+++ /dev/null
-/* Copyright 2013 Google Inc. All Rights Reserved.\r
-\r
- Distributed under MIT license.\r
- See file LICENSE for detail or copy at https://opensource.org/licenses/MIT\r
-*/\r
-\r
-#include "./static_dict.h"\r
-\r
-#include "../common/dictionary.h"\r
-#include "../common/platform.h"\r
-#include "../common/transform.h"\r
-#include "./encoder_dict.h"\r
-#include "./find_match_length.h"\r
-\r
-#if defined(__cplusplus) || defined(c_plusplus)\r
-extern "C" {\r
-#endif\r
-\r
-static BROTLI_INLINE uint32_t Hash(const uint8_t* data) {\r
- uint32_t h = BROTLI_UNALIGNED_LOAD32LE(data) * kDictHashMul32;\r
- /* The higher bits contain more mixture from the multiplication,\r
- so we take our results from there. */\r
- return h >> (32 - kDictNumBits);\r
-}\r
-\r
-static BROTLI_INLINE void AddMatch(size_t distance, size_t len, size_t len_code,\r
- uint32_t* matches) {\r
- uint32_t match = (uint32_t)((distance << 5) + len_code);\r
- matches[len] = BROTLI_MIN(uint32_t, matches[len], match);\r
-}\r
-\r
-static BROTLI_INLINE size_t DictMatchLength(const BrotliDictionary* dictionary,\r
- const uint8_t* data,\r
- size_t id,\r
- size_t len,\r
- size_t maxlen) {\r
- const size_t offset = dictionary->offsets_by_length[len] + len * id;\r
- return FindMatchLengthWithLimit(&dictionary->data[offset], data,\r
- BROTLI_MIN(size_t, len, maxlen));\r
-}\r
-\r
-static BROTLI_INLINE BROTLI_BOOL IsMatch(const BrotliDictionary* dictionary,\r
- DictWord w, const uint8_t* data, size_t max_length) {\r
- if (w.len > max_length) {\r
- return BROTLI_FALSE;\r
- } else {\r
- const size_t offset = dictionary->offsets_by_length[w.len] +\r
- (size_t)w.len * (size_t)w.idx;\r
- const uint8_t* dict = &dictionary->data[offset];\r
- if (w.transform == 0) {\r
- /* Match against base dictionary word. */\r
- return\r
- TO_BROTLI_BOOL(FindMatchLengthWithLimit(dict, data, w.len) == w.len);\r
- } else if (w.transform == 10) {\r
- /* Match against uppercase first transform.\r
- Note that there are only ASCII uppercase words in the lookup table. */\r
- return TO_BROTLI_BOOL(dict[0] >= 'a' && dict[0] <= 'z' &&\r
- (dict[0] ^ 32) == data[0] &&\r
- FindMatchLengthWithLimit(&dict[1], &data[1], w.len - 1u) ==\r
- w.len - 1u);\r
- } else {\r
- /* Match against uppercase all transform.\r
- Note that there are only ASCII uppercase words in the lookup table. */\r
- size_t i;\r
- for (i = 0; i < w.len; ++i) {\r
- if (dict[i] >= 'a' && dict[i] <= 'z') {\r
- if ((dict[i] ^ 32) != data[i]) return BROTLI_FALSE;\r
- } else {\r
- if (dict[i] != data[i]) return BROTLI_FALSE;\r
- }\r
- }\r
- return BROTLI_TRUE;\r
- }\r
- }\r
-}\r
-\r
-BROTLI_BOOL BrotliFindAllStaticDictionaryMatches(\r
- const BrotliEncoderDictionary* dictionary, const uint8_t* data,\r
