]>
Commit | Line | Data |
---|---|---|
11b7501a SB |
1 | /* Copyright 2013 Google Inc. All Rights Reserved.\r |
2 | \r | |
3 | Distributed under MIT license.\r | |
4 | See file LICENSE for detail or copy at https://opensource.org/licenses/MIT\r | |
5 | */\r | |
6 | \r | |
7 | #include "./static_dict.h"\r | |
8 | \r | |
9 | #include "../common/dictionary.h"\r | |
10 | #include "./find_match_length.h"\r | |
11 | #include "./port.h"\r | |
12 | #include "./static_dict_lut.h"\r | |
13 | \r | |
14 | #if defined(__cplusplus) || defined(c_plusplus)\r | |
15 | extern "C" {\r | |
16 | #endif\r | |
17 | \r | |
18 | static const uint8_t kUppercaseFirst = 10;\r | |
19 | static const uint8_t kOmitLastNTransforms[10] = {\r | |
20 | 0, 12, 27, 23, 42, 63, 56, 48, 59, 64,\r | |
21 | };\r | |
22 | \r | |
23 | static BROTLI_INLINE uint32_t Hash(const uint8_t *data) {\r | |
24 | uint32_t h = BROTLI_UNALIGNED_LOAD32(data) * kDictHashMul32;\r | |
25 | /* The higher bits contain more mixture from the multiplication,\r | |
26 | so we take our results from there. */\r | |
27 | return h >> (32 - kDictNumBits);\r | |
28 | }\r | |
29 | \r | |
30 | static BROTLI_INLINE void AddMatch(size_t distance, size_t len, size_t len_code,\r | |
31 | uint32_t* matches) {\r | |
32 | uint32_t match = (uint32_t)((distance << 5) + len_code);\r | |
33 | matches[len] = BROTLI_MIN(uint32_t, matches[len], match);\r | |
34 | }\r | |
35 | \r | |
36 | static BROTLI_INLINE size_t DictMatchLength(const uint8_t* data,\r | |
37 | size_t id,\r | |
38 | size_t len,\r | |
39 | size_t maxlen) {\r | |
40 | const size_t offset = kBrotliDictionaryOffsetsByLength[len] + len * id;\r | |
41 | return FindMatchLengthWithLimit(&kBrotliDictionary[offset], data,\r | |
42 | BROTLI_MIN(size_t, len, maxlen));\r | |
43 | }\r | |
44 | \r | |
45 | static BROTLI_INLINE BROTLI_BOOL IsMatch(\r | |
46 | DictWord w, const uint8_t* data, size_t max_length) {\r | |
47 | if (w.len > max_length) {\r | |
48 | return BROTLI_FALSE;\r | |
49 | } else {\r | |
50 | const size_t offset = kBrotliDictionaryOffsetsByLength[w.len] +\r | |
51 | (size_t)w.len * (size_t)w.idx;\r | |
52 | const uint8_t* dict = &kBrotliDictionary[offset];\r | |
53 | if (w.transform == 0) {\r | |
54 | /* Match against base dictionary word. */\r | |
55 | return\r | |
56 | TO_BROTLI_BOOL(FindMatchLengthWithLimit(dict, data, w.len) == w.len);\r | |
57 | } else if (w.transform == 10) {\r | |
58 | /* Match against uppercase first transform.\r | |
59 | Note that there are only ASCII uppercase words in the lookup table. */\r | |
60 | return TO_BROTLI_BOOL(dict[0] >= 'a' && dict[0] <= 'z' &&\r | |
61 | (dict[0] ^ 32) == data[0] &&\r | |
62 | FindMatchLengthWithLimit(&dict[1], &data[1], w.len - 1u) ==\r | |
63 | w.len - 1u);\r | |
64 | } else {\r | |
65 | /* Match against uppercase all transform.\r | |
