1 /* $Id: html.c,v 1.255 2019/04/30 15:53:00 schwarze Exp $ */
3 * Copyright (c) 2008-2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2011-2015, 2017-2019 Ingo Schwarze <schwarze@openbsd.org>
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.
10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS 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.
20 #include <sys/types.h>
33 #include "mandoc_aux.h"
34 #include "mandoc_ohash.h"
45 #define HTML_NOSTACK (1 << 0)
46 #define HTML_AUTOCLOSE (1 << 1)
47 #define HTML_NLBEFORE (1 << 2)
48 #define HTML_NLBEGIN (1 << 3)
49 #define HTML_NLEND (1 << 4)
50 #define HTML_NLAFTER (1 << 5)
51 #define HTML_NLAROUND (HTML_NLBEFORE | HTML_NLAFTER)
52 #define HTML_NLINSIDE (HTML_NLBEGIN | HTML_NLEND)
53 #define HTML_NLALL (HTML_NLAROUND | HTML_NLINSIDE)
54 #define HTML_INDENT (1 << 6)
55 #define HTML_NOINDENT (1 << 7)
58 static const struct htmldata htmltags[TAG_MAX] = {
60 {"head", HTML_NLALL | HTML_INDENT},
62 {"meta", HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL},
63 {"title", HTML_NLAROUND},
64 {"div", HTML_NLAROUND},
66 {"section", HTML_NLALL},
67 {"h1", HTML_NLAROUND},
68 {"h2", HTML_NLAROUND},
70 {"link", HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL},
71 {"br", HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL},
73 {"table", HTML_NLALL | HTML_INDENT},
74 {"tr", HTML_NLALL | HTML_INDENT},
75 {"td", HTML_NLAROUND},
76 {"li", HTML_NLAROUND | HTML_INDENT},
77 {"ul", HTML_NLALL | HTML_INDENT},
78 {"ol", HTML_NLALL | HTML_INDENT},
79 {"dl", HTML_NLALL | HTML_INDENT},
80 {"dt", HTML_NLAROUND},
81 {"dd", HTML_NLAROUND | HTML_INDENT},
82 {"p", HTML_NLAROUND | HTML_INDENT},
83 {"pre", HTML_NLALL | HTML_NOINDENT},
90 {"style", HTML_NLALL | HTML_INDENT},
91 {"math", HTML_NLALL | HTML_INDENT},
110 /* Avoid duplicate HTML id= attributes. */
111 static struct ohash id_unique;
113 static void html_reset_internal(struct html *);
114 static void print_byte(struct html *, char);
115 static void print_endword(struct html *);
116 static void print_indent(struct html *);
117 static void print_word(struct html *, const char *);
119 static void print_ctag(struct html *, struct tag *);
120 static int print_escape(struct html *, char);
121 static int print_encode(struct html *, const char *, const char *, int);
122 static void print_href(struct html *, const char *, const char *, int);
123 static void print_metaf(struct html *);
127 html_alloc(const struct manoutput *outopts)
131 h = mandoc_calloc(1, sizeof(struct html));
134 h->style = outopts->style;
135 if ((h->base_man1 = outopts->man) == NULL)
137 else if ((h->base_man2 = strchr(h->base_man1, ';')) != NULL)
138 *h->base_man2++ = '\0';
139 h->base_includes = outopts->includes;
140 if (outopts->fragment)
141 h->oflags |= HTML_FRAGMENT;
143 h->oflags |= HTML_TOC;
145 mandoc_ohash_init(&id_unique, 4, 0);
151 html_reset_internal(struct html *h)
157 while ((tag = h->tag) != NULL) {
161 cp = ohash_first(&id_unique, &slot);
164 cp = ohash_next(&id_unique, &slot);
166 ohash_delete(&id_unique);
172 html_reset_internal(p);
173 mandoc_ohash_init(&id_unique, 4, 0);
179 html_reset_internal(p);
184 print_gen_head(struct html *h)
188 print_otag(h, TAG_META, "?", "charset", "utf-8");
189 if (h->style != NULL) {
190 print_otag(h, TAG_LINK, "?h??", "rel", "stylesheet",
191 h->style, "type", "text/css", "media", "all");
196 * Print a minimal embedded style sheet.
