man/mantohtml.c

   1 /*
   2  * "$Id: mantohtml.c 7720 2008-07-11 22:46:21Z mike $"
   3  *
   4  *   Man page to HTML conversion program.
   5  *
   6  *   Copyright 2007-2009 by Apple Inc.
   7  *   Copyright 2004-2006 by Easy Software Products.
   8  *
   9  *   These coded instructions, statements, and computer programs are the
  10  *   property of Apple Inc. and are protected by Federal copyright
  11  *   law.  Distribution and use rights are outlined in the file "LICENSE.txt"
  12  *   which should have been included with this file.  If this file is
  13  *   file is missing or damaged, see the license at "http://www.cups.org/".
  14  *
  15  * Contents:
  16  *
  17  *   main()        - Convert a man page to HTML.
  18  *   putc_entity() - Put a single character, using entities as needed.
  19  *   strmove()     - Move characters within a string.
  20  */
  21
  22 /*
  23  * Include necessary headers.
  24  */
  25
  26 #include <cups/string.h>
  27 #include <stdlib.h>
  28 #include <unistd.h>
  29
  30
  31 /*
  32  * Local functions...
  33  */
  34
  35 static void     putc_entity(int ch, FILE *fp);
  36 static void     strmove(char *d, const char *s);
  37
  38
  39 /*
  40  * 'main()' - Convert a man page to HTML.
  41  */
  42
  43 int                                     /* O - Exit status */
  44 main(int  argc,                         /* I - Number of command-line args */
  45      char *argv[])                      /* I - Command-line arguments */
  46 {
  47   FILE          *infile,                /* Input file */
  48                 *outfile;               /* Output file */
  49   char          line[1024],             /* Line from file */
  50                 *lineptr,               /* Pointer into line */
  51                 *endptr,                /* Pointer to end of current */
  52                 endchar,                /* End character */
  53                 *paren,                 /* Pointer to parenthesis */
  54                 name[1024];             /* Man page name */
  55   int           section,                /* Man page section */
  56                 pre,                    /* Preformatted */
  57                 font,                   /* Current font */
  58                 blist,                  /* In a bullet list? */
  59                 list,                   /* In a list? */
  60                 linenum;                /* Current line number */
  61   const char    *post;                  /* Text to add after the current line */
  62   static const char                     /* Start/end tags for fonts */
  63         * const start_fonts[] = { "", "<b>", "<i>" },
  64         * const end_fonts[] = { "", "</b>", "</i>" };
  65
  66  /*
  67   * Check arguments...
  68   */
  69
  70   if (argc > 3)
  71   {
  72     fputs("Usage: mantohtml [filename.man [filename.html]]\n", stderr);
  73     return (1);
  74   }
  75
  76  /*
  77   * Open files as needed...
  78   */
  79
  80   if (argc > 1)
  81   {
  82     if ((infile = fopen(argv[1], "r")) == NULL)
  83     {
  84       perror(argv[1]);
  85       return (1);
  86     }
  87   }
  88   else
  89     infile = stdin;
  90
  91   if (argc > 2)
  92   {
  93     if ((outfile = fopen(argv[2], "w")) == NULL)
  94     {
  95       perror(argv[2]);
  96       fclose(infile);
  97       return (1);
  98     }
  99   }
 100   else
 101     outfile = stdout;
 102
 103  /*
 104   * Read from input and write the output...
 105   */
 106
 107   fputs("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" "
 108         "\"http://www.w3.org/TR/html4/loose.dtd\">\n"
 109         "<html>\n"
 110         "<!-- SECTION: Man Pages -->\n"
 111         "<head>\n"
 112         "\t<link rel=\"stylesheet\" type=\"text/css\" "
 113         "href=\"../cups-printable.css\">\n", outfile);
 114
 115   blist   = 0;
 116   font    = 0;
 117   list    = 0;
 118   linenum = 0;
 119   pre     = 0;
 120   post    = NULL;
 121   section = -1;
 122
 123   while (fgets(line, sizeof(line), infile))
 124   {
 125     linenum ++;
 126
 127     if (line[0] == '.')
