]>
Commit | Line | Data |
---|---|---|
2730470f LG |
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 "./transform.h"\r | |
8 | \r | |
9 | #if defined(__cplusplus) || defined(c_plusplus)\r | |
10 | extern "C" {\r | |
11 | #endif\r | |
12 | \r | |
13 | /* RFC 7932 transforms string data */\r | |
14 | static const char kPrefixSuffix[] =\r | |
15 | "\1 \2, \10 of the \4 of \2s \1.\5 and \4 "\r | |
16 | /* 0x _0 _2 __5 _E _3 _6 _8 _E */\r | |
17 | "in \1\"\4 to \2\">\1\n\2. \1]\5 for \3 a \6 "\r | |
18 | /* 2x _3_ _5 _A_ _D_ _F _2 _4 _A _E */\r | |
19 | "that \1\'\6 with \6 from \4 by \1(\6. T"\r | |
20 | /* 4x _5_ _7 _E _5 _A _C */\r | |
21 | "he \4 on \4 as \4 is \4ing \2\n\t\1:\3ed "\r | |
22 | /* 6x _3 _8 _D _2 _7_ _ _A _C */\r | |
23 | "\2=\"\4 at \3ly \1,\2=\'\5.com/\7. This \5"\r | |
24 | /* 8x _0 _ _3 _8 _C _E _ _1 _7 _F */\r | |
25 | " not \3er \3al \4ful \4ive \5less \4es"\r | |
26 | /* Ax _5 _9 _D _2 _7 _D */\r | |
27 | "t \4ize \2\xc2\xa0\4ous \5 the \2e \0";\r | |
28 | /* Cx _2 _7___ ___ _A _F _5 _8 */\r | |
29 | \r | |
30 | static const uint16_t kPrefixSuffixMap[50] = {\r | |
31 | 0x00, 0x02, 0x05, 0x0E, 0x13, 0x16, 0x18, 0x1E, 0x23, 0x25,\r | |
32 | 0x2A, 0x2D, 0x2F, 0x32, 0x34, 0x3A, 0x3E, 0x45, 0x47, 0x4E,\r | |
33 | 0x55, 0x5A, 0x5C, 0x63, 0x68, 0x6D, 0x72, 0x77, 0x7A, 0x7C,\r | |
34 | 0x80, 0x83, 0x88, 0x8C, 0x8E, 0x91, 0x97, 0x9F, 0xA5, 0xA9,\r | |
35 | 0xAD, 0xB2, 0xB7, 0xBD, 0xC2, 0xC7, 0xCA, 0xCF, 0xD5, 0xD8\r | |
36 | };\r | |
37 | \r | |
38 | /* RFC 7932 transforms */\r | |
39 | static const uint8_t kTransformsData[] = {\r | |
40 | 49, BROTLI_TRANSFORM_IDENTITY, 49,\r | |
41 | 49, BROTLI_TRANSFORM_IDENTITY, 0,\r | |
42 | 0, BROTLI_TRANSFORM_IDENTITY, 0,\r | |
43 | 49, BROTLI_TRANSFORM_OMIT_FIRST_1, 49,\r | |
44 | 49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 0,\r | |
45 | 49, BROTLI_TRANSFORM_IDENTITY, 47,\r | |
46 | 0, BROTLI_TRANSFORM_IDENTITY, 49,\r | |
47 | 4, BROTLI_TRANSFORM_IDENTITY, 0,\r | |
48 | 49, BROTLI_TRANSFORM_IDENTITY, 3,\r | |
49 | 49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 49,\r | |
50 | 49, BROTLI_TRANSFORM_IDENTITY, 6,\r | |
51 | 49, BROTLI_TRANSFORM_OMIT_FIRST_2, 49,\r | |
52 | 49, BROTLI_TRANSFORM_OMIT_LAST_1, 49,\r | |
53 | 1, BROTLI_TRANSFORM_IDENTITY, 0,\r | |
54 | 49, BROTLI_TRANSFORM_IDENTITY, 1,\r | |
55 | 0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 0,\r | |
56 | 49, BROTLI_TRANSFORM_IDENTITY, 7,\r | |
57 | 49, BROTLI_TRANSFORM_IDENTITY, 9,\r | |
58 | 48, BROTLI_TRANSFORM_IDENTITY, 0,\r | |
59 | 49, BROTLI_TRANSFORM_IDENTITY, 8,\r | |
60 | 49, BROTLI_TRANSFORM_IDENTITY, 5,\r | |
