]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/blob - contrib/mdocml/html.c
MFV r276568:
[FreeBSD/FreeBSD.git] / contrib / mdocml / html.c
1 /*      $Id: html.c,v 1.183 2014/12/02 10:08:06 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008-2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
5  *
6  * Permission to use, copy, modify, and distribute this software for any
7  * purpose with or without fee is hereby granted, provided that the above
8  * copyright notice and this permission notice appear in all copies.
9  *
10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17  */
18 #include "config.h"
19
20 #include <sys/types.h>
21
22 #include <assert.h>
23 #include <ctype.h>
24 #include <stdarg.h>
25 #include <stdio.h>
26 #include <stdint.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <unistd.h>
30
31 #include "mandoc.h"
32 #include "mandoc_aux.h"
33 #include "out.h"
34 #include "html.h"
35 #include "main.h"
36
37 struct  htmldata {
38         const char       *name;
39         int               flags;
40 #define HTML_CLRLINE     (1 << 0)
41 #define HTML_NOSTACK     (1 << 1)
42 #define HTML_AUTOCLOSE   (1 << 2) /* Tag has auto-closure. */
43 };
44
45 static  const struct htmldata htmltags[TAG_MAX] = {
46         {"html",        HTML_CLRLINE}, /* TAG_HTML */
47         {"head",        HTML_CLRLINE}, /* TAG_HEAD */
48         {"body",        HTML_CLRLINE}, /* TAG_BODY */
49         {"meta",        HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_META */
50         {"title",       HTML_CLRLINE}, /* TAG_TITLE */
51         {"div",         HTML_CLRLINE}, /* TAG_DIV */
52         {"h1",          0}, /* TAG_H1 */
53         {"h2",          0}, /* TAG_H2 */
54         {"span",        0}, /* TAG_SPAN */
55         {"link",        HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_LINK */
56         {"br",          HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_BR */
57         {"a",           0}, /* TAG_A */
58         {"table",       HTML_CLRLINE}, /* TAG_TABLE */
59         {"tbody",       HTML_CLRLINE}, /* TAG_TBODY */
60         {"col",         HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_COL */
61         {"tr",          HTML_CLRLINE}, /* TAG_TR */
62         {"td",          HTML_CLRLINE}, /* TAG_TD */
63         {"li",          HTML_CLRLINE}, /* TAG_LI */
64         {"ul",          HTML_CLRLINE}, /* TAG_UL */
65         {"ol",          HTML_CLRLINE}, /* TAG_OL */
66         {"dl",          HTML_CLRLINE}, /* TAG_DL */
67         {"dt",          HTML_CLRLINE}, /* TAG_DT */
68         {"dd",          HTML_CLRLINE}, /* TAG_DD */
69         {"blockquote",  HTML_CLRLINE}, /* TAG_BLOCKQUOTE */
70         {"pre",         HTML_CLRLINE }, /* TAG_PRE */
71         {"b",           0 }, /* TAG_B */
72         {"i",           0 }, /* TAG_I */
73         {"code",        0 }, /* TAG_CODE */
74         {"small",       0 }, /* TAG_SMALL */
75         {"style",       HTML_CLRLINE}, /* TAG_STYLE */
76         {"math",        HTML_CLRLINE}, /* TAG_MATH */
77         {"mrow",        0}, /* TAG_MROW */
78         {"mi",          0}, /* TAG_MI */
79         {"mo",          0}, /* TAG_MO */
80         {"msup",        0}, /* TAG_MSUP */
81         {"msub",        0}, /* TAG_MSUB */
82         {"msubsup",     0}, /* TAG_MSUBSUP */
83         {"mfrac",       0}, /* TAG_MFRAC */
84         {"msqrt",       0}, /* TAG_MSQRT */
85         {"mfenced",     0}, /* TAG_MFENCED */
