rosapps/smartpdf/poppler/poppler/PDFDoc.cc

   1 //========================================================================
   2 //
   3 // PDFDoc.cc
   4 //
   5 // Copyright 1996-2003 Glyph & Cog, LLC
   6 //
   7 //========================================================================
   8
   9 #include <config.h>
  10
  11 #ifdef USE_GCC_PRAGMAS
  12 #pragma implementation
  13 #endif
  14
  15 #include <locale.h>
  16 #include <stdio.h>
  17 #include <stdlib.h>
  18 #include <stddef.h>
  19 #include <string.h>
  20 #ifdef WIN32
  21 #  include <windows.h>
  22 #endif
  23 #include "goo/GooString.h"
  24 #include "poppler-config.h"
  25 #include "GlobalParams.h"
  26 #include "Page.h"
  27 #include "Catalog.h"
  28 #include "Stream.h"
  29 #include "XRef.h"
  30 #include "Link.h"
  31 #include "OutputDev.h"
  32 #include "Error.h"
  33 #include "ErrorCodes.h"
  34 #include "Lexer.h"
  35 #include "Parser.h"
  36 #include "SecurityHandler.h"
  37 #ifndef DISABLE_OUTLINE
  38 #include "Outline.h"
  39 #endif
  40 #include "PDFDoc.h"
  41 #include "UGooString.h"
  42
  43 //------------------------------------------------------------------------
  44
  45 #define headerSearchSize 1024   // read this many bytes at beginning of
  46                                 //   file to look for '%PDF'
  47
  48 //------------------------------------------------------------------------
  49 // PDFDoc
  50 //------------------------------------------------------------------------
  51
  52 PDFDoc::PDFDoc(GooString *fileNameA, GooString *ownerPassword,
  53                GooString *userPassword, void *guiDataA) {
  54   Object obj;
  55   GooString *fileName1, *fileName2;
  56
  57   ok = gFalse;
  58   errCode = errNone;
  59
  60   guiData = guiDataA;
  61
  62   file = NULL;
  63   str = NULL;
  64   xref = NULL;
  65   catalog = NULL;
  66   links = NULL;
  67 #ifndef DISABLE_OUTLINE
  68   outline = NULL;
  69 #endif
  70
  71   fileName = fileNameA;
  72   fileName1 = fileName;
  73
  74
  75   // try to open file
  76   fileName2 = NULL;
  77 #ifdef VMS
  78   if (!(file = fopen(fileName1->getCString(), "rb", "ctx=stm"))) {
  79     error(-1, "Couldn't open file '%s'", fileName1->getCString());
  80     errCode = errOpenFile;
  81     return;
  82   }
  83 #else
  84   if (!(file = fopen(fileName1->getCString(), "rb"))) {
  85     fileName2 = fileName->copy();
  86     fileName2->lowerCase();
  87     if (!(file = fopen(fileName2->getCString(), "rb"))) {
  88       fileName2->upperCase();
  89       if (!(file = fopen(fileName2->getCString(), "rb"))) {
  90         error(-1, "Couldn't open file '%s'", fileName->getCString());
  91         delete fileName2;
  92         errCode = errOpenFile;
  93         return;
  94       }
  95     }
  96     delete fileName2;
  97   }
  98 #endif
  99
 100   // create stream
 101   obj.initNull();
 102   str = new FileStream(file, 0, gFalse, 0, &obj);
 103
 104   ok = setup(ownerPassword, userPassword);
 105 }
 106
 107 #ifdef WIN32
 108 PDFDoc::PDFDoc(wchar_t *fileNameA, int fileNameLen, GooString *ownerPassword,
 109                GooString *userPassword, void *guiDataA) {
 110   OSVERSIONINFO version;
 111   wchar_t fileName2[_MAX_PATH + 1];
 112   Object obj;
 113   int i;
 114
 115   ok = gFalse;
 116   errCode = errNone;
 117
 118   guiData = guiDataA;
 119
 120   file = NULL;
 121   str = NULL;
 122   xref = NULL;
 123   catalog = NULL;
 124   links = NULL;
 125 #ifndef DISABLE_OUTLINE
 126   outline = NULL;
 127 #endif
 128
 129   //~ file name should be stored in Unicode (?)
