[thirdparty/cups.git] / standards / rfctohtml.c

/*
 * "$Id: rfctohtml.c 4966 2006-01-23 00:41:22Z mike $"
 *
 *   RFC file to HTML conversion program.
 *
 *   Copyright 2006 by Easy Software Products.
 *
 *   These coded instructions, statements, and computer programs are the
 *   property of Easy Software Products and are protected by Federal
 *   copyright law.  Distribution and use rights are outlined in the file
 *   "LICENSE.txt" which should have been included with this file.  If this
 *   file is missing or damaged please contact Easy Software Products
 *   at:
 *
 *       Attn: CUPS Licensing Information
 *       Easy Software Products
 *       44141 Airport View Drive, Suite 204
 *       Hollywood, Maryland 20636 USA
 *
 *       Voice: (301) 373-9600
 *       EMail: cups-info@cups.org
 *         WWW: http://www.cups.org
 *
 * Contents:
 *
 */

/*
 * Include necessary headers.
 */

#include <cups/string.h>
#include <stdlib.h>
#include <cups/file.h>


/*
 * Local functions...
 */

void	put_entity(cups_file_t *fp, int ch);
void	put_line(cups_file_t *fp, const char *line);


/*
 * 'main()' - Convert a man page to HTML.
 */

int					/* O - Exit status */
main(int  argc,				/* I - Number of command-line args */
     char *argv[])			/* I - Command-line arguments */
{
  cups_file_t	*infile,		/* Input file */
		*outfile;		/* Output file */
  char		line[1024],		/* Line from file */
		*lineptr,		/* Pointer into line */
		name[1024],		/* Heading anchor name */
		*nameptr;		/* Pointer into anchor name */
  int		inheading,		/* Inside a heading? */
		inpre,			/* Inside preformatted text? */
		intoc,			/* Inside table-of-contents? */
		toclevel,		/* Current table-of-contents level */
		linenum;		/* Current line on page */


 /*
  * Check arguments...
  */

  if (argc > 3)
  {
    fputs("Usage: rfctohtml [rfcNNNN.txt [rfcNNNN.html]]\n", stderr);
    return (1);
  }

 /*
  * Open files as needed...
  */

  if (argc > 1)
  {
    if ((infile = cupsFileOpen(argv[1], "r")) == NULL)
    {
      perror(argv[1]);
      return (1);
    }
  }
  else
    infile = cupsFileOpenFd(0, "r");

  if (argc > 2)
  {
    if ((outfile = cupsFileOpen(argv[2], "w")) == NULL)
    {
      perror(argv[2]);
      cupsFileClose(infile);
      return (1);
    }
  }
  else
    outfile = cupsFileOpenFd(1, "w");

 /*
  * Read from input and write the output...
  */

  cupsFilePuts(outfile,
	       "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0 Transitional//EN\" "
	       "\"http://www.w3.org/TR/REC-html40/loose.dtd\">\n"
	       "<html>\n"
	       "<!-- SECTION: Specifications -->\n"
	       "<head>\n"
	       "\t<style type='text/css'><!--\n"
	       "\th1, h2, h3 { font-family: sans-serif; }\n"
	       "\tp, pre { font-family: monospace; }\n"
	       "\th1.title, h2.title, h3.title { border-bottom: solid "
	       "2px #000000; }\n"
	       "\t--></style>\n");

 /*
  * Skip the initial header stuff (working group ID, RFC #, authors, and
  * copyright...
  */

  linenum = 0;

  while (cupsFileGets(infile, line, sizeof(line)))
  {
    linenum ++;

    if (line[0])
      break;
  }

  while (cupsFileGets(infile, line, sizeof(line)))
  {
    linenum ++;

    if (!line[0])
      break;
  }

 /*
  * Read the document title...
  */

  while (cupsFileGets(infile, line, sizeof(line)))
  {
    linenum ++;

    if (line[0])
      break;
  }

  for (lineptr = line; isspace(*lineptr & 255); lineptr ++);

  cupsFilePrintf(outfile, "<title>%s", lineptr);

  while (cupsFileGets(infile, line, sizeof(line)))
  {
    linenum ++;

    if (!line[0])
      break;
    else
    {
      for (lineptr = line; isspace(*lineptr & 255); lineptr ++);
      cupsFilePrintf(outfile, " %s", lineptr);
    }
  }

  cupsFilePuts(outfile, "</title>\n"
		        "</head>\n"
		        "<body>\n");