86         {"mtable",      0}, /* TAG_MTABLE */
87         {"mtr",         0}, /* TAG_MTR */
88         {"mtd",         0}, /* TAG_MTD */
89         {"munderover",  0}, /* TAG_MUNDEROVER */
90         {"munder",      0}, /* TAG_MUNDER*/
91         {"mover",       0}, /* TAG_MOVER*/
92 };
93
94 static  const char      *const htmlattrs[ATTR_MAX] = {
95         "name", /* ATTR_NAME */
96         "rel", /* ATTR_REL */
97         "href", /* ATTR_HREF */
98         "type", /* ATTR_TYPE */
99         "media", /* ATTR_MEDIA */
100         "class", /* ATTR_CLASS */
101         "style", /* ATTR_STYLE */
102         "id", /* ATTR_ID */
103         "colspan", /* ATTR_COLSPAN */
104         "charset", /* ATTR_CHARSET */
105         "open", /* ATTR_OPEN */
106         "close", /* ATTR_CLOSE */
107         "mathvariant", /* ATTR_MATHVARIANT */
108 };
109
110 static  const char      *const roffscales[SCALE_MAX] = {
111         "cm", /* SCALE_CM */
112         "in", /* SCALE_IN */
113         "pc", /* SCALE_PC */
114         "pt", /* SCALE_PT */
115         "em", /* SCALE_EM */
116         "em", /* SCALE_MM */
117         "ex", /* SCALE_EN */
118         "ex", /* SCALE_BU */
119         "em", /* SCALE_VS */
120         "ex", /* SCALE_FS */
121 };
122
123 static  void     bufncat(struct html *, const char *, size_t);
124 static  void     print_ctag(struct html *, enum htmltag);
125 static  int      print_escape(char);
126 static  int      print_encode(struct html *, const char *, int);
127 static  void     print_metaf(struct html *, enum mandoc_esc);
128 static  void     print_attr(struct html *, const char *, const char *);
129
130
131 void *
132 html_alloc(const struct mchars *mchars, char *outopts)
133 {
134         struct html     *h;
135         const char      *toks[5];
136         char            *v;
137
138         toks[0] = "style";
139         toks[1] = "man";
140         toks[2] = "includes";
141         toks[3] = "fragment";
142         toks[4] = NULL;
143
144         h = mandoc_calloc(1, sizeof(struct html));
145
146         h->tags.head = NULL;
147         h->symtab = mchars;
148
149         while (outopts && *outopts)
150                 switch (getsubopt(&outopts, UNCONST(toks), &v)) {
151                 case 0:
152                         h->style = v;
153                         break;
154                 case 1:
155                         h->base_man = v;
156                         break;
157                 case 2:
158                         h->base_includes = v;
159                         break;
160                 case 3:
161                         h->oflags |= HTML_FRAGMENT;
162                         break;
163                 default:
164                         break;
165                 }
166
167         return(h);
168 }
169
170 void
171 html_free(void *p)
172 {
173         struct tag      *tag;
174         struct html     *h;
175
176         h = (struct html *)p;
177
178         while ((tag = h->tags.head) != NULL) {
179                 h->tags.head = tag->next;
180                 free(tag);
181         }
182
183         free(h);
184 }
185
186 void
187 print_gen_head(struct html *h)
188 {
189         struct htmlpair  tag[4];
190         struct tag      *t;
191
192         tag[0].key = ATTR_CHARSET;
193         tag[0].val = "utf-8";
194         print_otag(h, TAG_META, 1, tag);
195
196         /*
197          * Print a default style-sheet.
