1 // Copyright 2017 The PDFium Authors 2 // Use of this source code is governed by a BSD-style license that can be 3 // found in the LICENSE file. 4 5 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com 6 7 #include "core/fxcrt/cfx_utf8decoder.h" 8 9 #include <utility> 10 CFX_UTF8Decoder(ByteStringView input)11CFX_UTF8Decoder::CFX_UTF8Decoder(ByteStringView input) { 12 for (char c : input) { 13 ProcessByte(c); 14 } 15 } 16 17 CFX_UTF8Decoder::~CFX_UTF8Decoder() = default; 18 TakeResult()19WideString CFX_UTF8Decoder::TakeResult() { 20 return std::move(m_Buffer); 21 } 22 AppendCodePoint(uint32_t ch)23void CFX_UTF8Decoder::AppendCodePoint(uint32_t ch) { 24 m_Buffer += static_cast<wchar_t>(ch); 25 } 26 ProcessByte(uint8_t byte)27void CFX_UTF8Decoder::ProcessByte(uint8_t byte) { 28 if (byte < 0x80) { 29 m_PendingBytes = 0; 30 AppendCodePoint(byte); 31 } else if (byte < 0xc0) { 32 if (m_PendingBytes == 0) { 33 return; 34 } 35 m_PendingBytes--; 36 m_PendingChar |= (byte & 0x3f) << (m_PendingBytes * 6); 37 if (m_PendingBytes == 0) { 38 AppendCodePoint(m_PendingChar); 39 } 40 } else if (byte < 0xe0) { 41 m_PendingBytes = 1; 42 m_PendingChar = (byte & 0x1f) << 6; 43 } else if (byte < 0xf0) { 44 m_PendingBytes = 2; 45 m_PendingChar = (byte & 0x0f) << 12; 46 } else if (byte < 0xf8) { 47 m_PendingBytes = 3; 48 m_PendingChar = (byte & 0x07) << 18; 49 } else if (byte < 0xfc) { 50 m_PendingBytes = 4; 51 m_PendingChar = (byte & 0x03) << 24; 52 } else if (byte < 0xfe) { 53 m_PendingBytes = 5; 54 m_PendingChar = (byte & 0x01) << 30; 55 } else { 56 m_PendingBytes = 0; 57 } 58 } 59