1 // Copyright 2016 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_FPDFAPI_PARSER_CPDF_SYNTAX_PARSER_H_
8 #define CORE_FPDFAPI_PARSER_CPDF_SYNTAX_PARSER_H_
9 
10 #include <memory>
11 
12 #include "core/fxcrt/cfx_string_pool_template.h"
13 #include "core/fxcrt/cfx_weak_ptr.h"
14 #include "core/fxcrt/fx_basic.h"
15 
16 class CPDF_CryptoHandler;
17 class CPDF_Dictionary;
18 class CPDF_IndirectObjectHolder;
19 class CPDF_Object;
20 class CPDF_Stream;
21 class IFX_SeekableReadStream;
22 
23 class CPDF_SyntaxParser {
24  public:
25   CPDF_SyntaxParser();
26   explicit CPDF_SyntaxParser(const CFX_WeakPtr<CFX_ByteStringPool>& pPool);
27   ~CPDF_SyntaxParser();
28 
29   void InitParser(const CFX_RetainPtr<IFX_SeekableReadStream>& pFileAccess,
30                   uint32_t HeaderOffset);
31 
SavePos()32   FX_FILESIZE SavePos() const { return m_Pos; }
RestorePos(FX_FILESIZE pos)33   void RestorePos(FX_FILESIZE pos) { m_Pos = pos; }
34 
35   std::unique_ptr<CPDF_Object> GetObject(CPDF_IndirectObjectHolder* pObjList,
36                                          uint32_t objnum,
37                                          uint32_t gennum,
38                                          bool bDecrypt);
39 
40   std::unique_ptr<CPDF_Object> GetObjectForStrict(
41       CPDF_IndirectObjectHolder* pObjList,
42       uint32_t objnum,
43       uint32_t gennum);
44 
45   CFX_ByteString GetKeyword();
46   void ToNextLine();
47   void ToNextWord();
48   bool SearchWord(const CFX_ByteStringC& word,
49                   bool bWholeWord,
50                   bool bForward,
51                   FX_FILESIZE limit);
52 
53   FX_FILESIZE FindTag(const CFX_ByteStringC& tag, FX_FILESIZE limit);
54   void SetEncrypt(std::unique_ptr<CPDF_CryptoHandler> pCryptoHandler);
55   bool ReadBlock(uint8_t* pBuf, uint32_t size);
56   bool GetCharAt(FX_FILESIZE pos, uint8_t& ch);
57   CFX_ByteString GetNextWord(bool* bIsNumber);
58 
59  private:
60   friend class CPDF_Parser;
61   friend class CPDF_DataAvail;
62   friend class cpdf_syntax_parser_ReadHexString_Test;
63 
64   static const int kParserMaxRecursionDepth = 64;
65   static int s_CurrentRecursionDepth;
66 
67   uint32_t GetDirectNum();
68   bool ReadChar(FX_FILESIZE read_pos, uint32_t read_size);
69   bool GetNextChar(uint8_t& ch);
70   bool GetCharAtBackward(FX_FILESIZE pos, uint8_t& ch);
71   void GetNextWordInternal(bool* bIsNumber);
72   bool IsWholeWord(FX_FILESIZE startpos,
73                    FX_FILESIZE limit,
74                    const CFX_ByteStringC& tag,
75                    bool checkKeyword);
76 
77   CFX_ByteString ReadString();
78   CFX_ByteString ReadHexString();
79   unsigned int ReadEOLMarkers(FX_FILESIZE pos);
80   std::unique_ptr<CPDF_Stream> ReadStream(
81       std::unique_ptr<CPDF_Dictionary> pDict,
82       uint32_t objnum,
83       uint32_t gennum);
84 
CheckPosition(FX_FILESIZE pos)85   inline bool CheckPosition(FX_FILESIZE pos) {
86     return m_BufOffset >= pos ||
87            static_cast<FX_FILESIZE>(m_BufOffset + m_BufSize) <= pos;
88   }
89 
90   FX_FILESIZE m_Pos;
91   uint32_t m_MetadataObjnum;
92   CFX_RetainPtr<IFX_SeekableReadStream> m_pFileAccess;
93   FX_FILESIZE m_HeaderOffset;
94   FX_FILESIZE m_FileLen;
95   uint8_t* m_pFileBuf;
96   uint32_t m_BufSize;
97   FX_FILESIZE m_BufOffset;
98   std::unique_ptr<CPDF_CryptoHandler> m_pCryptoHandler;
99   uint8_t m_WordBuffer[257];
100   uint32_t m_WordSize;
101   CFX_WeakPtr<CFX_ByteStringPool> m_pPool;
102 };
103 
104 #endif  // CORE_FPDFAPI_PARSER_CPDF_SYNTAX_PARSER_H_
105