66 | Note that there are only ASCII uppercase words in the lookup table. */\r | |
67 | size_t i;\r | |
68 | for (i = 0; i < w.len; ++i) {\r | |
69 | if (dict[i] >= 'a' && dict[i] <= 'z') {\r | |
70 | if ((dict[i] ^ 32) != data[i]) return BROTLI_FALSE;\r | |
71 | } else {\r | |
72 | if (dict[i] != data[i]) return BROTLI_FALSE;\r | |
73 | }\r | |
74 | }\r | |
75 | return BROTLI_TRUE;\r | |
76 | }\r | |
77 | }\r | |
78 | }\r | |
79 | \r | |
80 | BROTLI_BOOL BrotliFindAllStaticDictionaryMatches(\r | |
81 | const uint8_t* data, size_t min_length, size_t max_length,\r | |
82 | uint32_t* matches) {\r | |
83 | BROTLI_BOOL has_found_match = BROTLI_FALSE;\r | |
84 | {\r | |
85 | size_t offset = kStaticDictionaryBuckets[Hash(data)];\r | |
86 | BROTLI_BOOL end = !offset;\r | |
87 | while (!end) {\r | |
88 | DictWord w = kStaticDictionaryWords[offset++];\r | |
89 | const size_t l = w.len & 0x7F;\r | |
90 | const size_t n = (size_t)1 << kBrotliDictionarySizeBitsByLength[l];\r | |
91 | const size_t id = w.idx;\r | |
92 | end = !!(w.len & 0x80);\r | |
93 | w.len = (uint8_t)l;\r | |
94 | if (w.transform == 0) {\r | |
95 | const size_t matchlen = DictMatchLength(data, id, l, max_length);\r | |
96 | const uint8_t* s;\r | |
97 | size_t minlen;\r | |
98 | size_t maxlen;\r | |
99 | size_t len;\r | |
100 | /* Transform "" + kIdentity + "" */\r | |
101 | if (matchlen == l) {\r | |
102 | AddMatch(id, l, l, matches);\r | |
103 | has_found_match = BROTLI_TRUE;\r | |
104 | }\r | |
105 | /* Transforms "" + kOmitLast1 + "" and "" + kOmitLast1 + "ing " */\r | |
106 | if (matchlen >= l - 1) {\r | |
107 | AddMatch(id + 12 * n, l - 1, l, matches);\r | |
108 | if (l + 2 < max_length &&\r | |
109 | data[l - 1] == 'i' && data[l] == 'n' && data[l + 1] == 'g' &&\r | |
110 | data[l + 2] == ' ') {\r | |
111 | AddMatch(id + 49 * n, l + 3, l, matches);\r | |
112 | }\r | |
113 | has_found_match = BROTLI_TRUE;\r | |
114 | }\r | |
115 | /* Transform "" + kOmitLastN + "" (N = 2 .. 9) */\r | |
116 | minlen = min_length;\r | |
117 | if (l > 9) minlen = BROTLI_MAX(size_t, minlen, l - 9);\r | |
118 | maxlen = BROTLI_MIN(size_t, matchlen, l - 2);\r | |
119 | for (len = minlen; len <= maxlen; ++len) {\r | |
120 | AddMatch(id + kOmitLastNTransforms[l - len] * n, len, l, matches);\r | |
121 | has_found_match = BROTLI_TRUE;\r | |
122 | }\r | |
123 | if (matchlen < l || l + 6 >= max_length) {\r | |
124 | continue;\r | |
125 | }\r | |
126 | s = &data[l];\r | |
127 | /* Transforms "" + kIdentity + <suffix> */\r | |
128 | if (s[0] == ' ') {\r | |
129 | AddMatch(id + n, l + 1, l, matches);\r | |
130 | if (s[1] == 'a') {\r | |
131 | if (s[2] == ' ') {\r | |
132 | AddMatch(id + 28 * n, l + 3, l, matches);\r | |
133 | } else if (s[2] == 's') {\r | |
134 | if (s[3] == ' ') AddMatch(id + 46 * n, l + 4, l, matches);\r | |
