1 /* Copyright 2013 Google Inc. All Rights Reserved.
3 Distributed under MIT license.
4 See file LICENSE for detail or copy at https://opensource.org/licenses/MIT
7 #include "./transform.h"
9 #if defined(__cplusplus) || defined(c_plusplus)
13 /* RFC 7932 transforms string data */
14 static const char kPrefixSuffix
[] =
15 "\1 \2, \10 of the \4 of \2s \1.\5 and \4 "
16 /* 0x _0 _2 __5 _E _3 _6 _8 _E */
17 "in \1\"\4 to \2\">\1\n\2. \1]\5 for \3 a \6 "
18 /* 2x _3_ _5 _A_ _D_ _F _2 _4 _A _E */
19 "that \1\'\6 with \6 from \4 by \1(\6. T"
20 /* 4x _5_ _7 _E _5 _A _C */
21 "he \4 on \4 as \4 is \4ing \2\n\t\1:\3ed "
22 /* 6x _3 _8 _D _2 _7_ _ _A _C */
23 "\2=\"\4 at \3ly \1,\2=\'\5.com/\7. This \5"
24 /* 8x _0 _ _3 _8 _C _E _ _1 _7 _F */
25 " not \3er \3al \4ful \4ive \5less \4es"
26 /* Ax _5 _9 _D _2 _7 _D */
27 "t \4ize \2\xc2\xa0\4ous \5 the \2e \0";
28 /* Cx _2 _7___ ___ _A _F _5 _8 */
30 static const uint16_t kPrefixSuffixMap
[50] = {
31 0x00, 0x02, 0x05, 0x0E, 0x13, 0x16, 0x18, 0x1E, 0x23, 0x25,
32 0x2A, 0x2D, 0x2F, 0x32, 0x34, 0x3A, 0x3E, 0x45, 0x47, 0x4E,
33 0x55, 0x5A, 0x5C, 0x63, 0x68, 0x6D, 0x72, 0x77, 0x7A, 0x7C,
34 0x80, 0x83, 0x88, 0x8C, 0x8E, 0x91, 0x97, 0x9F, 0xA5, 0xA9,
35 0xAD, 0xB2, 0xB7, 0xBD, 0xC2, 0xC7, 0xCA, 0xCF, 0xD5, 0xD8
38 /* RFC 7932 transforms */
39 static const uint8_t kTransformsData
[] = {
40 49, BROTLI_TRANSFORM_IDENTITY
, 49,
41 49, BROTLI_TRANSFORM_IDENTITY
, 0,
42 0, BROTLI_TRANSFORM_IDENTITY
, 0,
43 49, BROTLI_TRANSFORM_OMIT_FIRST_1
, 49,
44 49, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 0,
45 49, BROTLI_TRANSFORM_IDENTITY
, 47,
46 0, BROTLI_TRANSFORM_IDENTITY
, 49,
47 4, BROTLI_TRANSFORM_IDENTITY
, 0,
48 49, BROTLI_TRANSFORM_IDENTITY
, 3,
49 49, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 49,
50 49, BROTLI_TRANSFORM_IDENTITY
, 6,
51 49, BROTLI_TRANSFORM_OMIT_FIRST_2
, 49,
52 49, BROTLI_TRANSFORM_OMIT_LAST_1
, 49,
53 1, BROTLI_TRANSFORM_IDENTITY
, 0,
54 49, BROTLI_TRANSFORM_IDENTITY
, 1,
55 0, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 0,
56 49, BROTLI_TRANSFORM_IDENTITY
, 7,
57 49, BROTLI_TRANSFORM_IDENTITY
, 9,
58 48, BROTLI_TRANSFORM_IDENTITY
, 0,
59 49, BROTLI_TRANSFORM_IDENTITY
, 8,
60 49, BROTLI_TRANSFORM_IDENTITY
, 5,
61 49, BROTLI_TRANSFORM_IDENTITY
, 10,
62 49, BROTLI_TRANSFORM_IDENTITY
, 11,
63 49, BROTLI_TRANSFORM_OMIT_LAST_3
, 49,
64 49, BROTLI_TRANSFORM_IDENTITY
, 13,
65 49, BROTLI_TRANSFORM_IDENTITY
, 14,
66 49, BROTLI_TRANSFORM_OMIT_FIRST_3
, 49,
67 49, BROTLI_TRANSFORM_OMIT_LAST_2
, 49,
68 49, BROTLI_TRANSFORM_IDENTITY
, 15,
69 49, BROTLI_TRANSFORM_IDENTITY
, 16,
70 0, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 49,
71 49, BROTLI_TRANSFORM_IDENTITY
, 12,
72 5, BROTLI_TRANSFORM_IDENTITY
, 49,
73 0, BROTLI_TRANSFORM_IDENTITY
, 1,
