1 // Copyright 2014 PDFium 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 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com 6 7 #ifndef CORE_SRC_FXCRT_XML_INT_H_ 8 #define CORE_SRC_FXCRT_XML_INT_H_ 9 10 #include <algorithm> 11 12 #include "core/include/fxcrt/fx_stream.h" 13 14 class CFX_UTF8Decoder; 15 class CXML_Element; 16 17 class CXML_DataBufAcc : public IFX_BufferRead { 18 public: CXML_DataBufAcc(const uint8_t * pBuffer,size_t size)19 CXML_DataBufAcc(const uint8_t* pBuffer, size_t size) 20 : m_pBuffer(pBuffer), m_dwSize(size), m_dwCurPos(0) {} ~CXML_DataBufAcc()21 ~CXML_DataBufAcc() override {} 22 23 // IFX_BufferRead Release()24 void Release() override { delete this; } IsEOF()25 FX_BOOL IsEOF() override { return m_dwCurPos >= m_dwSize; } GetPosition()26 FX_FILESIZE GetPosition() override { return (FX_FILESIZE)m_dwCurPos; } ReadBlock(void * buffer,size_t size)27 size_t ReadBlock(void* buffer, size_t size) override { return 0; } 28 FX_BOOL ReadNextBlock(FX_BOOL bRestart = FALSE) override { 29 if (bRestart) { 30 m_dwCurPos = 0; 31 } 32 if (m_dwCurPos < m_dwSize) { 33 m_dwCurPos = m_dwSize; 34 return TRUE; 35 } 36 return FALSE; 37 } GetBlockBuffer()38 const uint8_t* GetBlockBuffer() override { return m_pBuffer; } GetBlockSize()39 size_t GetBlockSize() override { return m_dwSize; } GetBlockOffset()40 FX_FILESIZE GetBlockOffset() override { return 0; } 41 42 protected: 43 const uint8_t* m_pBuffer; 44 size_t m_dwSize; 45 size_t m_dwCurPos; 46 }; 47 48 class CXML_DataStmAcc : public IFX_BufferRead { 49 public: CXML_DataStmAcc(IFX_FileRead * pFileRead)50 explicit CXML_DataStmAcc(IFX_FileRead* pFileRead) 51 : m_pFileRead(pFileRead), m_pBuffer(NULL), m_nStart(0), m_dwSize(0) { 52 FXSYS_assert(m_pFileRead); 53 } ~CXML_DataStmAcc()54 ~CXML_DataStmAcc() override { FX_Free(m_pBuffer); } 55 Release()56 void Release() override { delete this; } IsEOF()57 FX_BOOL IsEOF() override { 58 return m_nStart + (FX_FILESIZE)m_dwSize >= m_pFileRead->GetSize(); 59 } GetPosition()60 FX_FILESIZE GetPosition() override { 61 return m_nStart + (FX_FILESIZE)m_dwSize; 62 } ReadBlock(void * buffer,size_t size)63 size_t ReadBlock(void* buffer, size_t size) override { return 0; } 64 FX_BOOL ReadNextBlock(FX_BOOL bRestart = FALSE) override { 65 if (bRestart) { 66 m_nStart = 0; 67 } 68 FX_FILESIZE nLength = m_pFileRead->GetSize(); 69 m_nStart += (FX_FILESIZE)m_dwSize; 70 if (m_nStart >= nLength) { 71 return FALSE; 72 } 73 static const FX_FILESIZE FX_XMLDATASTREAM_BufferSize = 32 * 1024; 74 m_dwSize = static_cast<size_t>( 75 std::min(FX_XMLDATASTREAM_BufferSize, nLength - m_nStart)); 76 if (!m_pBuffer) { 77 m_pBuffer = FX_Alloc(uint8_t, m_dwSize); 78 } 79 return m_pFileRead->ReadBlock(m_pBuffer, m_nStart, m_dwSize); 80 } GetBlockBuffer()81 const uint8_t* GetBlockBuffer() override { return (const uint8_t*)m_pBuffer; } GetBlockSize()82 size_t GetBlockSize() override { return m_dwSize; } GetBlockOffset()83 FX_FILESIZE GetBlockOffset() override { return m_nStart; } 84 85 protected: 86 IFX_FileRead* m_pFileRead; 87 uint8_t* m_pBuffer; 88 FX_FILESIZE m_nStart; 89 size_t m_dwSize; 90 }; 91 92 class CXML_Parser { 93 public: 94 ~CXML_Parser(); 95 IFX_BufferRead* m_pDataAcc; 96 FX_BOOL m_bOwnedStream; 97 FX_FILESIZE m_nOffset; 98 FX_BOOL m_bSaveSpaceChars; 99 const uint8_t* m_pBuffer; 100 size_t m_dwBufferSize; 101 FX_FILESIZE m_nBufferOffset; 102 size_t m_dwIndex; 103 FX_BOOL Init(uint8_t* pBuffer, size_t size); 104 FX_BOOL Init(IFX_FileRead* pFileRead); 105 FX_BOOL Init(IFX_BufferRead* pBuffer); 106 FX_BOOL Init(FX_BOOL bOwndedStream); 107 FX_BOOL ReadNextBlock(); 108 FX_BOOL IsEOF(); 109 FX_BOOL HaveAvailData(); 110 void SkipWhiteSpaces(); 111 void GetName(CFX_ByteString& space, CFX_ByteString& name); 112 void GetAttrValue(CFX_WideString& value); 113 FX_DWORD GetCharRef(); 114 void GetTagName(CFX_ByteString& space, 115 CFX_ByteString& name, 116 FX_BOOL& bEndTag, 117 FX_BOOL bStartTag = FALSE); 118 void SkipLiterals(const CFX_ByteStringC& str); 119 CXML_Element* ParseElement(CXML_Element* pParent, FX_BOOL bStartTag = FALSE); 120 void InsertContentSegment(FX_BOOL bCDATA, 121 const CFX_WideStringC& content, 122 CXML_Element* pElement); 123 void InsertCDATASegment(CFX_UTF8Decoder& decoder, CXML_Element* pElement); 124 }; 125 126 void FX_XML_SplitQualifiedName(const CFX_ByteStringC& bsFullName, 127 CFX_ByteStringC& bsSpace, 128 CFX_ByteStringC& bsName); 129 130 #endif // CORE_SRC_FXCRT_XML_INT_H_ 131