0 | AddMatch(id + 86 * n, l + 2, l, matches); |
233 | 0 | } |
234 | 0 | } else if (s[0] == 'a') { |
235 | 0 | if (s[1] == 'l' && s[2] == ' ') { |
236 | 0 | AddMatch(id + 84 * n, l + 3, l, matches); |
237 | 0 | } |
238 | 0 | } else if (s[0] == 'e') { |
239 | 0 | if (s[1] == 'd') { |
240 | 0 | if (s[2] == ' ') AddMatch(id + 53 * n, l + 3, l, matches); |
241 | 0 | } else if (s[1] == 'r') { |
242 | 0 | if (s[2] == ' ') AddMatch(id + 82 * n, l + 3, l, matches); |
243 | 0 | } else if (s[1] == 's') { |
244 | 0 | if (s[2] == 't' && s[3] == ' ') { |
245 | 0 | AddMatch(id + 95 * n, l + 4, l, matches); |
246 | 0 | } |
247 | 0 | } |
248 | 0 | } else if (s[0] == 'f') { |
249 | 0 | if (s[1] == 'u' && s[2] == 'l' && s[3] == ' ') { |
250 | 0 | AddMatch(id + 90 * n, l + 4, l, matches); |
251 | 0 | } |
252 | 0 | } else if (s[0] == 'i') { |
253 | 0 | if (s[1] == 'v') { |
254 | 0 | if (s[2] == 'e' && s[3] == ' ') { |
255 | 0 | AddMatch(id + 92 * n, l + 4, l, matches); |
256 | 0 | } |
257 | 0 | } else if (s[1] == 'z') { |
258 | 0 | if (s[2] == 'e' && s[3] == ' ') { |
259 | 0 | AddMatch(id + 100 * n, l + 4, l, matches); |
260 | 0 | } |
261 | 0 | } |
262 | 0 | } else if (s[0] == 'l') { |
263 | 0 | if (s[1] == 'e') { |
264 | 0 | if (s[2] == 's' && s[3] == 's' && s[4] == ' ') { |
265 | 0 | AddMatch(id + 93 * n, l + 5, l, matches); |
266 | 0 | } |
267 | 0 | } else if (s[1] == 'y') { |
268 | 0 | if (s[2] == ' ') AddMatch(id + 61 * n, l + 3, l, matches); |
269 | 0 | } |
270 | 0 | } else if (s[0] == 'o') { |
271 | 0 | if (s[1] == 'u' && s[2] == 's' && s[3] == ' ') { |
272 | 0 | AddMatch(id + 106 * n, l + 4, l, matches); |
273 | 0 | } |
274 | 0 | } |
275 | 0 | } else { |
276 | | /* Set is_all_caps=0 for kUppercaseFirst and |
277 | | is_all_caps=1 otherwise (kUppercaseAll) transform. */ |
278 | 0 | const BROTLI_BOOL is_all_caps = |
279 | 0 | TO_BROTLI_BOOL(w.transform != kUppercaseFirst); |
280 | 0 | const uint8_t* s; |
281 | 0 | if (!IsMatch(dictionary, w, data, max_length)) { |
282 | 0 | continue; |
283 | 0 | } |
284 | | /* Transform "" + kUppercase{First,All} + "" */ |
285 | 0 | AddMatch(id + (is_all_caps ? 44 : 9) * n, l, l, matches); |
286 | 0 | has_found_match = BROTLI_TRUE; |
287 | 0 | if (l + 1 >= max_length) { |
288 | 0 | continue; |
289 | 0 | } |
290 | | /* Transforms "" + kUppercase{First,All} + <suffix> */ |
291 | 0 | s = &data[l]; |
292 | 0 | if (s[0] == ' ') { |
293 | 0 | AddMatch(id + (is_all_caps ? 68 : 4) * n, l + 1, l, matches); |
294 | 0 | } else if (s[0] == '"') { |
295 | 0 | AddMatch(id + (is_all_caps ? 87 : 66) * n, l + 1, l, matches); |
296 | 0 | if (s[1] == '>') { |