135 | } else if (s[2] == 't') {\r | |
136 | if (s[3] == ' ') AddMatch(id + 60 * n, l + 4, l, matches);\r | |
137 | } else if (s[2] == 'n') {\r | |
138 | if (s[3] == 'd' && s[4] == ' ') {\r | |
139 | AddMatch(id + 10 * n, l + 5, l, matches);\r | |
140 | }\r | |
141 | }\r | |
142 | } else if (s[1] == 'b') {\r | |
143 | if (s[2] == 'y' && s[3] == ' ') {\r | |
144 | AddMatch(id + 38 * n, l + 4, l, matches);\r | |
145 | }\r | |
146 | } else if (s[1] == 'i') {\r | |
147 | if (s[2] == 'n') {\r | |
148 | if (s[3] == ' ') AddMatch(id + 16 * n, l + 4, l, matches);\r | |
149 | } else if (s[2] == 's') {\r | |
150 | if (s[3] == ' ') AddMatch(id + 47 * n, l + 4, l, matches);\r | |
151 | }\r | |
152 | } else if (s[1] == 'f') {\r | |
153 | if (s[2] == 'o') {\r | |
154 | if (s[3] == 'r' && s[4] == ' ') {\r | |
155 | AddMatch(id + 25 * n, l + 5, l, matches);\r | |
156 | }\r | |
157 | } else if (s[2] == 'r') {\r | |
158 | if (s[3] == 'o' && s[4] == 'm' && s[5] == ' ') {\r | |
159 | AddMatch(id + 37 * n, l + 6, l, matches);\r | |
160 | }\r | |
161 | }\r | |
162 | } else if (s[1] == 'o') {\r | |
163 | if (s[2] == 'f') {\r | |
164 | if (s[3] == ' ') AddMatch(id + 8 * n, l + 4, l, matches);\r | |
165 | } else if (s[2] == 'n') {\r | |
166 | if (s[3] == ' ') AddMatch(id + 45 * n, l + 4, l, matches);\r | |
167 | }\r | |
168 | } else if (s[1] == 'n') {\r | |
169 | if (s[2] == 'o' && s[3] == 't' && s[4] == ' ') {\r | |
170 | AddMatch(id + 80 * n, l + 5, l, matches);\r | |
171 | }\r | |
172 | } else if (s[1] == 't') {\r | |
173 | if (s[2] == 'h') {\r | |
174 | if (s[3] == 'e') {\r | |
175 | if (s[4] == ' ') AddMatch(id + 5 * n, l + 5, l, matches);\r | |
176 | } else if (s[3] == 'a') {\r | |
177 | if (s[4] == 't' && s[5] == ' ') {\r | |
178 | AddMatch(id + 29 * n, l + 6, l, matches);\r | |
179 | }\r | |
180 | }\r | |
181 | } else if (s[2] == 'o') {\r | |
182 | if (s[3] == ' ') AddMatch(id + 17 * n, l + 4, l, matches);\r | |
183 | }\r | |
184 | } else if (s[1] == 'w') {\r | |
185 | if (s[2] == 'i' && s[3] == 't' && s[4] == 'h' && s[5] == ' ') {\r | |
186 | AddMatch(id + 35 * n, l + 6, l, matches);\r | |
187 | }\r | |
188 | }\r | |
189 | } else if (s[0] == '"') {\r | |
190 | AddMatch(id + 19 * n, l + 1, l, matches);\r | |
191 | if (s[1] == '>') {\r | |
192 | AddMatch(id + 21 * n, l + 2, l, matches);\r | |
193 | }\r | |
194 | } else if (s[0] == '.') {\r | |
195 | AddMatch(id + 20 * n, l + 1, l, matches);\r | |
196 | if (s[1] == ' ') {\r | |
197 | AddMatch(id + 31 * n, l + 2, l, matches);\r | |
198 | if (s[2] == 'T' && s[3] == 'h') {\r | |
199 | if (s[4] == 'e') {\r | |
200 | if (s[5] == ' ') AddMatch(id + 43 * n, l + 6, l, matches);\r | |
201 | } else if (s[4] == 'i') {\r | |
202 | if (s[5] == 's' && s[6] == ' ') {\r | |
203 | AddMatch(id + 75 * n, l + 7, l, matches);\r | |
204 | }\r | |
205 | }\r | |