 /*
  * Read the rest of the file...
  */

  inheading = 0;
  inpre     = 0;
  intoc     = 0;
  toclevel  = 0;

  while (cupsFileGets(infile, line, sizeof(line)))
  {
    linenum ++;

    if (!line[0])
    {
      if (linenum > 50)
        continue;

      if (inpre)
      {
	cupsFilePuts(outfile, "</pre>\n");
	inpre = 0;
      }

      if (inheading)
      {
	if (inheading < 0)
	  cupsFilePuts(outfile, "</h1>\n");
	else
	  cupsFilePrintf(outfile, "</a></h%d>\n", inheading);

        inheading = 0;
      }
    }
    else if ((line[0] == ' ' ||
              (!isupper(line[0] & 255) && !isdigit(line[0] & 255) &&
	       !strstr(line, "[Page "))) && !inheading)
    {
      if (inheading)
      {
	if (inheading < 0)
	  cupsFilePuts(outfile, "</h1>\n");
	else
	  cupsFilePrintf(outfile, "</a></h%d>\n", inheading);

        inheading = 0;
      }

      for (lineptr = line; *lineptr == ' '; lineptr ++);

      if (intoc)
      {
        char	*temp;			/* Temporary pointer into line */
	int	level;			/* Heading level */


        if (isdigit(*lineptr & 255))
	{
	  strlcpy(name, lineptr, sizeof(name));

	  for (nameptr = name, level = -1; *nameptr;)
	    if (isdigit(*nameptr & 255))
	    {
	      while (isdigit(*nameptr & 255))
		nameptr ++;

	      level ++;
	    }
	    else if (*nameptr == ' ')
	    {
	      *nameptr = '\0';
	      break;
	    }
	    else
	      nameptr ++;

	  while (toclevel > level)
	  {
	    cupsFilePuts(outfile, "\n</ul>");
	    toclevel --;
	  }

	  while (toclevel < level)
	  {
	    cupsFilePuts(outfile, "\n<ul style=\"list-style-type: none;\">\n");
	    toclevel ++;
	  }

	  cupsFilePrintf(outfile, "\n<%s><a href=\"#s%s\">", toclevel ? "li" : "p",
	                 name);
	}

        temp = lineptr + strlen(lineptr) - 1;

	while (temp > lineptr)
	  if (*temp == ' ' || !isdigit(*temp & 255))
	    break;
	  else
	    temp --;

        if (*temp == ' ')
	{
	  while (temp > lineptr)
	    if (*temp != ' ' && *temp != '.')
	      break;
	    else
	      *temp-- = '\0';
	}
	else
	  temp = NULL;

        if (isdigit(*lineptr & 255))
          put_line(outfile, lineptr);
	else
	  put_line(outfile, lineptr - 1);

        if (temp)
	  cupsFilePuts(outfile, "</a>");
      }
      else if (!inpre)
      {
	cupsFilePuts(outfile, "\n<pre>");
	put_line(outfile, line);
	inpre = 1;
      }
      else
      {
	cupsFilePutChar(outfile, '\n');
	put_line(outfile, line);
      }
    }
    else if (strstr(line, "[Page "))
    {
     /*
      * Skip page footer and header...
      */

      cupsFileGets(infile, line, sizeof(line));
      cupsFileGets(infile, line, sizeof(line));
      cupsFileGets(infile, line, sizeof(line));
      cupsFileGets(infile, line, sizeof(line));
      linenum = 3;
    }
    else if (isdigit(line[0] & 255) && !inheading)
    {
      int level;			/* Heading level */


