man/mantohtml.c

   1 /*
   2  * "$Id: mantohtml.c 5142 2006-02-21 14:52:34Z mike $"
   3  *
   4  *   Man page to HTML conversion program.
   5  *
   6  *   Copyright 2004-2006 by Easy Software Products.
   7  *
   8  *   These coded instructions, statements, and computer programs are the
   9  *   property of Easy Software Products and are protected by Federal
  10  *   copyright law.  Distribution and use rights are outlined in the file
  11  *   "LICENSE.txt" which should have been included with this file.  If this
  12  *   file is missing or damaged please contact Easy Software Products
  13  *   at:
  14  *
  15  *       Attn: CUPS Licensing Information
  16  *       Easy Software Products
  17  *       44141 Airport View Drive, Suite 204
  18  *       Hollywood, Maryland 20636 USA
  19  *
  20  *       Voice: (301) 373-9600
  21  *       EMail: cups-info@cups.org
  22  *         WWW: http://www.cups.org
  23  *
  24  * Contents:
  25  *
  26  *   main()        - Convert a man page to HTML.
  27  *   putc_entity() - Put a single character, using entities as needed.
  28  *   strmove()     - Move characters within a string.
  29  */
  30
  31 /*
  32  * Include necessary headers.
  33  */
  34
  35 #include <cups/string.h>
  36 #include <stdlib.h>
  37 #include <unistd.h>
  38
  39
  40 /*
  41  * Local functions...
  42  */
  43
  44 static void     putc_entity(int ch, FILE *fp);
  45 static void     strmove(char *d, const char *s);
  46
  47
  48 /*
  49  * 'main()' - Convert a man page to HTML.
  50  */
  51
  52 int                                     /* O - Exit status */
  53 main(int  argc,                         /* I - Number of command-line args */
  54      char *argv[])                      /* I - Command-line arguments */
  55 {
  56   FILE          *infile,                /* Input file */
  57                 *outfile;               /* Output file */
  58   char          line[1024],             /* Line from file */
  59                 *lineptr,               /* Pointer into line */
  60                 *endptr,                /* Pointer to end of current */
  61                 endchar,                /* End character */
  62                 *paren,                 /* Pointer to parenthesis */
  63                 name[1024];             /* Man page name */
  64   int           section,                /* Man page section */
  65                 pre,                    /* Preformatted */
  66                 font,                   /* Current font */
  67                 blist,                  /* In a bullet list? */
  68                 list,                   /* In a list? */
  69                 linenum;                /* Current line number */
  70   const char    *post;                  /* Text to add after the current line */
  71   static const char                     /* Start/end tags for fonts */
  72         * const start_fonts[] = { "", "<b>", "<i>" },
  73         * const end_fonts[] = { "", "</b>", "</i>" };
  74
  75  /*
  76   * Check arguments...
  77   */
  78
  79   if (argc > 3)
  80   {
  81     fputs("Usage: mantohtml [filename.man [filename.html]]\n", stderr);
  82     return (1);
  83   }
  84
  85  /*
  86   * Open files as needed...
  87   */
  88
  89   if (argc > 1)
  90   {
  91     if ((infile = fopen(argv[1], "r")) == NULL)
  92     {
  93       perror(argv[1]);
  94       return (1);
  95     }
  96   }
  97   else
  98     infile = stdin;
  99
 100   if (argc > 2)
 101   {
 102     if ((outfile = fopen(argv[2], "w")) == NULL)
 103     {
 104       perror(argv[2]);
 105       fclose(infile);
 106       return (1);
 107     }
 108   }
 109   else
 110     outfile = stdout;
 111
 112  /*
 113   * Read from input and write the output...