74 49, BROTLI_TRANSFORM_OMIT_FIRST_4
, 49,
75 49, BROTLI_TRANSFORM_IDENTITY
, 18,
76 49, BROTLI_TRANSFORM_IDENTITY
, 17,
77 49, BROTLI_TRANSFORM_IDENTITY
, 19,
78 49, BROTLI_TRANSFORM_IDENTITY
, 20,
79 49, BROTLI_TRANSFORM_OMIT_FIRST_5
, 49,
80 49, BROTLI_TRANSFORM_OMIT_FIRST_6
, 49,
81 47, BROTLI_TRANSFORM_IDENTITY
, 49,
82 49, BROTLI_TRANSFORM_OMIT_LAST_4
, 49,
83 49, BROTLI_TRANSFORM_IDENTITY
, 22,
84 49, BROTLI_TRANSFORM_UPPERCASE_ALL
, 49,
85 49, BROTLI_TRANSFORM_IDENTITY
, 23,
86 49, BROTLI_TRANSFORM_IDENTITY
, 24,
87 49, BROTLI_TRANSFORM_IDENTITY
, 25,
88 49, BROTLI_TRANSFORM_OMIT_LAST_7
, 49,
89 49, BROTLI_TRANSFORM_OMIT_LAST_1
, 26,
90 49, BROTLI_TRANSFORM_IDENTITY
, 27,
91 49, BROTLI_TRANSFORM_IDENTITY
, 28,
92 0, BROTLI_TRANSFORM_IDENTITY
, 12,
93 49, BROTLI_TRANSFORM_IDENTITY
, 29,
94 49, BROTLI_TRANSFORM_OMIT_FIRST_9
, 49,
95 49, BROTLI_TRANSFORM_OMIT_FIRST_7
, 49,
96 49, BROTLI_TRANSFORM_OMIT_LAST_6
, 49,
97 49, BROTLI_TRANSFORM_IDENTITY
, 21,
98 49, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 1,
99 49, BROTLI_TRANSFORM_OMIT_LAST_8
, 49,
100 49, BROTLI_TRANSFORM_IDENTITY
, 31,
101 49, BROTLI_TRANSFORM_IDENTITY
, 32,
102 47, BROTLI_TRANSFORM_IDENTITY
, 3,
103 49, BROTLI_TRANSFORM_OMIT_LAST_5
, 49,
104 49, BROTLI_TRANSFORM_OMIT_LAST_9
, 49,
105 0, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 1,
106 49, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 8,
107 5, BROTLI_TRANSFORM_IDENTITY
, 21,
108 49, BROTLI_TRANSFORM_UPPERCASE_ALL
, 0,
109 49, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 10,
110 49, BROTLI_TRANSFORM_IDENTITY
, 30,
111 0, BROTLI_TRANSFORM_IDENTITY
, 5,
112 35, BROTLI_TRANSFORM_IDENTITY
, 49,
113 47, BROTLI_TRANSFORM_IDENTITY
, 2,
114 49, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 17,
115 49, BROTLI_TRANSFORM_IDENTITY
, 36,
116 49, BROTLI_TRANSFORM_IDENTITY
, 33,
117 5, BROTLI_TRANSFORM_IDENTITY
, 0,
118 49, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 21,
119 49, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 5,
120 49, BROTLI_TRANSFORM_IDENTITY
, 37,
121 0, BROTLI_TRANSFORM_IDENTITY
, 30,
122 49, BROTLI_TRANSFORM_IDENTITY
, 38,
123 0, BROTLI_TRANSFORM_UPPERCASE_ALL
, 0,
124 49, BROTLI_TRANSFORM_IDENTITY
, 39,
125 0, BROTLI_TRANSFORM_UPPERCASE_ALL
, 49,
126 49, BROTLI_TRANSFORM_IDENTITY
, 34,
127 49, BROTLI_TRANSFORM_UPPERCASE_ALL
, 8,
128 49, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 12,
129 0, BROTLI_TRANSFORM_IDENTITY
, 21,
130 49, BROTLI_TRANSFORM_IDENTITY
, 40,
131 0, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 12,
132 49, BROTLI_TRANSFORM_IDENTITY
, 41,
133 49, BROTLI_TRANSFORM_IDENTITY
, 42,
134 49, BROTLI_TRANSFORM_UPPERCASE_ALL
, 17,
135 49, BROTLI_TRANSFORM_IDENTITY
, 43,
136 0, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 5,