198          */
199         t = print_otag(h, TAG_STYLE, 0, NULL);
200         print_text(h, "table.head, table.foot { width: 100%; }\n"
201               "td.head-rtitle, td.foot-os { text-align: right; }\n"
202               "td.head-vol { text-align: center; }\n"
203               "table.foot td { width: 50%; }\n"
204               "table.head td { width: 33%; }\n"
205               "div.spacer { margin: 1em 0; }\n");
206         print_tagq(h, t);
207
208         if (h->style) {
209                 tag[0].key = ATTR_REL;
210                 tag[0].val = "stylesheet";
211                 tag[1].key = ATTR_HREF;
212                 tag[1].val = h->style;
213                 tag[2].key = ATTR_TYPE;
214                 tag[2].val = "text/css";
215                 tag[3].key = ATTR_MEDIA;
216                 tag[3].val = "all";
217                 print_otag(h, TAG_LINK, 4, tag);
218         }
219 }
220
221 static void
222 print_metaf(struct html *h, enum mandoc_esc deco)
223 {
224         enum htmlfont    font;
225
226         switch (deco) {
227         case ESCAPE_FONTPREV:
228                 font = h->metal;
229                 break;
230         case ESCAPE_FONTITALIC:
231                 font = HTMLFONT_ITALIC;
232                 break;
233         case ESCAPE_FONTBOLD:
234                 font = HTMLFONT_BOLD;
235                 break;
236         case ESCAPE_FONTBI:
237                 font = HTMLFONT_BI;
238                 break;
239         case ESCAPE_FONT:
240                 /* FALLTHROUGH */
241         case ESCAPE_FONTROMAN:
242                 font = HTMLFONT_NONE;
243                 break;
244         default:
245                 abort();
246                 /* NOTREACHED */
247         }
248
249         if (h->metaf) {
250                 print_tagq(h, h->metaf);
251                 h->metaf = NULL;
252         }
253
254         h->metal = h->metac;
255         h->metac = font;
256
257         switch (font) {
258         case HTMLFONT_ITALIC:
259                 h->metaf = print_otag(h, TAG_I, 0, NULL);
260                 break;
261         case HTMLFONT_BOLD:
262                 h->metaf = print_otag(h, TAG_B, 0, NULL);
263                 break;
264         case HTMLFONT_BI:
265                 h->metaf = print_otag(h, TAG_B, 0, NULL);
266                 print_otag(h, TAG_I, 0, NULL);
267                 break;
268         default:
269                 break;
270         }
271 }
272
273 int
274 html_strlen(const char *cp)
275 {
276         size_t           rsz;
277         int              skip, sz;
278
279         /*
280          * Account for escaped sequences within string length
281          * calculations.  This follows the logic in term_strlen() as we
282          * must calculate the width of produced strings.
283          * Assume that characters are always width of "1".  This is
284          * hacky, but it gets the job done for approximation of widths.
285          */
286
287         sz = 0;
288         skip = 0;
289         while (1) {
290                 rsz = strcspn(cp, "\\");
291                 if (rsz) {
292                         cp += rsz;
293                         if (skip) {
294                                 skip = 0;
295                                 rsz--;
296                         }
297                         sz += rsz;
298                 }
299                 if ('\0' == *cp)
300                         break;
301                 cp++;
302                 switch (mandoc_escape(&cp, NULL, NULL)) {
303                 case ESCAPE_ERROR:
304                         return(sz);
305                 case ESCAPE_UNICODE:
306                         /* FALLTHROUGH */
307                 case ESCAPE_NUMBERED:
308                         /* FALLTHROUGH */
309                 case ESCAPE_SPECIAL:
310                         if (skip)
311                                 skip = 0;
312                         else
313                                 sz++;
314                         break;
315                 case ESCAPE_SKIPCHAR:
316                         skip = 1;
317                         break;
318                 default:
319                         break;
320                 }
321         }
322         return(sz);
323 }
324
325 static int
326 print_escape(char c)
327 {
328
329         switch (c) {
330         case '<':
331                 printf("&lt;");
332                 break;
333         case '>':
334                 printf("&gt;");
335                 break;