      if (intoc)
      {
        while (toclevel > 0)
	{
	  cupsFilePuts(outfile, "\n</ul>");
	  toclevel --;
	}

	cupsFilePutChar(outfile, '\n');
	intoc = 0;
      }

      if (inpre)
      {
	cupsFilePuts(outfile, "</pre>\n");
	inpre = 0;
      }

      strlcpy(name, line, sizeof(name));
      for (nameptr = name, level = 0; *nameptr;)
        if (isdigit(*nameptr & 255))
	{
	  while (isdigit(*nameptr & 255))
	    nameptr ++;

          level ++;
	}
	else if (*nameptr == ' ')
	{
	  *nameptr = '\0';
	  break;
	}
	else
	  nameptr ++;

      cupsFilePrintf(outfile, "\n<h%d class='title'><a name='s%s'>", level, name);
      put_line(outfile, line);

      intoc     = 0;
      inheading = level;
    }
    else
    {
      if (intoc)
      {
        while (toclevel > 0)
	{
	  cupsFilePuts(outfile, "\n</ul>");
	  toclevel --;
	}

	cupsFilePutChar(outfile, '\n');
	intoc = 0;
      }

      if (!inheading)
      {
        cupsFilePuts(outfile, "\n<h1 class='title'>");
        inheading = -1;
      }

      put_line(outfile, line);

      intoc    = !strcasecmp(line, "Table of Contents");
      toclevel = 0;
    }
  }

  if (inpre)
    cupsFilePuts(outfile, "</pre>\n");

  if (inheading)
  {
    if (inheading < 0)
      cupsFilePuts(outfile, "</h1>\n");
    else
      cupsFilePrintf(outfile, "</a></h%d>\n", inheading);
  }

  cupsFilePuts(outfile, "</body>\n"
                        "</html>\n");

 /*
  * Close files...
  */

  cupsFileClose(infile);
  cupsFileClose(outfile);

 /*
  * Return with no errors...
  */

  return (0);
}


/*
 * 'put_entity()' - Put a single character, using entities as needed.
 */

void
put_entity(cups_file_t *fp,		/* I - File */
           int         ch)		/* I - Character */
{
  if (ch == '&')
    cupsFilePuts(fp, "&amp;");
  else if (ch == '<')
    cupsFilePuts(fp, "&lt;");
  else
    cupsFilePutChar(fp, ch);
}


/*
 * 'put_line()' - Put a whole string for a line.
 */

void
put_line(cups_file_t *fp,		/* I - File */
         const char  *s)		/* I - String */
{
  int	whitespace,			/* Saw whitespace */
	i,				/* Looping var */
	len;				/* Length of keyword */
  static const char * const keywords[] =/* Special keywords to boldface */
  {
    "MAY",
    "MUST",
    "NOT",
    "SHALL",
    "SHOULD"
  };


  whitespace = 1;

  while (*s)
  {
    if (*s == ' ')
      whitespace = 1;

    if (whitespace && isupper(*s & 255))
    {
      whitespace = 0;

      for (i = 0; i < (int)(sizeof(keywords) / sizeof(sizeof(keywords[0]))); i ++)
      {
        len = strlen(keywords[i]);
	if (!strncmp(s, keywords[i], len) && (isspace(s[len] & 255) || !*s))
	{
	  cupsFilePrintf(fp, "<b>%s</b>", keywords[i]);
	  s += len;
	  break;
	}
      }

      if (i >= (int)(sizeof(keywords) / sizeof(sizeof(keywords[0]))))
        put_entity(fp, *s++);
    }
    else
    {
      if (*s != ' ')
        whitespace = 0;

