2 * HTML backend for Halibut
8 * - I'm never entirely convinced that having a fragment link to
9 * come in at the start of the real text in the file is
10 * sensible. Perhaps for the topmost section in the file, no
11 * fragment should be used? (Though it should probably still be
12 * _there_ even if unused.)
21 #define is_heading_type(type) ( (type) == para_Title || \
22 (type) == para_Chapter || \
23 (type) == para_Appendix || \
24 (type) == para_UnnumberedChapter || \
25 (type) == para_Heading || \
26 (type) == para_Subsect)
28 #define heading_depth(p) ( (p)->type == para_Subsect ? (p)->aux + 1 : \
29 (p)->type == para_Heading ? 1 : \
30 (p)->type == para_Title ? -1 : 0 )
34 wchar_t *number_suffix
;
39 sectlevel achapter
, *asect
;
40 int *contents_depths
; /* 0=main, 1=chapter, 2=sect etc */
42 int address_section
, visible_version_id
;
43 int leaf_contains_contents
, leaf_smallest_contents
;
44 char *contents_filename
;
46 char *template_filename
;
47 char *single_filename
;
48 char *template_fragment
;
49 char *head_end
, *body_start
, *body_end
, *addr_start
, *addr_end
;
50 char *body_tag
, *nav_attr
;
51 wchar_t *author
, *description
;
52 wchar_t *index_text
, *contents_text
, *preamble_text
, *title_separator
;
53 wchar_t *nav_prev_text
, *nav_next_text
, *nav_separator
;
54 wchar_t *index_main_sep
, *index_multi_sep
;
55 wchar_t *pre_versionid
, *post_versionid
;
56 int restrict_charset
, output_charset
;
58 HTML_3_2
, HTML_4
, ISO_HTML
,
59 XHTML_1_0_TRANSITIONAL
, XHTML_1_0_STRICT
61 wchar_t *lquote
, *rquote
;
65 #define contents_depth(conf, level) \
66 ( (conf).ncdepths > (level) ? (conf).contents_depths[level] : (level)+2 )
68 #define is_xhtml(ver) ((ver) >= XHTML_1_0_TRANSITIONAL)
70 typedef struct htmlfile htmlfile
;
71 typedef struct htmlsect htmlsect
;
76 int last_fragment_number
;
77 int min_heading_depth
;
78 htmlsect
*first
, *last
; /* first/last highest-level sections */
82 htmlsect
*next
, *parent
;
84 paragraph
*title
, *text
;
85 enum { NORMAL
, TOP
, INDEX
} type
;
91 htmlfile
*head
, *tail
;
92 htmlfile
*single
, *index
;
97 htmlsect
*head
, *tail
;
113 int generated
, referenced
;
118 * This level deals with charset conversion, starting and
119 * ending tags, and writing to the file. It's the lexical
123 int charset
, restrict_charset
;
124 charset_state cstate
;
127 HO_NEUTRAL
, HO_IN_TAG
, HO_IN_EMPTY_TAG
, HO_IN_TEXT
130 * Stuff beyond here deals with the higher syntactic level: it
131 * tracks how many levels of <ul> are currently open when
132 * producing a contents list, for example.
