arospdf/xpdf/pdffonts.cc

   1 //========================================================================
   2 //
   3 // pdffonts.cc
   4 //
   5 // Copyright 2001-2007 Glyph & Cog, LLC
   6 //
   7 //========================================================================
   8
   9 #include <aconf.h>
  10 #include <stdio.h>
  11 #include <stdlib.h>
  12 #include <stddef.h>
  13 #include <string.h>
  14 #include <math.h>
  15 #include "parseargs.h"
  16 #include "GString.h"
  17 #include "gmem.h"
  18 #include "GlobalParams.h"
  19 #include "Error.h"
  20 #include "Object.h"
  21 #include "Dict.h"
  22 #include "GfxFont.h"
  23 #include "Annot.h"
  24 #include "PDFDoc.h"
  25 #include "config.h"
  26
  27 // NB: this must match the definition of GfxFontType in GfxFont.h.
  28 static char *fontTypeNames[] = {
  29   "unknown",
  30   "Type 1",
  31   "Type 1C",
  32   "Type 1C (OT)",
  33   "Type 3",
  34   "TrueType",
  35   "TrueType (OT)",
  36   "CID Type 0",
  37   "CID Type 0C",
  38   "CID Type 0C (OT)",
  39   "CID TrueType",
  40   "CID TrueType (OT)"
  41 };
  42
  43 static void scanFonts(Dict *resDict, PDFDoc *doc);
  44 static void scanFont(GfxFont *font, PDFDoc *doc);
  45
  46 static int firstPage = 1;
  47 static int lastPage = 0;
  48 static char ownerPassword[33] = "\001";
  49 static char userPassword[33] = "\001";
  50 static char cfgFileName[256] = "";
  51 static GBool printVersion = gFalse;
  52 static GBool printHelp = gFalse;
  53
  54 static ArgDesc argDesc[] = {
  55   {"-f",      argInt,      &firstPage,     0,
  56    "first page to examine"},
  57   {"-l",      argInt,      &lastPage,      0,
  58    "last page to examine"},
  59   {"-opw",    argString,   ownerPassword,  sizeof(ownerPassword),
  60    "owner password (for encrypted files)"},
  61   {"-upw",    argString,   userPassword,   sizeof(userPassword),
  62    "user password (for encrypted files)"},
  63   {"-cfg",        argString,      cfgFileName,    sizeof(cfgFileName),
  64    "configuration file to use in place of .xpdfrc"},
  65   {"-v",      argFlag,     &printVersion,  0,
  66    "print copyright and version info"},
  67   {"-h",      argFlag,     &printHelp,     0,
  68    "print usage information"},
  69   {"-help",   argFlag,     &printHelp,     0,
  70    "print usage information"},
  71   {"--help",  argFlag,     &printHelp,     0,
  72    "print usage information"},
  73   {"-?",      argFlag,     &printHelp,     0,
  74    "print usage information"},
  75   {NULL}
  76 };
  77
  78 static Ref *fonts;
  79 static int fontsLen;
  80 static int fontsSize;
  81
  82 int main(int argc, char *argv[]) {
  83   PDFDoc *doc;
  84   GString *fileName;
  85   GString *ownerPW, *userPW;
  86   GBool ok;
  87   Page *page;
  88   Dict *resDict;
  89   Annots *annots;
  90   xObject obj1, obj2;
  91   int pg, i;
  92   int exitCode;
  93
  94   exitCode = 99;
  95
  96   // parse args
  97   ok = parseArgs(argDesc, &argc, argv);
  98   if (!ok || argc != 2 || printVersion || printHelp) {
  99     fprintf(stderr, "pdffonts version %s\n", xpdfVersion);
 100     fprintf(stderr, "%s\n", xpdfCopyright);
 101     if (!printVersion) {
 102       printUsage("pdffonts", "<PDF-file>", argDesc);
 103     }
 104     goto err0;
 105   }
 106   fileName = new GString(argv[1]);
 107
 108   // read config file
 109   globalParams = new GlobalParams(cfgFileName);
 110
 111   // open PDF file
 112   if (ownerPassword[0] != '\001') {
 113     ownerPW = new GString(ownerPassword);
 114   } else {
 115     ownerPW = NULL;
 116   }
 117   if (userPassword[0] != '\001') {
 118     userPW = new GString(userPassword);
 119   } else {
 120     userPW = NULL;
 121   }
 122   doc = new PDFDoc(fileName, ownerPW, userPW);
 123   if (userPW) {
 124     delete userPW;
 125   }
 126   if (ownerPW) {
 127     delete ownerPW;
 128   }
 129   if (!doc->isOk()) {
 130     exitCode = 1;
 131     goto err1;
 132   }
 133
 134   // get page range
 135   if (firstPage < 1) {
 136     firstPage = 1;
 137   }
 138   if (lastPage < 1 || lastPage > doc->getNumPages()) {
 139     lastPage = doc->getNumPages();
 140   }
 141
 142   // scan the fonts
 143   printf("name                                 type              emb sub uni xObject ID\n");
 144   printf("------------------------------------ ----------------- --- --- --- ---------\n");
 145   fonts = NULL;
 146   fontsLen = fontsSize = 0;
 147   for (pg = firstPage; pg <= lastPage; ++pg) {
 148     page = doc->getCatalog()->getPage(pg);
