standards/rfctohtml.c

   1 /*
   2  * "$Id: rfctohtml.c 5143 2006-02-21 19:13:01Z mike $"
   3  *
   4  *   RFC file to HTML conversion program.
   5  *
   6  *   Copyright 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  *   put_entity() - Put a single character, using entities as needed.
  28  *   put_line()   - Put a whole string for a line.
  29  */
  30
  31 /*
  32  * Include necessary headers.
  33  */
  34
  35 #include <cups/string.h>
  36 #include <stdlib.h>
  37 #include <cups/file.h>
  38
  39
  40 /*
  41  * Local functions...
  42  */
  43
  44 void    put_entity(cups_file_t *fp, int ch);
  45 void    put_line(cups_file_t *fp, const char *line);
  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   cups_file_t   *infile,                /* Input file */
  57                 *outfile;               /* Output file */
  58   char          line[1024],             /* Line from file */
  59                 *lineptr,               /* Pointer into line */
  60                 name[1024],             /* Heading anchor name */
  61                 *nameptr;               /* Pointer into anchor name */
  62   int           rfc,                    /* RFC # */
  63                 inheading,              /* Inside a heading? */
  64                 inpre,                  /* Inside preformatted text? */
  65                 intoc,                  /* Inside table-of-contents? */
  66                 toclevel,               /* Current table-of-contents level */
  67                 linenum;                /* Current line on page */
  68
  69
  70  /*
  71   * Check arguments...
  72   */
  73
  74   if (argc > 3)
  75   {
  76     fputs("Usage: rfctohtml [rfcNNNN.txt [rfcNNNN.html]]\n", stderr);
  77     return (1);
  78   }
  79
  80  /*
  81   * Open files as needed...
  82   */
  83
  84   if (argc > 1)
  85   {
  86     if ((infile = cupsFileOpen(argv[1], "r")) == NULL)
  87     {
  88       perror(argv[1]);
  89       return (1);
  90     }
  91   }
  92   else
  93     infile = cupsFileOpenFd(0, "r");
  94
  95   if (argc > 2)
  96   {
  97     if ((outfile = cupsFileOpen(argv[2], "w")) == NULL)
  98     {
  99       perror(argv[2]);
 100       cupsFileClose(infile);
 101       return (1);
 102     }
 103   }
 104   else
 105     outfile = cupsFileOpenFd(1, "w");
 106
 107  /*
 108   * Read from input and write the output...
 109   */
 110
 111   cupsFilePuts(outfile,
 112                "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0 Transitional//EN\" "
 113                "\"http://www.w3.org/TR/REC-html40/loose.dtd\">\n"
 114                "<html>\n"
 115                "<!-- SECTION: Specifications -->\n"
 116                "<head>\n"
 117                "\t<style type='text/css'><!--\n"
 118                "\th1, h2, h3 { font-family: sans-serif; }\n"
 119                "\tp, pre { font-family: monospace; }\n"
 120                "\th2.title, h3.title, h3.title { border-bottom: solid "
 121                "2px #000000; }\n"
 122                "\t--></style>\n");
 123
 124  /*
 125   * Skip the initial header stuff (working group ID, RFC #, authors, and
 126   * copyright...
 127   */
 128
 129   linenum = 0;
 130   rfc     = 0;
 131
 132   while (cupsFileGets(infile, line, sizeof(line)))
 133   {
 134     linenum ++;
 135
 136     if (line[0])
 137       break;
 138   }
 139
 140   while (cupsFileGets(infile, line, sizeof(line)))
 141   {
 142     linenum ++;
 143
 144     if (!line[0])
 145       break;
 146     else if (!strncasecmp(line, "Request for Comments:", 21))
 147       rfc = atoi(line + 21);
 148   }
 149
 150  /*
 151   * Read the document title...
 152   */
 153
 154   while (cupsFileGets(infile, line, sizeof(line)))
 155   {
 156     linenum ++;
 157
 158     if (line[0])
 159       break;
 160   }
 161
 162   for (lineptr = line; isspace(*lineptr & 255); lineptr ++);
 163
 164   cupsFilePrintf(outfile, "<title>RFC %d: %s", rfc, lineptr);
 165
 166   while (cupsFileGets(infile, line, sizeof(line)))
 167   {
 168     linenum ++;
 169
 170     if (!line[0])
 171       break;
 172     else
 173     {
 174       for (lineptr = line; isspace(*lineptr & 255); lineptr ++);
 175       cupsFilePrintf(outfile, " %s", lineptr);
 176     }
 177   }
 178
 179   cupsFilePuts(outfile, "</title>\n"
 180                         "</head>\n"
 181                         "<body>\n");
 182
 183  /*
 184   * Read the rest of the file...
