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_FPDFAPI_FPDF_PARSER_INCLUDE_CPDF_DATA_AVAIL_H_ | |
8 #define CORE_FPDFAPI_FPDF_PARSER_INCLUDE_CPDF_DATA_AVAIL_H_ | |
9 | |
10 #include "core/fpdfapi/fpdf_parser/cpdf_syntax_parser.h" | |
11 #include "core/fpdfapi/fpdf_parser/include/cpdf_parser.h" | |
12 #include "core/fxcrt/include/fx_basic.h" | |
13 | |
14 class CPDF_Dictionary; | |
15 class CPDF_HintTables; | |
16 class CPDF_IndirectObjectHolder; | |
17 class CPDF_Parser; | |
18 | |
19 enum PDF_DATAAVAIL_STATUS { | |
20 PDF_DATAAVAIL_HEADER = 0, | |
21 PDF_DATAAVAIL_FIRSTPAGE, | |
22 PDF_DATAAVAIL_FIRSTPAGE_PREPARE, | |
23 PDF_DATAAVAIL_HINTTABLE, | |
24 PDF_DATAAVAIL_END, | |
25 PDF_DATAAVAIL_CROSSREF, | |
26 PDF_DATAAVAIL_CROSSREF_ITEM, | |
27 PDF_DATAAVAIL_CROSSREF_STREAM, | |
28 PDF_DATAAVAIL_TRAILER, | |
29 PDF_DATAAVAIL_LOADALLCROSSREF, | |
30 PDF_DATAAVAIL_ROOT, | |
31 PDF_DATAAVAIL_INFO, | |
32 PDF_DATAAVAIL_ACROFORM, | |
33 PDF_DATAAVAIL_ACROFORM_SUBOBJECT, | |
34 PDF_DATAAVAIL_PAGETREE, | |
35 PDF_DATAAVAIL_PAGE, | |
36 PDF_DATAAVAIL_PAGE_LATERLOAD, | |
37 PDF_DATAAVAIL_RESOURCES, | |
38 PDF_DATAAVAIL_DONE, | |
39 PDF_DATAAVAIL_ERROR, | |
40 PDF_DATAAVAIL_LOADALLFILE, | |
41 PDF_DATAAVAIL_TRAILER_APPEND | |
42 }; | |
43 | |
44 enum PDF_PAGENODE_TYPE { | |
45 PDF_PAGENODE_UNKNOWN = 0, | |
46 PDF_PAGENODE_PAGE, | |
47 PDF_PAGENODE_PAGES, | |
48 PDF_PAGENODE_ARRAY, | |
49 }; | |
50 | |
51 class CPDF_DataAvail final { | |
52 public: | |
53 // Must match PDF_DATA_* definitions in public/fpdf_dataavail.h, but cannot | |
54 // #include that header. fpdfsdk/fpdf_dataavail.cpp has static_asserts | |
55 // to make sure the two sets of values match. | |
56 enum DocAvailStatus { | |
57 DataError = -1, // PDF_DATA_ERROR | |
58 DataNotAvailable = 0, // PDF_DATA_NOTAVAIL | |
59 DataAvailable = 1, // PDF_DATA_AVAIL | |
60 }; | |
61 | |
62 // Must match PDF_*LINEAR* definitions in public/fpdf_dataavail.h, but cannot | |
63 // #include that header. fpdfsdk/fpdf_dataavail.cpp has static_asserts | |
64 // to make sure the two sets of values match. | |
65 enum DocLinearizationStatus { | |
66 LinearizationUnknown = -1, // PDF_LINEARIZATION_UNKNOWN | |
67 NotLinearized = 0, // PDF_NOT_LINEARIZED | |
68 Linearized = 1, // PDF_LINEARIZED | |
69 }; | |
70 | |
71 // Must match PDF_FORM_* definitions in public/fpdf_dataavail.h, but cannot | |
72 // #include that header. fpdfsdk/fpdf_dataavail.cpp has static_asserts | |
73 // to make sure the two sets of values match. | |
74 enum DocFormStatus { | |
75 FormError = -1, // PDF_FORM_ERROR | |
76 FormNotAvailable = 0, // PDF_FORM_NOTAVAIL | |
77 FormAvailable = 1, // PDF_FORM_AVAIL | |
78 FormNotExist = 2, // PDF_FORM_NOTEXIST | |
79 }; | |
80 | |
81 class FileAvail { | |
82 public: | |
83 virtual ~FileAvail(); | |
84 virtual FX_BOOL IsDataAvail(FX_FILESIZE offset, uint32_t size) = 0; | |
85 }; | |
86 | |
87 class DownloadHints { | |
88 public: | |
89 virtual ~DownloadHints(); | |
90 virtual void AddSegment(FX_FILESIZE offset, uint32_t size) = 0; | |
91 }; | |
92 | |
93 CPDF_DataAvail(FileAvail* pFileAvail, | |
94 IFX_FileRead* pFileRead, | |
95 FX_BOOL bSupportHintTable); | |
96 ~CPDF_DataAvail(); | |
97 | |
98 DocAvailStatus IsDocAvail(DownloadHints* pHints); | |