297 | 0 | AddMatch(id + (is_all_caps ? 97 : 69) * n, l + 2, l, matches); |
298 | 0 | } |
299 | 0 | } else if (s[0] == '.') { |
300 | 0 | AddMatch(id + (is_all_caps ? 101 : 79) * n, l + 1, l, matches); |
301 | 0 | if (s[1] == ' ') { |
302 | 0 | AddMatch(id + (is_all_caps ? 114 : 88) * n, l + 2, l, matches); |
303 | 0 | } |
304 | 0 | } else if (s[0] == ',') { |
305 | 0 | AddMatch(id + (is_all_caps ? 112 : 99) * n, l + 1, l, matches); |
306 | 0 | if (s[1] == ' ') { |
307 | 0 | AddMatch(id + (is_all_caps ? 107 : 58) * n, l + 2, l, matches); |
308 | 0 | } |
309 | 0 | } else if (s[0] == '\'') { |
310 | 0 | AddMatch(id + (is_all_caps ? 94 : 74) * n, l + 1, l, matches); |
311 | 0 | } else if (s[0] == '(') { |
312 | 0 | AddMatch(id + (is_all_caps ? 113 : 78) * n, l + 1, l, matches); |
313 | 0 | } else if (s[0] == '=') { |
314 | 0 | if (s[1] == '"') { |
315 | 0 | AddMatch(id + (is_all_caps ? 105 : 104) * n, l + 2, l, matches); |
316 | 0 | } else if (s[1] == '\'') { |
317 | 0 | AddMatch(id + (is_all_caps ? 116 : 108) * n, l + 2, l, matches); |
318 | 0 | } |
319 | 0 | } |
320 | 0 | } |
321 | 0 | } |
322 | 0 | } |
323 | | /* Transforms with prefixes " " and "." */ |
324 | 0 | if (max_length >= 5 && (data[0] == ' ' || data[0] == '.')) { |
325 | 0 | BROTLI_BOOL is_space = TO_BROTLI_BOOL(data[0] == ' '); |
326 | 0 | size_t offset = kStaticDictionaryBuckets[Hash(&data[1])]; |
327 | 0 | BROTLI_BOOL end = !offset; |
328 | 0 | while (!end) { |
329 | 0 | DictWord w = kStaticDictionaryWords[offset++]; |
330 | 0 | const size_t l = w.len & 0x1F; |
331 | 0 | const size_t n = (size_t)1 << dictionary->size_bits_by_length[l]; |
332 | 0 | const size_t id = w.idx; |
333 | 0 | end = !!(w.len & 0x80); |
334 | 0 | w.len = (uint8_t)l; |
335 | 0 | if (w.transform == 0) { |
336 | 0 | const uint8_t* s; |
337 | 0 | if (!IsMatch(dictionary, w, &data[1], max_length - 1)) { |
338 | 0 | continue; |
339 | 0 | } |
340 | | /* Transforms " " + kIdentity + "" and "." + kIdentity + "" */ |
341 | 0 | AddMatch(id + (is_space ? 6 : 32) * n, l + 1, l, matches); |
342 | 0 | has_found_match = BROTLI_TRUE; |
343 | 0 | if (l + 2 >= max_length) { |
344 | 0 | continue; |
345 | 0 | } |
346 | | /* Transforms " " + kIdentity + <suffix> and "." + kIdentity + <suffix> |
347 | | */ |
348 | 0 | s = &data[l + 1]; |
349 | 0 | if (s[0] == ' ') { |
350 | 0 | AddMatch(id + (is_space ? 2 : 77) * n, l + 2, l, matches); |
351 | 0 | } else if (s[0] == '(') { |
352 | 0 | AddMatch(id + (is_space ? 89 : 67) * n, l + 2, l, matches); |
353 | 0 | } else if (is_space) { |
354 | 0 | if (s[0] == ',') { |
