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