 130   fileName = new GooString();
 131   for (i = 0; i < fileNameLen; ++i) {
 132     fileName->append((char)fileNameA[i]);
 133   }
 134
 135   // zero-terminate the file name string
 136   for (i = 0; i < fileNameLen && i < _MAX_PATH; ++i) {
 137     fileName2[i] = fileNameA[i];
 138   }
 139   fileName2[i] = 0;
 140
 141   // try to open file
 142   // NB: _wfopen is only available in NT
 143   version.dwOSVersionInfoSize = sizeof(version);
 144   GetVersionEx(&version);
 145   if (version.dwPlatformId == VER_PLATFORM_WIN32_NT) {
 146     file = _wfopen(fileName2, L"rb");
 147   } else {
 148     file = fopen(fileName->getCString(), "rb");
 149   }
 150   if (!file) {
 151     error(-1, "Couldn't open file '%s'", fileName->getCString());
 152     errCode = errOpenFile;
 153     return;
 154   }
 155
 156   // create stream
 157   obj.initNull();
 158   str = new FileStream(file, 0, gFalse, 0, &obj);
 159
 160   ok = setup(ownerPassword, userPassword);
 161 }
 162 #endif
 163
 164 PDFDoc::PDFDoc(BaseStream *strA, GooString *ownerPassword,
 165                GooString *userPassword, void *guiDataA) {
 166   ok = gFalse;
 167   errCode = errNone;
 168   guiData = guiDataA;
 169   fileName = NULL;
 170   file = NULL;
 171   str = strA;
 172   xref = NULL;
 173   catalog = NULL;
 174   links = NULL;
 175 #ifndef DISABLE_OUTLINE
 176   outline = NULL;
 177 #endif
 178   ok = setup(ownerPassword, userPassword);
 179 }
 180
 181 GBool PDFDoc::setup(GooString *ownerPassword, GooString *userPassword) {
 182   str->reset();
 183
 184   // check footer
 185   if (!checkFooter()) return gFalse;
 186
 187   // check header
 188   checkHeader();
 189
 190   // read xref table
 191   xref = new XRef(str);
 192   if (!xref->isOk()) {
 193     error(-1, "Couldn't read xref table");
 194     errCode = xref->getErrorCode();
 195     return gFalse;
 196   }
 197
 198   // check for encryption
 199   if (!checkEncryption(ownerPassword, userPassword)) {
 200     errCode = errEncrypted;
 201     return gFalse;
 202   }
 203
 204   // read catalog
 205   catalog = new Catalog(xref);
 206   if (!catalog->isOk()) {
 207     error(-1, "Couldn't read page catalog");
 208     errCode = errBadCatalog;
 209     return gFalse;
 210   }
 211
 212 #ifndef DISABLE_OUTLINE
 213   // read outline
 214   outline = new Outline(catalog->getOutline(), xref);
 215 #endif
 216
 217   // done
 218   return gTrue;
 219 }
 220
 221 PDFDoc::~PDFDoc() {
 222 #ifndef DISABLE_OUTLINE
 223   if (outline) {
 224     delete outline;
 225   }
 226 #endif
 227   if (catalog) {
 228     delete catalog;
 229   }
 230   if (xref) {
 231     delete xref;
 232   }
 233   if (str) {
 234     delete str;
 235   }
 236   if (file) {
 237     fclose(file);
 238   }
 239   if (fileName) {
 240     delete fileName;
 241   }
 242   if (links) {
 243     delete links;
 244   }
 245 }
 246
 247
 248 // Check for a %%EOF at the end of this stream
 249 GBool PDFDoc::checkFooter() {
 250   // we look in the last 1024 chars because Adobe does the same
 251   char *eof = new char[1025];
 252   int pos = str->getPos();
 253   str->setPos(1024, -1);
 254   int i, ch;
 255   for (i = 0; i < 1024; i++)
 256   {
 257     ch = str->getChar();
 258     if (ch == EOF)
 259       break;
 260     eof[i] = ch;
 261   }
 262   eof[i] = '\0';
 263
 264   bool found = false;
 265   for (i = i - 5; i >= 0; i--) {
 266     if (strncmp (&eof[i], "%%EOF", 5) == 0) {
 267       found = true;
 268       break;
 269     }
 270   }
 271   if (!found)
 272   {
 273     error(-1, "Document has not the mandatory ending %%EOF");
 274     errCode = errDamaged;
 275     delete[] eof;
 276     return gFalse;
 277   }
 278   delete[] eof;
 279   str->setPos(pos);
 280   return gTrue;
 281 }
 282
 283 // Check for a PDF header on this stream.  Skip past some garbage
 284 // if necessary.