137 static int html_fragment_compare(void *av
, void *bv
)
139 htmlfragment
*a
= (htmlfragment
*)av
;
140 htmlfragment
*b
= (htmlfragment
*)bv
;
143 if ((cmp
= strcmp(a
->file
->filename
, b
->file
->filename
)) != 0)
146 return strcmp(a
->fragment
, b
->fragment
);
149 static void html_file_section(htmlconfig
*cfg
, htmlfilelist
*files
,
150 htmlsect
*sect
, int depth
);
152 static htmlfile
*html_new_file(htmlfilelist
*list
, char *filename
);
153 static htmlsect
*html_new_sect(htmlsectlist
*list
, paragraph
*title
);
155 /* Flags for html_words() flags parameter */
159 #define INDEXENTS 0x04
161 static void html_words(htmloutput
*ho
, word
*words
, int flags
,
162 htmlfile
*file
, keywordlist
*keywords
, htmlconfig
*cfg
);
163 static void html_codepara(htmloutput
*ho
, word
*words
);
165 static void element_open(htmloutput
*ho
, char const *name
);
166 static void element_close(htmloutput
*ho
, char const *name
);
167 static void element_empty(htmloutput
*ho
, char const *name
);
168 static void element_attr(htmloutput
*ho
, char const *name
, char const *value
);
169 static void element_attr_w(htmloutput
*ho
, char const *name
,
170 wchar_t const *value
);
171 static void html_text(htmloutput
*ho
, wchar_t const *str
);
172 static void html_text_nbsp(htmloutput
*ho
, wchar_t const *str
);
173 static void html_text_limit(htmloutput
*ho
, wchar_t const *str
, int maxlen
);
174 static void html_text_limit_internal(htmloutput
*ho
, wchar_t const *text
,
175 int maxlen
, int quote_quotes
, int nbsp
);
176 static void html_nl(htmloutput
*ho
);
177 static void html_raw(htmloutput
*ho
, char *text
);
178 static void html_raw_as_attr(htmloutput
*ho
, char *text
);
179 static void cleanup(htmloutput
*ho
);
181 static void html_href(htmloutput
*ho
, htmlfile
*thisfile
,
182 htmlfile
*targetfile
, char *targetfrag
);
183 static void html_fragment(htmloutput
*ho
, char const *fragment
);
185 static char *html_format(paragraph
*p
, char *template_string
);
186 static char *html_sanitise_fragment(htmlfilelist
*files
, htmlfile
*file
,
189 static void html_contents_entry(htmloutput
*ho
, int depth
, htmlsect
*s
,
190 htmlfile
*thisfile
, keywordlist
*keywords
,
192 static void html_section_title(htmloutput
*ho
, htmlsect
*s
,
193 htmlfile
*thisfile
, keywordlist
*keywords
,
194 htmlconfig
*cfg
, int real
);
196 static htmlconfig
html_configure(paragraph
*source
) {
204 ret
.achapter
.just_numbers
= FALSE
;
205 ret
.achapter
.number_suffix
= L
": ";
207 ret
.asect
= snewn(ret
.nasect
, sectlevel
);
208 ret
.asect
[0].just_numbers
= TRUE
;
209 ret
.asect
[0].number_suffix
= L
" ";
211 ret
.contents_depths
= 0;
212 ret
.visible_version_id
= TRUE
;
213 ret
.address_section
= TRUE
;
214 ret
.leaf_contains_contents
= FALSE
;
215 ret
.leaf_smallest_contents
= 4;
216 ret
.single_filename
= dupstr("Manual.html");
217 ret
.contents_filename
= dupstr("Contents.html");
218 ret
.index_filename
= dupstr("IndexPage.html");
219 ret
.template_filename
= dupstr("%n.html");
220 ret
.template_fragment
= dupstr("%b");
221 ret
.head_end
= ret
.body_tag
= ret
.body_start
= ret
.body_end
=
222 ret
.addr_start
= ret
.addr_end
= ret
.nav_attr
= NULL
;
223 ret
.author
= ret
.description
= NULL
;
224 ret
.restrict_charset
= CS_UTF8
;
225 ret
.output_charset
= CS_ASCII
;
226 ret
.htmlver
= HTML_4
;
227 ret
.index_text
= L
"Index";
228 ret
.contents_text
= L
"Contents";
229 ret
.preamble_text
= L
"Preamble";
230 ret
.title_separator
= L
" - ";
231 ret
.nav_prev_text
= L
"Previous";
232 ret
.nav_next_text
= L
"Next";
233 ret
.nav_separator
= L
" | ";
234 ret
.index_main_sep
= L
": ";
235 ret
.index_multi_sep
= L
", ";
236 ret
.pre_versionid
= L
"[";
237 ret
.post_versionid
= L
"]";
239 * Default quote characters are Unicode matched single quotes,
240 * falling back to ordinary ASCII ".
