Emily Bernier | d0a1eb7 | 2015-03-24 16:35:39 -0400 | [diff] [blame^] | 1 | // Copyright 2014 the V8 project authors. All rights reserved. |
| 2 | // Use of this source code is governed by a BSD-style license that can be |
| 3 | // found in the LICENSE file. |
| 4 | |
| 5 | |
| 6 | #include "src/unicode-inl.h" |
| 7 | #include "src/unicode-decoder.h" |
| 8 | #include <stdio.h> |
| 9 | #include <stdlib.h> |
| 10 | |
| 11 | namespace unibrow { |
| 12 | |
| 13 | void Utf8DecoderBase::Reset(uint16_t* buffer, unsigned buffer_length, |
| 14 | const uint8_t* stream, unsigned stream_length) { |
| 15 | // Assume everything will fit in the buffer and stream won't be needed. |
| 16 | last_byte_of_buffer_unused_ = false; |
| 17 | unbuffered_start_ = NULL; |
| 18 | bool writing_to_buffer = true; |
| 19 | // Loop until stream is read, writing to buffer as long as buffer has space. |
| 20 | unsigned utf16_length = 0; |
| 21 | while (stream_length != 0) { |
| 22 | unsigned cursor = 0; |
| 23 | uint32_t character = Utf8::ValueOf(stream, stream_length, &cursor); |
| 24 | DCHECK(cursor > 0 && cursor <= stream_length); |
| 25 | stream += cursor; |
| 26 | stream_length -= cursor; |
| 27 | bool is_two_characters = character > Utf16::kMaxNonSurrogateCharCode; |
| 28 | utf16_length += is_two_characters ? 2 : 1; |
| 29 | // Don't need to write to the buffer, but still need utf16_length. |
| 30 | if (!writing_to_buffer) continue; |
| 31 | // Write out the characters to the buffer. |
| 32 | // Must check for equality with buffer_length as we've already updated it. |
| 33 | if (utf16_length <= buffer_length) { |
| 34 | if (is_two_characters) { |
| 35 | *buffer++ = Utf16::LeadSurrogate(character); |
| 36 | *buffer++ = Utf16::TrailSurrogate(character); |
| 37 | } else { |
| 38 | *buffer++ = character; |
| 39 | } |
| 40 | if (utf16_length == buffer_length) { |
| 41 | // Just wrote last character of buffer |
| 42 | writing_to_buffer = false; |
| 43 | unbuffered_start_ = stream; |
| 44 | } |
| 45 | continue; |
| 46 | } |
| 47 | // Have gone over buffer. |
| 48 | // Last char of buffer is unused, set cursor back. |
| 49 | DCHECK(is_two_characters); |
| 50 | writing_to_buffer = false; |
| 51 | last_byte_of_buffer_unused_ = true; |
| 52 | unbuffered_start_ = stream - cursor; |
| 53 | } |
| 54 | utf16_length_ = utf16_length; |
| 55 | } |
| 56 | |
| 57 | |
| 58 | void Utf8DecoderBase::WriteUtf16Slow(const uint8_t* stream, uint16_t* data, |
| 59 | unsigned data_length) { |
| 60 | while (data_length != 0) { |
| 61 | unsigned cursor = 0; |
| 62 | uint32_t character = Utf8::ValueOf(stream, Utf8::kMaxEncodedSize, &cursor); |
| 63 | // There's a total lack of bounds checking for stream |
| 64 | // as it was already done in Reset. |
| 65 | stream += cursor; |
| 66 | if (character > unibrow::Utf16::kMaxNonSurrogateCharCode) { |
| 67 | *data++ = Utf16::LeadSurrogate(character); |
| 68 | *data++ = Utf16::TrailSurrogate(character); |
| 69 | DCHECK(data_length > 1); |
| 70 | data_length -= 2; |
| 71 | } else { |
| 72 | *data++ = character; |
| 73 | data_length -= 1; |
| 74 | } |
| 75 | } |
| 76 | } |
| 77 | |
| 78 | } // namespace unibrow |