199 t = print_otag(h, TAG_STYLE, "");
200 print_text(h, "table.head, table.foot { width: 100%; }");
202 print_text(h, "td.head-rtitle, td.foot-os { text-align: right; }");
204 print_text(h, "td.head-vol { text-align: center; }");
206 print_text(h, "div.Pp { margin: 1ex 0ex; }");
208 print_text(h, "div.Nd, div.Bf, div.Op { display: inline; }");
210 print_text(h, "span.Pa, span.Ad { font-style: italic; }");
212 print_text(h, "span.Ms { font-weight: bold; }");
214 print_text(h, "dl.Bl-diag ");
216 print_text(h, " dt { font-weight: bold; }");
218 print_text(h, "code.Nm, code.Fl, code.Cm, code.Ic, "
219 "code.In, code.Fd, code.Fn,");
221 print_text(h, "code.Cd { font-weight: bold; "
222 "font-family: inherit; }");
227 html_setfont(struct html *h, enum mandoc_esc font)
230 case ESCAPE_FONTPREV:
233 case ESCAPE_FONTITALIC:
234 case ESCAPE_FONTBOLD:
237 case ESCAPE_FONTROMAN:
240 font = ESCAPE_FONTROMAN;
251 print_metaf(struct html *h)
254 print_tagq(h, h->metaf);
258 case ESCAPE_FONTITALIC:
259 h->metaf = print_otag(h, TAG_I, "");
261 case ESCAPE_FONTBOLD:
262 h->metaf = print_otag(h, TAG_B, "");
265 h->metaf = print_otag(h, TAG_B, "");
266 print_otag(h, TAG_I, "");
269 h->metaf = print_otag(h, TAG_SPAN, "c", "Li");
277 html_close_paragraph(struct html *h)
281 for (t = h->tag; t != NULL && t->closed == 0; t = t->next) {
298 * ROFF_nf switches to no-fill mode, ROFF_fi to fill mode.
299 * TOKEN_NONE does not switch. The old mode is returned.
302 html_fillmode(struct html *h, enum roff_tok want)
307 for (t = h->tag; t != NULL; t = t->next)
308 if (t->tag == TAG_PRE)
311 had = t == NULL ? ROFF_fi : ROFF_nf;
319 html_close_paragraph(h);
320 print_otag(h, TAG_PRE, "");
332 html_make_id(const struct roff_node *n, int unique)
334 const struct roff_node *nch;
335 char *buf, *bufs, *cp;
339 for (nch = n->child; nch != NULL; nch = nch->next)
340 if (nch->type != ROFFT_TEXT)
349 * In ID attributes, only use ASCII characters that are
350 * permitted in URL-fragment strings according to the
352 * https://url.spec.whatwg.org/#url-fragment-string
355 for (cp = buf; *cp != '\0'; cp++)
356 if (isalnum((unsigned char)*cp) == 0 &&
357 strchr("!$&'()*+,-./:;=?@_~", *cp) == NULL)
363 /* Avoid duplicate HTML id= attributes. */
367 slot = ohash_qlookup(&id_unique, buf);
368 cp = ohash_find(&id_unique, slot);
372 if (++suffix > 127) {
376 mandoc_asprintf(&bufs, "%s_%d", buf, suffix);
377 slot = ohash_qlookup(&id_unique, bufs);
378 cp = ohash_find(&id_unique, slot);
383 ohash_insert(&id_unique, slot, buf);
388 print_escape(struct html *h, char c)
393 print_word(h, "<");
396 print_word(h, ">");
399 print_word(h, "&");
402 print_word(h, """);
405 print_word(h, " ");
419 print_encode(struct html *h, const char *p, const char *pend, int norecurse)
424 int c, len, breakline, nospace;
426 static const char rejs[10] = { ' ', '\\', '<', '>', '&', '"',
427 ASCII_NBRSP, ASCII_HYPH, ASCII_BREAK, '\0' };
430 pend = strchr(p, '\0');
436 if (HTML_SKIPCHAR & h->flags && '\\' != *p) {
437 h->flags &= ~HTML_SKIPCHAR;
442 for (sz = strcspn(p, rejs); sz-- && p < pend; p++)
446 (p >= pend || *p == ' ' || *p == ASCII_NBRSP)) {
447 print_otag(h, TAG_BR, "");
449 while (p < pend && (*p == ' ' || *p == ASCII_NBRSP))
463 if (print_escape(h, *p++))
