]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/blob - contrib/mdocml/html.c
MFV r288408:
[FreeBSD/FreeBSD.git] / contrib / mdocml / html.c
1 /*      $Id: html.c,v 1.185 2015/01/21 20:33:25 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008-2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2011-2015 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 *, struct tag *);
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                         /* FALLTHROUGH */
311                 case ESCAPE_OVERSTRIKE:
312                         if (skip)
313                                 skip = 0;
314                         else
315                                 sz++;
316                         break;
317                 case ESCAPE_SKIPCHAR:
318                         skip = 1;
319                         break;
320                 default:
321                         break;
322                 }
323         }
324         return(sz);
325 }
326
327 static int
328 print_escape(char c)
329 {
330
331         switch (c) {
332         case '<':
333                 printf("&lt;");
334                 break;
335         case '>':
336                 printf("&gt;");
337                 break;
338         case '&':
339                 printf("&amp;");
340                 break;
341         case '"':
342                 printf("&quot;");
343                 break;
344         case ASCII_NBRSP:
345                 putchar('-');
346                 break;
347         case ASCII_HYPH:
348                 putchar('-');
349                 /* FALLTHROUGH */
350         case ASCII_BREAK:
351                 break;
352         default:
353                 return(0);
354         }
355         return(1);
356 }
357
358 static int
359 print_encode(struct html *h, const char *p, int norecurse)
360 {
361         size_t           sz;
362         int              c, len, nospace;
363         const char      *seq;
364         enum mandoc_esc  esc;
365         static const char rejs[9] = { '\\', '<', '>', '&', '"',
366                 ASCII_NBRSP, ASCII_HYPH, ASCII_BREAK, '\0' };
367
368         nospace = 0;
369
370         while ('\0' != *p) {
371                 if (HTML_SKIPCHAR & h->flags && '\\' != *p) {
372                         h->flags &= ~HTML_SKIPCHAR;
373                         p++;
374                         continue;
375                 }
376
377                 sz = strcspn(p, rejs);
378
379                 fwrite(p, 1, sz, stdout);
380                 p += (int)sz;
381
382                 if ('\0' == *p)
383                         break;
384
385                 if (print_escape(*p++))
386                         continue;
387
388                 esc = mandoc_escape(&p, &seq, &len);
389                 if (ESCAPE_ERROR == esc)
390                         break;
391
392                 switch (esc) {
393                 case ESCAPE_FONT:
394                         /* FALLTHROUGH */
395                 case ESCAPE_FONTPREV:
396                         /* FALLTHROUGH */
397                 case ESCAPE_FONTBOLD:
398                         /* FALLTHROUGH */
399                 case ESCAPE_FONTITALIC:
400                         /* FALLTHROUGH */
401                 case ESCAPE_FONTBI:
402                         /* FALLTHROUGH */
403                 case ESCAPE_FONTROMAN:
404                         if (0 == norecurse)
405                                 print_metaf(h, esc);
406                         continue;
407                 case ESCAPE_SKIPCHAR:
408                         h->flags |= HTML_SKIPCHAR;
409                         continue;
410                 default:
411                         break;
412                 }
413
414                 if (h->flags & HTML_SKIPCHAR) {
415                         h->flags &= ~HTML_SKIPCHAR;
416                         continue;
417                 }
418
419                 switch (esc) {
420                 case ESCAPE_UNICODE:
421                         /* Skip past "u" header. */
422                         c = mchars_num2uc(seq + 1, len - 1);
423                         break;
424                 case ESCAPE_NUMBERED:
425                         c = mchars_num2char(seq, len);
426                         if (c < 0)
427                                 continue;
428                         break;
429                 case ESCAPE_SPECIAL:
430                         c = mchars_spec2cp(h->symtab, seq, len);
431                         if (c <= 0)
432                                 continue;
433                         break;
434                 case ESCAPE_NOSPACE:
435                         if ('\0' == *p)