- size_t min_length, size_t max_length, uint32_t* matches) {\r
- BROTLI_BOOL has_found_match = BROTLI_FALSE;\r
- {\r
- size_t offset = dictionary->buckets[Hash(data)];\r
- BROTLI_BOOL end = !offset;\r
- while (!end) {\r
- DictWord w = dictionary->dict_words[offset++];\r
- const size_t l = w.len & 0x1F;\r
- const size_t n = (size_t)1 << dictionary->words->size_bits_by_length[l];\r
- const size_t id = w.idx;\r
- end = !!(w.len & 0x80);\r
- w.len = (uint8_t)l;\r
- if (w.transform == 0) {\r
- const size_t matchlen =\r
- DictMatchLength(dictionary->words, data, id, l, max_length);\r
- const uint8_t* s;\r
- size_t minlen;\r
- size_t maxlen;\r
- size_t len;\r
- /* Transform "" + BROTLI_TRANSFORM_IDENTITY + "" */\r
- if (matchlen == l) {\r
- AddMatch(id, l, l, matches);\r
- has_found_match = BROTLI_TRUE;\r
- }\r
- /* Transforms "" + BROTLI_TRANSFORM_OMIT_LAST_1 + "" and\r
- "" + BROTLI_TRANSFORM_OMIT_LAST_1 + "ing " */\r
- if (matchlen >= l - 1) {\r
- AddMatch(id + 12 * n, l - 1, l, matches);\r
- if (l + 2 < max_length &&\r
- data[l - 1] == 'i' && data[l] == 'n' && data[l + 1] == 'g' &&\r
- data[l + 2] == ' ') {\r
- AddMatch(id + 49 * n, l + 3, l, matches);\r
- }\r
- has_found_match = BROTLI_TRUE;\r
- }\r
- /* Transform "" + BROTLI_TRANSFORM_OMIT_LAST_# + "" (# = 2 .. 9) */\r
- minlen = min_length;\r
- if (l > 9) minlen = BROTLI_MAX(size_t, minlen, l - 9);\r
- maxlen = BROTLI_MIN(size_t, matchlen, l - 2);\r
- for (len = minlen; len <= maxlen; ++len) {\r
- size_t cut = l - len;\r
- size_t transform_id = (cut << 2) +\r
- (size_t)((dictionary->cutoffTransforms >> (cut * 6)) & 0x3F);\r
- AddMatch(id + transform_id * n, len, l, matches);\r
- has_found_match = BROTLI_TRUE;\r
- }\r
- if (matchlen < l || l + 6 >= max_length) {\r
- continue;\r
- }\r
- s = &data[l];\r
- /* Transforms "" + BROTLI_TRANSFORM_IDENTITY + <suffix> */\r
- if (s[0] == ' ') {\r
- AddMatch(id + n, l + 1, l, matches);\r
- if (s[1] == 'a') {\r
- if (s[2] == ' ') {\r
- AddMatch(id + 28 * n, l + 3, l, matches);\r
- } else if (s[2] == 's') {\r
- if (s[3] == ' ') AddMatch(id + 46 * n, l + 4, l, matches);\r
- } else if (s[2] == 't') {\r
- if (s[3] == ' ') AddMatch(id + 60 * n, l + 4, l, matches);\r
- } else if (s[2] == 'n') {\r
- if (s[3] == 'd' && s[4] == ' ') {\r
- AddMatch(id + 10 * n, l + 5, l, matches);\r
- }\r
- }\r
- } else if (s[1] == 'b') {\r
- if (s[2] == 'y' && s[3] == ' ') {\r
- AddMatch(id + 38 * n, l + 4, l, matches);\r
- }\r
- } else if (s[1] == 'i') {\r
- if (s[2] == 'n') {\r
- if (s[3] == ' ') AddMatch(id + 16 * n, l + 4, l, matches);\r
- } else if (s[2] == 's') {\r
- if (s[3] == ' ') AddMatch(id + 47 * n, l + 4, l, matches);\r
- }\r
- } else if (s[1] == 'f') {\r
- if (s[2] == 'o') {\r
- if (s[3] == 'r' && s[4] == ' ') {\r