336         case '&':
337                 printf("&amp;");
338                 break;
339         case '"':
340                 printf("&quot;");
341                 break;
342         case ASCII_NBRSP:
343                 putchar('-');
344                 break;
345         case ASCII_HYPH:
346                 putchar('-');
347                 /* FALLTHROUGH */
348         case ASCII_BREAK:
349                 break;
350         default:
351                 return(0);
352         }
353         return(1);
354 }
355
356 static int
357 print_encode(struct html *h, const char *p, int norecurse)
358 {
359         size_t           sz;
360         int              c, len, nospace;
361         const char      *seq;
362         enum mandoc_esc  esc;
363         static const char rejs[9] = { '\\', '<', '>', '&', '"',
364                 ASCII_NBRSP, ASCII_HYPH, ASCII_BREAK, '\0' };
365
366         nospace = 0;
367
368         while ('\0' != *p) {
369                 if (HTML_SKIPCHAR & h->flags && '\\' != *p) {
370                         h->flags &= ~HTML_SKIPCHAR;
371                         p++;
372                         continue;
373                 }
374
375                 sz = strcspn(p, rejs);
376
377                 fwrite(p, 1, sz, stdout);
378                 p += (int)sz;
379
380                 if ('\0' == *p)
381                         break;
382
383                 if (print_escape(*p++))
384                         continue;
385
386                 esc = mandoc_escape(&p, &seq, &len);
387                 if (ESCAPE_ERROR == esc)
388                         break;
389
390                 switch (esc) {
391                 case ESCAPE_FONT:
392                         /* FALLTHROUGH */
393                 case ESCAPE_FONTPREV:
394                         /* FALLTHROUGH */
395                 case ESCAPE_FONTBOLD:
396                         /* FALLTHROUGH */
397                 case ESCAPE_FONTITALIC:
398                         /* FALLTHROUGH */
399                 case ESCAPE_FONTBI:
400                         /* FALLTHROUGH */
401                 case ESCAPE_FONTROMAN:
402                         if (0 == norecurse)
403                                 print_metaf(h, esc);
404                         continue;
405                 case ESCAPE_SKIPCHAR:
406                         h->flags |= HTML_SKIPCHAR;
407                         continue;
408                 default:
409                         break;
410                 }
411
412                 if (h->flags & HTML_SKIPCHAR) {
413                         h->flags &= ~HTML_SKIPCHAR;
414                         continue;
415                 }
416
417                 switch (esc) {
418                 case ESCAPE_UNICODE:
419                         /* Skip past "u" header. */
420                         c = mchars_num2uc(seq + 1, len - 1);
421                         break;
422                 case ESCAPE_NUMBERED:
423                         c = mchars_num2char(seq, len);
424                         if (c < 0)
425                                 continue;
426                         break;
427                 case ESCAPE_SPECIAL:
428                         c = mchars_spec2cp(h->symtab, seq, len);
429                         if (c <= 0)
430                                 continue;
431                         break;
432                 case ESCAPE_NOSPACE:
433                         if ('\0' == *p)
434                                 nospace = 1;
435                         continue;
436                 default:
437                         continue;
438                 }
439                 if ((c < 0x20 && c != 0x09) ||
440                     (c > 0x7E && c < 0xA0))
441                         c = 0xFFFD;
442                 if (c > 0x7E)
443                         printf("&#%d;", c);
444                 else if ( ! print_escape(c))
445                         putchar(c);
446         }
447
448         return(nospace);
449 }
450
451 static void
452 print_attr(struct html *h, const char *key, const char *val)
453 {
454         printf(" %s=\"", key);
455         (void)print_encode(h, val, 1);
456         putchar('\"');
457 }
458
459 struct tag *
460 print_otag(struct html *h, enum htmltag tag,
461                 int sz, const struct htmlpair *p)
462 {
463         int              i;
464         struct tag      *t;
465
466         /* Push this tags onto the stack of open scopes. */
467
468         if ( ! (HTML_NOSTACK & htmltags[tag].flags)) {