      put_entity(fp, *s++);
    }
  }
}


/*
 * End of "$Id: rfctohtml.c 4966 2006-01-23 00:41:22Z mike $".
 */
Commit	Line	Data
fa73b229	1	/*
	2	* "$Id: rfctohtml.c 4966 2006-01-23 00:41:22Z 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	*/
	27
	28	/*
	29	* Include necessary headers.
	30	*/
	31
	32	#include <cups/string.h>
	33	#include <stdlib.h>
	34	#include <cups/file.h>
	35
	36
	37	/*
	38	* Local functions...
	39	*/
	40
	41	void put_entity(cups_file_t *fp, int ch);
	42	void put_line(cups_file_t fp, const char line);
	43
	44
	45	/*
	46	* 'main()' - Convert a man page to HTML.
	47	*/
	48
	49	int /* O - Exit status */
	50	main(int argc, /* I - Number of command-line args */
	51	char argv[]) / I - Command-line arguments */
	52	{
	53	cups_file_t infile, / Input file */
	54	outfile; / Output file */
	55	char line[1024], /* Line from file */
	56	lineptr, / Pointer into line */
	57	name[1024], /* Heading anchor name */
	58	nameptr; / Pointer into anchor name */
	59	int inheading, /* Inside a heading? */
	60	inpre, /* Inside preformatted text? */
	61	intoc, /* Inside table-of-contents? */
	62	toclevel, /* Current table-of-contents level */
	63	linenum; /* Current line on page */
	64
65
66	/*
67	* Check arguments...
68	*/
69
70	if (argc > 3)
71	{
72	fputs("Usage: rfctohtml [rfcNNNN.txt [rfcNNNN.html]]\n", stderr);
73	return (1);
74	}
75
76	/*
77	* Open files as needed...
78	*/
79
80	if (argc > 1)
81	{
82	if ((infile = cupsFileOpen(argv[1], "r")) == NULL)
83	{
84	perror(argv[1]);
85	return (1);
86	}
87	}
88	else
89	infile = cupsFileOpenFd(0, "r");
90
91	if (argc > 2)
92	{
93	if ((outfile = cupsFileOpen(argv[2], "w")) == NULL)
94	{
95	perror(argv[2]);
96	cupsFileClose(infile);
97	return (1);
98	}
99	}
100	else
101	outfile = cupsFileOpenFd(1, "w");
102
103	/*
104	* Read from input and write the output...
105	*/
106
107	cupsFilePuts(outfile,
108	"<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0 Transitional//EN\" "
109	"\"http://www.w3.org/TR/REC-html40/loose.dtd\">\n"
110	"<html>\n"
111	"<!-- SECTION: Specifications -->\n"
112	"<head>\n"
113	"\t<style type='text/css'><!--\n"
114	"\th1, h2, h3 { font-family: sans-serif; }\n"
115	"\tp, pre { font-family: monospace; }\n"
116	"\th1.title, h2.title, h3.title { border-bottom: solid "
117	"2px #000000; }\n"
118	"\t--></style>\n");
119
120	/*
121	* Skip the initial header stuff (working group ID, RFC #, authors, and
122	* copyright...
123	*/
124
125	linenum = 0;
126
127	while (cupsFileGets(infile, line, sizeof(line)))
128	{
129	linenum ++;
130
131	if (line[0])
132	break;
133	}
134
135	while (cupsFileGets(infile, line, sizeof(line)))
136	{
137	linenum ++;
138
139	if (!line[0])
140	break;
141	}
142
143	/*
144	* Read the document title...
145	*/
146
147	while (cupsFileGets(infile, line, sizeof(line)))
148	{
149	linenum ++;
150
151	if (line[0])
152	break;
153	}
154
155	for (lineptr = line; isspace(*lineptr & 255); lineptr ++);
156
157	cupsFilePrintf(outfile, "<title>%s", lineptr);
158
159	while (cupsFileGets(infile, line, sizeof(line)))
160	{
161	linenum ++;
162
163	if (!line[0])