- AddMatch(id + 25 * n, l + 5, l, matches);\r
- }\r
- } else if (s[2] == 'r') {\r
- if (s[3] == 'o' && s[4] == 'm' && s[5] == ' ') {\r
- AddMatch(id + 37 * n, l + 6, l, matches);\r
- }\r
- }\r
- } else if (s[1] == 'o') {\r
- if (s[2] == 'f') {\r
- if (s[3] == ' ') AddMatch(id + 8 * n, l + 4, l, matches);\r
- } else if (s[2] == 'n') {\r
- if (s[3] == ' ') AddMatch(id + 45 * n, l + 4, l, matches);\r
- }\r
- } else if (s[1] == 'n') {\r
- if (s[2] == 'o' && s[3] == 't' && s[4] == ' ') {\r
- AddMatch(id + 80 * n, l + 5, l, matches);\r
- }\r
- } else if (s[1] == 't') {\r
- if (s[2] == 'h') {\r
- if (s[3] == 'e') {\r
- if (s[4] == ' ') AddMatch(id + 5 * n, l + 5, l, matches);\r
- } else if (s[3] == 'a') {\r
- if (s[4] == 't' && s[5] == ' ') {\r
- AddMatch(id + 29 * n, l + 6, l, matches);\r
- }\r
- }\r
- } else if (s[2] == 'o') {\r
- if (s[3] == ' ') AddMatch(id + 17 * n, l + 4, l, matches);\r
- }\r
- } else if (s[1] == 'w') {\r
- if (s[2] == 'i' && s[3] == 't' && s[4] == 'h' && s[5] == ' ') {\r
- AddMatch(id + 35 * n, l + 6, l, matches);\r
- }\r
- }\r
- } else if (s[0] == '"') {\r
- AddMatch(id + 19 * n, l + 1, l, matches);\r
- if (s[1] == '>') {\r
- AddMatch(id + 21 * n, l + 2, l, matches);\r
- }\r
- } else if (s[0] == '.') {\r
- AddMatch(id + 20 * n, l + 1, l, matches);\r
- if (s[1] == ' ') {\r
- AddMatch(id + 31 * n, l + 2, l, matches);\r
- if (s[2] == 'T' && s[3] == 'h') {\r
- if (s[4] == 'e') {\r
- if (s[5] == ' ') AddMatch(id + 43 * n, l + 6, l, matches);\r
- } else if (s[4] == 'i') {\r
- if (s[5] == 's' && s[6] == ' ') {\r
- AddMatch(id + 75 * n, l + 7, l, matches);\r
- }\r
- }\r
- }\r
- }\r
- } else if (s[0] == ',') {\r
- AddMatch(id + 76 * n, l + 1, l, matches);\r
- if (s[1] == ' ') {\r
- AddMatch(id + 14 * n, l + 2, l, matches);\r
- }\r
- } else if (s[0] == '\n') {\r
- AddMatch(id + 22 * n, l + 1, l, matches);\r
- if (s[1] == '\t') {\r
- AddMatch(id + 50 * n, l + 2, l, matches);\r
- }\r
- } else if (s[0] == ']') {\r
- AddMatch(id + 24 * n, l + 1, l, matches);\r
- } else if (s[0] == '\'') {\r
- AddMatch(id + 36 * n, l + 1, l, matches);\r
- } else if (s[0] == ':') {\r
- AddMatch(id + 51 * n, l + 1, l, matches);\r
- } else if (s[0] == '(') {\r
- AddMatch(id + 57 * n, l + 1, l, matches);\r
- } else if (s[0] == '=') {\r
- if (s[1] == '"') {\r
- AddMatch(id + 70 * n, l + 2, l, matches);\r
- } else if (s[1] == '\'') {\r
- AddMatch(id + 86 * n, l + 2, l, matches);\r
- }\r
- } else if (s[0] == 'a') {\r
- if (s[1] == 'l' && s[2] == ' ') {\r
- AddMatch(id + 84 * n, l + 3, l, matches);\r
- }\r
- } else if (s[0] == 'e') {\r
- if (s[1] == 'd') {\r
- if (s[2] == ' ') AddMatch(id + 53 * n, l + 3, l, matches);\r
- } else if (s[1] == 'r') {\r
- if (s[2] == ' ') AddMatch(id + 82 * n, l + 3, l, matches);\r
- } else if (s[1] == 's') {\r
