Home | History | Annotate | Download | only in parser
      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_DOCUMENT_H_
      8 #define CORE_FPDFAPI_PARSER_CPDF_DOCUMENT_H_
      9 
     10 #include <functional>
     11 #include <memory>
     12 #include <set>
     13 #include <utility>
     14 #include <vector>
     15 
     16 #include "core/fpdfapi/parser/cpdf_indirect_object_holder.h"
     17 #include "core/fpdfapi/parser/cpdf_object.h"
     18 #include "core/fpdfdoc/cpdf_linklist.h"
     19 #include "core/fxcrt/fx_basic.h"
     20 
     21 class CFX_Font;
     22 class CFX_Matrix;
     23 class CPDF_ColorSpace;
     24 class CPDF_DocPageData;
     25 class CPDF_DocRenderData;
     26 class CPDF_Font;
     27 class CPDF_FontEncoding;
     28 class CPDF_IccProfile;
     29 class CPDF_Image;
     30 class CPDF_LinearizedHeader;
     31 class CPDF_Parser;
     32 class CPDF_Pattern;
     33 class CPDF_StreamAcc;
     34 class JBig2_DocumentContext;
     35 
     36 #define FPDFPERM_PRINT 0x0004
     37 #define FPDFPERM_MODIFY 0x0008
     38 #define FPDFPERM_EXTRACT 0x0010
     39 #define FPDFPERM_ANNOT_FORM 0x0020
     40 #define FPDFPERM_FILL_FORM 0x0100
     41 #define FPDFPERM_EXTRACT_ACCESS 0x0200
     42 #define FPDFPERM_ASSEMBLE 0x0400
     43 #define FPDFPERM_PRINT_HIGH 0x0800
     44 #define FPDF_PAGE_MAX_NUM 0xFFFFF
     45 
     46 class CPDF_Document : public CPDF_IndirectObjectHolder {
     47  public:
     48   explicit CPDF_Document(std::unique_ptr<CPDF_Parser> pParser);
     49   ~CPDF_Document() override;
     50 
     51   CPDF_Parser* GetParser() const { return m_pParser.get(); }
     52   CPDF_Dictionary* GetRoot() const { return m_pRootDict; }
     53   CPDF_Dictionary* GetInfo() const { return m_pInfoDict; }
     54 
     55   void DeletePage(int iPage);
     56   int GetPageCount() const;
     57   bool IsPageLoaded(int iPage) const;
     58   CPDF_Dictionary* GetPage(int iPage);
     59   int GetPageIndex(uint32_t objnum);
     60   uint32_t GetUserPermissions() const;
     61   CPDF_DocPageData* GetPageData() const { return m_pDocPage; }
     62 
     63   void SetPageObjNum(int iPage, uint32_t objNum);
     64 
     65   std::unique_ptr<JBig2_DocumentContext>* CodecContext() {
     66     return &m_pCodecContext;
     67   }
     68   std::unique_ptr<CPDF_LinkList>* LinksContext() { return &m_pLinksContext; }
     69 
     70   CPDF_DocRenderData* GetRenderData() const { return m_pDocRender.get(); }
     71 
     72   // |pFontDict| must not be null.