469                 t = mandoc_malloc(sizeof(struct tag));
470                 t->tag = tag;
471                 t->next = h->tags.head;
472                 h->tags.head = t;
473         } else
474                 t = NULL;
475
476         if ( ! (HTML_NOSPACE & h->flags))
477                 if ( ! (HTML_CLRLINE & htmltags[tag].flags)) {
478                         /* Manage keeps! */
479                         if ( ! (HTML_KEEP & h->flags)) {
480                                 if (HTML_PREKEEP & h->flags)
481                                         h->flags |= HTML_KEEP;
482                                 putchar(' ');
483                         } else
484                                 printf("&#160;");
485                 }
486
487         if ( ! (h->flags & HTML_NONOSPACE))
488                 h->flags &= ~HTML_NOSPACE;
489         else
490                 h->flags |= HTML_NOSPACE;
491
492         /* Print out the tag name and attributes. */
493
494         printf("<%s", htmltags[tag].name);
495         for (i = 0; i < sz; i++)
496                 print_attr(h, htmlattrs[p[i].key], p[i].val);
497
498         /* Accommodate for "well-formed" singleton escaping. */
499
500         if (HTML_AUTOCLOSE & htmltags[tag].flags)
501                 putchar('/');
502
503         putchar('>');
504
505         h->flags |= HTML_NOSPACE;
506
507         if ((HTML_AUTOCLOSE | HTML_CLRLINE) & htmltags[tag].flags)
508                 putchar('\n');
509
510         return(t);
511 }
512
513 static void
514 print_ctag(struct html *h, enum htmltag tag)
515 {
516
517         printf("</%s>", htmltags[tag].name);
518         if (HTML_CLRLINE & htmltags[tag].flags) {
519                 h->flags |= HTML_NOSPACE;
520                 putchar('\n');
521         }
522 }
523
524 void
525 print_gen_decls(struct html *h)
526 {
527
528         puts("<!DOCTYPE html>");
529 }
530
531 void
532 print_text(struct html *h, const char *word)
533 {
534
535         if ( ! (HTML_NOSPACE & h->flags)) {
536                 /* Manage keeps! */
537                 if ( ! (HTML_KEEP & h->flags)) {
538                         if (HTML_PREKEEP & h->flags)
539                                 h->flags |= HTML_KEEP;
540                         putchar(' ');
541                 } else
542                         printf("&#160;");
543         }
544
545         assert(NULL == h->metaf);
546         switch (h->metac) {
547         case HTMLFONT_ITALIC:
548                 h->metaf = print_otag(h, TAG_I, 0, NULL);
549                 break;
550         case HTMLFONT_BOLD:
551                 h->metaf = print_otag(h, TAG_B, 0, NULL);
552                 break;
553         case HTMLFONT_BI:
554                 h->metaf = print_otag(h, TAG_B, 0, NULL);
555                 print_otag(h, TAG_I, 0, NULL);
556                 break;
557         default:
558                 break;
559         }
560
561         assert(word);
562         if ( ! print_encode(h, word, 0)) {
563                 if ( ! (h->flags & HTML_NONOSPACE))
564                         h->flags &= ~HTML_NOSPACE;
565                 h->flags &= ~HTML_NONEWLINE;
566         } else
567                 h->flags |= HTML_NOSPACE | HTML_NONEWLINE;
568
569         if (h->metaf) {
570                 print_tagq(h, h->metaf);
571                 h->metaf = NULL;
572         }
573
574         h->flags &= ~HTML_IGNDELIM;
575 }
576
577 void
578 print_tagq(struct html *h, const struct tag *until)
579 {
580         struct tag      *tag;
581
582         while ((tag = h->tags.head) != NULL) {
583                 /*
584                  * Remember to close out and nullify the current
585                  * meta-font and table, if applicable.
586                  */
587                 if (tag == h->metaf)
588                         h->metaf = NULL;
589                 if (tag == h->tblt)
590                         h->tblt = NULL;
591                 print_ctag(h, tag->tag);
592                 h->tags.head = tag->next;
593                 free(tag);
594                 if (until && tag == until)
595                         return;
596         }
597 }
598
599 void
600 print_stagq(struct html *h, const struct tag *suntil)
601 {
602         struct tag      *tag;
603
604         while ((tag = h->tags.head) != NULL) {
605                 if (suntil && tag == suntil)
606                         return;
607                 /*
608                  * Remember to close out and nullify the current
609                  * meta-font and table, if applicable.
