Split CPDF_SyntaxParser into its own named .cpp/.h files.

core/include/fpdfapi/fpdf_parser.h

 // Copyright 2014 PDFium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
 
 #ifndef CORE_INCLUDE_FPDFAPI_FPDF_PARSER_H_
 #define CORE_INCLUDE_FPDFAPI_FPDF_PARSER_H_
 
 #include <map>
@@ skipping 10 matching lines @@
 class CPDF_DocPageData;
 class CPDF_DocRenderData;
 class CPDF_Font;
 class CPDF_FontEncoding;
 class CPDF_IccProfile;
 class CPDF_Image;
 class CPDF_Object;
 class CPDF_Parser;
 class CPDF_Pattern;
 class CPDF_StandardSecurityHandler;
+class CPDF_SyntaxParser;

[dsinclair, 2016/03/08 01:15:05]

What's our thoughts on forward declarations? The style guide says to minimize
them (https://google.github.io/styleguide/cppguide.html#Forward_Declarations)
but we've got a lot of them  already.

[Tom Sepez, 2016/03/08 19:35:41]

In a header file, if the class is only used as a pointer or reference, then use
a forward declaration if it avoids including a larger header, whose details
aren't needed.  That's my take. YMMV.

 class IPDF_SecurityHandler;
 
 #define FPDFPERM_PRINT 0x0004
 #define FPDFPERM_MODIFY 0x0008
 #define FPDFPERM_EXTRACT 0x0010
 #define FPDFPERM_ANNOT_FORM 0x0020
 #define FPDFPERM_FILL_FORM 0x0100
 #define FPDFPERM_EXTRACT_ACCESS 0x0200
 #define FPDFPERM_ASSEMBLE 0x0400
 #define FPDFPERM_PRINT_HIGH 0x0800
@@ skipping 183 matching lines @@
   FX_DWORD GetCurPos() const { return m_dwCurPos; }
 
  private:
   void ParseWord(const uint8_t*& pStart, FX_DWORD& dwSize);
 
   const uint8_t* m_pData;
   FX_DWORD m_dwSize;
   FX_DWORD m_dwCurPos;
 };
 
-class CPDF_SyntaxParser {
- public:
-  CPDF_SyntaxParser();
-  ~CPDF_SyntaxParser();
-
-  void InitParser(IFX_FileRead* pFileAccess, FX_DWORD HeaderOffset);
-
-  FX_FILESIZE SavePos() const { return m_Pos; }
-  void RestorePos(FX_FILESIZE pos) { m_Pos = pos; }
-
-  CPDF_Object* GetObject(CPDF_IndirectObjectHolder* pObjList,
-                         FX_DWORD objnum,
-                         FX_DWORD gennum,
-                         FX_BOOL bDecrypt);
-  CPDF_Object* GetObjectByStrict(CPDF_IndirectObjectHolder* pObjList,
-                                 FX_DWORD objnum,
-                                 FX_DWORD gennum);
-  CFX_ByteString GetKeyword();
-
-  void ToNextLine();
-  void ToNextWord();
-
-  FX_BOOL SearchWord(const CFX_ByteStringC& word,
-                     FX_BOOL bWholeWord,
-                     FX_BOOL bForward,
-                     FX_FILESIZE limit);
-  int SearchMultiWord(const CFX_ByteStringC& words,
-                      FX_BOOL bWholeWord,
-                      FX_FILESIZE limit);
-  FX_FILESIZE FindTag(const CFX_ByteStringC& tag, FX_FILESIZE limit);
-
-  void SetEncrypt(std::unique_ptr<CPDF_CryptoHandler> pCryptoHandler);
-
-  FX_BOOL ReadBlock(uint8_t* pBuf, FX_DWORD size);
-  FX_BOOL GetCharAt(FX_FILESIZE pos, uint8_t& ch);
-  CFX_ByteString GetNextWord(bool* bIsNumber);
-
- private:
-  friend class CPDF_Parser;
-  friend class CPDF_DataAvail;
-  friend class fpdf_parser_parser_ReadHexString_Test;
-
-  static const int kParserMaxRecursionDepth = 64;
-  static int s_CurrentRecursionDepth;
-
-  uint32_t GetDirectNum();
-
-  FX_BOOL GetNextChar(uint8_t& ch);
-  FX_BOOL GetCharAtBackward(FX_FILESIZE pos, uint8_t& ch);
-  void GetNextWordInternal(bool* bIsNumber);
-  bool IsWholeWord(FX_FILESIZE startpos,
-                   FX_FILESIZE limit,
-                   const CFX_ByteStringC& tag,
-                   FX_BOOL checkKeyword);
-
-  CFX_ByteString ReadString();
-  CFX_ByteString ReadHexString();
-  unsigned int ReadEOLMarkers(FX_FILESIZE pos);
-  CPDF_Stream* ReadStream(CPDF_Dictionary* pDict,
-                          FX_DWORD objnum,
-                          FX_DWORD gennum);
-
-  FX_FILESIZE m_Pos;
-  int m_MetadataObjnum;
-  IFX_FileRead* m_pFileAccess;
-  FX_DWORD m_HeaderOffset;
-  FX_FILESIZE m_FileLen;
-  uint8_t* m_pFileBuf;
-  FX_DWORD m_BufSize;
-  FX_FILESIZE m_BufOffset;
-  std::unique_ptr<CPDF_CryptoHandler> m_pCryptoHandler;
-  uint8_t m_WordBuffer[257];
-  FX_DWORD m_WordSize;
-};
-
 class CPDF_Parser {
  public:
   enum Error {
     SUCCESS = 0,
     FILE_ERROR,
     FORMAT_ERROR,
     PASSWORD_ERROR,
     HANDLER_ERROR
   };
 