137 49, BROTLI_TRANSFORM_UPPERCASE_ALL
, 10,
138 0, BROTLI_TRANSFORM_IDENTITY
, 34,
139 49, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 33,
140 49, BROTLI_TRANSFORM_IDENTITY
, 44,
141 49, BROTLI_TRANSFORM_UPPERCASE_ALL
, 5,
142 45, BROTLI_TRANSFORM_IDENTITY
, 49,
143 0, BROTLI_TRANSFORM_IDENTITY
, 33,
144 49, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 30,
145 49, BROTLI_TRANSFORM_UPPERCASE_ALL
, 30,
146 49, BROTLI_TRANSFORM_IDENTITY
, 46,
147 49, BROTLI_TRANSFORM_UPPERCASE_ALL
, 1,
148 49, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 34,
149 0, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 33,
150 0, BROTLI_TRANSFORM_UPPERCASE_ALL
, 30,
151 0, BROTLI_TRANSFORM_UPPERCASE_ALL
, 1,
152 49, BROTLI_TRANSFORM_UPPERCASE_ALL
, 33,
153 49, BROTLI_TRANSFORM_UPPERCASE_ALL
, 21,
154 49, BROTLI_TRANSFORM_UPPERCASE_ALL
, 12,
155 0, BROTLI_TRANSFORM_UPPERCASE_ALL
, 5,
156 49, BROTLI_TRANSFORM_UPPERCASE_ALL
, 34,
157 0, BROTLI_TRANSFORM_UPPERCASE_ALL
, 12,
158 0, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 30,
159 0, BROTLI_TRANSFORM_UPPERCASE_ALL
, 34,
160 0, BROTLI_TRANSFORM_UPPERCASE_FIRST
, 34,
163 static BrotliTransforms kBrotliTransforms
= {
164 sizeof(kPrefixSuffix
),
165 (const uint8_t*)kPrefixSuffix
,
167 sizeof(kTransformsData
) / (3 * sizeof(kTransformsData
[0])),
169 {0, 12, 27, 23, 42, 63, 56, 48, 59, 64}
172 const BrotliTransforms
* BrotliGetTransforms(void) {
173 return &kBrotliTransforms
;
176 static int ToUpperCase(uint8_t* p
) {
178 if (p
[0] >= 'a' && p
[0] <= 'z') {
183 /* An overly simplified uppercasing model for UTF-8. */
188 /* An arbitrary transform for three byte characters. */
193 int BrotliTransformDictionaryWord(uint8_t* dst
, const uint8_t* word
, int len
,
194 const BrotliTransforms
* transforms
, int transfom_idx
) {
196 const uint8_t* prefix
= BROTLI_TRANSFORM_PREFIX(transforms
, transfom_idx
);
197 uint8_t type
= BROTLI_TRANSFORM_TYPE(transforms
, transfom_idx
);
198 const uint8_t* suffix
= BROTLI_TRANSFORM_SUFFIX(transforms
, transfom_idx
);
200 int prefix_len
= *prefix
++;
201 while (prefix_len
--) { dst
[idx
++] = *prefix
++; }
206 if (t
<= BROTLI_TRANSFORM_OMIT_LAST_9
) {
208 } else if (t
>= BROTLI_TRANSFORM_OMIT_FIRST_1
209 && t
<= BROTLI_TRANSFORM_OMIT_FIRST_9
) {
210 int skip
= t
- (BROTLI_TRANSFORM_OMIT_FIRST_1
- 1);
214 while (i
< len
) { dst
[idx
++] = word
[i
++]; }
215 if (t
== BROTLI_TRANSFORM_UPPERCASE_FIRST
) {
216 ToUpperCase(&dst
[idx
- len
]);
217 } else if (t
== BROTLI_TRANSFORM_UPPERCASE_ALL
) {
218 uint8_t* uppercase
= &dst
[idx
- len
];
220 int step
= ToUpperCase(uppercase
);
227 int suffix_len
= *suffix
++;
228 while (suffix_len
--) { dst
[idx
++] = *suffix
++; }
233 #if defined(__cplusplus) || defined(c_plusplus)