355 | 0 | AddMatch(id + 103 * n, l + 2, l, matches); |
356 | 0 | if (s[1] == ' ') { |
357 | 0 | AddMatch(id + 33 * n, l + 3, l, matches); |
358 | 0 | } |
359 | 0 | } else if (s[0] == '.') { |
360 | 0 | AddMatch(id + 71 * n, l + 2, l, matches); |
361 | 0 | if (s[1] == ' ') { |
362 | 0 | AddMatch(id + 52 * n, l + 3, l, matches); |
363 | 0 | } |
364 | 0 | } else if (s[0] == '=') { |
365 | 0 | if (s[1] == '"') { |
366 | 0 | AddMatch(id + 81 * n, l + 3, l, matches); |
367 | 0 | } else if (s[1] == '\'') { |
368 | 0 | AddMatch(id + 98 * n, l + 3, l, matches); |
369 | 0 | } |
370 | 0 | } |
371 | 0 | } |
372 | 0 | } else if (is_space) { |
373 | | /* Set is_all_caps=0 for kUppercaseFirst and |
374 | | is_all_caps=1 otherwise (kUppercaseAll) transform. */ |
375 | 0 | const BROTLI_BOOL is_all_caps = |
376 | 0 | TO_BROTLI_BOOL(w.transform != kUppercaseFirst); |
377 | 0 | const uint8_t* s; |
378 | 0 | if (!IsMatch(dictionary, w, &data[1], max_length - 1)) { |
379 | 0 | continue; |
380 | 0 | } |
381 | | /* Transforms " " + kUppercase{First,All} + "" */ |
382 | 0 | AddMatch(id + (is_all_caps ? 85 : 30) * n, l + 1, l, matches); |
383 | 0 | has_found_match = BROTLI_TRUE; |
384 | 0 | if (l + 2 >= max_length) { |
385 | 0 | continue; |
386 | 0 | } |
387 | | /* Transforms " " + kUppercase{First,All} + <suffix> */ |
388 | 0 | s = &data[l + 1]; |
389 | 0 | if (s[0] == ' ') { |
390 | 0 | AddMatch(id + (is_all_caps ? 83 : 15) * n, l + 2, l, matches); |
391 | 0 | } else if (s[0] == ',') { |
392 | 0 | if (!is_all_caps) { |
393 | 0 | AddMatch(id + 109 * n, l + 2, l, matches); |
394 | 0 | } |
395 | 0 | if (s[1] == ' ') { |
396 | 0 | AddMatch(id + (is_all_caps ? 111 : 65) * n, l + 3, l, matches); |
397 | 0 | } |
398 | 0 | } else if (s[0] == '.') { |
399 | 0 | AddMatch(id + (is_all_caps ? 115 : 96) * n, l + 2, l, matches); |
400 | 0 | if (s[1] == ' ') { |
401 | 0 | AddMatch(id + (is_all_caps ? 117 : 91) * n, l + 3, l, matches); |
402 | 0 | } |
403 | 0 | } else if (s[0] == '=') { |
404 | 0 | if (s[1] == '"') { |
405 | 0 | AddMatch(id + (is_all_caps ? 110 : 118) * n, l + 3, l, matches); |
406 | 0 | } else if (s[1] == '\'') { |
407 | 0 | AddMatch(id + (is_all_caps ? 119 : 120) * n, l + 3, l, matches); |
408 | 0 | } |
409 | 0 | } |
410 | 0 | } |
411 | 0 | } |
412 | 0 | } |
413 | 0 | if (max_length >= 6) { |
414 | | /* Transforms with prefixes "e ", "s ", ", " and "\xc2\xa0" */ |
415 | 0 | if ((data[1] == ' ' && |
416 | 0 | (data[0] == 'e' || data[0] == 's' || data[0] == ',')) || |
417 | 0 | (data[0] == 0xc2 && data[1] == 0xa0)) { |
418 | 0 | size_t offset = kStaticDictionaryBuckets[Hash(&data[2])]; |
