pdf/pdfium/pdfium_page.h

   1 // Copyright (c) 2010 The Chromium 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 #ifndef PDF_PDFIUM_PDFIUM_PAGE_H_
   6 #define PDF_PDFIUM_PDFIUM_PAGE_H_
   7
   8 #include <string>
   9 #include <vector>
  10
  11 #include "base/strings/string16.h"
  12 #include "ppapi/cpp/rect.h"
  13 #include "third_party/pdfium/fpdfsdk/include/fpdfdoc.h"
  14 #include "third_party/pdfium/fpdfsdk/include/fpdfformfill.h"
  15 #include "third_party/pdfium/fpdfsdk/include/fpdftext.h"
  16
  17 namespace base {
  18 class Value;
  19 }
  20
  21 namespace chrome_pdf {
  22
  23 class PDFiumEngine;
  24
  25 // Wrapper around a page from the document.
  26 class PDFiumPage {
  27  public:
  28   PDFiumPage(PDFiumEngine* engine,
  29              int i,
  30              const pp::Rect& r,
  31              bool available);
  32   ~PDFiumPage();
  33   // Unloads the PDFium data for this page from memory.
  34   void Unload();
  35   // Gets the FPDF_PAGE for this page, loading and parsing it if necessary.
  36   FPDF_PAGE GetPage();
  37   //Get the FPDF_PAGE for printing.
  38   FPDF_PAGE GetPrintPage();
  39   //Close the printing page.
  40   void ClosePrintPage();
  41
  42   // Returns FPDF_TEXTPAGE for the page, loading and parsing it if necessary.
  43   FPDF_TEXTPAGE GetTextPage();
  44
  45   // Returns a DictionaryValue version of the page.
  46   base::Value* GetAccessibleContentAsValue(int rotation);
  47
  48   enum Area {
  49     NONSELECTABLE_AREA,
  50     TEXT_AREA,
  51     WEBLINK_AREA,  // Area is a hyperlink.
  52     DOCLINK_AREA,  // Area is a link to a different part of the same document.
  53   };
  54
  55   struct LinkTarget {
  56     // We are using std::string here which have a copy contructor.
  57     // That prevents us from using union here.
  58     std::string url;  // Valid for WEBLINK_AREA only.
  59     int page;         // Valid for DOCLINK_AREA only.
  60   };
  61
  62   // Given a point in the document that's in this page, returns its character
  63   // index if it's near a character, and also the type of text.
  64   // Target is optional. It will be filled in for WEBLINK_AREA or
  65   // DOCLINK_AREA only.
  66   Area GetCharIndex(const pp::Point& point, int rotation, int* char_index,
  67                     LinkTarget* target);
  68
  69   // Gets the character at the given index.
  70   base::char16 GetCharAtIndex(int index);
  71
  72   // Gets the number of characters in the page.
  73   int GetCharCount();
  74
  75   // Converts from page coordinates to screen coordinates.
  76   pp::Rect PageToScreen(const pp::Point& offset,
  77                         double zoom,
  78                         double left,
  79                         double top,
  80                         double right,
  81                         double bottom,
  82                         int rotation);
  83
  84   int index() const { return index_; }
  85   pp::Rect rect() const { return rect_; }
  86   void set_rect(const pp::Rect& r) { rect_ = r; }
  87   bool available() const { return available_; }
  88   void set_available(bool available) { available_ = available; }
  89   void set_calculated_links(bool calculated_links) {
  90      calculated_links_ = calculated_links;
  91   }
  92
  93  private:
  94   // Returns a link index if the given character index is over a link, or -1
  95   // otherwise.
  96   int GetLink(int char_index, LinkTarget* target);
  97   // Returns the link indices if the given rect intersects a link rect, or an
  98   // empty vector otherwise.
  99   std::vector<int> GetLinks(pp::Rect text_area,
 100                             std::vector<LinkTarget>* targets);
 101   // Calculate the locations of any links on the page.
 102   void CalculateLinks();
 103   // Returns link type and target associated with a link. Returns
 104   // NONSELECTABLE_AREA if link detection failed.
 105   Area GetLinkTarget(FPDF_LINK link, LinkTarget* target);
 106   // Returns target associated with a destination.
 107   Area GetDestinationTarget(FPDF_DEST destination, LinkTarget* target);
 108   // Returns the text in the supplied box as a Value Node
 109   base::Value* GetTextBoxAsValue(double page_height, double left, double top,
 110                                  double right, double bottom, int rotation);
 111   // Helper functions for JSON generation
 112   base::Value* CreateTextNode(std::string text);
 113   base::Value* CreateURLNode(std::string text, std::string url);
 114
 115   struct Link {
 116     Link();
 117     ~Link();
 118
 119     std::string url;
 120     // Bounding rectangles of characters.
 121     std::vector<pp::Rect> rects;
 122   };
 123
 124   PDFiumEngine* engine_;
 125   FPDF_PAGE page_;
 126   FPDF_TEXTPAGE text_page_;
 127   int index_;
 128   pp::Rect rect_;
 129   bool calculated_links_;
 130   std::vector<Link> links_;
 131   bool available_;
 132 };
 133
 134 }  // namespace chrome_pdf
 135
 136 #endif  // PDF_PDFIUM_PDFIUM_PAGE_H_