242 ret
.lquote
= L
"\x2018\0\x2019\0\"\0\"\0\0";
243 ret
.rquote
= uadv(ret
.lquote
);
246 * Two-pass configuration so that we can pick up global config
247 * (e.g. `quotes') before having it overridden by specific
248 * config (`html-quotes'), irrespective of the order in which
251 for (p
= source
; p
; p
= p
->next
) {
252 if (p
->type
== para_Config
) {
253 if (!ustricmp(p
->keyword
, L
"quotes")) {
254 if (*uadv(p
->keyword
) && *uadv(uadv(p
->keyword
))) {
255 ret
.lquote
= uadv(p
->keyword
);
256 ret
.rquote
= uadv(ret
.lquote
);
262 for (p
= source
; p
; p
= p
->next
) {
263 if (p
->type
== para_Config
) {
264 wchar_t *k
= p
->keyword
;
266 if (!ustrnicmp(k
, L
"xhtml-", 6))
267 k
++; /* treat `xhtml-' and `html-' the same */
269 if (!ustricmp(k
, L
"html-restrict-charset")) {
270 ret
.restrict_charset
= charset_from_ustr(&p
->fpos
, uadv(k
));
271 } else if (!ustricmp(k
, L
"html-output-charset")) {
272 ret
.output_charset
= charset_from_ustr(&p
->fpos
, uadv(k
));
273 } else if (!ustricmp(k
, L
"html-version")) {
274 wchar_t *vername
= uadv(k
);
275 static const struct {
279 {L
"html3.2", HTML_3_2
},
281 {L
"iso-html", ISO_HTML
},
282 {L
"xhtml1.0transitional", XHTML_1_0_TRANSITIONAL
},
283 {L
"xhtml1.0strict", XHTML_1_0_STRICT
}
287 for (i
= 0; i
< (int)lenof(versions
); i
++)
288 if (!ustricmp(versions
[i
].name
, vername
))
291 if (i
== lenof(versions
))
292 error(err_htmlver
, &p
->fpos
, vername
);
294 ret
.htmlver
= versions
[i
].ver
;
295 } else if (!ustricmp(k
, L
"html-single-filename")) {
296 sfree(ret
.single_filename
);
297 ret
.single_filename
= dupstr(adv(p
->origkeyword
));
298 } else if (!ustricmp(k
, L
"html-contents-filename")) {
299 sfree(ret
.contents_filename
);
300 ret
.contents_filename
= dupstr(adv(p
->origkeyword
));
301 } else if (!ustricmp(k
, L
"html-index-filename")) {
302 sfree(ret
.index_filename
);
303 ret
.index_filename
= dupstr(adv(p
->origkeyword
));
304 } else if (!ustricmp(k
, L
"html-template-filename")) {
305 sfree(ret
.template_filename
);
306 ret
.template_filename
= dupstr(adv(p
->origkeyword
));
307 } else if (!ustricmp(k
, L
"html-template-fragment")) {
308 sfree(ret
.template_fragment
);
309 ret
.template_fragment
= dupstr(adv(p
->origkeyword
));
310 } else if (!ustricmp(k
, L
"html-chapter-numeric")) {
311 ret
.achapter
.just_numbers
= utob(uadv(k
));
312 } else if (!ustricmp(k
, L
"html-chapter-suffix")) {
313 ret
.achapter
.number_suffix
= uadv(k
);
314 } else if (!ustricmp(k
, L
"html-leaf-level")) {
315 ret
.leaf_level
= utoi(uadv(k
));
316 } else if (!ustricmp(k
, L
"html-section-numeric")) {
317 wchar_t *q
= uadv(k
);
323 if (n
>= ret
.nasect
) {
325 ret
.asect
= sresize(ret
.asect
, n
+1, sectlevel
);
326 for (i
= ret
.nasect
; i
<= n
; i
++)
327 ret
.asect
[i
] = ret
.asect
[ret
.nasect
-1];
330 ret
.asect
[n
].just_numbers
= utob(q
);
331 } else if (!ustricmp(k
, L
"html-section-suffix")) {
332 wchar_t *q
= uadv(k
);
338 if (n
>= ret
.nasect
) {
340 ret
.asect
= sresize(ret
.asect
, n
+1, sectlevel
);
341 for (i
= ret
.nasect
; i
<= n
; i
++) {
342 ret
.asect
[i
] = ret
.asect
[ret
.nasect
-1];
346 ret
.asect