610                  */
611                 if (tag == h->metaf)
612                         h->metaf = NULL;
613                 if (tag == h->tblt)
614                         h->tblt = NULL;
615                 print_ctag(h, tag->tag);
616                 h->tags.head = tag->next;
617                 free(tag);
618         }
619 }
620
621 void
622 print_paragraph(struct html *h)
623 {
624         struct tag      *t;
625         struct htmlpair  tag;
626
627         PAIR_CLASS_INIT(&tag, "spacer");
628         t = print_otag(h, TAG_DIV, 1, &tag);
629         print_tagq(h, t);
630 }
631
632
633 void
634 bufinit(struct html *h)
635 {
636
637         h->buf[0] = '\0';
638         h->buflen = 0;
639 }
640
641 void
642 bufcat_style(struct html *h, const char *key, const char *val)
643 {
644
645         bufcat(h, key);
646         bufcat(h, ":");
647         bufcat(h, val);
648         bufcat(h, ";");
649 }
650
651 void
652 bufcat(struct html *h, const char *p)
653 {
654
655         /*
656          * XXX This is broken and not easy to fix.
657          * When using the -Oincludes option, buffmt_includes()
658          * may pass in strings overrunning BUFSIZ, causing a crash.
659          */
660
661         h->buflen = strlcat(h->buf, p, BUFSIZ);
662         assert(h->buflen < BUFSIZ);
663 }
664
665 void
666 bufcat_fmt(struct html *h, const char *fmt, ...)
667 {
668         va_list          ap;
669
670         va_start(ap, fmt);
671         (void)vsnprintf(h->buf + (int)h->buflen,
672             BUFSIZ - h->buflen - 1, fmt, ap);
673         va_end(ap);
674         h->buflen = strlen(h->buf);
675 }
676
677 static void
678 bufncat(struct html *h, const char *p, size_t sz)
679 {
680
681         assert(h->buflen + sz + 1 < BUFSIZ);
682         strncat(h->buf, p, sz);
683         h->buflen += sz;
684 }
685
686 void
687 buffmt_includes(struct html *h, const char *name)
688 {
689         const char      *p, *pp;
690
691         pp = h->base_includes;
692
693         bufinit(h);
694         while (NULL != (p = strchr(pp, '%'))) {
695                 bufncat(h, pp, (size_t)(p - pp));
696                 switch (*(p + 1)) {
697                 case'I':
698                         bufcat(h, name);
699                         break;
700                 default:
701                         bufncat(h, p, 2);
702                         break;
703                 }
704                 pp = p + 2;
705         }
706         if (pp)
707                 bufcat(h, pp);
708 }
709
710 void
711 buffmt_man(struct html *h, const char *name, const char *sec)
712 {
713         const char      *p, *pp;
714
715         pp = h->base_man;
716
717         bufinit(h);
718         while (NULL != (p = strchr(pp, '%'))) {
719                 bufncat(h, pp, (size_t)(p - pp));
720                 switch (*(p + 1)) {
721                 case 'S':
722                         bufcat(h, sec ? sec : "1");
723                         break;
724                 case 'N':
725                         bufcat_fmt(h, "%s", name);
726                         break;
727                 default:
728                         bufncat(h, p, 2);
729                         break;
730                 }
731                 pp = p + 2;
732         }
733         if (pp)
734                 bufcat(h, pp);
735 }
736
737 void
738 bufcat_su(struct html *h, const char *p, const struct roffsu *su)
739 {
740         double           v;
741
742         v = su->scale;
743         if (SCALE_MM == su->unit && 0.0 == (v /= 100.0))
744                 v = 1.0;
745         else if (SCALE_BU == su->unit)
746                 v /= 24.0;
747
748         bufcat_fmt(h, "%s: %.2f%s;", p, v, roffscales[su->unit]);
749 }
750
751 void
752 bufcat_id(struct html *h, const char *src)
753 {
754
755         /* Cf. <http://www.w3.org/TR/html4/types.html#h-6.2>. */
756
757         while ('\0' != *src)
758                 bufcat_fmt(h, "%.2x", *src++);
759 }