 114   */
 115
 116   fputs("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0 Transitional//EN\" "
 117         "\"http://www.w3.org/TR/REC-html40/loose.dtd\">\n"
 118         "<html>\n"
 119         "<!-- SECTION: Man Pages -->\n"
 120         "<head>\n"
 121         "\t<style type='text/css'><!--\n"
 122         "\th1, h2, h3, p { font-family: sans-serif; text-align: justify; }\n"
 123         "\ttt, pre a:link, pre a:visited, tt a:link, tt a:visited { font-weight: bold; color: #7f0000; }\n"
 124         "\tpre { font-weight: bold; color: #7f0000; margin-left: 2em; }\n"
 125         "\th1.title, h2.title, h3.title { border-bottom: solid 2px #000000; }\n"
 126         "\t--></style>\n", outfile);
 127
 128   blist   = 0;
 129   font    = 0;
 130   list    = 0;
 131   linenum = 0;
 132   pre     = 0;
 133   post    = NULL;
 134   section = -1;
 135
 136   while (fgets(line, sizeof(line), infile))
 137   {
 138     linenum ++;
 139
 140     if (line[0] == '.')
 141     {
 142      /*
 143       * Strip leading whitespace...
 144       */
 145
 146       while (line[1] == ' ' || line[1] == '\t')
 147         strmove(line + 1, line + 2);
 148
 149      /*
 150       * Process man page commands...
 151       */
 152
 153       if (!strncmp(line, ".TH ", 4) && section < 0)
 154       {
 155        /*
 156         * Grab man page title...
 157         */
 158
 159         sscanf(line + 4, "%s%d", name, &section);
 160
 161         fprintf(outfile,
 162                 "\t<title>%s(%d)</title>\n"
 163                 "</head>\n"
 164                 "<body>\n"
 165                 "%s",
 166                 name, section, start_fonts[font]);
 167       }
 168       else if (section < 0)
 169         continue;
 170       else if (!strncmp(line, ".SH ", 4) || !strncmp(line, ".Sh ", 4))
 171       {
 172        /*
 173         * Grab heading...
 174         */
 175
 176         int first = 1;
 177
 178         fputs(end_fonts[font], outfile);
 179
 180         if (blist)
 181         {
 182           fputs("</li>\n</ul>\n", outfile);
 183           blist = 0;
 184         }
 185
 186         if (list)
 187         {
 188           if (list == 1)
 189             fputs("</dt>\n", outfile);
 190           else if (list)
 191             fputs("</dd>\n", outfile);
 192
 193           fputs("</dl>\n", outfile);
 194           list = 0;
 195         }
 196
 197         line[strlen(line) - 1] = '\0';  /* Strip LF */
 198
 199         if (line[2] == 'H')
 200           fputs("<h2><a name='", outfile);
 201         else
 202           fputs("<h3><a name='", outfile);
 203
 204         for (lineptr = line + 4; *lineptr; lineptr ++)
 205           if (*lineptr == '\"')
 206             continue;
 207           else if (*lineptr == ' ')
 208             putc_entity('_', outfile);
 209           else
 210             putc_entity(*lineptr, outfile);
 211
 212         fputs("'>", outfile);
 213
 214         for (lineptr = line + 4; *lineptr; lineptr ++)
 215           if (*lineptr == '\"')
 216             continue;
 217           else if (*lineptr == ' ')
 218           {
 219             putc_entity(' ', outfile);
 220
 221             first = 1;
 222           }
 223           else
 224           {
 225             if (first)
 226               putc_entity(*lineptr, outfile);
 227             else
 228               putc_entity(tolower(*lineptr), outfile);
 229
 230             first = 0;
 231           }
 232
 233         if (line[2] == 'H')
 234           fprintf(outfile, "</a></h2>\n%s", start_fonts[font]);
 235         else
 236           fprintf(outfile, "</a></h3>\n%s", start_fonts[font]);
 237       }
 238       else if (!strncmp(line, ".LP", 3) || !strncmp(line, ".PP", 3))
 239       {
 240        /*
 241         * New paragraph...
 242         */
 243
 244         fputs(end_fonts[font], outfile);
 245
 246         if (blist)
 247         {
 248           fputs("</li>\n</ul>\n", outfile);
 249           blist = 0;
 250         }
 251
 252         if (list)
 253         {
 254           if (list == 1)
 255             fputs("</dt>\n", outfile);
 256           else if (list)
 257             fputs("</dd>\n", outfile);
 258
 259           fputs("</dl>\n", outfile);
 260           list = 0;
 261         }
 262
 263         fputs("<p>", outfile);
 264         font = 0;
 265       }
 266       else if (!strncmp(line, ".TP ", 4))
 267       {
 268        /*
 269         * Grab list...