[n
].number_suffix
= q
;
347 } else if (!ustricmp(k
, L
"html-contents-depth") ||
348 !ustrnicmp(k
, L
"html-contents-depth-", 20)) {
350 * Relic of old implementation: this directive used
351 * to be written as \cfg{html-contents-depth-3}{2}
352 * rather than the usual Halibut convention of
353 * \cfg{html-contents-depth}{3}{2}. We therefore
356 wchar_t *q
= k
[19] ? k
+20 : uadv(k
);
362 if (n
>= ret
.ncdepths
) {
364 ret
.contents_depths
=
365 sresize(ret
.contents_depths
, n
+1, int);
366 for (i
= ret
.ncdepths
; i
<= n
; i
++) {
367 ret
.contents_depths
[i
] = i
+2;
371 ret
.contents_depths
[n
] = utoi(q
);
372 } else if (!ustricmp(k
, L
"html-head-end")) {
373 ret
.head_end
= adv(p
->origkeyword
);
374 } else if (!ustricmp(k
, L
"html-body-tag")) {
375 ret
.body_tag
= adv(p
->origkeyword
);
376 } else if (!ustricmp(k
, L
"html-body-start")) {
377 ret
.body_start
= adv(p
->origkeyword
);
378 } else if (!ustricmp(k
, L
"html-body-end")) {
379 ret
.body_end
= adv(p
->origkeyword
);
380 } else if (!ustricmp(k
, L
"html-address-start")) {
381 ret
.addr_start
= adv(p
->origkeyword
);
382 } else if (!ustricmp(k
, L
"html-address-end")) {
383 ret
.addr_end
= adv(p
->origkeyword
);
384 } else if (!ustricmp(k
, L
"html-navigation-attributes")) {
385 ret
.nav_attr
= adv(p
->origkeyword
);
386 } else if (!ustricmp(k
, L
"html-author")) {
387 ret
.author
= uadv(k
);
388 } else if (!ustricmp(k
, L
"html-description")) {
389 ret
.description
= uadv(k
);
390 } else if (!ustricmp(k
, L
"html-suppress-address")) {
391 ret
.address_section
= !utob(uadv(k
));
392 } else if (!ustricmp(k
, L
"html-versionid")) {
393 ret
.visible_version_id
= utob(uadv(k
));
394 } else if (!ustricmp(k
, L
"html-quotes")) {
395 if (*uadv(k
) && *uadv(uadv(k
))) {
396 ret
.lquote
= uadv(k
);
397 ret
.rquote
= uadv(ret
.lquote
);
399 } else if (!ustricmp(k
, L
"html-leaf-contains-contents")) {
400 ret
.leaf_contains_contents
= utob(uadv(k
));
401 } else if (!ustricmp(k
, L
"html-leaf-smallest-contents")) {
402 ret
.leaf_smallest_contents
= utoi(uadv(k
));
403 } else if (!ustricmp(k
, L
"html-index-text")) {
404 ret
.index_text
= uadv(k
);
405 } else if (!ustricmp(k
, L
"html-contents-text")) {
406 ret
.contents_text
= uadv(k
);
407 } else if (!ustricmp(k
, L
"html-preamble-text")) {
408 ret
.preamble_text
= uadv(k
);
409 } else if (!ustricmp(k
, L
"html-title-separator")) {
410 ret
.title_separator
= uadv(k
);
411 } else if (!ustricmp(k
, L
"html-nav-prev-text")) {
412 ret
.nav_prev_text
= uadv(k
);
413 } else if (!ustricmp(k
, L
"html-nav-next-text")) {
414 ret
.nav_next_text
= uadv(k
);
415 } else if (!ustricmp(k
, L
"html-nav-separator")) {
416 ret
.nav_separator
= uadv(k
);
417 } else if (!ustricmp(k
, L
"html-index-main-separator")) {
418 ret
.index_main_sep
= uadv(k
);
419 } else if (!ustricmp(k
, L
"html-index-multiple-separator")) {
420 ret
.index_multi_sep
= uadv(k
);
421 } else if (!ustricmp(k
, L
"html-pre-versionid")) {
422 ret
.pre_versionid
= uadv(k
);
423 } else if (!ustricmp(k
, L
"html-post-versionid")) {
424 ret
.post_versionid
= uadv(k
);
430 * Now process fallbacks on quote characters.