61 | 49, BROTLI_TRANSFORM_IDENTITY, 10,\r | |
62 | 49, BROTLI_TRANSFORM_IDENTITY, 11,\r | |
63 | 49, BROTLI_TRANSFORM_OMIT_LAST_3, 49,\r | |
64 | 49, BROTLI_TRANSFORM_IDENTITY, 13,\r | |
65 | 49, BROTLI_TRANSFORM_IDENTITY, 14,\r | |
66 | 49, BROTLI_TRANSFORM_OMIT_FIRST_3, 49,\r | |
67 | 49, BROTLI_TRANSFORM_OMIT_LAST_2, 49,\r | |
68 | 49, BROTLI_TRANSFORM_IDENTITY, 15,\r | |
69 | 49, BROTLI_TRANSFORM_IDENTITY, 16,\r | |
70 | 0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 49,\r | |
71 | 49, BROTLI_TRANSFORM_IDENTITY, 12,\r | |
72 | 5, BROTLI_TRANSFORM_IDENTITY, 49,\r | |
73 | 0, BROTLI_TRANSFORM_IDENTITY, 1,\r | |
74 | 49, BROTLI_TRANSFORM_OMIT_FIRST_4, 49,\r | |
75 | 49, BROTLI_TRANSFORM_IDENTITY, 18,\r | |
76 | 49, BROTLI_TRANSFORM_IDENTITY, 17,\r | |
77 | 49, BROTLI_TRANSFORM_IDENTITY, 19,\r | |
78 | 49, BROTLI_TRANSFORM_IDENTITY, 20,\r | |
79 | 49, BROTLI_TRANSFORM_OMIT_FIRST_5, 49,\r | |
80 | 49, BROTLI_TRANSFORM_OMIT_FIRST_6, 49,\r | |
81 | 47, BROTLI_TRANSFORM_IDENTITY, 49,\r | |
82 | 49, BROTLI_TRANSFORM_OMIT_LAST_4, 49,\r | |
83 | 49, BROTLI_TRANSFORM_IDENTITY, 22,\r | |
84 | 49, BROTLI_TRANSFORM_UPPERCASE_ALL, 49,\r | |
85 | 49, BROTLI_TRANSFORM_IDENTITY, 23,\r | |
86 | 49, BROTLI_TRANSFORM_IDENTITY, 24,\r | |
87 | 49, BROTLI_TRANSFORM_IDENTITY, 25,\r | |
88 | 49, BROTLI_TRANSFORM_OMIT_LAST_7, 49,\r | |
89 | 49, BROTLI_TRANSFORM_OMIT_LAST_1, 26,\r | |
90 | 49, BROTLI_TRANSFORM_IDENTITY, 27,\r | |
91 | 49, BROTLI_TRANSFORM_IDENTITY, 28,\r | |
92 | 0, BROTLI_TRANSFORM_IDENTITY, 12,\r | |
93 | 49, BROTLI_TRANSFORM_IDENTITY, 29,\r | |
94 | 49, BROTLI_TRANSFORM_OMIT_FIRST_9, 49,\r | |
95 | 49, BROTLI_TRANSFORM_OMIT_FIRST_7, 49,\r | |
96 | 49, BROTLI_TRANSFORM_OMIT_LAST_6, 49,\r | |
97 | 49, BROTLI_TRANSFORM_IDENTITY, 21,\r | |
98 | 49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 1,\r | |
99 | 49, BROTLI_TRANSFORM_OMIT_LAST_8, 49,\r | |
100 | 49, BROTLI_TRANSFORM_IDENTITY, 31,\r | |
101 | 49, BROTLI_TRANSFORM_IDENTITY, 32,\r | |
102 | 47, BROTLI_TRANSFORM_IDENTITY, 3,\r | |
103 | 49, BROTLI_TRANSFORM_OMIT_LAST_5, 49,\r | |
104 | 49, BROTLI_TRANSFORM_OMIT_LAST_9, 49,\r | |
105 | 0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 1,\r | |
106 | 49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 8,\r | |
107 | 5, BROTLI_TRANSFORM_IDENTITY, 21,\r | |
108 | 49, BROTLI_TRANSFORM_UPPERCASE_ALL, 0,\r | |
109 | 49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 10,\r | |
110 | 49, BROTLI_TRANSFORM_IDENTITY, 30,\r | |
111 | 0, BROTLI_TRANSFORM_IDENTITY, 5,\r | |
112 | 35, BROTLI_TRANSFORM_IDENTITY, 49,\r | |
113 | 47, BROTLI_TRANSFORM_IDENTITY, 2,\r | |
