| Index: third_party/brotli/enc/utf8_util.c
|
| diff --git a/third_party/brotli/enc/utf8_util.cc b/third_party/brotli/enc/utf8_util.c
|
| similarity index 62%
|
| rename from third_party/brotli/enc/utf8_util.cc
|
| rename to third_party/brotli/enc/utf8_util.c
|
| index a2b5c3a6776477af0b1132667fb8d6b05e4ccee6..a334927f369132816a66b0762c24dbe556736d09 100644
|
| --- a/third_party/brotli/enc/utf8_util.cc
|
| +++ b/third_party/brotli/enc/utf8_util.c
|
| @@ -4,25 +4,26 @@
|
| See file LICENSE for detail or copy at https://opensource.org/licenses/MIT
|
| */
|
|
|
| -// Heuristics for deciding about the UTF8-ness of strings.
|
| +/* Heuristics for deciding about the UTF8-ness of strings. */
|
|
|
| #include "./utf8_util.h"
|
|
|
| -#include "./types.h"
|
| +#include <brotli/types.h>
|
|
|
| -namespace brotli {
|
| +#if defined(__cplusplus) || defined(c_plusplus)
|
| +extern "C" {
|
| +#endif
|
|
|
| -namespace {
|
| -
|
| -size_t ParseAsUTF8(int* symbol, const uint8_t* input, size_t size) {
|
| - // ASCII
|
| +static size_t BrotliParseAsUTF8(
|
| + int* symbol, const uint8_t* input, size_t size) {
|
| + /* ASCII */
|
| if ((input[0] & 0x80) == 0) {
|
| *symbol = input[0];
|
| if (*symbol > 0) {
|
| return 1;
|
| }
|
| }
|
| - // 2-byte UTF8
|
| + /* 2-byte UTF8 */
|
| if (size > 1u &&
|
| (input[0] & 0xe0) == 0xc0 &&
|
| (input[1] & 0xc0) == 0x80) {
|
| @@ -32,7 +33,7 @@ size_t ParseAsUTF8(int* symbol, const uint8_t* input, size_t size) {
|
| return 2;
|
| }
|
| }
|
| - // 3-byte UFT8
|
| + /* 3-byte UFT8 */
|
| if (size > 2u &&
|
| (input[0] & 0xf0) == 0xe0 &&
|
| (input[1] & 0xc0) == 0x80 &&
|
| @@ -44,7 +45,7 @@ size_t ParseAsUTF8(int* symbol, const uint8_t* input, size_t size) {
|
| return 3;
|
| }
|
| }
|
| - // 4-byte UFT8
|
| + /* 4-byte UFT8 */
|
| if (size > 3u &&
|
| (input[0] & 0xf8) == 0xf0 &&
|
| (input[1] & 0xc0) == 0x80 &&
|
| @@ -58,26 +59,27 @@ size_t ParseAsUTF8(int* symbol, const uint8_t* input, size_t size) {
|
| return 4;
|
| }
|
| }
|
| - // Not UTF8, emit a special symbol above the UTF8-code space
|
| + /* 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) {
|
| +/* Returns 1 if at least min_fraction of the data is UTF8-encoded.*/
|
| +BROTLI_BOOL BrotliIsMostlyUTF8(
|
| + 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;
|
| - size_t bytes_read = ParseAsUTF8(
|
| - &symbol, &data[(pos + i) & mask], length - i);
|
| + size_t bytes_read =
|
| + BrotliParseAsUTF8(&symbol, &data[(pos + i) & mask], length - i);
|
| i += bytes_read;
|
| if (symbol < 0x110000) size_utf8 += bytes_read;
|
| }
|
| - return size_utf8 > min_fraction * static_cast<double>(length);
|
| + return TO_BROTLI_BOOL(size_utf8 > min_fraction * (double)length);
|
| }
|
|
|
| -} // namespace brotli
|
| +#if defined(__cplusplus) || defined(c_plusplus)
|
| +} /* extern "C" */
|
| +#endif
|
|
|