206 | }\r | |
207 | }\r | |
208 | } else if (s[0] == ',') {\r | |
209 | AddMatch(id + 76 * n, l + 1, l, matches);\r | |
210 | if (s[1] == ' ') {\r | |
211 | AddMatch(id + 14 * n, l + 2, l, matches);\r | |
212 | }\r | |
213 | } else if (s[0] == '\n') {\r | |
214 | AddMatch(id + 22 * n, l + 1, l, matches);\r | |
215 | if (s[1] == '\t') {\r | |
216 | AddMatch(id + 50 * n, l + 2, l, matches);\r | |
217 | }\r | |
218 | } else if (s[0] == ']') {\r | |
219 | AddMatch(id + 24 * n, l + 1, l, matches);\r | |
220 | } else if (s[0] == '\'') {\r | |
221 | AddMatch(id + 36 * n, l + 1, l, matches);\r | |
222 | } else if (s[0] == ':') {\r | |
223 | AddMatch(id + 51 * n, l + 1, l, matches);\r | |
224 | } else if (s[0] == '(') {\r | |
225 | AddMatch(id + 57 * n, l + 1, l, matches);\r | |
226 | } else if (s[0] == '=') {\r | |
227 | if (s[1] == '"') {\r | |
228 | AddMatch(id + 70 * n, l + 2, l, matches);\r | |
229 | } else if (s[1] == '\'') {\r | |
230 | AddMatch(id + 86 * n, l + 2, l, matches);\r | |
231 | }\r | |
232 | } else if (s[0] == 'a') {\r | |
233 | if (s[1] == 'l' && s[2] == ' ') {\r | |
234 | AddMatch(id + 84 * n, l + 3, l, matches);\r | |
235 | }\r | |
236 | } else if (s[0] == 'e') {\r | |
237 | if (s[1] == 'd') {\r | |
238 | if (s[2] == ' ') AddMatch(id + 53 * n, l + 3, l, matches);\r | |
239 | } else if (s[1] == 'r') {\r | |
240 | if (s[2] == ' ') AddMatch(id + 82 * n, l + 3, l, matches);\r | |
241 | } else if (s[1] == 's') {\r | |
242 | if (s[2] == 't' && s[3] == ' ') {\r | |
243 | AddMatch(id + 95 * n, l + 4, l, matches);\r | |
244 | }\r | |
245 | }\r | |
246 | } else if (s[0] == 'f') {\r | |
247 | if (s[1] == 'u' && s[2] == 'l' && s[3] == ' ') {\r | |
248 | AddMatch(id + 90 * n, l + 4, l, matches);\r | |
249 | }\r | |
250 | } else if (s[0] == 'i') {\r | |
251 | if (s[1] == 'v') {\r | |
252 | if (s[2] == 'e' && s[3] == ' ') {\r | |
253 | AddMatch(id + 92 * n, l + 4, l, matches);\r | |
254 | }\r | |
255 | } else if (s[1] == 'z') {\r | |
256 | if (s[2] == 'e' && s[3] == ' ') {\r | |
257 | AddMatch(id + 100 * n, l + 4, l, matches);\r | |
258 | }\r | |
259 | }\r | |
260 | } else if (s[0] == 'l') {\r | |
261 | if (s[1] == 'e') {\r | |
262 | if (s[2] == 's' && s[3] == 's' && s[4] == ' ') {\r | |
263 | AddMatch(id + 93 * n, l + 5, l, matches);\r | |
264 | }\r | |
265 | } else if (s[1] == 'y') {\r | |
266 | if (s[2] == ' ') AddMatch(id + 61 * n, l + 3, l, matches);\r | |
267 | }\r | |
268 | } else if (s[0] == 'o') {\r | |
269 | if (s[1] == 'u' && s[2] == 's' && s[3] == ' ') {\r | |
270 | AddMatch(id + 106 * n, l + 4, l, matches);\r | |
271 | }\r | |
272 | }\r | |
273 | } else {\r | |
274 | /* Set is_all_caps=0 for kUppercaseFirst and\r | |
275 | is_all_caps=1 otherwise (kUppercaseAll) transform. */\r | |
276 | const BROTLI_BOOL is_all_caps =\r | |
277 | TO_BROTLI_BOOL(w.transform != kUppercaseFirst);\r | |