114 | 49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 17,\r | |
115 | 49, BROTLI_TRANSFORM_IDENTITY, 36,\r | |
116 | 49, BROTLI_TRANSFORM_IDENTITY, 33,\r | |
117 | 5, BROTLI_TRANSFORM_IDENTITY, 0,\r | |
118 | 49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 21,\r | |
119 | 49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 5,\r | |
120 | 49, BROTLI_TRANSFORM_IDENTITY, 37,\r | |
121 | 0, BROTLI_TRANSFORM_IDENTITY, 30,\r | |
122 | 49, BROTLI_TRANSFORM_IDENTITY, 38,\r | |
123 | 0, BROTLI_TRANSFORM_UPPERCASE_ALL, 0,\r | |
124 | 49, BROTLI_TRANSFORM_IDENTITY, 39,\r | |
125 | 0, BROTLI_TRANSFORM_UPPERCASE_ALL, 49,\r | |
126 | 49, BROTLI_TRANSFORM_IDENTITY, 34,\r | |
127 | 49, BROTLI_TRANSFORM_UPPERCASE_ALL, 8,\r | |
128 | 49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 12,\r | |
129 | 0, BROTLI_TRANSFORM_IDENTITY, 21,\r | |
130 | 49, BROTLI_TRANSFORM_IDENTITY, 40,\r | |
131 | 0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 12,\r | |
132 | 49, BROTLI_TRANSFORM_IDENTITY, 41,\r | |
133 | 49, BROTLI_TRANSFORM_IDENTITY, 42,\r | |
134 | 49, BROTLI_TRANSFORM_UPPERCASE_ALL, 17,\r | |
135 | 49, BROTLI_TRANSFORM_IDENTITY, 43,\r | |
136 | 0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 5,\r | |
137 | 49, BROTLI_TRANSFORM_UPPERCASE_ALL, 10,\r | |
138 | 0, BROTLI_TRANSFORM_IDENTITY, 34,\r | |
139 | 49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 33,\r | |
140 | 49, BROTLI_TRANSFORM_IDENTITY, 44,\r | |
141 | 49, BROTLI_TRANSFORM_UPPERCASE_ALL, 5,\r | |
142 | 45, BROTLI_TRANSFORM_IDENTITY, 49,\r | |
143 | 0, BROTLI_TRANSFORM_IDENTITY, 33,\r | |
144 | 49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 30,\r | |
145 | 49, BROTLI_TRANSFORM_UPPERCASE_ALL, 30,\r | |
146 | 49, BROTLI_TRANSFORM_IDENTITY, 46,\r | |
147 | 49, BROTLI_TRANSFORM_UPPERCASE_ALL, 1,\r | |
148 | 49, BROTLI_TRANSFORM_UPPERCASE_FIRST, 34,\r | |
149 | 0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 33,\r | |
150 | 0, BROTLI_TRANSFORM_UPPERCASE_ALL, 30,\r | |
151 | 0, BROTLI_TRANSFORM_UPPERCASE_ALL, 1,\r | |
152 | 49, BROTLI_TRANSFORM_UPPERCASE_ALL, 33,\r | |
153 | 49, BROTLI_TRANSFORM_UPPERCASE_ALL, 21,\r | |
154 | 49, BROTLI_TRANSFORM_UPPERCASE_ALL, 12,\r | |
155 | 0, BROTLI_TRANSFORM_UPPERCASE_ALL, 5,\r | |
156 | 49, BROTLI_TRANSFORM_UPPERCASE_ALL, 34,\r | |
157 | 0, BROTLI_TRANSFORM_UPPERCASE_ALL, 12,\r | |
158 | 0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 30,\r | |
159 | 0, BROTLI_TRANSFORM_UPPERCASE_ALL, 34,\r | |
160 | 0, BROTLI_TRANSFORM_UPPERCASE_FIRST, 34,\r | |
161 | };\r | |
162 | \r | |
163 | static BrotliTransforms kBrotliTransforms = {\r | |