466 esc = mandoc_escape(&p, &seq, &len);
469 case ESCAPE_FONTPREV:
470 case ESCAPE_FONTBOLD:
471 case ESCAPE_FONTITALIC:
474 case ESCAPE_FONTROMAN:
475 if (0 == norecurse) {
476 h->flags |= HTML_NOSPACE;
477 if (html_setfont(h, esc))
479 h->flags &= ~HTML_NOSPACE;
482 case ESCAPE_SKIPCHAR:
483 h->flags |= HTML_SKIPCHAR;
491 if (h->flags & HTML_SKIPCHAR) {
492 h->flags &= ~HTML_SKIPCHAR;
498 /* Skip past "u" header. */
499 c = mchars_num2uc(seq + 1, len - 1);
501 case ESCAPE_NUMBERED:
502 c = mchars_num2char(seq, len);
507 c = mchars_spec2cp(seq, len);
515 print_word(h, "html");
524 case ESCAPE_OVERSTRIKE:
532 if ((c < 0x20 && c != 0x09) ||
533 (c > 0x7E && c < 0xA0))
536 (void)snprintf(numbuf, sizeof(numbuf), "&#x%.4X;", c);
537 print_word(h, numbuf);
538 } else if (print_escape(h, c) == 0)
546 print_href(struct html *h, const char *name, const char *sec, int man)
554 if (h->base_man2 != NULL) {
555 mandoc_asprintf(&filename, "%s.%s", name, sec);
556 if (stat(filename, &sb) == -1)
561 pp = h->base_includes;
563 while ((p = strchr(pp, '%')) != NULL) {
564 print_encode(h, pp, p, 1);
565 if (man && p[1] == 'S') {
569 print_encode(h, sec, NULL, 1);
570 } else if ((man && p[1] == 'N') ||
571 (man == 0 && p[1] == 'I'))
572 print_encode(h, name, NULL, 1);
574 print_encode(h, p, p + 2, 1);
578 print_encode(h, pp, NULL, 1);
582 print_otag(struct html *h, enum htmltag tag, const char *fmt, ...)
588 int style_written, tflags;
590 tflags = htmltags[tag].flags;
592 /* Push this tag onto the stack of open scopes. */
594 if ((tflags & HTML_NOSTACK) == 0) {
595 t = mandoc_malloc(sizeof(struct tag));
604 if (tflags & HTML_NLBEFORE)
608 else if ((h->flags & HTML_NOSPACE) == 0) {
609 if (h->flags & HTML_KEEP)
610 print_word(h, " ");
612 if (h->flags & HTML_PREKEEP)
613 h->flags |= HTML_KEEP;
618 if ( ! (h->flags & HTML_NONOSPACE))
619 h->flags &= ~HTML_NOSPACE;
621 h->flags |= HTML_NOSPACE;
623 /* Print out the tag name and attributes. */
626 print_word(h, htmltags[tag].name);
630 while (*fmt != '\0' && *fmt != 's') {
632 /* Parse attributes and arguments. */
634 arg1 = va_arg(ap, char *);
648 arg1 = va_arg(ap, char *);
654 arg2 = va_arg(ap, char *);
658 /* Print the attributes. */
666 print_href(h, arg1, NULL, 0);
670 print_href(h, arg1, arg2, 1);
675 print_encode(h, arg1, NULL, 1);
679 print_encode(h, arg1, NULL, 1);
686 while (*fmt++ == 's') {
687 arg1 = va_arg(ap, char *);
688 arg2 = va_arg(ap, char *);
692 if (style_written == 0) {
693 print_word(h, "style=\"");
707 /* Accommodate for "well-formed" singleton escaping. */
709 if (HTML_AUTOCLOSE & htmltags[tag].flags)
714 if (tflags & HTML_NLBEGIN)
717 h->flags |= HTML_NOSPACE;
719 if (tflags & HTML_INDENT)
721 if (tflags & HTML_NOINDENT)
728 print_ctag(struct html *h, struct tag *tag)
732 if (tag->closed == 0) {
739 tflags = htmltags[tag->tag].flags;
740 if (tflags & HTML_INDENT)
742 if (tflags & HTML_NOINDENT)
744 if (tflags & HTML_NLEND)
749 print_word(h, htmltags[tag->tag].name);
751 if (tflags & HTML_NLAFTER)
754 if (tag->refcnt == 0) {
761 print_gen_decls(struct html *h)
763 print_word(h, "<!DOCTYPE html>");
768 print_gen_comment(struct html *h, struct roff_node *n)
772 print_word(h, "<!-- This is an automatically generated file."
