arospdf/xpdf/pdfinfo.cc

   1 //========================================================================
   2 //
   3 // pdfinfo.cc
   4 //
   5 // Copyright 1998-2003 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 <time.h>
  15 #include <math.h>
  16 #include "parseargs.h"
  17 #include "GString.h"
  18 #include "gmem.h"
  19 #include "GlobalParams.h"
  20 #include "Object.h"
  21 #include "Stream.h"
  22 #include "Array.h"
  23 #include "Dict.h"
  24 #include "XRef.h"
  25 #include "Catalog.h"
  26 #include "Page.h"
  27 #include "PDFDoc.h"
  28 #include "CharTypes.h"
  29 #include "UnicodeMap.h"
  30 #include "PDFDocEncoding.h"
  31 #include "Error.h"
  32 #include "config.h"
  33
  34 static void printInfoString(Dict *infoDict, char *key, char *text,
  35                             UnicodeMap *uMap);
  36 static void printInfoDate(Dict *infoDict, char *key, char *text);
  37 static void printBox(char *text, PDFRectangle *box);
  38
  39 static int firstPage = 1;
  40 static int lastPage = 0;
  41 static GBool printBoxes = gFalse;
  42 static GBool printMetadata = gFalse;
  43 static char textEncName[128] = "";
  44 static char ownerPassword[33] = "\001";
  45 static char userPassword[33] = "\001";
  46 static char cfgFileName[256] = "";
  47 static GBool printVersion = gFalse;
  48 static GBool printHelp = gFalse;
  49
  50 static ArgDesc argDesc[] = {
  51   {"-f",      argInt,      &firstPage,        0,
  52    "first page to convert"},
  53   {"-l",      argInt,      &lastPage,         0,
  54    "last page to convert"},
  55   {"-box",    argFlag,     &printBoxes,       0,
  56    "print the page bounding boxes"},
  57   {"-meta",   argFlag,     &printMetadata,    0,
  58    "print the document metadata (XML)"},
  59   {"-enc",    argString,   textEncName,    sizeof(textEncName),
  60    "output text encoding name"},
  61   {"-opw",    argString,   ownerPassword,  sizeof(ownerPassword),
  62    "owner password (for encrypted files)"},
  63   {"-upw",    argString,   userPassword,   sizeof(userPassword),
  64    "user password (for encrypted files)"},
  65   {"-cfg",        argString,      cfgFileName,    sizeof(cfgFileName),
  66    "configuration file to use in place of .xpdfrc"},
  67   {"-v",      argFlag,     &printVersion,  0,
  68    "print copyright and version info"},
  69   {"-h",      argFlag,     &printHelp,     0,
  70    "print usage information"},
  71   {"-help",   argFlag,     &printHelp,     0,
  72    "print usage information"},
  73   {"--help",  argFlag,     &printHelp,     0,
  74    "print usage information"},
  75   {"-?",      argFlag,     &printHelp,     0,
  76    "print usage information"},
  77   {NULL}
  78 };
  79
  80 int main(int argc, char *argv[]) {
  81   PDFDoc *doc;
  82   GString *fileName;
  83   GString *ownerPW, *userPW;
  84   UnicodeMap *uMap;
  85   Page *page;
  86   xObject info;
  87   char buf[256];
  88   double w, h, wISO, hISO;
  89   FILE *f;
  90   GString *metadata;
  91   GBool ok;
  92   int exitCode;
  93   int pg, i;
  94   GBool multiPage;
  95
  96   exitCode = 99;
  97
  98   // parse args
  99   ok = parseArgs(argDesc, &argc, argv);
 100   if (!ok || argc != 2 || printVersion || printHelp) {
 101     fprintf(stderr, "pdfinfo version %s\n", xpdfVersion);
 102     fprintf(stderr, "%s\n", xpdfCopyright);
 103     if (!printVersion) {
 104       printUsage("pdfinfo", "<PDF-file>", argDesc);
 105     }
 106     goto err0;
 107   }
 108   fileName = new GString(argv[1]);
 109
 110   // read config file
 111   globalParams = new GlobalParams(cfgFileName);
 112   if (textEncName[0]) {
 113     globalParams->setTextEncoding(textEncName);
 114   }
 115
 116   // get mapping to output encoding
 117   if (!(uMap = globalParams->getTextEncoding())) {
 118     error(-1, "Couldn't get text encoding");
 119     delete fileName;
 120     goto err1;
 121   }
 122
 123   // open PDF file
 124   if (ownerPassword[0] != '\001') {
 125     ownerPW = new GString(ownerPassword);
 126   } else {
 127     ownerPW = NULL;
 128   }
 129   if (userPassword[0] != '\001') {
