Chromium Code Reviews| OLD | NEW |
|---|---|
| 1 // Copyright 2014 PDFium Authors. All rights reserved. | 1 // Copyright 2014 PDFium Authors. All rights reserved. |
| 2 // Use of this source code is governed by a BSD-style license that can be | 2 // Use of this source code is governed by a BSD-style license that can be |
| 3 // found in the LICENSE file. | 3 // found in the LICENSE file. |
| 4 | 4 |
| 5 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com | 5 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com |
| 6 | 6 |
| 7 #include "core/fxcrt/xml_int.h" | 7 #include "core/fxcrt/xml_int.h" |
| 8 | 8 |
| 9 #include <vector> | 9 #include <vector> |
| 10 | 10 |
| 11 #include "core/fxcrt/fx_ext.h" | 11 #include "core/fxcrt/fx_ext.h" |
| 12 #include "core/fxcrt/fx_xml.h" | 12 #include "core/fxcrt/fx_xml.h" |
| 13 #include "third_party/base/ptr_util.h" | 13 #include "third_party/base/ptr_util.h" |
| 14 #include "third_party/base/stl_util.h" | 14 #include "third_party/base/stl_util.h" |
| 15 | 15 |
| 16 namespace { | |
| 17 | |
| 18 #define FXCRTM_XML_CHARTYPE_Normal 0x00 | |
|
npm
2016/10/26 18:26:04
Some of these appear to be unused. Remove them
Tom Sepez
2016/10/26 20:47:03
Yes, but they document what the bits are in the by
| |
| 19 #define FXCRTM_XML_CHARTYPE_SpaceChar 0x01 | |
| 20 #define FXCRTM_XML_CHARTYPE_Letter 0x02 | |
| 21 #define FXCRTM_XML_CHARTYPE_Digital 0x04 | |
| 22 #define FXCRTM_XML_CHARTYPE_NameIntro 0x08 | |
| 23 #define FXCRTM_XML_CHARTYPE_NameChar 0x10 | |
| 24 #define FXCRTM_XML_CHARTYPE_HexDigital 0x20 | |
| 25 #define FXCRTM_XML_CHARTYPE_HexLowerLetter 0x40 | |
| 26 #define FXCRTM_XML_CHARTYPE_HexUpperLetter 0x60 | |
| 27 #define FXCRTM_XML_CHARTYPE_HexChar 0x60 | |
| 28 | |
| 29 const uint8_t g_FXCRT_XML_ByteTypes[256] = { | |
| 30 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, | |
| 31 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, | |
| 32 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x00, 0x00, 0x00, | |
| 33 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x10, 0x10, 0x00, | |
| 34 0x34, 0x34, 0x34, 0x34, 0x34, 0x34, 0x34, 0x34, 0x34, 0x34, 0x08, 0x00, | |
| 35 0x00, 0x00, 0x00, 0x00, 0x00, 0x7A, 0x7A, 0x7A, 0x7A, 0x7A, 0x7A, 0x1A, | |
| 36 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 37 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x00, 0x00, 0x00, 0x00, 0x18, | |
| 38 0x00, 0x5A, 0x5A, 0x5A, 0x5A, 0x5A, 0x5A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 39 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 40 0x1A, 0x1A, 0x1A, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 41 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 42 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 43 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 44 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 45 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 46 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 47 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 48 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 49 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 50 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 51 0x1A, 0x1A, 0x01, 0x01, | |
| 52 }; | |
| 53 | |
| 54 bool g_FXCRT_XML_IsWhiteSpace(uint8_t ch) { | |
| 55 return !!(g_FXCRT_XML_ByteTypes[ch] & FXCRTM_XML_CHARTYPE_SpaceChar); | |
| 56 } | |
| 57 | |
| 58 bool g_FXCRT_XML_IsDigital(uint8_t ch) { | |
| 59 return !!(g_FXCRT_XML_ByteTypes[ch] & FXCRTM_XML_CHARTYPE_Digital); | |
| 60 } | |
| 61 | |
| 62 bool g_FXCRT_XML_IsNameIntro(uint8_t ch) { | |
| 63 return !!(g_FXCRT_XML_ByteTypes[ch] & FXCRTM_XML_CHARTYPE_NameIntro); | |
| 64 } | |
| 65 | |
| 66 bool g_FXCRT_XML_IsNameChar(uint8_t ch) { | |
| 67 return !!(g_FXCRT_XML_ByteTypes[ch] & FXCRTM_XML_CHARTYPE_NameChar); | |
| 68 } | |
| 69 | |