419 | 0 | BROTLI_BOOL end = !offset; |
420 | 0 | while (!end) { |
421 | 0 | DictWord w = kStaticDictionaryWords[offset++]; |
422 | 0 | const size_t l = w.len & 0x1F; |
423 | 0 | const size_t n = (size_t)1 << dictionary->size_bits_by_length[l]; |
424 | 0 | const size_t id = w.idx; |
425 | 0 | end = !!(w.len & 0x80); |
426 | 0 | w.len = (uint8_t)l; |
427 | 0 | if (w.transform == 0 && |
428 | 0 | IsMatch(dictionary, w, &data[2], max_length - 2)) { |
429 | 0 | if (data[0] == 0xc2) { |
430 | 0 | AddMatch(id + 102 * n, l + 2, l, matches); |
431 | 0 | has_found_match = BROTLI_TRUE; |
432 | 0 | } else if (l + 2 < max_length && data[l + 2] == ' ') { |
433 | 0 | size_t t = data[0] == 'e' ? 18 : (data[0] == 's' ? 7 : 13); |
434 | 0 | AddMatch(id + t * n, l + 3, l, matches); |
435 | 0 | has_found_match = BROTLI_TRUE; |
436 | 0 | } |
437 | 0 | } |
438 | 0 | } |
439 | 0 | } |
440 | 0 | } |
441 | 0 | if (max_length >= 9) { |
442 | | /* Transforms with prefixes " the " and ".com/" */ |
443 | 0 | if ((data[0] == ' ' && data[1] == 't' && data[2] == 'h' && |
444 | 0 | data[3] == 'e' && data[4] == ' ') || |
445 | 0 | (data[0] == '.' && data[1] == 'c' && data[2] == 'o' && |
446 | 0 | data[3] == 'm' && data[4] == '/')) { |
447 | 0 | size_t offset = kStaticDictionaryBuckets[Hash(&data[5])]; |
448 | 0 | BROTLI_BOOL end = !offset; |
449 | 0 | while (!end) { |
450 | 0 | DictWord w = kStaticDictionaryWords[offset++]; |
451 | 0 | const size_t l = w.len & 0x1F; |
452 | 0 | const size_t n = (size_t)1 << dictionary->size_bits_by_length[l]; |
453 | 0 | const size_t id = w.idx; |
454 | 0 | end = !!(w.len & 0x80); |
455 | 0 | w.len = (uint8_t)l; |
456 | 0 | if (w.transform == 0 && |
457 | 0 | IsMatch(dictionary, w, &data[5], max_length - 5)) { |
458 | 0 | AddMatch(id + (data[0] == ' ' ? 41 : 72) * n, l + 5, l, matches); |
459 | 0 | has_found_match = BROTLI_TRUE; |
460 | 0 | if (l + 5 < max_length) { |
461 | 0 | const uint8_t* s = &data[l + 5]; |
462 | 0 | if (data[0] == ' ') { |
463 | 0 | if (l + 8 < max_length && |
464 | 0 | s[0] == ' ' && s[1] == 'o' && s[2] == 'f' && s[3] == ' ') { |
465 | 0 | AddMatch(id + 62 * n, l + 9, l, matches); |
466 | 0 | if (l + 12 < max_length && |
467 | 0 | s[4] == 't' && s[5] == 'h' && s[6] == 'e' && s[7] == ' ') { |
468 | 0 | AddMatch(id + 73 * n, l + 13, l, matches); |
469 | 0 | } |
470 | 0 | } |
471 | 0 | } |
472 | 0 | } |
473 | 0 | } |
474 | 0 | } |
475 | 0 | } |
476 | 0 | } |
477 | 0 | return has_found_match; |
478 | 0 | } |
479 | | |
480 | | #if defined(__cplusplus) || defined(c_plusplus) |
481 | | } /* extern "C" */ |
482 | | #endif |