776 while (n != NULL && n->type == ROFFT_COMMENT) {
777 if (strstr(n->string, "-->") == NULL &&
778 (wantblank || *n->string != '\0')) {
781 print_word(h, n->string);
782 wantblank = *n->string != '\0';
788 print_word(h, " -->");
794 print_text(struct html *h, const char *word)
796 if (h->col && (h->flags & HTML_NOSPACE) == 0) {
797 if ( ! (HTML_KEEP & h->flags)) {
798 if (HTML_PREKEEP & h->flags)
799 h->flags |= HTML_KEEP;
802 print_word(h, " ");
805 assert(h->metaf == NULL);
808 if ( ! print_encode(h, word, NULL, 0)) {
809 if ( ! (h->flags & HTML_NONOSPACE))
810 h->flags &= ~HTML_NOSPACE;
811 h->flags &= ~HTML_NONEWLINE;
813 h->flags |= HTML_NOSPACE | HTML_NONEWLINE;
815 if (h->metaf != NULL) {
816 print_tagq(h, h->metaf);
820 h->flags &= ~HTML_IGNDELIM;
824 print_tagq(struct html *h, const struct tag *until)
826 struct tag *this, *next;
828 for (this = h->tag; this != NULL; this = next) {
829 next = this == until ? NULL : this->next;
835 * Close out all open elements up to but excluding suntil.
836 * Note that a paragraph just inside stays open together with it
837 * because paragraphs include subsequent phrasing content.
840 print_stagq(struct html *h, const struct tag *suntil)
842 struct tag *this, *next;
844 for (this = h->tag; this != NULL; this = next) {
846 if (this == suntil || (next == suntil &&
847 (this->tag == TAG_P || this->tag == TAG_PRE)))
854 /***********************************************************************
855 * Low level output functions.
856 * They implement line breaking using a short static buffer.
857 ***********************************************************************/
860 * Buffer one HTML output byte.
861 * If the buffer is full, flush and deactivate it and start a new line.
862 * If the buffer is inactive, print directly.
865 print_byte(struct html *h, char c)
867 if ((h->flags & HTML_BUFFER) == 0) {
873 if (h->col + h->bufcol < sizeof(h->buf)) {
874 h->buf[h->bufcol++] = c;
883 fwrite(h->buf, h->bufcol, 1, stdout);
885 h->col = (h->indent + 1) * 2 + h->bufcol + 1;
887 h->flags &= ~HTML_BUFFER;
891 * If something was printed on the current output line, end it.
892 * Not to be called right after print_indent().
895 print_endline(struct html *h)
902 fwrite(h->buf, h->bufcol, 1, stdout);
907 h->flags |= HTML_NOSPACE;
908 h->flags &= ~HTML_BUFFER;
912 * Flush the HTML output buffer.
913 * If it is inactive, activate it.
916 print_endword(struct html *h)
923 if ((h->flags & HTML_BUFFER) == 0) {
925 h->flags |= HTML_BUFFER;
926 } else if (h->bufcol) {
928 fwrite(h->buf, h->bufcol, 1, stdout);
929 h->col += h->bufcol + 1;
935 * If at the beginning of a new output line,
936 * perform indentation and mark the line as containing output.
937 * Make sure to really produce some output right afterwards,
938 * but do not use print_otag() for producing it.
941 print_indent(struct html *h)
948 if (h->noindent == 0) {
949 h->col = h->indent * 2;
950 for (i = 0; i < h->col; i++)
953 h->flags &= ~HTML_NOSPACE;
957 * Print or buffer some characters
958 * depending on the current HTML output buffer state.
961 print_word(struct html *h, const char *cp)
964 print_byte(h, *cp++);