2015-11-27 10:27:11 +00:00
|
|
|
/* Copyright 2013 Google Inc. All Rights Reserved.
|
|
|
|
|
2015-12-11 10:11:51 +00:00
|
|
|
Distributed under MIT license.
|
2015-11-27 10:27:11 +00:00
|
|
|
See file LICENSE for detail or copy at https://opensource.org/licenses/MIT
|
|
|
|
*/
|
|
|
|
|
2015-10-01 13:10:42 +00:00
|
|
|
// Heuristics for deciding about the UTF8-ness of strings.
|
|
|
|
|
|
|
|
#include "./utf8_util.h"
|
|
|
|
|
|
|
|
#include "./types.h"
|
|
|
|
|
|
|
|
namespace brotli {
|
|
|
|
|
|
|
|
namespace {
|
|
|
|
|
2016-01-07 15:27:49 +00:00
|
|
|
size_t ParseAsUTF8(int* symbol, const uint8_t* input, size_t size) {
|
2015-10-01 13:10:42 +00:00
|
|
|
// ASCII
|
|
|
|
if ((input[0] & 0x80) == 0) {
|
|
|
|
*symbol = input[0];
|
|
|
|
if (*symbol > 0) {
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
// 2-byte UTF8
|
2015-10-28 16:44:47 +00:00
|
|
|
if (size > 1u &&
|
2015-10-01 13:10:42 +00:00
|
|
|
(input[0] & 0xe0) == 0xc0 &&
|
|
|
|
(input[1] & 0xc0) == 0x80) {
|
|
|
|
*symbol = (((input[0] & 0x1f) << 6) |
|
|
|
|
(input[1] & 0x3f));
|
|
|
|
if (*symbol > 0x7f) {
|
|
|
|
return 2;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
// 3-byte UFT8
|
2015-10-28 16:44:47 +00:00
|
|
|
if (size > 2u &&
|
2015-10-01 13:10:42 +00:00
|
|
|
(input[0] & 0xf0) == 0xe0 &&
|
|
|
|
(input[1] & 0xc0) == 0x80 &&
|
|
|
|
(input[2] & 0xc0) == 0x80) {
|
|
|
|
*symbol = (((input[0] & 0x0f) << 12) |
|
|
|
|
((input[1] & 0x3f) << 6) |
|
|
|
|
(input[2] & 0x3f));
|
|
|
|
if (*symbol > 0x7ff) {
|
|
|
|
return 3;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
// 4-byte UFT8
|
2015-10-28 16:44:47 +00:00
|
|
|
if (size > 3u &&
|
2015-10-01 13:10:42 +00:00
|
|
|
(input[0] & 0xf8) == 0xf0 &&
|
|
|
|
(input[1] & 0xc0) == 0x80 &&
|
|
|
|
(input[2] & 0xc0) == 0x80 &&
|
|
|
|
(input[3] & 0xc0) == 0x80) {
|
|
|
|
*symbol = (((input[0] & 0x07) << 18) |
|
|
|
|
((input[1] & 0x3f) << 12) |
|
|
|
|
((input[2] & 0x3f) << 6) |
|
|
|
|
(input[3] & 0x3f));
|
|
|
|
if (*symbol > 0xffff && *symbol <= 0x10ffff) {
|
|
|
|
return 4;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
// Not UTF8, emit a special symbol above the UTF8-code space
|
|
|
|
*symbol = 0x110000 | input[0];
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
} // namespace
|
|
|
|
|
|
|
|
// Returns true if at least min_fraction of the data is UTF8-encoded.
|
|
|
|
bool IsMostlyUTF8(const uint8_t* data, const size_t pos, const size_t mask,
|
|
|
|
const size_t length, const double min_fraction) {
|
|
|
|
size_t size_utf8 = 0;
|
|
|
|
size_t i = 0;
|
|
|
|
while (i < length) {
|
|
|
|
int symbol;
|
2016-01-07 15:27:49 +00:00
|
|
|
size_t bytes_read = ParseAsUTF8(
|
|
|
|
&symbol, &data[(pos + i) & mask], length - i);
|
2015-10-01 13:10:42 +00:00
|
|
|
i += bytes_read;
|
|
|
|
if (symbol < 0x110000) size_utf8 += bytes_read;
|
|
|
|
}
|
2015-10-28 16:44:47 +00:00
|
|
|
return size_utf8 > min_fraction * static_cast<double>(length);
|
2015-10-01 13:10:42 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
} // namespace brotli
|