 270         */
 271
 272         fputs(end_fonts[font], outfile);
 273
 274         if (blist)
 275         {
 276           fputs("</li>\n</ul>\n", outfile);
 277           blist = 0;
 278         }
 279
 280         if (!list)
 281           fputs("<dl>\n", outfile);
 282         else if (list == 1)
 283           fputs("</dt>\n", outfile);
 284         else if (list)
 285           fputs("</dd>\n", outfile);
 286
 287         fputs("<dt>", outfile);
 288         list = 1;
 289         font = 0;
 290       }
 291       else if (!strncmp(line, ".br", 3))
 292       {
 293        /*
 294         * Grab line break...
 295         */
 296
 297         if (list == 1)
 298         {
 299           fputs("</dt>\n<dd>", outfile);
 300           list = 2;
 301         }
 302         else if (list)
 303           fputs("</dd>\n<dd>", outfile);
 304         else
 305           fputs("<br>\n", outfile);
 306       }
 307       else if (!strncmp(line, ".de ", 4))
 308       {
 309        /*
 310         * Define macro - ignore...
 311         */
 312
 313         while (fgets(line, sizeof(line), infile))
 314         {
 315           linenum ++;
 316
 317           if (!strncmp(line, "..", 2))
 318             break;
 319         }
 320       }
 321       else if (!strncmp(line, ".RS", 3))
 322       {
 323        /*
 324         * Indent...
 325         */
 326
 327         fputs("<div style='margin-left: 3em;'>\n", outfile);
 328       }
 329       else if (!strncmp(line, ".RE", 3))
 330       {
 331        /*
 332         * Unindent...
 333         */
 334
 335         fputs("</div>\n", outfile);
 336       }
 337       else if (!strncmp(line, ".ds ", 4) || !strncmp(line, ".rm ", 4) ||
 338                !strncmp(line, ".tr ", 4) || !strncmp(line, ".hy ", 4) ||
 339                !strncmp(line, ".IX ", 4) || !strncmp(line, ".PD", 3) ||
 340                !strncmp(line, ".Sp", 3))
 341       {
 342        /*
 343         * Ignore unused commands...
 344         */
 345       }
 346       else if (!strncmp(line, ".Vb", 3) || !strncmp(line, ".nf", 3))
 347       {
 348        /*
 349         * Start preformatted...
 350         */
 351
 352         pre = 1;
 353         fputs("<pre>\n", outfile);
 354       }
 355       else if (!strncmp(line, ".Ve", 3) || !strncmp(line, ".fi", 3))
 356       {
 357        /*
 358         * End preformatted...
 359         */
 360
 361         if (pre)
 362         {
 363           pre = 0;
 364           fputs("</pre>\n", outfile);
 365         }
 366       }
 367       else if (!strncmp(line, ".IP \\(bu", 8))
 368       {
 369        /*
 370         * Bullet list...
 371         */
 372
 373         if (blist)
 374           fputs("</li>\n", outfile);
 375         else
 376         {
 377           fputs("<ul>\n", outfile);
 378           blist = 1;
 379         }
 380
 381         fputs("<li>", outfile);
 382       }
 383       else if (!strncmp(line, ".IP ", 4))
 384       {
 385        /*
 386         * Indented paragraph...
 387         */
 388
 389         if (blist)
 390         {
 391           fputs("</li>\n</ul>\n", outfile);
 392           blist = 0;
 393         }
 394
 395         fputs("<p style='margin-left: 3em;'>", outfile);
 396
 397         for (lineptr = line + 4; isspace(*lineptr); lineptr ++);
 398
 399         if (*lineptr == '\"')
 400         {
 401           strmove(line, lineptr + 1);
 402
 403           if ((lineptr = strchr(line, '\"')) != NULL)
 404             *lineptr = '\0';
 405         }
 406         else
 407         {
 408           strmove(line, lineptr);
 409
 410           if ((lineptr = strchr(line, ' ')) != NULL)
 411             *lineptr = '\0';
 412         }
 413
 414        /*
 415         * Process the text as if it was in-line...
 416         */
 417
 418         post = "\n<br />\n<br />";
 419         goto process_text;
 420       }
 421       else if (!strncmp(line, ".\\}", 3))
 422       {
 423        /*
 424         * Ignore close block...
