Traverse PDF page tree only once in CPDF_Document Try 3

core/fpdfapi/parser/cpdf_document.h

 // Copyright 2016 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_FPDFAPI_PARSER_CPDF_DOCUMENT_H_
 #define CORE_FPDFAPI_PARSER_CPDF_DOCUMENT_H_
 
 #include <functional>
@@ skipping 87 matching lines @@
                             bool bVert,
                             bool bTranslateName = false);
   CPDF_Font* AddWindowsFont(LOGFONTW* pLogFont,
                             bool bVert,
                             bool bTranslateName = false);
 #endif
 
  protected:
   // Retrieve page count information by getting count value from the tree nodes
   int RetrievePageCount() const;
-  CPDF_Dictionary* FindPDFPage(CPDF_Dictionary* pPages,
-                               int iPage,
-                               int nPagesToGo,
-                               int level);
+  CPDF_Dictionary* TraversePDFPages(int iPage, int& nPagesToGo, int level);

[Tom Sepez, 2016/11/03 22:31:00]

nit: the style guide would say int* nPagesToGo. 

[npm, 2016/11/04 19:33:33]

Done.

   int FindPageIndex(CPDF_Dictionary* pNode,
                     uint32_t& skip_count,
                     uint32_t objnum,
                     int& index,
                     int level = 0);
   CPDF_Object* ParseIndirectObject(uint32_t objnum) override;
   void LoadDocInternal();
   size_t CalculateEncodingDict(int charset, CPDF_Dictionary* pBaseDict);
   CPDF_Dictionary* GetPagesDict() const;
   CPDF_Dictionary* ProcessbCJK(
       CPDF_Dictionary* pBaseDict,
       int charset,
       bool bVert,
       CFX_ByteString basefont,
       std::function<void(FX_WCHAR, FX_WCHAR, CPDF_Array*)> Insert);
   bool InsertDeletePDFPage(CPDF_Dictionary* pPages,
                            int nPagesToGo,
                            CPDF_Dictionary* pPageDict,
                            bool bInsert,
                            std::set<CPDF_Dictionary*>* pVisited);
   bool InsertNewPage(int iPage, CPDF_Dictionary* pPageDict);
 
   std::unique_ptr<CPDF_Parser> m_pParser;
   CPDF_Dictionary* m_pRootDict;
   CPDF_Dictionary* m_pInfoDict;
+  // Vector of nodes to know current position in the page tree. Int is the index

[Lei Zhang, 2016/11/04 01:38:52]

Maybe mention the index into this vector is the level, and what both values in
the pair represent.

[npm, 2016/11/04 19:33:33]

Done.

+  // of the child being processed.

[Tom Sepez, 2016/11/03 22:31:00]

nit:  // of the child being processed within the dictionary's /Kids array.

[npm, 2016/11/04 19:33:33]

Done.

+  std::vector<std::pair<CPDF_Dictionary*, int>> m_pTreeTraversal;

[Lei Zhang, 2016/11/04 01:38:52]

Should the int also be a size_t here?

[npm, 2016/11/04 19:33:33]

Done.

+  // Index of the next page that will be traversed from the page tree.
+  int m_iNextPageToTraverse;
   bool m_bLinearized;
   int m_iFirstPageNo;
   uint32_t m_dwFirstPageObjNum;
   // TODO(thestig): Figure out why this cannot be a std::unique_ptr.
   CPDF_DocPageData* m_pDocPage;
   std::unique_ptr<CPDF_DocRenderData> m_pDocRender;
   std::unique_ptr<JBig2_DocumentContext> m_pCodecContext;
   std::unique_ptr<CPDF_LinkList> m_pLinksContext;
   CFX_ArrayTemplate<uint32_t> m_PageList;
   CFX_WeakPtr<CFX_ByteStringPool> m_pByteStringPool;
 };
 
 #endif  // CORE_FPDFAPI_PARSER_CPDF_DOCUMENT_H_