278 | const uint8_t* s;\r | |
279 | if (!IsMatch(w, data, max_length)) {\r | |
280 | continue;\r | |
281 | }\r | |
282 | /* Transform "" + kUppercase{First,All} + "" */\r | |
283 | AddMatch(id + (is_all_caps ? 44 : 9) * n, l, l, matches);\r | |
284 | has_found_match = BROTLI_TRUE;\r | |
285 | if (l + 1 >= max_length) {\r | |
286 | continue;\r | |
287 | }\r | |
288 | /* Transforms "" + kUppercase{First,All} + <suffix> */\r | |
289 | s = &data[l];\r | |
290 | if (s[0] == ' ') {\r | |
291 | AddMatch(id + (is_all_caps ? 68 : 4) * n, l + 1, l, matches);\r | |
292 | } else if (s[0] == '"') {\r | |
293 | AddMatch(id + (is_all_caps ? 87 : 66) * n, l + 1, l, matches);\r | |
294 | if (s[1] == '>') {\r | |
295 | AddMatch(id + (is_all_caps ? 97 : 69) * n, l + 2, l, matches);\r | |
296 | }\r | |
297 | } else if (s[0] == '.') {\r | |
298 | AddMatch(id + (is_all_caps ? 101 : 79) * n, l + 1, l, matches);\r | |
299 | if (s[1] == ' ') {\r | |
300 | AddMatch(id + (is_all_caps ? 114 : 88) * n, l + 2, l, matches);\r | |
301 | }\r | |
302 | } else if (s[0] == ',') {\r | |
303 | AddMatch(id + (is_all_caps ? 112 : 99) * n, l + 1, l, matches);\r | |
304 | if (s[1] == ' ') {\r | |
305 | AddMatch(id + (is_all_caps ? 107 : 58) * n, l + 2, l, matches);\r | |
306 | }\r | |
307 | } else if (s[0] == '\'') {\r | |
308 | AddMatch(id + (is_all_caps ? 94 : 74) * n, l + 1, l, matches);\r | |
309 | } else if (s[0] == '(') {\r | |
310 | AddMatch(id + (is_all_caps ? 113 : 78) * n, l + 1, l, matches);\r | |
311 | } else if (s[0] == '=') {\r | |
312 | if (s[1] == '"') {\r | |
313 | AddMatch(id + (is_all_caps ? 105 : 104) * n, l + 2, l, matches);\r | |
314 | } else if (s[1] == '\'') {\r | |
315 | AddMatch(id + (is_all_caps ? 116 : 108) * n, l + 2, l, matches);\r | |
316 | }\r | |
317 | }\r | |
318 | }\r | |
319 | }\r | |
320 | }\r | |
321 | /* Transforms with prefixes " " and "." */\r | |
322 | if (max_length >= 5 && (data[0] == ' ' || data[0] == '.')) {\r | |
323 | BROTLI_BOOL is_space = TO_BROTLI_BOOL(data[0] == ' ');\r | |
324 | size_t offset = kStaticDictionaryBuckets[Hash(&data[1])];\r | |
325 | BROTLI_BOOL end = !offset;\r | |
326 | while (!end) {\r | |
327 | DictWord w = kStaticDictionaryWords[offset++];\r | |
328 | const size_t l = w.len & 0x7F;\r | |
329 | const size_t n = (size_t)1 << kBrotliDictionarySizeBitsByLength[l];\r | |
330 | const size_t id = w.idx;\r | |
331 | end = !!(w.len & 0x80);\r | |
332 | w.len = (uint8_t)l;\r | |
333 | if (w.transform == 0) {\r | |
334 | const uint8_t* s;\r | |
335 | if (!IsMatch(w, &data[1], max_length - 1)) {\r | |
336 | continue;\r | |
337 | }\r | |
338 | /* Transforms " " + kIdentity + "" and "." + kIdentity + "" */\r | |
339 | AddMatch(id + (is_space ? 6 : 32) * n, l + 1, l, matches);\r | |
340 | has_found_match = BROTLI_TRUE;\r | |