 130     userPW = new GString(userPassword);
 131   } else {
 132     userPW = NULL;
 133   }
 134   doc = new PDFDoc(fileName, ownerPW, userPW);
 135   if (userPW) {
 136     delete userPW;
 137   }
 138   if (ownerPW) {
 139     delete ownerPW;
 140   }
 141   if (!doc->isOk()) {
 142     exitCode = 1;
 143     goto err2;
 144   }
 145
 146   // get page range
 147   if (firstPage < 1) {
 148     firstPage = 1;
 149   }
 150   if (lastPage == 0) {
 151     multiPage = gFalse;
 152     lastPage = 1;
 153   } else {
 154     multiPage = gTrue;
 155   }
 156   if (lastPage < 1 || lastPage > doc->getNumPages()) {
 157     lastPage = doc->getNumPages();
 158   }
 159
 160   // print doc info
 161   doc->getDocInfo(&info);
 162   if (info.isDict()) {
 163     printInfoString(info.getDict(), "Title",        "Title:          ", uMap);
 164     printInfoString(info.getDict(), "Subject",      "Subject:        ", uMap);
 165     printInfoString(info.getDict(), "Keywords",     "Keywords:       ", uMap);
 166     printInfoString(info.getDict(), "Author",       "Author:         ", uMap);
 167     printInfoString(info.getDict(), "Creator",      "Creator:        ", uMap);
 168     printInfoString(info.getDict(), "Producer",     "Producer:       ", uMap);
 169     printInfoDate(info.getDict(),   "CreationDate", "CreationDate:   ");
 170     printInfoDate(info.getDict(),   "ModDate",      "ModDate:        ");
 171   }
 172   info.free();
 173
 174   // print tagging info
 175   printf("Tagged:         %s\n",
 176          doc->getStructTreeRoot()->isDict() ? "yes" : "no");
 177
 178   // print page count
 179   printf("Pages:          %d\n", doc->getNumPages());
 180
 181   // print encryption info
 182   printf("Encrypted:      ");
 183   if (doc->isEncrypted()) {
 184     printf("yes (print:%s copy:%s change:%s addNotes:%s)\n",
 185            doc->okToPrint(gTrue) ? "yes" : "no",
 186            doc->okToCopy(gTrue) ? "yes" : "no",
 187            doc->okToChange(gTrue) ? "yes" : "no",
 188            doc->okToAddNotes(gTrue) ? "yes" : "no");
 189   } else {
 190     printf("no\n");
 191   }
 192
 193   // print page size
 194   for (pg = firstPage; pg <= lastPage; ++pg) {
 195     w = doc->getPageCropWidth(pg);
 196     h = doc->getPageCropHeight(pg);
 197     if (multiPage) {
 198       printf("Page %4d size: %g x %g pts", pg, w, h);
 199     } else {
 200       printf("Page size:      %g x %g pts", w, h);
 201     }
 202     if ((fabs(w - 612) < 0.1 && fabs(h - 792) < 0.1) ||
 203         (fabs(w - 792) < 0.1 && fabs(h - 612) < 0.1)) {
 204       printf(" (letter)");
 205     } else {
 206       hISO = sqrt(sqrt(2.0)) * 7200 / 2.54;
 207       wISO = hISO / sqrt(2.0);
 208       for (i = 0; i <= 6; ++i) {
 209         if ((fabs(w - wISO) < 1 && fabs(h - hISO) < 1) ||
 210             (fabs(w - hISO) < 1 && fabs(h - wISO) < 1)) {
 211           printf(" (A%d)", i);
 212           break;
 213         }
 214         hISO = wISO;
 215         wISO /= sqrt(2.0);
 216       }
 217     }
 218     printf("\n");
 219   }
 220
 221   // print the boxes
 222   if (printBoxes) {
 223     if (multiPage) {
 224       for (pg = firstPage; pg <= lastPage; ++pg) {
 225         page = doc->getCatalog()->getPage(pg);
 226         sprintf(buf, "Page %4d MediaBox: ", pg);
 227         printBox(buf, page->getMediaBox());
 228         sprintf(buf, "Page %4d CropBox:  ", pg);
 229         printBox(buf, page->getCropBox());
 230         sprintf(buf, "Page %4d BleedBox: ", pg);
 231         printBox(buf, page->getBleedBox());
 232         sprintf(buf, "Page %4d TrimBox:  ", pg);
 233         printBox(buf, page->getTrimBox());
 234         sprintf(buf, "Page %4d ArtBox:   ", pg);
 235         printBox(buf, page->getArtBox());
 236       }
 237     } else {
 238       page = doc->getCatalog()->getPage(firstPage);
 239       printBox("MediaBox:       ", page->getMediaBox());
 240       printBox("CropBox:        ", page->getCropBox());
 241       printBox("BleedBox:       ", page->getBleedBox());
 242       printBox("TrimBox:        ", page->getTrimBox());
