src/third_party/pdfium/fpdfsdk/src/fpdftext.cpp

   1 // Copyright 2014 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 #include "../include/fsdk_define.h"
   8 #include "../include/fpdftext.h"
   9
  10 #ifdef _WIN32
  11 #include <tchar.h>
  12 #endif
  13
  14         // jabdelmalek: commented out to build on Linux.  Not used.
  15         // extern HANDLE g_hModule;
  16
  17 DLLEXPORT FPDF_TEXTPAGE STDCALL FPDFText_LoadPage(FPDF_PAGE page)
  18 {
  19         if (!page) return NULL;
  20         IPDF_TextPage* textpage=NULL;
  21         try
  22         {
  23                 CPDF_ViewerPreferences viewRef(((CPDF_Page*)page)->m_pDocument);
  24                 textpage=IPDF_TextPage::CreateTextPage((CPDF_Page*)page,viewRef.IsDirectionR2L());
  25                 textpage->ParseTextPage();
  26         }
  27         catch (...)
  28         {
  29                 if (textpage)
  30                         delete textpage;
  31                 return NULL;
  32         }
  33         return textpage;
  34 }
  35 DLLEXPORT void STDCALL FPDFText_ClosePage(FPDF_TEXTPAGE text_page)
  36 {
  37         if (text_page){
  38                 IPDF_TextPage* textpage=(IPDF_TextPage*)text_page;
  39                 delete textpage;
  40                 text_page=NULL;
  41         }
  42 }
  43 DLLEXPORT int STDCALL FPDFText_CountChars(FPDF_TEXTPAGE text_page)
  44 {
  45         if (!text_page) return  -1;
  46         IPDF_TextPage* textpage=(IPDF_TextPage*)text_page;
  47         return  textpage->CountChars();
  48 }
  49 DLLEXPORT unsigned int STDCALL FPDFText_GetUnicode(FPDF_TEXTPAGE text_page, int index)
  50 {
  51         if (!text_page) return -1;
  52         IPDF_TextPage* textpage=(IPDF_TextPage*)text_page;
  53
  54         if (index<0 || index>=textpage->CountChars()) return 0;
  55
  56         FPDF_CHAR_INFO  charinfo;
  57         textpage->GetCharInfo(index,charinfo);
  58         return charinfo.m_Unicode;
  59 }
  60 DLLEXPORT double STDCALL FPDFText_GetFontSize(FPDF_TEXTPAGE text_page, int index)
  61 {
  62         if (!text_page) return 0;
  63         IPDF_TextPage* textpage=(IPDF_TextPage*)text_page;
  64
  65         if (index<0 || index>=textpage->CountChars()) return 0;
  66
  67         FPDF_CHAR_INFO  charinfo;
  68         textpage->GetCharInfo(index,charinfo);
  69         return charinfo.m_FontSize;
  70 }
  71
  72 DLLEXPORT void STDCALL FPDFText_GetCharBox(FPDF_TEXTPAGE text_page, int index,double* left,
  73                                                                                                         double* right, double* bottom, double* top)
  74 {
  75         if (!text_page) return;
  76         IPDF_TextPage* textpage=(IPDF_TextPage*)text_page;
  77
  78         if (index<0 || index>=textpage->CountChars()) return ;
  79         FPDF_CHAR_INFO  charinfo;
  80         textpage->GetCharInfo(index,charinfo);
  81         *left=charinfo.m_CharBox.left;
  82         *right=charinfo.m_CharBox.right;
  83         *bottom=charinfo.m_CharBox.bottom;
  84         *top=charinfo.m_CharBox.top;
  85 }
  86
  87 //select
  88 DLLEXPORT int STDCALL FPDFText_GetCharIndexAtPos(FPDF_TEXTPAGE text_page,double x,double y,double xTorelance,double yTorelance)
  89 {
  90         if (!text_page) return -3;
  91         IPDF_TextPage* textpage=(IPDF_TextPage*)text_page;