- if (s[2] == 't' && s[3] == ' ') {\r
- AddMatch(id + 95 * n, l + 4, l, matches);\r
- }\r
- }\r
- } else if (s[0] == 'f') {\r
- if (s[1] == 'u' && s[2] == 'l' && s[3] == ' ') {\r
- AddMatch(id + 90 * n, l + 4, l, matches);\r
- }\r
- } else if (s[0] == 'i') {\r
- if (s[1] == 'v') {\r
- if (s[2] == 'e' && s[3] == ' ') {\r
- AddMatch(id + 92 * n, l + 4, l, matches);\r
- }\r
- } else if (s[1] == 'z') {\r
- if (s[2] == 'e' && s[3] == ' ') {\r
- AddMatch(id + 100 * n, l + 4, l, matches);\r
- }\r
- }\r
- } else if (s[0] == 'l') {\r
- if (s[1] == 'e') {\r
- if (s[2] == 's' && s[3] == 's' && s[4] == ' ') {\r
- AddMatch(id + 93 * n, l + 5, l, matches);\r
- }\r
- } else if (s[1] == 'y') {\r
- if (s[2] == ' ') AddMatch(id + 61 * n, l + 3, l, matches);\r
- }\r
- } else if (s[0] == 'o') {\r
- if (s[1] == 'u' && s[2] == 's' && s[3] == ' ') {\r
- AddMatch(id + 106 * n, l + 4, l, matches);\r
- }\r
- }\r
- } else {\r
- /* Set is_all_caps=0 for BROTLI_TRANSFORM_UPPERCASE_FIRST and\r
- is_all_caps=1 otherwise (BROTLI_TRANSFORM_UPPERCASE_ALL)\r
- transform. */\r
- const BROTLI_BOOL is_all_caps =\r
- TO_BROTLI_BOOL(w.transform != BROTLI_TRANSFORM_UPPERCASE_FIRST);\r
- const uint8_t* s;\r
- if (!IsMatch(dictionary->words, w, data, max_length)) {\r
- continue;\r
- }\r
- /* Transform "" + kUppercase{First,All} + "" */\r
- AddMatch(id + (is_all_caps ? 44 : 9) * n, l, l, matches);\r
- has_found_match = BROTLI_TRUE;\r
- if (l + 1 >= max_length) {\r
- continue;\r
- }\r
- /* Transforms "" + kUppercase{First,All} + <suffix> */\r
- s = &data[l];\r
- if (s[0] == ' ') {\r
- AddMatch(id + (is_all_caps ? 68 : 4) * n, l + 1, l, matches);\r
- } else if (s[0] == '"') {\r
- AddMatch(id + (is_all_caps ? 87 : 66) * n, l + 1, l, matches);\r
- if (s[1] == '>') {\r
- AddMatch(id + (is_all_caps ? 97 : 69) * n, l + 2, l, matches);\r
- }\r
- } else if (s[0] == '.') {\r
- AddMatch(id + (is_all_caps ? 101 : 79) * n, l + 1, l, matches);\r
- if (s[1] == ' ') {\r
- AddMatch(id + (is_all_caps ? 114 : 88) * n, l + 2, l, matches);\r
- }\r
- } else if (s[0] == ',') {\r
- AddMatch(id + (is_all_caps ? 112 : 99) * n, l + 1, l, matches);\r
- if (s[1] == ' ') {\r
- AddMatch(id + (is_all_caps ? 107 : 58) * n, l + 2, l, matches);\r
- }\r
- } else if (s[0] == '\'') {\r
- AddMatch(id + (is_all_caps ? 94 : 74) * n, l + 1, l, matches);\r
- } else if (s[0] == '(') {\r
- AddMatch(id + (is_all_caps ? 113 : 78) * n, l + 1, l, matches);\r
- } else if (s[0] == '=') {\r
- if (s[1] == '"') {\r
- AddMatch(id + (is_all_caps ? 105 : 104) * n, l + 2, l, matches);\r
- } else if (s[1] == '\'') {\r
- AddMatch(id + (is_all_caps ? 116 : 108) * n, l + 2, l, matches);\r
- }\r
- }\r
- }\r
- }\r
- }\r
- /* Transforms with prefixes " " and "." */\r
- if (max_length >= 5 && (data[0] == ' ' || data[0] == '.')) {\r