164	break;
165	else
166	{
167	for (lineptr = line; isspace(*lineptr & 255); lineptr ++);
168	cupsFilePrintf(outfile, " %s", lineptr);
169	}
170	}
171
172	cupsFilePuts(outfile, "</title>\n"
173	"</head>\n"
174	"<body>\n");
175
176	/*
177	* Read the rest of the file...
178	*/
179
180	inheading = 0;
181	inpre = 0;
182	intoc = 0;
183	toclevel = 0;
184
185	while (cupsFileGets(infile, line, sizeof(line)))
186	{
187	linenum ++;
188
189	if (!line[0])
190	{
191	if (linenum > 50)
192	continue;
193
194	if (inpre)
195	{
196	cupsFilePuts(outfile, "</pre>\n");
197	inpre = 0;
198	}
199
200	if (inheading)
201	{
202	if (inheading < 0)
203	cupsFilePuts(outfile, "</h1>\n");
204	else
205	cupsFilePrintf(outfile, "</a></h%d>\n", inheading);
206
207	inheading = 0;
208	}
209	}
210	else if ((line[0] == ' ' \|\|
211	(!isupper(line[0] & 255) && !isdigit(line[0] & 255) &&
212	!strstr(line, "[Page "))) && !inheading)
213	{
214	if (inheading)
215	{
216	if (inheading < 0)
217	cupsFilePuts(outfile, "</h1>\n");
218	else
219	cupsFilePrintf(outfile, "</a></h%d>\n", inheading);
220
221	inheading = 0;
222	}
223
224	for (lineptr = line; *lineptr == ' '; lineptr ++);
225
226	if (intoc)
227	{
228	char temp; / Temporary pointer into line */
229	int level; /* Heading level */
230
231
232	if (isdigit(*lineptr & 255))
233	{
234	strlcpy(name, lineptr, sizeof(name));
235
236	for (nameptr = name, level = -1; *nameptr;)
237	if (isdigit(*nameptr & 255))
238	{
239	while (isdigit(*nameptr & 255))
240	nameptr ++;
241
242	level ++;
243	}
244	else if (*nameptr == ' ')
245	{
246	*nameptr = '\0';
247	break;
248	}
249	else
250	nameptr ++;
251
252	while (toclevel > level)
253	{
254	cupsFilePuts(outfile, "\n</ul>");
255	toclevel --;
256	}
257
258	while (toclevel < level)
259	{
260	cupsFilePuts(outfile, "\n<ul style=\"list-style-type: none;\">\n");
261	toclevel ++;
262	}
263
264	cupsFilePrintf(outfile, "\n<%s><a href=\"#s%s\">", toclevel ? "li" : "p",
265	name);
266	}
267
268	temp = lineptr + strlen(lineptr) - 1;
269
270	while (temp > lineptr)
271	if (temp == ' ' \|\| !isdigit(temp & 255))
272	break;
273	else
274	temp --;
275
276	if (*temp == ' ')
277	{
278	while (temp > lineptr)
279	if (temp != ' ' && temp != '.')
280	break;
281	else
282	*temp-- = '\0';
283	}
284	else
285	temp = NULL;
286
287	if (isdigit(*lineptr & 255))
288	put_line(outfile, lineptr);
289	else
290	put_line(outfile, lineptr - 1);
291
292	if (temp)
293	cupsFilePuts(outfile, "</a>");
294	}
295	else if (!inpre)
296	{
297	cupsFilePuts(outfile, "\n<pre>");
298	put_line(outfile, line);
299	inpre = 1;
300	}
301	else
302	{
303	cupsFilePutChar(outfile, '\n');
304	put_line(outfile, line);
305	}
306	}
307	else if (strstr(line, "[Page "))
308	{
309	/*
310	* Skip page footer and header...
311	*/
312
313	cupsFileGets(infile, line, sizeof(line));
314	cupsFileGets(infile, line, sizeof(line));
315	cupsFileGets(infile, line, sizeof(line));
316	cupsFileGets(infile, line, sizeof(line));
317	linenum = 3;
318	}
319	else if (isdigit(line[0] & 255) && !inheading)
320	{
321	int level; /* Heading level */
322
323
324	if (intoc)