  92         return textpage->GetIndexAtPos((FX_FLOAT)x,(FX_FLOAT)y,(FX_FLOAT)xTorelance,(FX_FLOAT)yTorelance);
  93 }
  94
  95 DLLEXPORT int STDCALL FPDFText_GetText(FPDF_TEXTPAGE text_page,int start,int count,unsigned short* result)
  96 {
  97         if (!text_page) return 0;
  98         IPDF_TextPage* textpage=(IPDF_TextPage*)text_page;
  99
 100         if (start>=textpage->CountChars()) return 0;
 101
 102         CFX_WideString str=textpage->GetPageText(start,count);
 103         if(str.GetLength()>count)
 104                 str = str.Left(count);
 105
 106         CFX_ByteString cbUTF16str = str.UTF16LE_Encode();
 107         FXSYS_memcpy(result,cbUTF16str.GetBuffer(cbUTF16str.GetLength()),cbUTF16str.GetLength());
 108         cbUTF16str.ReleaseBuffer(cbUTF16str.GetLength());
 109
 110         return cbUTF16str.GetLength()/sizeof(unsigned short);
 111 }
 112
 113 DLLEXPORT int STDCALL FPDFText_CountRects(FPDF_TEXTPAGE text_page,int start,int count)
 114 {
 115         if (!text_page) return 0;
 116         IPDF_TextPage* textpage=(IPDF_TextPage*)text_page;
 117         return  textpage->CountRects(start,count);
 118
 119 }
 120 DLLEXPORT void STDCALL FPDFText_GetRect(FPDF_TEXTPAGE text_page,int rect_index, double* left,double*  top,
 121                                                                                 double*  right, double*  bottom)
 122 {
 123         if (!text_page) return;
 124         IPDF_TextPage* textpage=(IPDF_TextPage*)text_page;
 125         CFX_FloatRect   rect;
 126         textpage->GetRect(rect_index,rect.left,rect.top,rect.right,rect.bottom);
 127         *left=rect.left;
 128         *top=rect.top;
 129         *right=rect.right;
 130         *bottom=rect.bottom;
 131 }
 132
 133 DLLEXPORT int STDCALL FPDFText_GetBoundedText(FPDF_TEXTPAGE text_page,double left, double top,
 134                                                                                           double right, double bottom,unsigned short* buffer,int buflen)
 135 {
 136         if (!text_page) return 0;
 137         IPDF_TextPage* textpage=(IPDF_TextPage*)text_page;
 138         CFX_FloatRect rect((FX_FLOAT)left,(FX_FLOAT)bottom,(FX_FLOAT)right,(FX_FLOAT)top);
 139         CFX_WideString str=textpage->GetTextByRect(rect);
 140
 141         if (buflen<=0 || buffer==NULL)
 142         {
 143                 return str.GetLength();
 144         }
 145
 146         CFX_ByteString cbUTF16Str = str.UTF16LE_Encode();
 147         int len = cbUTF16Str.GetLength()/sizeof(unsigned short);
 148         int size = buflen > len ? len : buflen;
 149         FXSYS_memcpy(buffer,cbUTF16Str.GetBuffer(size*sizeof(unsigned short)),size*sizeof(unsigned short));
 150         cbUTF16Str.ReleaseBuffer(size*sizeof(unsigned short));
 151
 152         return size;
 153
 154 }
 155
 156 //Search
 157 //-1 for end
 158 DLLEXPORT FPDF_SCHHANDLE STDCALL FPDFText_FindStart(FPDF_TEXTPAGE text_page,FPDF_WIDESTRING findwhat,unsigned long flags,int start_index)
 159 {
 160         if (!text_page) return NULL;
 161         IPDF_TextPageFind* textpageFind=NULL;
 162         try
 163         {
 164                 textpageFind=IPDF_TextPageFind::CreatePageFind((IPDF_TextPage*)text_page);
 165                 textpageFind->FindFirst(CFX_WideString::FromUTF16LE(findwhat),flags,start_index);
 166         }
 167         catch (...)