 128     {
 129      /*
 130       * Strip leading whitespace...
 131       */
 132
 133       while (line[1] == ' ' || line[1] == '\t')
 134         strmove(line + 1, line + 2);
 135
 136      /*
 137       * Process man page commands...
 138       */
 139
 140       if (!strncmp(line, ".TH ", 4) && section < 0)
 141       {
 142        /*
 143         * Grab man page title...
 144         */
 145
 146         sscanf(line + 4, "%s%d", name, &section);
 147
 148         fprintf(outfile,
 149                 "\t<title>%s(%d)</title>\n"
 150                 "</head>\n"
 151                 "<body>\n"
 152                 "<h1 class=\"title\">%s(%d)</h1>\n"
 153                 "%s",
 154                 name, section, name, section, start_fonts[font]);
 155       }
 156       else if (section < 0)
 157         continue;
 158       else if (!strncmp(line, ".SH ", 4) || !strncmp(line, ".SS ", 4))
 159       {
 160        /*
 161         * Grab heading...
 162         */
 163
 164         int first = 1;
 165
 166         fputs(end_fonts[font], outfile);
 167
 168         if (blist)
 169         {
 170           fputs("</li>\n</ul>\n", outfile);
 171           blist = 0;
 172         }
 173
 174         if (list)
 175         {
 176           if (list == 1)
 177             fputs("</dt>\n", outfile);
 178           else if (list)
 179             fputs("</dd>\n", outfile);
 180
 181           fputs("</dl>\n", outfile);
 182           list = 0;
 183         }
 184
 185         line[strlen(line) - 1] = '\0';  /* Strip LF */
 186
 187         if (line[2] == 'H')
 188           fputs("<h2 class=\"title\"><a name=\"", outfile);
 189         else
 190           fputs("<h3><a name=\"", outfile);
 191
 192         for (lineptr = line + 4; *lineptr; lineptr ++)
 193           if (*lineptr == '\"')
 194             continue;
 195           else if (*lineptr == ' ')
 196             putc_entity('_', outfile);
 197           else
 198             putc_entity(*lineptr, outfile);
 199
 200         fputs("\">", outfile);
 201
 202         for (lineptr = line + 4; *lineptr; lineptr ++)
 203           if (*lineptr == '\"')
 204             continue;
 205           else if (*lineptr == ' ')
 206           {
 207             putc_entity(' ', outfile);
 208
 209             first = 1;
 210           }
 211           else
 212           {
 213             if (first)
 214               putc_entity(*lineptr, outfile);
 215             else
 216               putc_entity(tolower(*lineptr), outfile);
 217
 218             first = 0;
 219           }
 220
 221         if (line[2] == 'H')
 222           fprintf(outfile, "</a></h2>\n%s", start_fonts[font]);
 223         else
 224           fprintf(outfile, "</a></h3>\n%s", start_fonts[font]);
 225       }
 226       else if (!strncmp(line, ".LP", 3) || !strncmp(line, ".PP", 3))
 227       {
 228        /*
 229         * New paragraph...
 230         */
 231
 232         fputs(end_fonts[font], outfile);
 233
 234         if (blist)
 235         {
 236           fputs("</li>\n</ul>\n", outfile);
 237           blist = 0;
 238         }
 239
 240         if (list)
 241         {
 242           if (list == 1)
 243             fputs("</dt>\n", outfile);
 244           else if (list)
 245             fputs("</dd>\n", outfile);
 246
 247           fputs("</dl>\n", outfile);
 248           list = 0;
 249         }
 250
 251         fputs("<p>", outfile);
 252         font = 0;
 253       }
 254       else if (!strncmp(line, ".TP ", 4))
 255       {
 256        /*
 257         * Grab list...