325	{
326	while (toclevel > 0)
327	{
328	cupsFilePuts(outfile, "\n</ul>");
329	toclevel --;
330	}
331
332	cupsFilePutChar(outfile, '\n');
333	intoc = 0;
334	}
335
336	if (inpre)
337	{
338	cupsFilePuts(outfile, "</pre>\n");
339	inpre = 0;
340	}
341
342	strlcpy(name, line, sizeof(name));
343	for (nameptr = name, level = 0; *nameptr;)
344	if (isdigit(*nameptr & 255))
345	{
346	while (isdigit(*nameptr & 255))
347	nameptr ++;
348
349	level ++;
350	}
351	else if (*nameptr == ' ')
352	{
353	*nameptr = '\0';
354	break;
355	}
356	else
357	nameptr ++;
358
359	cupsFilePrintf(outfile, "\n<h%d class='title'><a name='s%s'>", level, name);
360	put_line(outfile, line);
361
362	intoc = 0;
363	inheading = level;
364	}
365	else
366	{
367	if (intoc)
368	{
369	while (toclevel > 0)
370	{
371	cupsFilePuts(outfile, "\n</ul>");
372	toclevel --;
373	}
374
375	cupsFilePutChar(outfile, '\n');
376	intoc = 0;
377	}
378
379	if (!inheading)
380	{
381	cupsFilePuts(outfile, "\n<h1 class='title'>");
382	inheading = -1;
383	}
384
385	put_line(outfile, line);
386
387	intoc = !strcasecmp(line, "Table of Contents");
388	toclevel = 0;
389	}
390	}
391
392	if (inpre)
393	cupsFilePuts(outfile, "</pre>\n");
394
395	if (inheading)
396	{
397	if (inheading < 0)
398	cupsFilePuts(outfile, "</h1>\n");
399	else
400	cupsFilePrintf(outfile, "</a></h%d>\n", inheading);
401	}
402
403	cupsFilePuts(outfile, "</body>\n"
404	"</html>\n");
405
406	/*
407	* Close files...
408	*/
409
410	cupsFileClose(infile);
411	cupsFileClose(outfile);
412
413	/*
414	* Return with no errors...
415	*/
416
417	return (0);
418	}
419
420
421	/*
422	* 'put_entity()' - Put a single character, using entities as needed.
423	*/
424
425	void
426	put_entity(cups_file_t fp, / I - File */
427	int ch) /* I - Character */
428	{
429	if (ch == '&')
430	cupsFilePuts(fp, "&");
431	else if (ch == '<')
432	cupsFilePuts(fp, "<");
433	else
434	cupsFilePutChar(fp, ch);
435	}
436
437
438	/*
439	* 'put_line()' - Put a whole string for a line.
440	*/
441
442	void
443	put_line(cups_file_t fp, / I - File */
444	const char s) / I - String */
445	{
446	int whitespace, /* Saw whitespace */
447	i, /* Looping var */
448	len; /* Length of keyword */
449	static const char * const keywords[] =/* Special keywords to boldface */
450	{
451	"MAY",
452	"MUST",
453	"NOT",
454	"SHALL",
455	"SHOULD"
456	};
457
458
459	whitespace = 1;
460
461	while (*s)
462	{
463	if (*s == ' ')
464	whitespace = 1;
465
466	if (whitespace && isupper(*s & 255))
467	{
468	whitespace = 0;
469
470	for (i = 0; i < (int)(sizeof(keywords) / sizeof(sizeof(keywords[0]))); i ++)
471	{
472	len = strlen(keywords[i]);
473	if (!strncmp(s, keywords[i], len) && (isspace(s[len] & 255) \|\| !*s))
474	{
475	cupsFilePrintf(fp, "<b>%s</b>", keywords[i]);
476	s += len;
477	break;
478	}
479	}
480
481	if (i >= (int)(sizeof(keywords) / sizeof(sizeof(keywords[0]))))
482	put_entity(fp, *s++);
483	}
484	else
485	{
486	if (*s != ' ')
487	whitespace = 0;
488
489	put_entity(fp, *s++);
490	}
491	}
492	}
493
494
495	/*
496	* End of "$Id: rfctohtml.c 4966 2006-01-23 00:41:22Z mike $".
497	*/