 185   */
 186
 187   inheading = 0;
 188   inpre     = 0;
 189   intoc     = 0;
 190   toclevel  = 0;
 191
 192   while (cupsFileGets(infile, line, sizeof(line)))
 193   {
 194     linenum ++;
 195
 196     if (!line[0])
 197     {
 198       if (linenum > 50)
 199         continue;
 200
 201       if (inpre)
 202       {
 203         cupsFilePuts(outfile, "</pre>\n");
 204         inpre = 0;
 205       }
 206
 207       if (inheading)
 208       {
 209         if (inheading < 0)
 210           cupsFilePuts(outfile, "</h1>\n");
 211         else
 212           cupsFilePrintf(outfile, "</a></h%d>\n", inheading);
 213
 214         inheading = 0;
 215       }
 216     }
 217     else if ((line[0] == ' ' ||
 218               (!isupper(line[0] & 255) && !isdigit(line[0] & 255) &&
 219                !strstr(line, "[Page "))) && !inheading)
 220     {
 221       if (inheading)
 222       {
 223         if (inheading < 0)
 224           cupsFilePuts(outfile, "</h1>\n");
 225         else
 226           cupsFilePrintf(outfile, "</a></h%d>\n", inheading);
 227
 228         inheading = 0;
 229       }
 230
 231       for (lineptr = line; *lineptr == ' '; lineptr ++);
 232
 233       if (intoc)
 234       {
 235         char    *temp;                  /* Temporary pointer into line */
 236         int     level;                  /* Heading level */
 237
 238
 239         if (isdigit(*lineptr & 255))
 240         {
 241           strlcpy(name, lineptr, sizeof(name));
 242
 243           for (nameptr = name, level = -1; *nameptr;)
 244             if (isdigit(*nameptr & 255))
 245             {
 246               while (isdigit(*nameptr & 255))
 247                 nameptr ++;
 248
 249               level ++;
 250             }
 251             else if (*nameptr == ' ')
 252             {
 253               *nameptr = '\0';
 254               break;
 255             }
 256             else
 257               nameptr ++;
 258
 259           while (toclevel > level)
 260           {
 261             cupsFilePuts(outfile, "\n</ul>");
 262             toclevel --;
 263           }
 264
 265           while (toclevel < level)
 266           {
 267             cupsFilePuts(outfile, "\n<ul style=\"list-style-type: none;\">\n");
 268             toclevel ++;
 269           }
 270
 271           cupsFilePrintf(outfile, "\n<%s><a href=\"#s%s\">", toclevel ? "li" : "p",
 272                          name);
 273         }
 274
 275         temp = lineptr + strlen(lineptr) - 1;
 276
 277         while (temp > lineptr)
 278           if (*temp == ' ' || !isdigit(*temp & 255))
 279             break;
 280           else
 281             temp --;
 282
 283         if (*temp == ' ')
 284         {
 285           while (temp > lineptr)
 286             if (*temp != ' ' && *temp != '.')
 287               break;
 288             else
 289               *temp-- = '\0';
 290         }
 291         else
 292           temp = NULL;
 293
 294         if (isdigit(*lineptr & 255))
 295           put_line(outfile, lineptr);
 296         else
 297           put_line(outfile, lineptr - 1);
 298
 299         if (temp)
 300           cupsFilePuts(outfile, "</a>");
 301       }
 302       else if (!inpre)
 303       {
 304         cupsFilePuts(outfile, "\n<pre>");
 305         put_line(outfile, line);
 306         inpre = 1;
 307       }
 308       else
 309       {
 310         cupsFilePutChar(outfile, '\n');
 311         put_line(outfile, line);
 312       }
 313     }
 314     else if (strstr(line, "[Page "))
 315     {
 316      /*
 317       * Skip page footer and header...