 258         */
 259
 260         fputs(end_fonts[font], outfile);
 261
 262         if (blist)
 263         {
 264           fputs("</li>\n</ul>\n", outfile);
 265           blist = 0;
 266         }
 267
 268         if (!list)
 269           fputs("<dl>\n", outfile);
 270         else if (list == 1)
 271           fputs("</dt>\n", outfile);
 272         else if (list)
 273           fputs("</dd>\n", outfile);
 274
 275         fputs("<dt>", outfile);
 276         list = 1;
 277         font = 0;
 278       }
 279       else if (!strncmp(line, ".br", 3))
 280       {
 281        /*
 282         * Grab line break...
 283         */
 284
 285         if (list == 1)
 286         {
 287           fputs("</dt>\n<dd>", outfile);
 288           list = 2;
 289         }
 290         else if (list)
 291           fputs("</dd>\n<dd>", outfile);
 292         else
 293           fputs("<br>\n", outfile);
 294       }
 295       else if (!strncmp(line, ".de ", 4))
 296       {
 297        /*
 298         * Define macro - ignore...
 299         */
 300
 301         while (fgets(line, sizeof(line), infile))
 302         {
 303           linenum ++;
 304
 305           if (!strncmp(line, "..", 2))
 306             break;
 307         }
 308       }
 309       else if (!strncmp(line, ".RS", 3))
 310       {
 311        /*
 312         * Indent...
 313         */
 314
 315         fputs("<div style='margin-left: 3em;'>\n", outfile);
 316       }
 317       else if (!strncmp(line, ".RE", 3))
 318       {
 319        /*
 320         * Unindent...
 321         */
 322
 323         fputs("</div>\n", outfile);
 324       }
 325       else if (!strncmp(line, ".ds ", 4) || !strncmp(line, ".rm ", 4) ||
 326                !strncmp(line, ".tr ", 4) || !strncmp(line, ".hy ", 4) ||
 327                !strncmp(line, ".IX ", 4) || !strncmp(line, ".PD", 3) ||
 328                !strncmp(line, ".Sp", 3))
 329       {
 330        /*
 331         * Ignore unused commands...
 332         */
 333       }
 334       else if (!strncmp(line, ".Vb", 3) || !strncmp(line, ".nf", 3))
 335       {
 336        /*
 337         * Start preformatted...
 338         */
 339
 340         pre = 1;
 341         fputs("<pre>\n", outfile);
 342       }
 343       else if (!strncmp(line, ".Ve", 3) || !strncmp(line, ".fi", 3))
 344       {
 345        /*
 346         * End preformatted...
 347         */
 348
 349         if (pre)
 350         {
 351           pre = 0;
 352           fputs("</pre>\n", outfile);
 353         }
 354       }
 355       else if (!strncmp(line, ".IP \\(bu", 8))
 356       {
 357        /*
 358         * Bullet list...
 359         */
 360
 361         if (blist)
 362           fputs("</li>\n", outfile);
 363         else
 364         {
 365           fputs("<ul>\n", outfile);
 366           blist = 1;
 367         }
 368
 369         fputs("<li>", outfile);
 370       }
 371       else if (!strncmp(line, ".IP ", 4))
 372       {
 373        /*
 374         * Indented paragraph...
 375         */
 376
 377         if (blist)
 378         {
 379           fputs("</li>\n</ul>\n", outfile);
 380           blist = 0;
 381         }
 382
 383         fputs("<p style='margin-left: 3em;'>", outfile);
 384
 385         for (lineptr = line + 4; isspace(*lineptr); lineptr ++);
 386
 387         if (*lineptr == '\"')
 388         {
 389           strmove(line, lineptr + 1);
 390
 391           if ((lineptr = strchr(line, '\"')) != NULL)
 392             *lineptr = '\0';
 393         }
 394         else
 395         {
 396           strmove(line, lineptr);
 397
 398           if ((lineptr = strchr(line, ' ')) != NULL)
 399             *lineptr = '\0';
 400         }
 401
 402        /*
 403         * Process the text as if it was in-line...
 404         */
 405
 406         post = "\n<br />\n<br />";
 407         goto process_text;
 408       }
 409       else if (!strncmp(line, ".\\}", 3))
 410       {
 411        /*
 412         * Ignore close block...
 413         */
 414       }
 415       else if (!strncmp(line, ".ie", 3) || !strncmp(line, ".if", 3) ||
 416                !strncmp(line, ".el", 3))
 417       {
 418        /*
 419         * If/else - ignore...