436                                 nospace = 1;
437                         continue;
438                 case ESCAPE_OVERSTRIKE:
439                         if (len == 0)
440                                 continue;
441                         c = seq[len - 1];
442                         break;
443                 default:
444                         continue;
445                 }
446                 if ((c < 0x20 && c != 0x09) ||
447                     (c > 0x7E && c < 0xA0))
448                         c = 0xFFFD;
449                 if (c > 0x7E)
450                         printf("&#%d;", c);
451                 else if ( ! print_escape(c))
452                         putchar(c);
453         }
454
455         return(nospace);
456 }
457
458 static void
459 print_attr(struct html *h, const char *key, const char *val)
460 {
461         printf(" %s=\"", key);
462         (void)print_encode(h, val, 1);
463         putchar('\"');
464 }
465
466 struct tag *
467 print_otag(struct html *h, enum htmltag tag,
468                 int sz, const struct htmlpair *p)
469 {
470         int              i;
471         struct tag      *t;
472
473         /* Push this tags onto the stack of open scopes. */
474
475         if ( ! (HTML_NOSTACK & htmltags[tag].flags)) {
476                 t = mandoc_malloc(sizeof(struct tag));
477                 t->tag = tag;
478                 t->next = h->tags.head;
479                 h->tags.head = t;
480         } else
481                 t = NULL;
482
483         if ( ! (HTML_NOSPACE & h->flags))
484                 if ( ! (HTML_CLRLINE & htmltags[tag].flags)) {
485                         /* Manage keeps! */
486                         if ( ! (HTML_KEEP & h->flags)) {
487                                 if (HTML_PREKEEP & h->flags)
488                                         h->flags |= HTML_KEEP;
489                                 putchar(' ');
490                         } else
491                                 printf("&#160;");
492                 }
493
494         if ( ! (h->flags & HTML_NONOSPACE))
495                 h->flags &= ~HTML_NOSPACE;
496         else
497                 h->flags |= HTML_NOSPACE;
498
499         /* Print out the tag name and attributes. */
500
501         printf("<%s", htmltags[tag].name);
502         for (i = 0; i < sz; i++)
503                 print_attr(h, htmlattrs[p[i].key], p[i].val);
504
505         /* Accommodate for "well-formed" singleton escaping. */
506
507         if (HTML_AUTOCLOSE & htmltags[tag].flags)
508                 putchar('/');
509
510         putchar('>');
511
512         h->flags |= HTML_NOSPACE;
513
514         if ((HTML_AUTOCLOSE | HTML_CLRLINE) & htmltags[tag].flags)
515                 putchar('\n');
516
517         return(t);
518 }
519
520 static void
521 print_ctag(struct html *h, struct tag *tag)
522 {
523
524         /*
525          * Remember to close out and nullify the current
526          * meta-font and table, if applicable.
527          */
528         if (tag == h->metaf)
529                 h->metaf = NULL;
530         if (tag == h->tblt)
531                 h->tblt = NULL;
532
533         printf("</%s>", htmltags[tag->tag].name);
534         if (HTML_CLRLINE & htmltags[tag->tag].flags) {
535                 h->flags |= HTML_NOSPACE;
536                 putchar('\n');
537         }
538
539         h->tags.head = tag->next;
540         free(tag);
541 }
542
543 void
544 print_gen_decls(struct html *h)
545 {
546
547         puts("<!DOCTYPE html>");
548 }
549
550 void
551 print_text(struct html *h, const char *word)
552 {
553
554         if ( ! (HTML_NOSPACE & h->flags)) {
555                 /* Manage keeps! */
556                 if ( ! (HTML_KEEP & h->flags)) {
557                         if (HTML_PREKEEP & h->flags)
558                                 h->flags |= HTML_KEEP;
559                         putchar(' ');
560                 } else
561                         printf("&#160;");
562         }
563
564         assert(NULL == h->metaf);
565         switch (h->metac) {
566         case HTMLFONT_ITALIC:
567                 h->metaf = print_otag(h, TAG_I, 0, NULL);
568                 break;
569         case HTMLFONT_BOLD:
570                 h->metaf = print_otag(h, TAG_B, 0, NULL);
571                 break;
572         case HTMLFONT_BI:
573                 h->metaf = print_otag(h, TAG_B, 0, NULL);
574                 print_otag(h, TAG_I, 0, NULL);
575                 break;
576         default:
577                 break;
578         }
579
580         assert(word);
581         if ( ! print_encode(h, word, 0)) {