341 | if (l + 2 >= max_length) {\r | |
342 | continue;\r | |
343 | }\r | |
344 | /* Transforms " " + kIdentity + <suffix> and "." + kIdentity + <suffix>\r | |
345 | */\r | |
346 | s = &data[l + 1];\r | |
347 | if (s[0] == ' ') {\r | |
348 | AddMatch(id + (is_space ? 2 : 77) * n, l + 2, l, matches);\r | |
349 | } else if (s[0] == '(') {\r | |
350 | AddMatch(id + (is_space ? 89 : 67) * n, l + 2, l, matches);\r | |
351 | } else if (is_space) {\r | |
352 | if (s[0] == ',') {\r | |
353 | AddMatch(id + 103 * n, l + 2, l, matches);\r | |
354 | if (s[1] == ' ') {\r | |
355 | AddMatch(id + 33 * n, l + 3, l, matches);\r | |
356 | }\r | |
357 | } else if (s[0] == '.') {\r | |
358 | AddMatch(id + 71 * n, l + 2, l, matches);\r | |
359 | if (s[1] == ' ') {\r | |
360 | AddMatch(id + 52 * n, l + 3, l, matches);\r | |
361 | }\r | |
362 | } else if (s[0] == '=') {\r | |
363 | if (s[1] == '"') {\r | |
364 | AddMatch(id + 81 * n, l + 3, l, matches);\r | |
365 | } else if (s[1] == '\'') {\r | |
366 | AddMatch(id + 98 * n, l + 3, l, matches);\r | |
367 | }\r | |
368 | }\r | |
369 | }\r | |
370 | } else if (is_space) {\r | |
371 | /* Set is_all_caps=0 for kUppercaseFirst and\r | |
372 | is_all_caps=1 otherwise (kUppercaseAll) transform. */\r | |
373 | const BROTLI_BOOL is_all_caps =\r | |
374 | TO_BROTLI_BOOL(w.transform != kUppercaseFirst);\r | |
375 | const uint8_t* s;\r | |
376 | if (!IsMatch(w, &data[1], max_length - 1)) {\r | |
377 | continue;\r | |
378 | }\r | |
379 | /* Transforms " " + kUppercase{First,All} + "" */\r | |
380 | AddMatch(id + (is_all_caps ? 85 : 30) * n, l + 1, l, matches);\r | |
381 | has_found_match = BROTLI_TRUE;\r | |
382 | if (l + 2 >= max_length) {\r | |
383 | continue;\r | |
384 | }\r | |
385 | /* Transforms " " + kUppercase{First,All} + <suffix> */\r | |
386 | s = &data[l + 1];\r | |
387 | if (s[0] == ' ') {\r | |
388 | AddMatch(id + (is_all_caps ? 83 : 15) * n, l + 2, l, matches);\r | |
389 | } else if (s[0] == ',') {\r | |
390 | if (!is_all_caps) {\r | |
391 | AddMatch(id + 109 * n, l + 2, l, matches);\r | |
392 | }\r | |
393 | if (s[1] == ' ') {\r | |
394 | AddMatch(id + (is_all_caps ? 111 : 65) * n, l + 3, l, matches);\r | |
395 | }\r | |
396 | } else if (s[0] == '.') {\r | |
397 | AddMatch(id + (is_all_caps ? 115 : 96) * n, l + 2, l, matches);\r | |
398 | if (s[1] == ' ') {\r | |
399 | AddMatch(id + (is_all_caps ? 117 : 91) * n, l + 3, l, matches);\r | |
400 | }\r | |
401 | } else if (s[0] == '=') {\r | |
402 | if (s[1] == '"') {\r | |
403 | AddMatch(id + (is_all_caps ? 110 : 118) * n, l + 3, l, matches);\r | |
404 | } else if (s[1] == '\'') {\r | |
405 | AddMatch(id + (is_all_caps ? 119 : 120) * n, l + 3, l, matches);\r | |
406 | }\r | |
407 | }\r | |
408 | }\r | |
409 | }\r | |
410 | }\r | |
411 | if (max_length >= 6) {\r | |