 318       */
 319
 320       cupsFileGets(infile, line, sizeof(line));
 321       cupsFileGets(infile, line, sizeof(line));
 322       cupsFileGets(infile, line, sizeof(line));
 323       cupsFileGets(infile, line, sizeof(line));
 324       linenum = 3;
 325     }
 326     else if (isdigit(line[0] & 255) && !inheading)
 327     {
 328       int level;                        /* Heading level */
 329
 330
 331       if (intoc)
 332       {
 333         while (toclevel > 0)
 334         {
 335           cupsFilePuts(outfile, "\n</ul>");
 336           toclevel --;
 337         }
 338
 339         cupsFilePutChar(outfile, '\n');
 340         intoc = 0;
 341       }
 342
 343       if (inpre)
 344       {
 345         cupsFilePuts(outfile, "</pre>\n");
 346         inpre = 0;
 347       }
 348
 349       strlcpy(name, line, sizeof(name));
 350       for (nameptr = name, level = 1; *nameptr;)
 351         if (isdigit(*nameptr & 255))
 352         {
 353           while (isdigit(*nameptr & 255))
 354             nameptr ++;
 355
 356           level ++;
 357         }
 358         else if (*nameptr == ' ')
 359         {
 360           *nameptr = '\0';
 361           break;
 362         }
 363         else
 364           nameptr ++;
 365
 366       cupsFilePrintf(outfile, "\n<h%d class='title'><a name='s%s'>", level,
 367                      name);
 368       put_line(outfile, line);
 369
 370       intoc     = 0;
 371       inheading = level;
 372     }
 373     else
 374     {
 375       if (intoc)
 376       {
 377         while (toclevel > 0)
 378         {
 379           cupsFilePuts(outfile, "\n</ul>");
 380           toclevel --;
 381         }
 382
 383         cupsFilePutChar(outfile, '\n');
 384         intoc = 0;
 385       }
 386
 387       if (!inheading)
 388       {
 389         cupsFilePuts(outfile, "\n<h2 class='title'>");
 390         inheading = -1;
 391       }
 392
 393       put_line(outfile, line);
 394
 395       intoc    = !strcasecmp(line, "Table of Contents");
 396       toclevel = 0;
 397     }
 398   }
 399
 400   if (inpre)
 401     cupsFilePuts(outfile, "</pre>\n");
 402
 403   if (inheading)
 404   {
 405     if (inheading < 0)
 406       cupsFilePuts(outfile, "</h2>\n");
 407     else
 408       cupsFilePrintf(outfile, "</a></h%d>\n", inheading);
 409   }
 410
 411   cupsFilePuts(outfile, "</body>\n"
 412                         "</html>\n");
 413
 414  /*
 415   * Close files...
 416   */
 417
 418   cupsFileClose(infile);
 419   cupsFileClose(outfile);
 420
 421  /*
 422   * Return with no errors...
 423   */
 424
 425   return (0);
 426 }
 427
 428
 429 /*
 430  * 'put_entity()' - Put a single character, using entities as needed.
 431  */
 432
 433 void
 434 put_entity(cups_file_t *fp,             /* I - File */
 435            int         ch)              /* I - Character */
 436 {
 437   if (ch == '&')
 438     cupsFilePuts(fp, "&amp;");
 439   else if (ch == '<')
 440     cupsFilePuts(fp, "&lt;");
 441   else
 442     cupsFilePutChar(fp, ch);
 443 }
 444
 445
 446 /*
 447  * 'put_line()' - Put a whole string for a line.
 448  */
 449
 450 void
 451 put_line(cups_file_t *fp,               /* I - File */
 452          const char  *s)                /* I - String */
 453 {
 454   int   whitespace,                     /* Saw whitespace */
 455         i,                              /* Looping var */
 456         len;                            /* Length of keyword */
 457   static const char * const keywords[] =/* Special keywords to boldface */
 458   {
 459     "MAY",
 460     "MUST",
 461     "NOT",
 462     "SHALL",
 463     "SHOULD"
 464   };
 465
 466
 467   whitespace = 1;
 468
 469   while (*s)
 470   {
 471     if (*s == ' ')
 472       whitespace = 1;
 473
 474     if (whitespace && isupper(*s & 255))
 475     {
 476       whitespace = 0;
 477
 478       for (i = 0; i < (int)(sizeof(keywords) / sizeof(sizeof(keywords[0]))); i ++)
 479       {
 480         len = strlen(keywords[i]);
 481         if (!strncmp(s, keywords[i], len) && (isspace(s[len] & 255) || !*s))
 482         {
 483           cupsFilePrintf(fp, "<b>%s</b>", keywords[i]);
 484           s += len;
 485           break;
 486         }
 487       }
 488
 489       if (i >= (int)(sizeof(keywords) / sizeof(sizeof(keywords[0]))))
 490         put_entity(fp, *s++);
 491     }
 492     else
 493     {
 494       if (*s != ' ')
 495         whitespace = 0;
 496
 497       put_entity(fp, *s++);
 498     }
 499   }
 500 }
 501
 502
 503 /*
 504  * End of "$Id: rfctohtml.c 5143 2006-02-21 19:13:01Z mike $".
 505  */