582                 if ( ! (h->flags & HTML_NONOSPACE))
583                         h->flags &= ~HTML_NOSPACE;
584                 h->flags &= ~HTML_NONEWLINE;
585         } else
586                 h->flags |= HTML_NOSPACE | HTML_NONEWLINE;
587
588         if (h->metaf) {
589                 print_tagq(h, h->metaf);
590                 h->metaf = NULL;
591         }
592
593         h->flags &= ~HTML_IGNDELIM;
594 }
595
596 void
597 print_tagq(struct html *h, const struct tag *until)
598 {
599         struct tag      *tag;
600
601         while ((tag = h->tags.head) != NULL) {
602                 print_ctag(h, tag);
603                 if (until && tag == until)
604                         return;
605         }
606 }
607
608 void
609 print_stagq(struct html *h, const struct tag *suntil)
610 {
611         struct tag      *tag;
612
613         while ((tag = h->tags.head) != NULL) {
614                 if (suntil && tag == suntil)
615                         return;
616                 print_ctag(h, tag);
617         }
618 }
619
620 void
621 print_paragraph(struct html *h)
622 {
623         struct tag      *t;
624         struct htmlpair  tag;
625
626         PAIR_CLASS_INIT(&tag, "spacer");
627         t = print_otag(h, TAG_DIV, 1, &tag);
628         print_tagq(h, t);
629 }
630
631
632 void
633 bufinit(struct html *h)
634 {
635
636         h->buf[0] = '\0';
637         h->buflen = 0;
638 }
639
640 void
641 bufcat_style(struct html *h, const char *key, const char *val)
642 {
643
644         bufcat(h, key);
645         bufcat(h, ":");
646         bufcat(h, val);
647         bufcat(h, ";");
648 }
649
650 void
651 bufcat(struct html *h, const char *p)
652 {
653
654         /*
655          * XXX This is broken and not easy to fix.
656          * When using the -Oincludes option, buffmt_includes()
657          * may pass in strings overrunning BUFSIZ, causing a crash.
658          */
659
660         h->buflen = strlcat(h->buf, p, BUFSIZ);
661         assert(h->buflen < BUFSIZ);
662 }
663
664 void
665 bufcat_fmt(struct html *h, const char *fmt, ...)
666 {
667         va_list          ap;
668
669         va_start(ap, fmt);
670         (void)vsnprintf(h->buf + (int)h->buflen,
671             BUFSIZ - h->buflen - 1, fmt, ap);
672         va_end(ap);
673         h->buflen = strlen(h->buf);
674 }
675
676 static void
677 bufncat(struct html *h, const char *p, size_t sz)
678 {
679
680         assert(h->buflen + sz + 1 < BUFSIZ);
681         strncat(h->buf, p, sz);
682         h->buflen += sz;
683 }
684
685 void
686 buffmt_includes(struct html *h, const char *name)
687 {
688         const char      *p, *pp;
689
690         pp = h->base_includes;
691
692         bufinit(h);
693         while (NULL != (p = strchr(pp, '%'))) {
694                 bufncat(h, pp, (size_t)(p - pp));
695                 switch (*(p + 1)) {
696                 case'I':
697                         bufcat(h, name);
698                         break;
699                 default:
700                         bufncat(h, p, 2);
701                         break;
702                 }
703                 pp = p + 2;
704         }
705         if (pp)
706                 bufcat(h, pp);
707 }
708
709 void
710 buffmt_man(struct html *h, const char *name, const char *sec)
711 {
712         const char      *p, *pp;
713
714         pp = h->base_man;
715
716         bufinit(h);
717         while (NULL != (p = strchr(pp, '%'))) {
718                 bufncat(h, pp, (size_t)(p - pp));
719                 switch (*(p + 1)) {
720                 case 'S':
721                         bufcat(h, sec ? sec : "1");
722                         break;
723                 case 'N':
724                         bufcat_fmt(h, "%s", name);
725                         break;
726                 default:
727                         bufncat(h, p, 2);
728                         break;
729                 }
730                 pp = p + 2;
731         }
732         if (pp)
733                 bufcat(h, pp);
734 }
735
736 void
737 bufcat_su(struct html *h, const char *p, const struct roffsu *su)
738 {
739         double           v;
740
741         v = su->scale;
742         if (SCALE_MM == su->unit && 0.0 == (v /= 100.0))
743                 v = 1.0;
744         else if (SCALE_BU == su->unit)
745                 v /= 24.0;
746
747         bufcat_fmt(h, "%s: %.2f%s;", p, v, roffscales[su->unit]);
748 }
749
750 void
751 bufcat_id(struct html *h, const char *src)
752 {
753
754         /* Cf. <http://www.w3.org/TR/html4/types.html#h-6.2>. */
755
756         while ('\0' != *src)
757                 bufcat_fmt(h, "%.2x", *src++);
758 }