| 70 } // namespace | |
| 71 | |
| 16 CXML_DataBufAcc::CXML_DataBufAcc(const uint8_t* pBuffer, size_t size) | 72 CXML_DataBufAcc::CXML_DataBufAcc(const uint8_t* pBuffer, size_t size) |
| 17 : m_pBuffer(pBuffer), m_dwSize(size), m_dwCurPos(0) {} | 73 : m_pBuffer(pBuffer), m_dwSize(size), m_dwCurPos(0) {} |
| 18 | 74 |
| 19 CXML_DataBufAcc::~CXML_DataBufAcc() {} | 75 CXML_DataBufAcc::~CXML_DataBufAcc() {} |
| 20 | 76 |
| 21 void CXML_DataBufAcc::Release() { | 77 void CXML_DataBufAcc::Release() { |
| 22 delete this; | 78 delete this; |
| 23 } | 79 } |
| 24 | 80 |
| 25 FX_BOOL CXML_DataBufAcc::IsEOF() { | 81 FX_BOOL CXML_DataBufAcc::IsEOF() { |
| (...skipping 95 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
| 121 m_dwBufferSize(0), | 177 m_dwBufferSize(0), |
| 122 m_nBufferOffset(0), | 178 m_nBufferOffset(0), |
| 123 m_dwIndex(0) {} | 179 m_dwIndex(0) {} |
| 124 | 180 |
| 125 CXML_Parser::~CXML_Parser() { | 181 CXML_Parser::~CXML_Parser() { |
| 126 if (m_bOwnedStream) { | 182 if (m_bOwnedStream) { |
| 127 m_pDataAcc->Release(); | 183 m_pDataAcc->Release(); |
| 128 } | 184 } |
| 129 } | 185 } |
| 130 | 186 |
| 131 FX_BOOL CXML_Parser::Init(uint8_t* pBuffer, size_t size) { | 187 bool CXML_Parser::Init(uint8_t* pBuffer, size_t size) { |
| 132 m_pDataAcc = new CXML_DataBufAcc(pBuffer, size); | 188 m_pDataAcc = new CXML_DataBufAcc(pBuffer, size); |
| 133 return Init(TRUE); | 189 return Init(true); |
| 134 } | 190 } |
| 135 FX_BOOL CXML_Parser::Init(IFX_SeekableReadStream* pFileRead) { | 191 |
| 192 bool CXML_Parser::Init(IFX_SeekableReadStream* pFileRead) { | |
| 136 m_pDataAcc = new CXML_DataStmAcc(pFileRead); | 193 m_pDataAcc = new CXML_DataStmAcc(pFileRead); |
| 137 return Init(TRUE); | 194 return Init(true); |
| 138 } | 195 } |
| 139 FX_BOOL CXML_Parser::Init(IFX_BufferRead* pBuffer) { | 196 |
| 140 if (!pBuffer) { | 197 bool CXML_Parser::Init(IFX_BufferRead* pBuffer) { |
| 141 return FALSE; | 198 if (!pBuffer) |
| 142 } | 199 return false; |
| 200 | |
| 143 m_pDataAcc = pBuffer; | 201 m_pDataAcc = pBuffer; |
| 144 return Init(FALSE); | 202 return Init(false); |
| 145 } | 203 } |
| 146 FX_BOOL CXML_Parser::Init(FX_BOOL bOwndedStream) { | 204 |
| 205 bool CXML_Parser::Init(bool bOwndedStream) { | |
| 147 m_bOwnedStream = bOwndedStream; | 206 m_bOwnedStream = bOwndedStream; |
| 148 m_nOffset = 0; | 207 m_nOffset = 0; |
| 149 return ReadNextBlock(); | 208 return ReadNextBlock(); |
| 150 } | 209 } |
| 151 FX_BOOL CXML_Parser::ReadNextBlock() { | 210 |
| 152 if (!m_pDataAcc->ReadNextBlock()) { | 211 bool CXML_Parser::ReadNextBlock() { |
| 153 return FALSE; | 212 if (!m_pDataAcc->ReadNextBlock()) |
| 154 } | 213 return false; |
| 214 | |
| 155 m_pBuffer = m_pDataAcc->GetBlockBuffer(); | 215 m_pBuffer = m_pDataAcc->GetBlockBuffer(); |
| 156 m_dwBufferSize = m_pDataAcc->GetBlockSize(); | 216 m_dwBufferSize = m_pDataAcc->GetBlockSize(); |
| 157 m_nBufferOffset = m_pDataAcc->GetBlockOffset(); | 217 m_nBufferOffset = m_pDataAcc->GetBlockOffset(); |
| 158 m_dwIndex = 0; | 218 m_dwIndex = 0; |
| 159 return m_dwBufferSize > 0; | 219 return m_dwBufferSize > 0; |
| 160 } | 220 } |
| 161 FX_BOOL CXML_Parser::IsEOF() { | 221 |
| 162 if (!m_pDataAcc->IsEOF()) { | 222 bool CXML_Parser::IsEOF() { |
| 163 return FALSE; | 223 return m_pDataAcc->IsEOF() && m_dwIndex >= m_dwBufferSize; |
| 164 } | |
| 165 return m_dwIndex >= m_dwBufferSize; | |
| 166 } | 224 } |
| 167 #define FXCRTM_XML_CHARTYPE_Normal 0x00 | 225 |
| 168 #define FXCRTM_XML_CHARTYPE_SpaceChar 0x01 | |
| 169 #define FXCRTM_XML_CHARTYPE_Letter 0x02 | |
| 170 #define FXCRTM_XML_CHARTYPE_Digital 0x04 | |
| 171 #define FXCRTM_XML_CHARTYPE_NameIntro 0x08 | |
| 172 #define FXCRTM_XML_CHARTYPE_NameChar 0x10 | |
| 173 #define FXCRTM_XML_CHARTYPE_HexDigital 0x20 | |
| 174 #define FXCRTM_XML_CHARTYPE_HexLowerLetter 0x40 | |