 425         */
 426       }
 427       else if (!strncmp(line, ".ie", 3) || !strncmp(line, ".if", 3) ||
 428                !strncmp(line, ".el", 3))
 429       {
 430        /*
 431         * If/else - ignore...
 432         */
 433
 434         if (strchr(line, '{') != NULL)
 435         {
 436          /*
 437           * Skip whole block...
 438           */
 439
 440           while (fgets(line, sizeof(line), infile))
 441           {
 442             linenum ++;
 443
 444             if (strchr(line, '}') != NULL)
 445               break;
 446           }
 447         }
 448       }
 449 #if 0
 450       else if (!strncmp(line, ". ", 4))
 451       {
 452        /*
 453         * Grab ...
 454         */
 455       }
 456 #endif /* 0 */
 457       else if (!strncmp(line, ".B ", 3))
 458       {
 459        /*
 460         * Grab bold text...
 461         */
 462
 463         fprintf(outfile, "%s<b>%s</b>%s", end_fonts[font], line + 3,
 464                 start_fonts[font]);
 465       }
 466       else if (!strncmp(line, ".I ", 3))
 467       {
 468        /*
 469         * Grab italic text...
 470         */
 471
 472         fprintf(outfile, "%s<i>%s</i>%s", end_fonts[font], line + 3,
 473                 start_fonts[font]);
 474       }
 475       else if (strncmp(line, ".\\\"", 3))
 476       {
 477        /*
 478         * Unknown...
 479         */
 480
 481         if ((lineptr = strchr(line, ' ')) != NULL)
 482           *lineptr = '\0';
 483         else if ((lineptr = strchr(line, '\n')) != NULL)
 484           *lineptr = '\0';
 485
 486         fprintf(stderr, "mantohtml: Unknown man page command \'%s\' on line %d!\n",
 487                 line, linenum);
 488       }
 489
 490      /*
 491       * Skip continuation lines...
 492       */
 493
 494       lineptr = line + strlen(line) - 2;
 495       if (lineptr >= line && *lineptr == '\\')
 496       {
 497         while (fgets(line, sizeof(line), infile))
 498         {
 499           linenum ++;
 500           lineptr = line + strlen(line) - 2;
 501
 502           if (lineptr < line || *lineptr != '\\')
 503             break;
 504         }
 505       }
 506     }
 507     else
 508     {
 509      /*
 510       * Process man page text...
 511       */
 512
 513 process_text:
 514
 515       for (lineptr = line; *lineptr; lineptr ++)
 516       {
 517         if (!strncmp(lineptr, "http://", 7))
 518         {
 519          /*
 520           * Embed URL...
 521           */
 522
 523           for (endptr = lineptr + 7;
 524                *endptr && !isspace(*endptr & 255);
 525                endptr ++);
 526
 527           endchar = *endptr;
 528           *endptr = '\0';
 529
 530           fprintf(outfile, "<a href='%s'>%s</a>", lineptr, lineptr);
 531           *endptr = endchar;
 532           lineptr = endptr - 1;
 533         }
 534         else if (!strncmp(lineptr, "\\fI", 3) &&
 535                  (endptr = strstr(lineptr, "\\fR")) != NULL &&
 536                  (paren = strchr(lineptr, '(')) != NULL &&
 537                  paren < endptr)
 538         {
 539          /*
 540           * Link to man page?
 541           */
 542
 543           char  manfile[1024],          /* Man page filename */
 544                 manurl[1024];           /* Man page URL */
 545
 546
 547          /*
 548           * See if the man file is available locally...
 549           */
 550
 551           lineptr += 3;
 552           endchar = *paren;
 553           *paren  = '\0';
 554
 555           snprintf(manfile, sizeof(manfile), "%s.man", lineptr);
 556           snprintf(manurl, sizeof(manurl), "man-%s.html?TOPIC=Man+Pages",
 557                    lineptr);
 558
 559           *paren  = endchar;
 560           endchar = *endptr;
 561           *endptr = '\0';
 562
 563           if (access(manfile, 0))
 564           {
 565            /*
 566             * Not a local man page, just do it italic...