412 | /* Transforms with prefixes "e ", "s ", ", " and "\xc2\xa0" */\r | |
413 | if ((data[1] == ' ' &&\r | |
414 | (data[0] == 'e' || data[0] == 's' || data[0] == ',')) ||\r | |
415 | (data[0] == 0xc2 && data[1] == 0xa0)) {\r | |
416 | size_t offset = kStaticDictionaryBuckets[Hash(&data[2])];\r | |
417 | BROTLI_BOOL end = !offset;\r | |
418 | while (!end) {\r | |
419 | DictWord w = kStaticDictionaryWords[offset++];\r | |
420 | const size_t l = w.len & 0x7F;\r | |
421 | const size_t n = (size_t)1 << kBrotliDictionarySizeBitsByLength[l];\r | |
422 | const size_t id = w.idx;\r | |
423 | end = !!(w.len & 0x80);\r | |
424 | w.len = (uint8_t)l;\r | |
425 | if (w.transform == 0 && IsMatch(w, &data[2], max_length - 2)) {\r | |
426 | if (data[0] == 0xc2) {\r | |
427 | AddMatch(id + 102 * n, l + 2, l, matches);\r | |
428 | has_found_match = BROTLI_TRUE;\r | |
429 | } else if (l + 2 < max_length && data[l + 2] == ' ') {\r | |
430 | size_t t = data[0] == 'e' ? 18 : (data[0] == 's' ? 7 : 13);\r | |
431 | AddMatch(id + t * n, l + 3, l, matches);\r | |
432 | has_found_match = BROTLI_TRUE;\r | |
433 | }\r | |
434 | }\r | |
435 | }\r | |
436 | }\r | |
437 | }\r | |
438 | if (max_length >= 9) {\r | |
439 | /* Transforms with prefixes " the " and ".com/" */\r | |
440 | if ((data[0] == ' ' && data[1] == 't' && data[2] == 'h' &&\r | |
441 | data[3] == 'e' && data[4] == ' ') ||\r | |
442 | (data[0] == '.' && data[1] == 'c' && data[2] == 'o' &&\r | |
443 | data[3] == 'm' && data[4] == '/')) {\r | |
444 | size_t offset = kStaticDictionaryBuckets[Hash(&data[5])];\r | |
445 | BROTLI_BOOL end = !offset;\r | |
446 | while (!end) {\r | |
447 | DictWord w = kStaticDictionaryWords[offset++];\r | |
448 | const size_t l = w.len & 0x7F;\r | |
449 | const size_t n = (size_t)1 << kBrotliDictionarySizeBitsByLength[l];\r | |
450 | const size_t id = w.idx;\r | |
451 | end = !!(w.len & 0x80);\r | |
452 | w.len = (uint8_t)l;\r | |
453 | if (w.transform == 0 && IsMatch(w, &data[5], max_length - 5)) {\r | |
454 | AddMatch(id + (data[0] == ' ' ? 41 : 72) * n, l + 5, l, matches);\r | |
455 | has_found_match = BROTLI_TRUE;\r | |
456 | if (l + 5 < max_length) {\r | |
457 | const uint8_t* s = &data[l + 5];\r | |
458 | if (data[0] == ' ') {\r | |
459 | if (l + 8 < max_length &&\r | |
460 | s[0] == ' ' && s[1] == 'o' && s[2] == 'f' && s[3] == ' ') {\r | |
461 | AddMatch(id + 62 * n, l + 9, l, matches);\r | |
462 | if (l + 12 < max_length &&\r | |
463 | s[4] == 't' && s[5] == 'h' && s[6] == 'e' && s[7] == ' ') {\r | |
464 | AddMatch(id + 73 * n, l + 13, l, matches);\r | |
465 | }\r | |
466 | }\r | |
467 | }\r | |
468 | }\r | |
469 | }\r | |
470 | }\r | |
471 | }\r | |
472 | }\r | |
473 | return has_found_match;\r | |
474 | }\r | |
475 | \r | |
476 | #if defined(__cplusplus) || defined(c_plusplus)\r | |
477 | } /* extern "C" */\r | |
478 | #endif\r |