 285 void PDFDoc::checkHeader() {
 286   char hdrBuf[headerSearchSize+1];
 287   char *p;
 288   int i;
 289
 290   pdfVersion = 0;
 291   for (i = 0; i < headerSearchSize; ++i) {
 292     hdrBuf[i] = str->getChar();
 293   }
 294   hdrBuf[headerSearchSize] = '\0';
 295   for (i = 0; i < headerSearchSize - 5; ++i) {
 296     if (!strncmp(&hdrBuf[i], "%PDF-", 5)) {
 297       break;
 298     }
 299   }
 300   if (i >= headerSearchSize - 5) {
 301     error(-1, "May not be a PDF file (continuing anyway)");
 302     return;
 303   }
 304   str->moveStart(i);
 305   if (!(p = strtok(&hdrBuf[i+5], " \t\n\r"))) {
 306     error(-1, "May not be a PDF file (continuing anyway)");
 307     return;
 308   }
 309   {
 310     char *theLocale = setlocale(LC_NUMERIC, "C");
 311     pdfVersion = atof(p);
 312     setlocale(LC_NUMERIC, theLocale);
 313   }
 314   // We don't do the version check. Don't add it back in.
 315 }
 316
 317 GBool PDFDoc::checkEncryption(GooString *ownerPassword, GooString *userPassword) {
 318   Object encrypt;
 319   GBool encrypted;
 320   SecurityHandler *secHdlr;
 321   GBool ret;
 322
 323   xref->getTrailerDict()->dictLookup("Encrypt", &encrypt);
 324   if ((encrypted = encrypt.isDict())) {
 325     if ((secHdlr = SecurityHandler::make(this, &encrypt))) {
 326       if (secHdlr->checkEncryption(ownerPassword, userPassword)) {
 327         // authorization succeeded
 328         xref->setEncryption(secHdlr->getPermissionFlags(),
 329                             secHdlr->getOwnerPasswordOk(),
 330                             secHdlr->getFileKey(),
 331                             secHdlr->getFileKeyLength(),
 332                             secHdlr->getEncVersion(),
 333                             secHdlr->getEncRevision());
 334         ret = gTrue;
 335       } else {
 336         // authorization failed
 337         ret = gFalse;
 338       }
 339       delete secHdlr;
 340     } else {
 341       // couldn't find the matching security handler
 342       ret = gFalse;
 343     }
 344   } else {
 345     // document is not encrypted
 346     ret = gTrue;
 347   }
 348   encrypt.free();
 349   return ret;
 350 }
 351
 352 void PDFDoc::displayPage(OutputDev *out, int page, double hDPI, double vDPI,
 353                          int rotate, GBool useMediaBox, GBool crop, GBool doLinks,
 354                          GBool (*abortCheckCbk)(void *data),
 355                          void *abortCheckCbkData,
 356                          GBool (*annotDisplayDecideCbk)(Annot *annot, void *user_data),
 357                          void *annotDisplayDecideCbkData) {
 358   Page *p;
 359
 360   if (globalParams->getPrintCommands()) {
 361     printf("***** page %d *****\n", page);
 362   }
 363   p = catalog->getPage(page);
 364   if (doLinks) {
 365     if (links) {
 366       delete links;
 367     }
 368     getLinks(p);
 369     p->display(out, hDPI, vDPI, rotate, useMediaBox, crop, links, catalog,
 370                abortCheckCbk, abortCheckCbkData,
 371                annotDisplayDecideCbk, annotDisplayDecideCbkData);
 372   } else {
 373     p->display(out, hDPI, vDPI, rotate, useMediaBox, crop, NULL, catalog,
 374                abortCheckCbk, abortCheckCbkData,
 375                annotDisplayDecideCbk, annotDisplayDecideCbkData);
 376   }
 377 }
 378
 379 void PDFDoc::displayPages(OutputDev *out, int firstPage, int lastPage,
 380                           double hDPI, double vDPI, int rotate, GBool useMediaBox,