99 void SetDocument(CPDF_Document* pDoc); | |
100 DocAvailStatus IsPageAvail(uint32_t dwPage, DownloadHints* pHints); | |
101 DocFormStatus IsFormAvail(DownloadHints* pHints); | |
102 DocLinearizationStatus IsLinearizedPDF(); | |
103 FX_BOOL IsLinearized(); | |
104 void GetLinearizedMainXRefInfo(FX_FILESIZE* pPos, uint32_t* pSize); | |
105 IFX_FileRead* GetFileRead() const { return m_pFileRead; } | |
106 int GetPageCount() const; | |
107 CPDF_Dictionary* GetPage(int index); | |
108 | |
109 friend class CPDF_HintTables; | |
110 | |
111 protected: | |
112 class PageNode { | |
113 public: | |
114 PageNode(); | |
115 ~PageNode(); | |
116 | |
117 PDF_PAGENODE_TYPE m_type; | |
118 uint32_t m_dwPageNo; | |
119 CFX_ArrayTemplate<PageNode*> m_childNode; | |
120 }; | |
121 | |
122 static const int kMaxDataAvailRecursionDepth = 64; | |
123 static int s_CurrentDataAvailRecursionDepth; | |
124 static const int kMaxPageRecursionDepth = 1024; | |
125 | |
126 uint32_t GetObjectSize(uint32_t objnum, FX_FILESIZE& offset); | |
127 FX_BOOL IsObjectsAvail(CFX_ArrayTemplate<CPDF_Object*>& obj_array, | |
128 FX_BOOL bParsePage, | |
129 DownloadHints* pHints, | |
130 CFX_ArrayTemplate<CPDF_Object*>& ret_array); | |
131 FX_BOOL CheckDocStatus(DownloadHints* pHints); | |
132 FX_BOOL CheckHeader(DownloadHints* pHints); | |
133 FX_BOOL CheckFirstPage(DownloadHints* pHints); | |
134 FX_BOOL CheckHintTables(DownloadHints* pHints); | |
135 FX_BOOL CheckEnd(DownloadHints* pHints); | |
136 FX_BOOL CheckCrossRef(DownloadHints* pHints); | |
137 FX_BOOL CheckCrossRefItem(DownloadHints* pHints); | |
138 FX_BOOL CheckTrailer(DownloadHints* pHints); | |
139 FX_BOOL CheckRoot(DownloadHints* pHints); | |
140 FX_BOOL CheckInfo(DownloadHints* pHints); | |
141 FX_BOOL CheckPages(DownloadHints* pHints); | |
142 FX_BOOL CheckPage(DownloadHints* pHints); | |
143 FX_BOOL CheckResources(DownloadHints* pHints); | |
144 FX_BOOL CheckAnnots(DownloadHints* pHints); | |
145 FX_BOOL CheckAcroForm(DownloadHints* pHints); | |
146 FX_BOOL CheckAcroFormSubObject(DownloadHints* pHints); | |
147 FX_BOOL CheckTrailerAppend(DownloadHints* pHints); | |
148 FX_BOOL CheckPageStatus(DownloadHints* pHints); | |
149 FX_BOOL CheckAllCrossRefStream(DownloadHints* pHints); | |
150 | |
151 int32_t CheckCrossRefStream(DownloadHints* pHints, FX_FILESIZE& xref_offset); | |
152 FX_BOOL IsLinearizedFile(uint8_t* pData, uint32_t dwLen); | |
153 void SetStartOffset(FX_FILESIZE dwOffset); | |
154 FX_BOOL GetNextToken(CFX_ByteString& token); | |
155 FX_BOOL GetNextChar(uint8_t& ch); | |
156 CPDF_Object* ParseIndirectObjectAt( | |
157 FX_FILESIZE pos, | |
158 uint32_t objnum, | |
159 CPDF_IndirectObjectHolder* pObjList = nullptr); | |
160 CPDF_Object* GetObject(uint32_t objnum, | |
161 DownloadHints* pHints, | |
162 FX_BOOL* pExistInFile); | |
163 FX_BOOL GetPageKids(CPDF_Parser* pParser, CPDF_Object* pPages); | |
164 FX_BOOL PreparePageItem(); | |
165 FX_BOOL LoadPages(DownloadHints* pHints); | |
166 FX_BOOL LoadAllXref(DownloadHints* pHints); | |
167 FX_BOOL LoadAllFile(DownloadHints* pHints); | |
168 DocAvailStatus CheckLinearizedData(DownloadHints* pHints); | |
169 FX_BOOL CheckPageAnnots(uint32_t dwPage, DownloadHints* pHints); | |
170 | |