 567             */
 568
 569             fputs("<i>", outfile);
 570             while (*lineptr)
 571               putc_entity(*lineptr++, outfile);
 572             fputs("</i>", outfile);
 573           }
 574           else
 575           {
 576            /*
 577             * Local man page, do a link...
 578             */
 579
 580             fprintf(outfile, "<a href='%s'>", manurl);
 581             while (*lineptr)
 582               putc_entity(*lineptr++, outfile);
 583             fputs("</a>", outfile);
 584           }
 585
 586           *endptr = endchar;
 587           lineptr = endptr + 2;
 588         }
 589         else if (*lineptr == '\\')
 590         {
 591           lineptr ++;
 592           if (!*lineptr)
 593             break;
 594           else if (isdigit(lineptr[0]) && isdigit(lineptr[1]) &&
 595                    isdigit(lineptr[2]))
 596           {
 597             fprintf(outfile, "&#%d;", ((lineptr[0] - '0') * 8 +
 598                                        lineptr[1] - '0') * 8 +
 599                                       lineptr[2] - '0');
 600             lineptr += 2;
 601           }
 602           else if (*lineptr == '&')
 603             continue;
 604           else if (*lineptr == 's')
 605           {
 606             while (lineptr[1] == '-' || isdigit(lineptr[1]))
 607               lineptr ++;
 608           }
 609           else if (*lineptr == '*')
 610           {
 611             lineptr += 2;
 612           }
 613           else if (*lineptr != 'f')
 614             putc_entity(*lineptr, outfile);
 615           else
 616           {
 617             lineptr ++;
 618             if (!*lineptr)
 619               break;
 620             else
 621             {
 622               fputs(end_fonts[font], outfile);
 623
 624               switch (*lineptr)
 625               {
 626                 default : /* Regular */
 627                     font = 0;
 628                     break;
 629                 case 'B' : /* Bold */
 630                 case 'b' :
 631                     font = 1;
 632                     break;
 633                 case 'I' : /* Italic */
 634                 case 'i' :
 635                     font = 2;
 636                     break;
 637               }
 638
 639               fputs(start_fonts[font], outfile);
 640             }
 641           }
 642         }
 643         else
 644           putc_entity(*lineptr, outfile);
 645       }
 646
 647       if (post)
 648       {
 649         fputs(post, outfile);
 650         post = NULL;
 651       }
 652     }
 653   }
 654
 655   fprintf(outfile, "%s\n", end_fonts[font]);
 656
 657   if (blist)
 658   {
 659     fputs("</li>\n</ul>\n", outfile);
 660     blist = 0;
 661   }
 662
 663   if (list)
 664   {
 665     if (list == 1)
 666       fputs("</dt>\n", outfile);
 667     else if (list)
 668       fputs("</dd>\n", outfile);
 669
 670     fputs("</dl>\n", outfile);
 671     list = 0;
 672   }
 673
 674   fputs("</body>\n"
 675         "</html>\n", outfile);
 676
 677  /*
 678   * Close files...
 679   */
 680
 681   if (infile != stdin)
 682     fclose(infile);
 683
 684   if (outfile != stdout)
 685     fclose(outfile);
 686
 687  /*
 688   * Return with no errors...
 689   */
 690
 691   return (0);
 692 }
 693
 694
 695 /*
 696  * 'putc_entity()' - Put a single character, using entities as needed.
 697  */
 698
 699 static void
 700 putc_entity(int  ch,                    /* I - Character */
 701             FILE *fp)                   /* I - File */
 702 {
 703   if (ch == '&')
 704     fputs("&amp;", fp);
 705   else if (ch == '<')
 706     fputs("&lt;", fp);
 707   else
 708     putc(ch, fp);
 709 }
 710
 711
 712 /*
 713  * 'strmove()' - Move characters within a string.
 714  */
 715
 716 static void
 717 strmove(char       *d,                  /* I - Destination */
 718         const char *s)                  /* I - Source */
 719 {
 720   while (*s)
 721     *d++ = *s++;
 722
 723   *d = '\0';
 724 }
 725
 726
 727 /*
 728  * End of "$Id: mantohtml.c 5142 2006-02-21 14:52:34Z mike $".
 729  */