 420         */
 421
 422         if (strchr(line, '{') != NULL)
 423         {
 424          /*
 425           * Skip whole block...
 426           */
 427
 428           while (fgets(line, sizeof(line), infile))
 429           {
 430             linenum ++;
 431
 432             if (strchr(line, '}') != NULL)
 433               break;
 434           }
 435         }
 436       }
 437 #if 0
 438       else if (!strncmp(line, ". ", 4))
 439       {
 440        /*
 441         * Grab ...
 442         */
 443       }
 444 #endif /* 0 */
 445       else if (!strncmp(line, ".B ", 3))
 446       {
 447        /*
 448         * Grab bold text...
 449         */
 450
 451         fprintf(outfile, "%s<b>%s</b>%s", end_fonts[font], line + 3,
 452                 start_fonts[font]);
 453       }
 454       else if (!strncmp(line, ".I ", 3))
 455       {
 456        /*
 457         * Grab italic text...
 458         */
 459
 460         fprintf(outfile, "%s<i>%s</i>%s", end_fonts[font], line + 3,
 461                 start_fonts[font]);
 462       }
 463       else if (strncmp(line, ".\\\"", 3))
 464       {
 465        /*
 466         * Unknown...
 467         */
 468
 469         if ((lineptr = strchr(line, ' ')) != NULL)
 470           *lineptr = '\0';
 471         else if ((lineptr = strchr(line, '\n')) != NULL)
 472           *lineptr = '\0';
 473
 474         fprintf(stderr, "mantohtml: Unknown man page command \'%s\' on line %d!\n",
 475                 line, linenum);
 476       }
 477
 478      /*
 479       * Skip continuation lines...
 480       */
 481
 482       lineptr = line + strlen(line) - 2;
 483       if (lineptr >= line && *lineptr == '\\')
 484       {
 485         while (fgets(line, sizeof(line), infile))
 486         {
 487           linenum ++;
 488           lineptr = line + strlen(line) - 2;
 489
 490           if (lineptr < line || *lineptr != '\\')
 491             break;
 492         }
 493       }
 494     }
 495     else
 496     {
 497      /*
 498       * Process man page text...
 499       */
 500
 501 process_text:
 502
 503       for (lineptr = line; *lineptr; lineptr ++)
 504       {
 505         if (!strncmp(lineptr, "http://", 7))
 506         {
 507          /*
 508           * Embed URL...
 509           */
 510
 511           for (endptr = lineptr + 7;
 512                *endptr && !isspace(*endptr & 255);
 513                endptr ++);
 514
 515           endchar = *endptr;
 516           *endptr = '\0';
 517
 518           fprintf(outfile, "<a href='%s'>%s</a>", lineptr, lineptr);
 519           *endptr = endchar;
 520           lineptr = endptr - 1;
 521         }
 522         else if (!strncmp(lineptr, "\\fI", 3) &&
 523                  (endptr = strstr(lineptr, "\\fR")) != NULL &&
 524                  (paren = strchr(lineptr, '(')) != NULL &&
 525                  paren < endptr)
 526         {
 527          /*
 528           * Link to man page?
 529           */
 530
 531           char  manfile[1024],          /* Man page filename */
 532                 manurl[1024];           /* Man page URL */
 533
 534
 535          /*
 536           * See if the man file is available locally...
 537           */
 538
 539           lineptr += 3;
 540           endchar = *paren;
 541           *paren  = '\0';
 542
 543           snprintf(manfile, sizeof(manfile), "%s.man", lineptr);
 544           snprintf(manurl, sizeof(manurl), "man-%s.html?TOPIC=Man+Pages",
 545                    lineptr);
 546
 547           *paren  = endchar;
 548           endchar = *endptr;
 549           *endptr = '\0';
 550
 551           if (access(manfile, 0))
 552           {
 553            /*
 554             * Not a local man page, just do it italic...
