OLD | NEW |
| (Empty) |
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_INCLUDE_FPDFAPI_CPDF_PARSER_H_ | |
8 #define CORE_INCLUDE_FPDFAPI_CPDF_PARSER_H_ | |
9 | |
10 #include <map> | |
11 #include <memory> | |
12 #include <set> | |
13 | |
14 #include "core/include/fxcrt/fx_basic.h" | |
15 | |
16 class CPDF_Array; | |
17 class CPDF_Dictionary; | |
18 class CPDF_Document; | |
19 class CPDF_IndirectObjectHolder; | |
20 class CPDF_Object; | |
21 class CPDF_StreamAcc; | |
22 class CPDF_SyntaxParser; | |
23 class IFX_FileRead; | |
24 class IPDF_CryptoHandler; | |
25 class IPDF_SecurityHandler; | |
26 | |
27 class CPDF_Parser { | |
28 public: | |
29 enum Error { | |
30 SUCCESS = 0, | |
31 FILE_ERROR, | |
32 FORMAT_ERROR, | |
33 PASSWORD_ERROR, | |
34 HANDLER_ERROR | |
35 }; | |
36 | |
37 CPDF_Parser(); | |
38 ~CPDF_Parser(); | |
39 | |
40 Error StartParse(IFX_FileRead* pFile); | |
41 FX_DWORD GetPermissions(FX_BOOL bCheckRevision = FALSE); | |
42 | |
43 void SetPassword(const FX_CHAR* password) { m_Password = password; } | |
44 CFX_ByteString GetPassword() { return m_Password; } | |
45 CPDF_Dictionary* GetTrailer() const { return m_pTrailer; } | |
46 FX_FILESIZE GetLastXRefOffset() const { return m_LastXRefOffset; } | |
47 CPDF_Document* GetDocument() const { return m_pDocument; } | |
48 | |
49 FX_DWORD GetRootObjNum(); | |
50 FX_DWORD GetInfoObjNum(); | |
51 CPDF_Array* GetIDArray(); | |
52 | |
53 CPDF_Dictionary* GetEncryptDict() const { return m_pEncryptDict; } | |
54 | |
55 CPDF_Object* ParseIndirectObject(CPDF_IndirectObjectHolder* pObjList, | |
56 FX_DWORD objnum); | |
57 | |
58 FX_DWORD GetLastObjNum() const; | |
59 bool IsValidObjectNumber(FX_DWORD objnum) const; | |
60 FX_FILESIZE GetObjectPositionOrZero(FX_DWORD objnum) const; | |
61 uint8_t GetObjectType(FX_DWORD objnum) const; | |
62 uint16_t GetObjectGenNum(FX_DWORD objnum) const; | |
63 bool IsVersionUpdated() const { return m_bVersionUpdated; } | |
64 bool IsObjectFreeOrNull(FX_DWORD objnum) const; | |
65 FX_BOOL IsFormStream(FX_DWORD objnum, FX_BOOL& bForm); | |
66 IPDF_CryptoHandler* GetCryptoHandler(); | |
67 IFX_FileRead* GetFileAccess() const; | |
68 | |
69 FX_FILESIZE GetObjectOffset(FX_DWORD objnum) const; | |
70 FX_FILESIZE GetObjectSize(FX_DWORD objnum) const; | |
71 | |
72 void GetIndirectBinary(FX_DWORD objnum, uint8_t*& pBuffer, FX_DWORD& size); | |
73 int GetFileVersion() const { return m_FileVersion; } | |
74 FX_BOOL IsXRefStream() const { return m_bXRefStream; } | |
75 | |
76 CPDF_Object* ParseIndirectObjectAt(CPDF_IndirectObjectHolder* pObjList, | |
77 FX_FILESIZE pos, | |
78 FX_DWORD objnum); | |
79 | |
80 CPDF_Object* ParseIndirectObjectAtByStrict( | |
81 CPDF_IndirectObjectHolder* pObjList, | |
82 FX_FILESIZE pos, | |
83 FX_DWORD objnum, | |
84 FX_FILESIZE* pResultPos); | |
85 | |
86 Error StartAsyncParse(IFX_FileRead* pFile); | |
87 | |
88 FX_DWORD GetFirstPageNo() const { return m_dwFirstPageNo; } | |
89 | |
90 protected: | |
91 struct ObjectInfo { | |
92 ObjectInfo() : pos(0), type(0), gennum(0) {} | |