 168         {
 169                 if (textpageFind)
 170                         delete textpageFind;
 171                 return NULL;
 172         }
 173         return textpageFind;
 174 }
 175 DLLEXPORT FPDF_BOOL     STDCALL FPDFText_FindNext(FPDF_SCHHANDLE handle)
 176 {
 177         if (!handle) return FALSE;
 178         IPDF_TextPageFind* textpageFind=(IPDF_TextPageFind*)handle;
 179         return  textpageFind->FindNext();
 180 }
 181 DLLEXPORT FPDF_BOOL STDCALL FPDFText_FindPrev(FPDF_SCHHANDLE handle)
 182 {
 183         if (!handle) return FALSE;
 184         IPDF_TextPageFind* textpageFind=(IPDF_TextPageFind*)handle;
 185         return  textpageFind->FindPrev();
 186 }
 187 DLLEXPORT int STDCALL FPDFText_GetSchResultIndex(FPDF_SCHHANDLE handle)
 188 {
 189         if (!handle) return 0;
 190         IPDF_TextPageFind* textpageFind=(IPDF_TextPageFind*)handle;
 191         return textpageFind->GetCurOrder();
 192 }
 193 DLLEXPORT int STDCALL FPDFText_GetSchCount(FPDF_SCHHANDLE handle)
 194 {
 195         if (!handle) return 0;
 196         IPDF_TextPageFind* textpageFind=(IPDF_TextPageFind*)handle;
 197         return textpageFind->GetMatchedCount();
 198 }
 199 DLLEXPORT void STDCALL FPDFText_FindClose(FPDF_SCHHANDLE handle)
 200 {
 201         if (!handle) return;
 202         IPDF_TextPageFind* textpageFind=(IPDF_TextPageFind*)handle;
 203         delete  textpageFind;
 204         handle=NULL;
 205 }
 206
 207 //web link
 208 DLLEXPORT FPDF_PAGELINK STDCALL FPDFLink_LoadWebLinks(FPDF_TEXTPAGE text_page)
 209 {
 210         if (!text_page) return NULL;
 211         IPDF_LinkExtract* pageLink=NULL;
 212         try
 213         {
 214                 pageLink=IPDF_LinkExtract::CreateLinkExtract();
 215                 pageLink->ExtractLinks((IPDF_TextPage*)text_page);
 216         }
 217         catch (...)
 218         {
 219                 if (pageLink)
 220                         delete pageLink;
 221                 return NULL;
 222         }
 223         return pageLink;
 224 }
 225 DLLEXPORT int STDCALL FPDFLink_CountWebLinks(FPDF_PAGELINK link_page)
 226 {
 227         if (!link_page) return 0;
 228         IPDF_LinkExtract* pageLink=(IPDF_LinkExtract*)link_page;
 229         return  pageLink->CountLinks();
 230 }
 231 DLLEXPORT int STDCALL FPDFLink_GetURL(FPDF_PAGELINK link_page,int link_index, unsigned short* buffer,int buflen)
 232 {
 233         if (!link_page) return 0;
 234         IPDF_LinkExtract* pageLink=(IPDF_LinkExtract*)link_page;
 235         CFX_WideString url=pageLink->GetURL(link_index);
 236
 237         CFX_ByteString cbUTF16URL = url.UTF16LE_Encode();
 238         int len= cbUTF16URL.GetLength()/sizeof(unsigned short);
 239         if (buffer==NULL || buflen<=0)
 240                 return len;
 241         int size=len<buflen ? len :buflen;
 242         if (size>0)
 243         {
 244                 FXSYS_memcpy(buffer,cbUTF16URL.GetBuffer(size*sizeof(unsigned short)),size*sizeof(unsigned short));
 245                 cbUTF16URL.ReleaseBuffer(size*sizeof(unsigned short));
 246         }
 247         return size;
 248 }
 249 DLLEXPORT int STDCALL FPDFLink_CountRects(FPDF_PAGELINK link_page,int link_index)
 250 {
 251         if (!link_page) return 0;
 252         IPDF_LinkExtract* pageLink=(IPDF_LinkExtract*)link_page;
 253         CFX_RectArray rectArray;
 254         pageLink->GetRects(link_index,rectArray);
 255         return rectArray.GetSize();
 256 }
 257 DLLEXPORT void STDCALL FPDFLink_GetRect(FPDF_PAGELINK link_page,int link_index,  int rect_index, double* left,
 258                                                                                 double*  top,double*  right, double*  bottom)
 259 {
 260         if (!link_page) return;
 261         IPDF_LinkExtract* pageLink=(IPDF_LinkExtract*)link_page;
 262         CFX_RectArray rectArray;
 263         pageLink->GetRects(link_index,rectArray);
 264         CFX_FloatRect rect;
 265         rect=rectArray.GetAt(rect_index);
 266         *left=rect.left;
 267         *right=rect.right;
 268         *top=rect.top;
 269         *bottom=rect.bottom;
 270 }
 271 DLLEXPORT void  STDCALL FPDFLink_CloseWebLinks(FPDF_PAGELINK link_page)
 272 {
 273         if (!link_page) return;
 274         IPDF_LinkExtract* pageLink=(IPDF_LinkExtract*)link_page;
 275         delete pageLink;
 276         pageLink =NULL;
 277 }
 278