     73   CPDF_Font* LoadFont(CPDF_Dictionary* pFontDict);
     74   CPDF_ColorSpace* LoadColorSpace(CPDF_Object* pCSObj,
     75                                   CPDF_Dictionary* pResources = nullptr);
     76 
     77   CPDF_Pattern* LoadPattern(CPDF_Object* pObj,
     78                             bool bShading,
     79                             const CFX_Matrix& matrix);
     80 
     81   CPDF_Image* LoadImageFromPageData(uint32_t dwStreamObjNum);
     82   CPDF_StreamAcc* LoadFontFile(CPDF_Stream* pStream);
     83   CPDF_IccProfile* LoadIccProfile(CPDF_Stream* pStream);
     84 
     85   void LoadDoc();
     86   void LoadLinearizedDoc(const CPDF_LinearizedHeader* pLinearizationParams);
     87   void LoadPages();
     88 
     89   void CreateNewDoc();
     90   CPDF_Dictionary* CreateNewPage(int iPage);
     91 
     92   CPDF_Font* AddStandardFont(const FX_CHAR* font, CPDF_FontEncoding* pEncoding);
     93   CPDF_Font* AddFont(CFX_Font* pFont, int charset, bool bVert);
     94 #if _FXM_PLATFORM_ == _FXM_PLATFORM_WINDOWS_
     95   CPDF_Font* AddWindowsFont(LOGFONTA* pLogFont,
     96                             bool bVert,
     97                             bool bTranslateName = false);
     98   CPDF_Font* AddWindowsFont(LOGFONTW* pLogFont,
     99                             bool bVert,
    100                             bool bTranslateName = false);
    101 #endif
    102 
    103  protected:
    104   // Retrieve page count information by getting count value from the tree nodes
    105   int RetrievePageCount() const;
    106   // When this method is called, m_pTreeTraversal[level] exists.
    107   CPDF_Dictionary* TraversePDFPages(int iPage, int* nPagesToGo, size_t level);
    108   int FindPageIndex(CPDF_Dictionary* pNode,
    109                     uint32_t* skip_count,
    110                     uint32_t objnum,
    111                     int* index,
    112                     int level = 0);
    113   std::unique_ptr<CPDF_Object> ParseIndirectObject(uint32_t objnum) override;
    114   void LoadDocInternal();
    115   size_t CalculateEncodingDict(int charset, CPDF_Dictionary* pBaseDict);
    116   CPDF_Dictionary* GetPagesDict() const;
    117   CPDF_Dictionary* ProcessbCJK(
    118       CPDF_Dictionary* pBaseDict,
    119       int charset,
    120       bool bVert,
    121       CFX_ByteString basefont,
    122       std::function<void(FX_WCHAR, FX_WCHAR, CPDF_Array*)> Insert);
    123   bool InsertDeletePDFPage(CPDF_Dictionary* pPages,
    124                            int nPagesToGo,
    125                            CPDF_Dictionary* pPageDict,
    126                            bool bInsert,
    127                            std::set<CPDF_Dictionary*>* pVisited);
    128   bool InsertNewPage(int iPage, CPDF_Dictionary* pPageDict);
    129   void ResetTraversal();
    130 
    131   std::unique_ptr<CPDF_Parser> m_pParser;
    132   CPDF_Dictionary* m_pRootDict;
    133   CPDF_Dictionary* m_pInfoDict;
    134   // Vector of pairs to know current position in the page tree. The index in the
    135   // vector corresponds to the level being described. The pair contains a
    136   // pointer to the dictionary being processed at the level, and an index of the
    137   // of the child being processed within the dictionary's /Kids array.
    138   std::vector<std::pair<CPDF_Dictionary*, size_t>> m_pTreeTraversal;
    139   // Index of the next page that will be traversed from the page tree.
    140   int m_iNextPageToTraverse;
    141   bool m_bReachedMaxPageLevel;
    142   bool m_bLinearized;
    143   int m_iFirstPageNo;
    144   uint32_t m_dwFirstPageObjNum;
    145   // TODO(thestig): Figure out why this cannot be a std::unique_ptr.
    146   CPDF_DocPageData* m_pDocPage;
    147   std::unique_ptr<CPDF_DocRenderData> m_pDocRender;
    148   std::unique_ptr<JBig2_DocumentContext> m_pCodecContext;
    149   std::unique_ptr<CPDF_LinkList> m_pLinksContext;
    150   std::vector<uint32_t> m_PageList;
    151 };
    152 
    153 #endif  // CORE_FPDFAPI_PARSER_CPDF_DOCUMENT_H_
    154