171 DocAvailStatus CheckLinearizedFirstPage(uint32_t dwPage, | |
172 DownloadHints* pHints); | |
173 FX_BOOL HaveResourceAncestor(CPDF_Dictionary* pDict); | |
174 FX_BOOL CheckPage(uint32_t dwPage, DownloadHints* pHints); | |
175 FX_BOOL LoadDocPages(DownloadHints* pHints); | |
176 FX_BOOL LoadDocPage(uint32_t dwPage, DownloadHints* pHints); | |
177 FX_BOOL CheckPageNode(PageNode& pageNodes, | |
178 int32_t iPage, | |
179 int32_t& iCount, | |
180 DownloadHints* pHints, | |
181 int level); | |
182 FX_BOOL CheckUnkownPageNode(uint32_t dwPageNo, | |
183 PageNode* pPageNode, | |
184 DownloadHints* pHints); | |
185 FX_BOOL CheckArrayPageNode(uint32_t dwPageNo, | |
186 PageNode* pPageNode, | |
187 DownloadHints* pHints); | |
188 FX_BOOL CheckPageCount(DownloadHints* pHints); | |
189 bool IsFirstCheck(uint32_t dwPage); | |
190 void ResetFirstCheck(uint32_t dwPage); | |
191 FX_BOOL IsDataAvail(FX_FILESIZE offset, uint32_t size, DownloadHints* pHints); | |
192 | |
193 FileAvail* const m_pFileAvail; | |
194 IFX_FileRead* const m_pFileRead; | |
195 CPDF_Parser m_parser; | |
196 CPDF_SyntaxParser m_syntaxParser; | |
197 CPDF_Object* m_pRoot; | |
198 uint32_t m_dwRootObjNum; | |
199 uint32_t m_dwInfoObjNum; | |
200 CPDF_Object* m_pLinearized; | |
201 CPDF_Object* m_pTrailer; | |
202 FX_BOOL m_bDocAvail; | |
203 FX_FILESIZE m_dwHeaderOffset; | |
204 FX_FILESIZE m_dwLastXRefOffset; | |
205 FX_FILESIZE m_dwXRefOffset; | |
206 FX_FILESIZE m_dwTrailerOffset; | |
207 FX_FILESIZE m_dwCurrentOffset; | |
208 PDF_DATAAVAIL_STATUS m_docStatus; | |
209 FX_FILESIZE m_dwFileLen; | |
210 CPDF_Document* m_pDocument; | |
211 std::set<uint32_t> m_ObjectSet; | |
212 CFX_ArrayTemplate<CPDF_Object*> m_objs_array; | |
213 FX_FILESIZE m_Pos; | |
214 FX_FILESIZE m_bufferOffset; | |
215 uint32_t m_bufferSize; | |
216 CFX_ByteString m_WordBuf; | |
217 uint8_t m_bufferData[512]; | |
218 CFX_ArrayTemplate<uint32_t> m_XRefStreamList; | |
219 CFX_ArrayTemplate<uint32_t> m_PageObjList; | |
220 uint32_t m_PagesObjNum; | |
221 FX_BOOL m_bLinearized; | |
222 uint32_t m_dwFirstPageNo; | |
223 FX_BOOL m_bLinearedDataOK; | |
224 FX_BOOL m_bMainXRefLoadTried; | |
225 FX_BOOL m_bMainXRefLoadedOK; | |
226 FX_BOOL m_bPagesTreeLoad; | |
227 FX_BOOL m_bPagesLoad; | |
228 CPDF_Parser* m_pCurrentParser; | |
229 FX_FILESIZE m_dwCurrentXRefSteam; | |
230 FX_BOOL m_bAnnotsLoad; | |
231 FX_BOOL m_bHaveAcroForm; | |
232 uint32_t m_dwAcroFormObjNum; | |
233 FX_BOOL m_bAcroFormLoad; | |
234 CPDF_Object* m_pAcroForm; | |
235 CFX_ArrayTemplate<CPDF_Object*> m_arrayAcroforms; | |
236 CPDF_Dictionary* m_pPageDict; | |
237 CPDF_Object* m_pPageResource; | |
238 FX_BOOL m_bNeedDownLoadResource; | |
239 FX_BOOL m_bPageLoadedOK; | |
240 FX_BOOL m_bLinearizedFormParamLoad; | |
241 CFX_ArrayTemplate<CPDF_Object*> m_PagesArray; | |
242 uint32_t m_dwEncryptObjNum; | |
243 FX_FILESIZE m_dwPrevXRefOffset; | |
244 FX_BOOL m_bTotalLoadPageTree; | |
245 FX_BOOL m_bCurPageDictLoadOK; | |
246 PageNode m_pageNodes; | |
247 std::set<uint32_t> m_pageMapCheckState; | |
248 std::set<uint32_t> m_pagesLoadState; | |
249 std::unique_ptr<CPDF_HintTables> m_pHintTables; | |
250 FX_BOOL m_bSupportHintTable; | |
251 }; | |
252 | |
253 #endif // CORE_FPDFAPI_FPDF_PARSER_INCLUDE_CPDF_DATA_AVAIL_H_ | |
OLD | NEW |