- BROTLI_BOOL is_space = TO_BROTLI_BOOL(data[0] == ' ');\r
- size_t offset = dictionary->buckets[Hash(&data[1])];\r
- BROTLI_BOOL end = !offset;\r
- while (!end) {\r
- DictWord w = dictionary->dict_words[offset++];\r
- const size_t l = w.len & 0x1F;\r
- const size_t n = (size_t)1 << dictionary->words->size_bits_by_length[l];\r
- const size_t id = w.idx;\r
- end = !!(w.len & 0x80);\r
- w.len = (uint8_t)l;\r
- if (w.transform == 0) {\r
- const uint8_t* s;\r
- if (!IsMatch(dictionary->words, w, &data[1], max_length - 1)) {\r
- continue;\r
- }\r
- /* Transforms " " + BROTLI_TRANSFORM_IDENTITY + "" and\r
- "." + BROTLI_TRANSFORM_IDENTITY + "" */\r
- AddMatch(id + (is_space ? 6 : 32) * n, l + 1, l, matches);\r
- has_found_match = BROTLI_TRUE;\r
- if (l + 2 >= max_length) {\r
- continue;\r
- }\r
- /* Transforms " " + BROTLI_TRANSFORM_IDENTITY + <suffix> and\r
- "." + BROTLI_TRANSFORM_IDENTITY + <suffix>\r
- */\r
- s = &data[l + 1];\r
- if (s[0] == ' ') {\r
- AddMatch(id + (is_space ? 2 : 77) * n, l + 2, l, matches);\r
- } else if (s[0] == '(') {\r
- AddMatch(id + (is_space ? 89 : 67) * n, l + 2, l, matches);\r
- } else if (is_space) {\r
- if (s[0] == ',') {\r
- AddMatch(id + 103 * n, l + 2, l, matches);\r
- if (s[1] == ' ') {\r
- AddMatch(id + 33 * n, l + 3, l, matches);\r
- }\r
- } else if (s[0] == '.') {\r
- AddMatch(id + 71 * n, l + 2, l, matches);\r
- if (s[1] == ' ') {\r
- AddMatch(id + 52 * n, l + 3, l, matches);\r
- }\r
- } else if (s[0] == '=') {\r
- if (s[1] == '"') {\r
- AddMatch(id + 81 * n, l + 3, l, matches);\r
- } else if (s[1] == '\'') {\r
- AddMatch(id + 98 * n, l + 3, l, matches);\r
- }\r
- }\r
- }\r
- } else if (is_space) {\r
- /* Set is_all_caps=0 for BROTLI_TRANSFORM_UPPERCASE_FIRST and\r
- is_all_caps=1 otherwise (BROTLI_TRANSFORM_UPPERCASE_ALL)\r
- transform. */\r
- const BROTLI_BOOL is_all_caps =\r
- TO_BROTLI_BOOL(w.transform != BROTLI_TRANSFORM_UPPERCASE_FIRST);\r
- const uint8_t* s;\r
- if (!IsMatch(dictionary->words, w, &data[1], max_length - 1)) {\r
- continue;\r
- }\r
- /* Transforms " " + kUppercase{First,All} + "" */\r
- AddMatch(id + (is_all_caps ? 85 : 30) * n, l + 1, l, matches);\r
- has_found_match = BROTLI_TRUE;\r
- if (l + 2 >= max_length) {\r
- continue;\r
- }\r
- /* Transforms " " + kUppercase{First,All} + <suffix> */\r
- s = &data[l + 1];\r
- if (s[0] == ' ') {\r
- AddMatch(id + (is_all_caps ? 83 : 15) * n, l + 2, l, matches);\r
- } else if (s[0] == ',') {\r
- if (!is_all_caps) {\r
- AddMatch(id + 109 * n, l + 2, l, matches);\r
- }\r
- if (s[1] == ' ') {\r
- AddMatch(id + (is_all_caps ? 111 : 65) * n, l + 3, l, matches);\r
- }\r
- } else if (s[0] == '.') {\r
- AddMatch(id + (is_all_caps ? 115 : 96) * n, l + 2, l, matches);\r
- if (s[1] == ' ') {\r