432 while (*uadv(ret
.rquote
) && *uadv(uadv(ret
.rquote
)) &&
433 (!cvt_ok(ret
.restrict_charset
, ret
.lquote
) ||
434 !cvt_ok(ret
.restrict_charset
, ret
.rquote
))) {
435 ret
.lquote
= uadv(ret
.rquote
);
436 ret
.rquote
= uadv(ret
.lquote
);
442 paragraph
*html_config_filename(char *filename
)
445 * If the user passes in a single filename as a parameter to
446 * the `--html' command-line option, then we should assume it
447 * to imply _two_ config directives:
448 * \cfg{html-single-filename}{whatever} and
449 * \cfg{html-leaf-level}{0}; the rationale being that the user
450 * wants their output _in that file_.
454 p
= cmdline_cfg_simple("html-single-filename", filename
, NULL
);
455 q
= cmdline_cfg_simple("html-leaf-level", "0", NULL
);
460 void html_backend(paragraph
*sourceform
, keywordlist
*keywords
,
461 indexdata
*idx
, void *unused
)
465 htmlfilelist files
= { NULL
, NULL
, NULL
, NULL
, NULL
};
466 htmlsectlist sects
= { NULL
, NULL
}, nonsects
= { NULL
, NULL
};
471 conf
= html_configure(sourceform
);
474 * We're going to make heavy use of paragraphs' private data
475 * fields in the forthcoming code. Clear them first, so we can
476 * reliably tell whether we have auxiliary data for a
477 * particular paragraph.
479 for (p
= sourceform
; p
; p
= p
->next
)
480 p
->private_data
= NULL
;
482 files
.frags
= newtree234(html_fragment_compare
);
485 * Start by figuring out into which file each piece of the
486 * document should be put. We'll do this by inventing an
487 * `htmlsect' structure and stashing it in the private_data
488 * field of each section paragraph; we also need one additional
489 * htmlsect for the document index, which won't show up in the
490 * source form but needs to be consistently mentioned in
493 * While we're here, we'll also invent the HTML fragment name
497 htmlsect
*topsect
, *sect
;
500 topsect
= html_new_sect(§s
, NULL
);
502 topsect
->title
= NULL
;
503 topsect
->text
= sourceform
;
504 topsect
->contents_depth
= contents_depth(conf
, 0);
505 html_file_section(&conf
, &files
, topsect
, -1);
506 topsect
->fragment
= NULL
;
508 for (p
= sourceform
; p
; p
= p
->next
)
509 if (is_heading_type(p
->type
)) {
510 d
= heading_depth(p
);
512 if (p
->type
== para_Title
) {
517 sect
= html_new_sect(§s
, p
);
518 sect
->text
= p
->next
;
520 sect
->contents_depth
= contents_depth(conf
, d
+1) - (d
+1);
523 sect
->parent
= (htmlsect
*)p
->parent
->private_data
;
524 assert(sect
->parent
!= NULL
);
526 sect
->parent
= topsect
;
527 p
->private_data
= sect
;
529 html_file_section(&conf
, &files
, sect
, d
);
531 sect
->fragment
= html_format(p
, conf
.template_fragment
);
532 sect
->fragment
= html_sanitise_fragment(&files
, sect
->file
,
536 /* And the index, if we have one. */
537 has_index
= (count234(idx
->entries
) > 0);
539 sect
= html_new_sect(§s
, NULL
);
542 sect
->parent
= topsect
;
543 html_file_section(&conf
, &files
, sect
, 0); /* peer of chapters */
544 sect
->fragment
= utoa_dup(conf
.index_text