 243       printBox("ArtBox:         ", page->getArtBox());
 244     }
 245   }
 246
 247   // print file size
 248 #ifdef VMS
 249   f = fopen(fileName->getCString(), "rb", "ctx=stm");
 250 #else
 251   f = fopen(fileName->getCString(), "rb");
 252 #endif
 253   if (f) {
 254 #if HAVE_FSEEKO
 255     fseeko(f, 0, SEEK_END);
 256     printf("File size:      %u bytes\n", (Guint)ftello(f));
 257 #elif HAVE_FSEEK64
 258     fseek64(f, 0, SEEK_END);
 259     printf("File size:      %u bytes\n", (Guint)ftell64(f));
 260 #else
 261     fseek(f, 0, SEEK_END);
 262     printf("File size:      %d bytes\n", (int)ftell(f));
 263 #endif
 264     fclose(f);
 265   }
 266
 267   // print linearization info
 268   printf("Optimized:      %s\n", doc->isLinearized() ? "yes" : "no");
 269
 270   // print PDF version
 271   printf("PDF version:    %.1f\n", doc->getPDFVersion());
 272
 273   // print the metadata
 274   if (printMetadata && (metadata = doc->readMetadata())) {
 275     fputs("Metadata:\n", stdout);
 276     fputs(metadata->getCString(), stdout);
 277     fputc('\n', stdout);
 278     delete metadata;
 279   }
 280
 281   exitCode = 0;
 282
 283   // clean up
 284  err2:
 285   uMap->decRefCnt();
 286   delete doc;
 287  err1:
 288   delete globalParams;
 289  err0:
 290
 291   // check for memory leaks
 292   xObject::memCheck(stderr);
 293   gMemReport(stderr);
 294
 295   return exitCode;
 296 }
 297
 298 static void printInfoString(Dict *infoDict, char *key, char *text,
 299                             UnicodeMap *uMap) {
 300   xObject obj;
 301   GString *s1;
 302   GBool isUnicode;
 303   Unicode u;
 304   char buf[8];
 305   int i, n;
 306
 307   if (infoDict->lookup(key, &obj)->isString()) {
 308     fputs(text, stdout);
 309     s1 = obj.getString();
 310     if ((s1->getChar(0) & 0xff) == 0xfe &&
 311         (s1->getChar(1) & 0xff) == 0xff) {
 312       isUnicode = gTrue;
 313       i = 2;
 314     } else {
 315       isUnicode = gFalse;
 316       i = 0;
 317     }
 318     while (i < obj.getString()->getLength()) {
 319       if (isUnicode) {
 320         u = ((s1->getChar(i) & 0xff) << 8) |
 321             (s1->getChar(i+1) & 0xff);
 322         i += 2;
 323       } else {
 324         u = pdfDocEncoding[s1->getChar(i) & 0xff];
 325         ++i;
 326       }
 327       n = uMap->mapUnicode(u, buf, sizeof(buf));
 328       fwrite(buf, 1, n, stdout);
 329     }
 330     fputc('\n', stdout);
 331   }
 332   obj.free();
 333 }
 334
 335 static void printInfoDate(Dict *infoDict, char *key, char *text) {
 336   xObject obj;
 337   char *s;
 338   int year, mon, day, hour, min, sec, n;
 339   struct tm tmStruct;
 340   char buf[256];
 341
 342   if (infoDict->lookup(key, &obj)->isString()) {
 343     fputs(text, stdout);
 344     s = obj.getString()->getCString();
 345     if (s[0] == 'D' && s[1] == ':') {
 346       s += 2;
 347     }
 348     if ((n = sscanf(s, "%4d%2d%2d%2d%2d%2d",
 349                     &year, &mon, &day, &hour, &min, &sec)) >= 1) {
 350       switch (n) {
 351       case 1: mon = 1;
 352       case 2: day = 1;
 353       case 3: hour = 0;
 354       case 4: min = 0;
 355       case 5: sec = 0;
 356       }
 357       tmStruct.tm_year = year - 1900;
 358       tmStruct.tm_mon = mon - 1;
 359       tmStruct.tm_mday = day;
 360       tmStruct.tm_hour = hour;
 361       tmStruct.tm_min = min;
 362       tmStruct.tm_sec = sec;
 363       tmStruct.tm_wday = -1;
 364       tmStruct.tm_yday = -1;
 365       tmStruct.tm_isdst = -1;
 366       // compute the tm_wday and tm_yday fields
 367       if (mktime(&tmStruct) != (time_t)-1 &&
 368           strftime(buf, sizeof(buf), "%c", &tmStruct)) {
 369         fputs(buf, stdout);
 370       } else {
 371         fputs(s, stdout);
 372       }
 373     } else {
 374       fputs(s, stdout);
 375     }
 376
 377
 378
 379     fputc('\n', stdout);
 380   }
 381   obj.free();
 382 }
 383
 384 static void printBox(char *text, PDFRectangle *box) {
 385   printf("%s%8.2f %8.2f %8.2f %8.2f\n",
 386          text, box->x1, box->y1, box->x2, box->y2);
 387 }