164 | sizeof(kPrefixSuffix),\r | |
165 | (const uint8_t*)kPrefixSuffix,\r | |
166 | kPrefixSuffixMap,\r | |
167 | sizeof(kTransformsData) / (3 * sizeof(kTransformsData[0])),\r | |
168 | kTransformsData,\r | |
169 | {0, 12, 27, 23, 42, 63, 56, 48, 59, 64}\r | |
170 | };\r | |
171 | \r | |
172 | const BrotliTransforms* BrotliGetTransforms(void) {\r | |
173 | return &kBrotliTransforms;\r | |
174 | }\r | |
175 | \r | |
176 | static int ToUpperCase(uint8_t* p) {\r | |
177 | if (p[0] < 0xC0) {\r | |
178 | if (p[0] >= 'a' && p[0] <= 'z') {\r | |
179 | p[0] ^= 32;\r | |
180 | }\r | |
181 | return 1;\r | |
182 | }\r | |
183 | /* An overly simplified uppercasing model for UTF-8. */\r | |
184 | if (p[0] < 0xE0) {\r | |
185 | p[1] ^= 32;\r | |
186 | return 2;\r | |
187 | }\r | |
188 | /* An arbitrary transform for three byte characters. */\r | |
189 | p[2] ^= 5;\r | |
190 | return 3;\r | |
191 | }\r | |
192 | \r | |
193 | int BrotliTransformDictionaryWord(uint8_t* dst, const uint8_t* word, int len,\r | |
194 | const BrotliTransforms* transforms, int transfom_idx) {\r | |
195 | int idx = 0;\r | |
196 | const uint8_t* prefix = BROTLI_TRANSFORM_PREFIX(transforms, transfom_idx);\r | |
197 | uint8_t type = BROTLI_TRANSFORM_TYPE(transforms, transfom_idx);\r | |
198 | const uint8_t* suffix = BROTLI_TRANSFORM_SUFFIX(transforms, transfom_idx);\r | |
199 | {\r | |
200 | int prefix_len = *prefix++;\r | |
201 | while (prefix_len--) { dst[idx++] = *prefix++; }\r | |
202 | }\r | |
203 | {\r | |
204 | const int t = type;\r | |
205 | int i = 0;\r | |
206 | if (t <= BROTLI_TRANSFORM_OMIT_LAST_9) {\r | |
207 | len -= t;\r | |
208 | } else if (t >= BROTLI_TRANSFORM_OMIT_FIRST_1\r | |
209 | && t <= BROTLI_TRANSFORM_OMIT_FIRST_9) {\r | |
210 | int skip = t - (BROTLI_TRANSFORM_OMIT_FIRST_1 - 1);\r | |
211 | word += skip;\r | |
212 | len -= skip;\r | |
213 | }\r | |
214 | while (i < len) { dst[idx++] = word[i++]; }\r | |
215 | if (t == BROTLI_TRANSFORM_UPPERCASE_FIRST) {\r | |
216 | ToUpperCase(&dst[idx - len]);\r | |
217 | } else if (t == BROTLI_TRANSFORM_UPPERCASE_ALL) {\r | |
218 | uint8_t* uppercase = &dst[idx - len];\r | |
219 | while (len > 0) {\r | |
220 | int step = ToUpperCase(uppercase);\r | |
221 | uppercase += step;\r | |
222 | len -= step;\r | |
223 | }\r | |
224 | }\r | |
225 | }\r | |
226 | {\r | |
227 | int suffix_len = *suffix++;\r | |
228 | while (suffix_len--) { dst[idx++] = *suffix++; }\r | |
229 | return idx;\r | |
230 | }\r | |
231 | }\r | |
232 | \r | |
233 | #if defined(__cplusplus) || defined(c_plusplus)\r | |
234 | } /* extern "C" */\r | |
235 | #endif\r |