93 | |
94 FX_FILESIZE pos; | |
95 uint8_t type; | |
96 uint16_t gennum; | |
97 }; | |
98 | |
99 void CloseParser(); | |
100 CPDF_Object* ParseDirect(CPDF_Object* pObj); | |
101 FX_BOOL LoadAllCrossRefV4(FX_FILESIZE pos); | |
102 FX_BOOL LoadAllCrossRefV5(FX_FILESIZE pos); | |
103 bool LoadCrossRefV4(FX_FILESIZE pos, FX_FILESIZE streampos, FX_BOOL bSkip); | |
104 FX_BOOL LoadCrossRefV5(FX_FILESIZE* pos, FX_BOOL bMainXRef); | |
105 CPDF_Dictionary* LoadTrailerV4(); | |
106 FX_BOOL RebuildCrossRef(); | |
107 Error SetEncryptHandler(); | |
108 void ReleaseEncryptHandler(); | |
109 FX_BOOL LoadLinearizedAllCrossRefV4(FX_FILESIZE pos, FX_DWORD dwObjCount); | |
110 FX_BOOL LoadLinearizedCrossRefV4(FX_FILESIZE pos, FX_DWORD dwObjCount); | |
111 FX_BOOL LoadLinearizedAllCrossRefV5(FX_FILESIZE pos); | |
112 Error LoadLinearizedMainXRefTable(); | |
113 CPDF_StreamAcc* GetObjectStream(FX_DWORD number); | |
114 FX_BOOL IsLinearizedFile(IFX_FileRead* pFileAccess, FX_DWORD offset); | |
115 void SetEncryptDictionary(CPDF_Dictionary* pDict); | |
116 void ShrinkObjectMap(FX_DWORD size); | |
117 | |
118 CPDF_Document* m_pDocument; | |
119 std::unique_ptr<CPDF_SyntaxParser> m_pSyntax; | |
120 bool m_bOwnFileRead; | |
121 int m_FileVersion; | |
122 CPDF_Dictionary* m_pTrailer; | |
123 CPDF_Dictionary* m_pEncryptDict; | |
124 FX_FILESIZE m_LastXRefOffset; | |
125 FX_BOOL m_bXRefStream; | |
126 std::unique_ptr<IPDF_SecurityHandler> m_pSecurityHandler; | |
127 CFX_ByteString m_bsRecipient; | |
128 CFX_ByteString m_FilePath; | |
129 CFX_ByteString m_Password; | |
130 std::map<FX_DWORD, ObjectInfo> m_ObjectInfo; | |
131 std::set<FX_FILESIZE> m_SortedOffset; | |
132 CFX_ArrayTemplate<CPDF_Dictionary*> m_Trailers; | |
133 bool m_bVersionUpdated; | |
134 CPDF_Object* m_pLinearized; | |
135 FX_DWORD m_dwFirstPageNo; | |
136 FX_DWORD m_dwXrefStartObjNum; | |
137 | |
138 // A map of object numbers to indirect streams. Map owns the streams. | |
139 std::map<FX_DWORD, std::unique_ptr<CPDF_StreamAcc>> m_ObjectStreamMap; | |
140 | |
141 // Mapping of object numbers to offsets. The offsets are relative to the first | |
142 // object in the stream. | |
143 using StreamObjectCache = std::map<FX_DWORD, FX_DWORD>; | |
144 | |
145 // Mapping of streams to their object caches. This is valid as long as the | |
146 // streams in |m_ObjectStreamMap| are valid. | |
147 std::map<CPDF_StreamAcc*, StreamObjectCache> m_ObjCache; | |
148 | |
149 // All indirect object numbers that are being parsed. | |
150 std::set<FX_DWORD> m_ParsingObjNums; | |
151 | |
152 friend class CPDF_DataAvail; | |
153 | |
154 private: | |
155 enum class ParserState { | |
156 kDefault, | |
157 kComment, | |
158 kWhitespace, | |
159 kString, | |
160 kHexString, | |
161 kEscapedString, | |
162 kXref, | |
163 kObjNum, | |
164 kPostObjNum, | |
165 kGenNum, | |
166 kPostGenNum, | |
167 kTrailer, | |
168 kBeginObj, | |
169 kEndObj | |
170 }; | |
171 }; | |
172 | |
173 #endif // CORE_INCLUDE_FPDFAPI_CPDF_PARSER_H_ | |
OLD | NEW |