- AddMatch(id + (is_all_caps ? 117 : 91) * n, l + 3, l, matches);\r
- }\r
- } else if (s[0] == '=') {\r
- if (s[1] == '"') {\r
- AddMatch(id + (is_all_caps ? 110 : 118) * n, l + 3, l, matches);\r
- } else if (s[1] == '\'') {\r
- AddMatch(id + (is_all_caps ? 119 : 120) * n, l + 3, l, matches);\r
- }\r
- }\r
- }\r
- }\r
- }\r
- if (max_length >= 6) {\r
- /* Transforms with prefixes "e ", "s ", ", " and "\xC2\xA0" */\r
- if ((data[1] == ' ' &&\r
- (data[0] == 'e' || data[0] == 's' || data[0] == ',')) ||\r
- (data[0] == 0xC2 && data[1] == 0xA0)) {\r
- size_t offset = dictionary->buckets[Hash(&data[2])];\r
- BROTLI_BOOL end = !offset;\r
- while (!end) {\r
- DictWord w = dictionary->dict_words[offset++];\r
- const size_t l = w.len & 0x1F;\r
- const size_t n = (size_t)1 << dictionary->words->size_bits_by_length[l];\r
- const size_t id = w.idx;\r
- end = !!(w.len & 0x80);\r
- w.len = (uint8_t)l;\r
- if (w.transform == 0 &&\r
- IsMatch(dictionary->words, w, &data[2], max_length - 2)) {\r
- if (data[0] == 0xC2) {\r
- AddMatch(id + 102 * n, l + 2, l, matches);\r
- has_found_match = BROTLI_TRUE;\r
- } else if (l + 2 < max_length && data[l + 2] == ' ') {\r
- size_t t = data[0] == 'e' ? 18 : (data[0] == 's' ? 7 : 13);\r
- AddMatch(id + t * n, l + 3, l, matches);\r
- has_found_match = BROTLI_TRUE;\r
- }\r
- }\r
- }\r
- }\r
- }\r
- if (max_length >= 9) {\r
- /* Transforms with prefixes " the " and ".com/" */\r
- if ((data[0] == ' ' && data[1] == 't' && data[2] == 'h' &&\r
- data[3] == 'e' && data[4] == ' ') ||\r
- (data[0] == '.' && data[1] == 'c' && data[2] == 'o' &&\r
- data[3] == 'm' && data[4] == '/')) {\r
- size_t offset = dictionary->buckets[Hash(&data[5])];\r
- BROTLI_BOOL end = !offset;\r
- while (!end) {\r
- DictWord w = dictionary->dict_words[offset++];\r
- const size_t l = w.len & 0x1F;\r
- const size_t n = (size_t)1 << dictionary->words->size_bits_by_length[l];\r
- const size_t id = w.idx;\r
- end = !!(w.len & 0x80);\r
- w.len = (uint8_t)l;\r
- if (w.transform == 0 &&\r
- IsMatch(dictionary->words, w, &data[5], max_length - 5)) {\r
- AddMatch(id + (data[0] == ' ' ? 41 : 72) * n, l + 5, l, matches);\r
- has_found_match = BROTLI_TRUE;\r
- if (l + 5 < max_length) {\r
- const uint8_t* s = &data[l + 5];\r
- if (data[0] == ' ') {\r
- if (l + 8 < max_length &&\r
- s[0] == ' ' && s[1] == 'o' && s[2] == 'f' && s[3] == ' ') {\r
- AddMatch(id + 62 * n, l + 9, l, matches);\r
- if (l + 12 < max_length &&\r
- s[4] == 't' && s[5] == 'h' && s[6] == 'e' && s[7] == ' ') {\r
- AddMatch(id + 73 * n, l + 13, l, matches);\r
- }\r
- }\r
- }\r
- }\r
- }\r
- }\r
- }\r
- }\r
- return has_found_match;\r
-}\r
-\r
-#if defined(__cplusplus) || defined(c_plusplus)\r
-} /* extern "C" */\r
-#endif\r
projects / mirror_edk2.git / blobdiff