 149     if ((resDict = page->getResourceDict())) {
 150       scanFonts(resDict, doc);
 151     }
 152     annots = new Annots(doc->getXRef(), doc->getCatalog(),
 153                         page->getAnnots(&obj1));
 154     obj1.free();
 155     for (i = 0; i < annots->getNumAnnots(); ++i) {
 156       if (annots->getAnnot(i)->getAppearance(&obj1)->isStream()) {
 157         obj1.streamGetDict()->lookup("Resources", &obj2);
 158         if (obj2.isDict()) {
 159           scanFonts(obj2.getDict(), doc);
 160         }
 161         obj2.free();
 162       }
 163       obj1.free();
 164     }
 165     delete annots;
 166   }
 167
 168   exitCode = 0;
 169
 170   // clean up
 171   gfree(fonts);
 172  err1:
 173   delete doc;
 174   delete globalParams;
 175  err0:
 176
 177   // check for memory leaks
 178   xObject::memCheck(stderr);
 179   gMemReport(stderr);
 180
 181   return exitCode;
 182 }
 183
 184 static void scanFonts(Dict *resDict, PDFDoc *doc) {
 185   xObject obj1, obj2, xObjDict, xObj, resObj;
 186   Ref r;
 187   GfxFontDict *gfxFontDict;
 188   GfxFont *font;
 189   int i;
 190
 191   // scan the fonts in this resource dictionary
 192   gfxFontDict = NULL;
 193   resDict->lookupNF("Font", &obj1);
 194   if (obj1.isRef()) {
 195     obj1.fetch(doc->getXRef(), &obj2);
 196     if (obj2.isDict()) {
 197       r = obj1.getRef();
 198       gfxFontDict = new GfxFontDict(doc->getXRef(), &r, obj2.getDict());
 199     }
 200     obj2.free();
 201   } else if (obj1.isDict()) {
 202     gfxFontDict = new GfxFontDict(doc->getXRef(), NULL, obj1.getDict());
 203   }
 204   if (gfxFontDict) {
 205     for (i = 0; i < gfxFontDict->getNumFonts(); ++i) {
 206       if ((font = gfxFontDict->getFont(i))) {
 207         scanFont(font, doc);
 208       }
 209     }
 210     delete gfxFontDict;
 211   }
 212   obj1.free();
 213
 214   // recursively scan any resource dictionaries in xObjects in this
 215   // resource dictionary
 216   resDict->lookup("XObject", &xObjDict);
 217   if (xObjDict.isDict()) {
 218     for (i = 0; i < xObjDict.dictGetLength(); ++i) {
 219       xObjDict.dictGetVal(i, &xObj);
 220       if (xObj.isStream()) {
 221         xObj.streamGetDict()->lookup("Resources", &resObj);
 222         if (resObj.isDict()) {
 223           scanFonts(resObj.getDict(), doc);
 224         }
 225         resObj.free();
 226       }
 227       xObj.free();
 228     }
 229   }
 230   xObjDict.free();
 231 }
 232
 233 static void scanFont(GfxFont *font, PDFDoc *doc) {
 234   Ref fontRef, embRef;
 235   xObject fontObj, toUnicodeObj;
 236   GString *name;
 237   GBool emb, subset, hasToUnicode;
 238   int i;
 239
 240   fontRef = *font->getID();
 241
 242   // check for an already-seen font
 243   for (i = 0; i < fontsLen; ++i) {
 244     if (fontRef.num == fonts[i].num && fontRef.gen == fonts[i].gen) {
 245       return;
 246     }
 247   }
 248
 249   // font name
 250   name = font->getOrigName();
 251
 252   // check for an embedded font
 253   if (font->getType() == fontType3) {
 254     emb = gTrue;
 255   } else {
 256     emb = font->getEmbeddedFontID(&embRef);
 257   }
 258
 259   // look for a ToUnicode map
 260   hasToUnicode = gFalse;
 261   if (doc->getXRef()->fetch(fontRef.num, fontRef.gen, &fontObj)->isDict()) {
 262     hasToUnicode = fontObj.dictLookup("ToUnicode", &toUnicodeObj)->isStream();
 263     toUnicodeObj.free();
 264   }
 265   fontObj.free();
 266
 267   // check for a font subset name: capital letters followed by a '+'
 268   // sign
 269   subset = gFalse;
 270   if (name) {
 271     for (i = 0; i < name->getLength(); ++i) {
 272       if (name->getChar(i) < 'A' || name->getChar(i) > 'Z') {
 273         break;
 274       }
 275     }
 276     subset = i > 0 && i < name->getLength() && name->getChar(i) == '+';
 277   }
 278
 279   // print the font info
 280   printf("%-36s %-17s %-3s %-3s %-3s",
 281          name ? name->getCString() : "[none]",
 282          fontTypeNames[font->getType()],
 283          emb ? "yes" : "no",
 284          subset ? "yes" : "no",
 285          hasToUnicode ? "yes" : "no");
 286   if (fontRef.gen >= 100000) {
 287     printf(" [none]\n");
 288   } else {
 289     printf(" %6d %2d\n", fontRef.num, fontRef.gen);
 290   }
 291
 292   // add this font to the list
 293   if (fontsLen == fontsSize) {
 294     fontsSize += 32;
 295     fonts = (Ref *)greallocn(fonts, fontsSize, sizeof(Ref));
 296   }
 297   fonts[fontsLen++] = *font->getID();
 298 }