 555             */
 556
 557             fputs("<i>", outfile);
 558             while (*lineptr)
 559               putc_entity(*lineptr++, outfile);
 560             fputs("</i>", outfile);
 561           }
 562           else
 563           {
 564            /*
 565             * Local man page, do a link...
 566             */
 567
 568             fprintf(outfile, "<a href='%s'>", manurl);
 569             while (*lineptr)
 570               putc_entity(*lineptr++, outfile);
 571             fputs("</a>", outfile);
 572           }
 573
 574           *endptr = endchar;
 575           lineptr = endptr + 2;
 576         }
 577         else if (*lineptr == '\\')
 578         {
 579           lineptr ++;
 580           if (!*lineptr)
 581             break;
 582           else if (isdigit(lineptr[0]) && isdigit(lineptr[1]) &&
 583                    isdigit(lineptr[2]))
 584           {
 585             fprintf(outfile, "&#%d;", ((lineptr[0] - '0') * 8 +
 586                                        lineptr[1] - '0') * 8 +
 587                                       lineptr[2] - '0');
 588             lineptr += 2;
 589           }
 590           else if (*lineptr == '&')
 591             continue;
 592           else if (*lineptr == 's')
 593           {
 594             while (lineptr[1] == '-' || isdigit(lineptr[1]))
 595               lineptr ++;
 596           }
 597           else if (*lineptr == '*')
 598           {
 599             lineptr += 2;
 600           }
 601           else if (*lineptr != 'f')
 602             putc_entity(*lineptr, outfile);
 603           else
 604           {
 605             lineptr ++;
 606             if (!*lineptr)
 607               break;
 608             else
 609             {
 610               fputs(end_fonts[font], outfile);
 611
 612               switch (*lineptr)
 613               {
 614                 default : /* Regular */
 615                     font = 0;
 616                     break;
 617                 case 'B' : /* Bold */
 618                 case 'b' :
 619                     font = 1;
 620                     break;
 621                 case 'I' : /* Italic */
 622                 case 'i' :
 623                     font = 2;
 624                     break;
 625               }
 626
 627               fputs(start_fonts[font], outfile);
 628             }
 629           }
 630         }
 631         else
 632           putc_entity(*lineptr, outfile);
 633       }
 634
 635       if (post)
 636       {
 637         fputs(post, outfile);
 638         post = NULL;
 639       }
 640     }
 641   }
 642
 643   fprintf(outfile, "%s\n", end_fonts[font]);
 644
 645   if (blist)
 646   {
 647     fputs("</li>\n</ul>\n", outfile);
 648   }
 649
 650   if (list)
 651   {
 652     if (list == 1)
 653       fputs("</dt>\n", outfile);
 654     else if (list)
 655       fputs("</dd>\n", outfile);
 656
 657     fputs("</dl>\n", outfile);
 658   }
 659
 660   fputs("</body>\n"
 661         "</html>\n", outfile);
 662
 663  /*
 664   * Close files...
 665   */
 666
 667   if (infile != stdin)
 668     fclose(infile);
 669
 670   if (outfile != stdout)
 671     fclose(outfile);
 672
 673  /*
 674   * Return with no errors...
 675   */
 676
 677   return (0);
 678 }
 679
 680
 681 /*
 682  * 'putc_entity()' - Put a single character, using entities as needed.
 683  */
 684
 685 static void
 686 putc_entity(int  ch,                    /* I - Character */
 687             FILE *fp)                   /* I - File */
 688 {
 689   if (ch == '&')
 690     fputs("&amp;", fp);
 691   else if (ch == '<')
 692     fputs("&lt;", fp);
 693   else
 694     putc(ch, fp);
 695 }
 696
 697
 698 /*
 699  * 'strmove()' - Move characters within a string.
 700  */
 701
 702 static void
 703 strmove(char       *d,                  /* I - Destination */
 704         const char *s)                  /* I - Source */
 705 {
 706   while (*s)
 707     *d++ = *s++;
 708
 709   *d = '\0';
 710 }
 711
 712
 713 /*
 714  * End of "$Id: mantohtml.c 7720 2008-07-11 22:46:21Z mike $".
 715  */