]>
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 | /* Heuristics for deciding about the UTF8-ness of strings. */\r | |
8 | \r | |
9 | #include "./utf8_util.h"\r | |
10 | \r | |
dd4f667e | 11 | #include <brotli/types.h>\r |
11b7501a SB |
12 | \r |
13 | #if defined(__cplusplus) || defined(c_plusplus)\r | |
14 | extern "C" {\r | |
15 | #endif\r | |
16 | \r | |
17 | static size_t BrotliParseAsUTF8(\r | |
18 | int* symbol, const uint8_t* input, size_t size) {\r | |
19 | /* ASCII */\r | |
20 | if ((input[0] & 0x80) == 0) {\r | |
21 | *symbol = input[0];\r | |
22 | if (*symbol > 0) {\r | |
23 | return 1;\r | |
24 | }\r | |
25 | }\r | |
26 | /* 2-byte UTF8 */\r | |
27 | if (size > 1u &&\r | |
dd4f667e LG |
28 | (input[0] & 0xE0) == 0xC0 &&\r |
29 | (input[1] & 0xC0) == 0x80) {\r | |
30 | *symbol = (((input[0] & 0x1F) << 6) |\r | |
31 | (input[1] & 0x3F));\r | |
32 | if (*symbol > 0x7F) {\r | |
11b7501a SB |
33 | return 2;\r |
34 | }\r | |
35 | }\r | |
36 | /* 3-byte UFT8 */\r | |
37 | if (size > 2u &&\r | |
dd4f667e LG |
38 | (input[0] & 0xF0) == 0xE0 &&\r |
39 | (input[1] & 0xC0) == 0x80 &&\r | |
40 | (input[2] & 0xC0) == 0x80) {\r | |
41 | *symbol = (((input[0] & 0x0F) << 12) |\r | |
42 | ((input[1] & 0x3F) << 6) |\r | |
43 | (input[2] & 0x3F));\r | |
44 | if (*symbol > 0x7FF) {\r | |
11b7501a SB |
45 | return 3;\r |
46 | }\r | |
47 | }\r | |
48 | /* 4-byte UFT8 */\r | |
49 | if (size > 3u &&\r | |
dd4f667e LG |
50 | (input[0] & 0xF8) == 0xF0 &&\r |
51 | (input[1] & 0xC0) == 0x80 &&\r | |
52 | (input[2] & 0xC0) == 0x80 &&\r | |
53 | (input[3] & 0xC0) == 0x80) {\r | |
11b7501a | 54 | *symbol = (((input[0] & 0x07) << 18) |\r |
dd4f667e LG |
55 | ((input[1] & 0x3F) << 12) |\r |
56 | ((input[2] & 0x3F) << 6) |\r | |
57 | (input[3] & 0x3F));\r | |
58 | if (*symbol > 0xFFFF && *symbol <= 0x10FFFF) {\r | |
11b7501a SB |
59 | return 4;\r |
60 | }\r | |
61 | }\r | |
62 | /* Not UTF8, emit a special symbol above the UTF8-code space */\r | |
63 | *symbol = 0x110000 | input[0];\r | |
64 | return 1;\r | |
65 | }\r | |
66 | \r | |
67 | /* Returns 1 if at least min_fraction of the data is UTF8-encoded.*/\r | |
68 | BROTLI_BOOL BrotliIsMostlyUTF8(\r | |
69 | const uint8_t* data, const size_t pos, const size_t mask,\r | |
70 | const size_t length, const double min_fraction) {\r | |
71 | size_t size_utf8 = 0;\r | |
72 | size_t i = 0;\r | |
73 | while (i < length) {\r | |
74 | int symbol;\r | |
75 | size_t bytes_read =\r | |
76 | BrotliParseAsUTF8(&symbol, &data[(pos + i) & mask], length - i);\r | |
77 | i += bytes_read;\r | |
78 | if (symbol < 0x110000) size_utf8 += bytes_read;\r | |
79 | }\r | |
80 | return TO_BROTLI_BOOL(size_utf8 > min_fraction * (double)length);\r | |
81 | }\r | |
82 | \r | |
83 | #if defined(__cplusplus) || defined(c_plusplus)\r | |
84 | } /* extern "C" */\r | |
85 | #endif\r |