| 175 #define FXCRTM_XML_CHARTYPE_HexUpperLetter 0x60 | |
| 176 #define FXCRTM_XML_CHARTYPE_HexChar 0x60 | |
| 177 uint8_t g_FXCRT_XML_ByteTypes[256] = { | |
| 178 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, | |
| 179 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, | |
| 180 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x00, 0x00, 0x00, | |
| 181 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x10, 0x10, 0x00, | |
| 182 0x34, 0x34, 0x34, 0x34, 0x34, 0x34, 0x34, 0x34, 0x34, 0x34, 0x08, 0x00, | |
| 183 0x00, 0x00, 0x00, 0x00, 0x00, 0x7A, 0x7A, 0x7A, 0x7A, 0x7A, 0x7A, 0x1A, | |
| 184 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 185 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x00, 0x00, 0x00, 0x00, 0x18, | |
| 186 0x00, 0x5A, 0x5A, 0x5A, 0x5A, 0x5A, 0x5A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 187 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 188 0x1A, 0x1A, 0x1A, 0x00, 0x00, 0x00, 0x00, 0x00, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 189 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 190 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 191 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 192 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 193 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 194 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 195 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 196 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 197 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 198 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, 0x1A, | |
| 199 0x1A, 0x1A, 0x01, 0x01, | |
| 200 }; | |
| 201 FX_BOOL g_FXCRT_XML_IsWhiteSpace(uint8_t ch) { | |
| 202 return (g_FXCRT_XML_ByteTypes[ch] & FXCRTM_XML_CHARTYPE_SpaceChar) != 0; | |
| 203 } | |
| 204 FX_BOOL g_FXCRT_XML_IsLetter(uint8_t ch) { | |
| 205 return (g_FXCRT_XML_ByteTypes[ch] & FXCRTM_XML_CHARTYPE_Letter) != 0; | |
| 206 } | |
| 207 FX_BOOL g_FXCRT_XML_IsDigital(uint8_t ch) { | |
| 208 return (g_FXCRT_XML_ByteTypes[ch] & FXCRTM_XML_CHARTYPE_Digital) != 0; | |
| 209 } | |
| 210 FX_BOOL g_FXCRT_XML_IsNameIntro(uint8_t ch) { | |
| 211 return (g_FXCRT_XML_ByteTypes[ch] & FXCRTM_XML_CHARTYPE_NameIntro) != 0; | |
| 212 } | |
| 213 FX_BOOL g_FXCRT_XML_IsNameChar(uint8_t ch) { | |
| 214 return (g_FXCRT_XML_ByteTypes[ch] & FXCRTM_XML_CHARTYPE_NameChar) != 0; | |
| 215 } | |
| 216 FX_BOOL g_FXCRT_XML_IsHexChar(uint8_t ch) { | |
| 217 return (g_FXCRT_XML_ByteTypes[ch] & FXCRTM_XML_CHARTYPE_HexChar) != 0; | |
| 218 } | |
| 219 void CXML_Parser::SkipWhiteSpaces() { | 226 void CXML_Parser::SkipWhiteSpaces() { |
| 220 m_nOffset = m_nBufferOffset + (FX_FILESIZE)m_dwIndex; | 227 m_nOffset = m_nBufferOffset + (FX_FILESIZE)m_dwIndex; |
| 221 if (IsEOF()) { | 228 if (IsEOF()) { |
| 222 return; | 229 return; |
| 223 } | 230 } |
| 224 do { | 231 do { |
| 225 while (m_dwIndex < m_dwBufferSize && | 232 while (m_dwIndex < m_dwBufferSize && |
| 226 g_FXCRT_XML_IsWhiteSpace(m_pBuffer[m_dwIndex])) { | 233 g_FXCRT_XML_IsWhiteSpace(m_pBuffer[m_dwIndex])) { |
| 227 m_dwIndex++; | 234 m_dwIndex++; |
| 228 } | 235 } |
| (...skipping 678 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
| 907 m_pMap->push_back({space, name, CFX_WideString(value)}); | 914 m_pMap->push_back({space, name, CFX_WideString(value)}); |
| 908 } | 915 } |
| 909 | 916 |
| 910 int CXML_AttrMap::GetSize() const { | 917 int CXML_AttrMap::GetSize() const { |
| 911 return m_pMap ? pdfium::CollectionSize<int>(*m_pMap) : 0; | 918 return m_pMap ? pdfium::CollectionSize<int>(*m_pMap) : 0; |
| 912 } | 919 } |
| 913 | 920 |
| 914 CXML_AttrItem& CXML_AttrMap::GetAt(int index) const { | 921 CXML_AttrItem& CXML_AttrMap::GetAt(int index) const { |
| 915 return (*m_pMap)[index]; | 922 return (*m_pMap)[index]; |
| 916 } | 923 } |
| OLD | NEW |