   CPDF_Parser();
   ~CPDF_Parser();
 
   Error StartParse(IFX_FileRead* pFile);
   FX_DWORD GetPermissions(FX_BOOL bCheckRevision = FALSE);
 
   void SetPassword(const FX_CHAR* password) { m_Password = password; }
   CFX_ByteString GetPassword() { return m_Password; }
-
-  CPDF_CryptoHandler* GetCryptoHandler() {
-    return m_Syntax.m_pCryptoHandler.get();
-  }
-
   CPDF_Dictionary* GetTrailer() const { return m_pTrailer; }
   FX_FILESIZE GetLastXRefOffset() const { return m_LastXRefOffset; }
   CPDF_Document* GetDocument() const { return m_pDocument; }
 
   FX_DWORD GetRootObjNum();
   FX_DWORD GetInfoObjNum();
   CPDF_Array* GetIDArray();
 
   CPDF_Dictionary* GetEncryptDict() const { return m_pEncryptDict; }
 
   CPDF_Object* ParseIndirectObject(CPDF_IndirectObjectHolder* pObjList,
                                    FX_DWORD objnum);
 
   FX_DWORD GetLastObjNum() const;
   bool IsValidObjectNumber(FX_DWORD objnum) const;
   FX_FILESIZE GetObjectPositionOrZero(FX_DWORD objnum) const;
   uint8_t GetObjectType(FX_DWORD objnum) const;
   uint16_t GetObjectGenNum(FX_DWORD objnum) const;
   bool IsVersionUpdated() const { return m_bVersionUpdated; }
   bool IsObjectFreeOrNull(FX_DWORD objnum) const;
   FX_BOOL IsFormStream(FX_DWORD objnum, FX_BOOL& bForm);
+  CPDF_CryptoHandler* GetCryptoHandler();
+  IFX_FileRead* GetFileAccess() const;
 
   FX_FILESIZE GetObjectOffset(FX_DWORD objnum) const;
   FX_FILESIZE GetObjectSize(FX_DWORD objnum) const;
 
   void GetIndirectBinary(FX_DWORD objnum, uint8_t*& pBuffer, FX_DWORD& size);
-  IFX_FileRead* GetFileAccess() const { return m_Syntax.m_pFileAccess; }
   int GetFileVersion() const { return m_FileVersion; }
   FX_BOOL IsXRefStream() const { return m_bXRefStream; }
 
   CPDF_Object* ParseIndirectObjectAt(CPDF_IndirectObjectHolder* pObjList,
                                      FX_FILESIZE pos,
                                      FX_DWORD objnum);
 
   CPDF_Object* ParseIndirectObjectAtByStrict(
       CPDF_IndirectObjectHolder* pObjList,
       FX_FILESIZE pos,
@@ skipping 26 matching lines @@
   FX_BOOL LoadLinearizedAllCrossRefV4(FX_FILESIZE pos, FX_DWORD dwObjCount);
   FX_BOOL LoadLinearizedCrossRefV4(FX_FILESIZE pos, FX_DWORD dwObjCount);
   FX_BOOL LoadLinearizedAllCrossRefV5(FX_FILESIZE pos);
   Error LoadLinearizedMainXRefTable();
   CPDF_StreamAcc* GetObjectStream(FX_DWORD number);
   FX_BOOL IsLinearizedFile(IFX_FileRead* pFileAccess, FX_DWORD offset);
   void SetEncryptDictionary(CPDF_Dictionary* pDict);
   void ShrinkObjectMap(FX_DWORD size);
 
   CPDF_Document* m_pDocument;
-  CPDF_SyntaxParser m_Syntax;
+  std::unique_ptr<CPDF_SyntaxParser> m_pSyntax;
   bool m_bOwnFileRead;
   int m_FileVersion;
   CPDF_Dictionary* m_pTrailer;
   CPDF_Dictionary* m_pEncryptDict;
   FX_FILESIZE m_LastXRefOffset;
   FX_BOOL m_bXRefStream;
   std::unique_ptr<IPDF_SecurityHandler> m_pSecurityHandler;
   CFX_ByteString m_bsRecipient;
   CFX_ByteString m_FilePath;
   CFX_ByteString m_Password;
@@ skipping 434 matching lines @@
                        FX_DWORD src_size,
                        const CPDF_Dictionary* pDict,
                        uint8_t*& dest_buf,
                        FX_DWORD& dest_size,
                        CFX_ByteString& ImageEncoding,
                        CPDF_Dictionary*& pImageParms,
                        FX_DWORD estimated_size,
                        FX_BOOL bImageAcc);
 
 #endif  // CORE_INCLUDE_FPDFAPI_FPDF_PARSER_H_