 381                           GBool crop, GBool doLinks,
 382                           GBool (*abortCheckCbk)(void *data),
 383                           void *abortCheckCbkData,
 384                           GBool (*annotDisplayDecideCbk)(Annot *annot, void *user_data),
 385                           void *annotDisplayDecideCbkData) {
 386   int page;
 387
 388   for (page = firstPage; page <= lastPage; ++page) {
 389     displayPage(out, page, hDPI, vDPI, rotate, useMediaBox, crop, doLinks,
 390                 abortCheckCbk, abortCheckCbkData,
 391                 annotDisplayDecideCbk, annotDisplayDecideCbkData);
 392   }
 393 }
 394
 395 void PDFDoc::displayPageSlice(OutputDev *out, int page,
 396                               double hDPI, double vDPI,
 397                               int rotate, GBool useMediaBox, GBool crop, GBool doLinks,
 398                               int sliceX, int sliceY, int sliceW, int sliceH,
 399                               GBool (*abortCheckCbk)(void *data),
 400                               void *abortCheckCbkData,
 401                               GBool (*annotDisplayDecideCbk)(Annot *annot, void *user_data),
 402                               void *annotDisplayDecideCbkData) {
 403   Page *p;
 404
 405   p = catalog->getPage(page);
 406   if (doLinks)
 407   {
 408     if (links) {
 409       delete links;
 410     }
 411     getLinks(p);
 412     p->displaySlice(out, hDPI, vDPI, rotate, useMediaBox, crop,
 413                   sliceX, sliceY, sliceW, sliceH,
 414                   links, catalog,
 415                   abortCheckCbk, abortCheckCbkData,
 416                   annotDisplayDecideCbk, annotDisplayDecideCbkData);
 417   } else {
 418     p->displaySlice(out, hDPI, vDPI, rotate, useMediaBox, crop,
 419                   sliceX, sliceY, sliceW, sliceH,
 420                   NULL, catalog,
 421                   abortCheckCbk, abortCheckCbkData,
 422                   annotDisplayDecideCbk, annotDisplayDecideCbkData);
 423   }
 424 }
 425
 426 Links *PDFDoc::takeLinks() {
 427   Links *ret;
 428
 429   ret = links;
 430   links = NULL;
 431   return ret;
 432 }
 433
 434 GBool PDFDoc::isLinearized() {
 435   Parser *parser;
 436   Object obj1, obj2, obj3, obj4, obj5;
 437   GBool lin;
 438
 439   lin = gFalse;
 440   obj1.initNull();
 441   parser = new Parser(xref,
 442              new Lexer(xref,
 443                str->makeSubStream(str->getStart(), gFalse, 0, &obj1)));
 444   parser->getObj(&obj1);
 445   parser->getObj(&obj2);
 446   parser->getObj(&obj3);
 447   parser->getObj(&obj4);
 448   if (obj1.isInt() && obj2.isInt() && obj3.isCmd("obj") &&
 449       obj4.isDict()) {
 450     obj4.dictLookup("Linearized", &obj5);
 451     if (obj5.isNum() && obj5.getNum() > 0) {
 452       lin = gTrue;
 453     }
 454     obj5.free();
 455   }
 456   obj4.free();
 457   obj3.free();
 458   obj2.free();
 459   obj1.free();
 460   delete parser;
 461   return lin;
 462 }
 463
 464 GBool PDFDoc::saveAs(GooString *name) {
 465   FILE *f;
 466   int c;
 467
 468   if (!(f = fopen(name->getCString(), "wb"))) {
 469     error(-1, "Couldn't open file '%s'", name->getCString());
 470     return gFalse;
 471   }
 472   str->reset();
 473   while ((c = str->getChar()) != EOF) {
 474     fputc(c, f);
 475   }
 476   str->close();
 477   fclose(f);
 478   return gTrue;
 479 }
 480
 481 void PDFDoc::getLinks(Page *page) {
 482   Object obj;
 483
 484   links = new Links(page->getAnnots(&obj), catalog->getBaseURI());
 485   obj.free();
 486 }