, CS_ASCII
);
545 sect
->fragment
= html_sanitise_fragment(&files
, sect
->file
,
547 files
.index
= sect
->file
;
552 * Go through the keyword list and sort out fragment IDs for
553 * all the potentially referenced paragraphs which _aren't_
561 for (i
= 0; (kw
= index234(keywords
->keys
, i
)) != NULL
; i
++) {
562 paragraph
*q
, *p
= kw
->para
;
564 if (!is_heading_type(p
->type
)) {
568 * Find the paragraph's parent htmlsect, to
569 * determine which file it will end up in.
574 * Preamble paragraphs have no parent. So if we
575 * have a non-heading with no parent, it must
576 * be preamble, and therefore its parent
577 * htmlsect must be the preamble one.
580 sects
.head
->type
== TOP
);
583 parent
= (htmlsect
*)q
->private_data
;
586 * Now we can construct an htmlsect for this
587 * paragraph itself, taking care to put it in the
588 * list of non-sections rather than the list of
589 * sections (so that traverses of the `sects' list
590 * won't attempt to add it to the contents or
591 * anything weird like that).
593 sect
= html_new_sect(&nonsects
, p
);
594 sect
->file
= parent
->file
;
595 sect
->parent
= parent
;
596 p
->private_data
= sect
;
599 * Fragment IDs for these paragraphs will simply be
600 * `p' followed by an integer.
602 sect
->fragment
= snewn(40, char);
603 sprintf(sect
->fragment
, "p%d",
604 sect
->file
->last_fragment_number
++);
605 sect
->fragment
= html_sanitise_fragment(&files
, sect
->file
,
612 * Reset the fragment numbers in each file. I've just used them
613 * to generate `p' fragment IDs for non-section paragraphs
614 * (numbered list elements, bibliocited), and now I want to use
615 * them for `i' fragment IDs for index entries.
619 for (file
= files
.head
; file
; file
= file
->next
)
620 file
->last_fragment_number
= 0;
624 * Now sort out the index. This involves:
626 * - For each index term, we set up an htmlindex structure to
627 * store all the references to that term.
629 * - Then we make a pass over the actual document, finding
630 * every word_IndexRef; for each one, we actually figure out
631 * the HTML filename/fragment pair we will use to reference
632 * it, store that information in the private data field of
633 * the word_IndexRef itself (so we can recreate it when the
634 * time comes to output our HTML), and add a reference to it
635 * to the index term in question.
644 * Set up the htmlindex structures.
647 for (i
= 0; (entry
= index234(idx
->entries
, i
)) != NULL
; i
++) {
648 htmlindex
*hi
= snew(htmlindex
);
650 hi
->nrefs
= hi
->refsize
= 0;
653 entry
->backend_data
= hi
;
657 * Run over the document inventing fragments. Each fragment
658 * is of the form `i' followed by an integer.
660 lastsect
= sects
.head
; /* this is always the top section */
661 for (p
= sourceform
; p
; p
= p
->next
) {
662 if (is_heading_type(p
->type
) && p
->type
!= para_Title
)
663 lastsect
= (htmlsect
*)p
->private_data
;
665 for (w
= p
->words
; w
; w
= w
->next
)
666 if (w
->type
== word_IndexRef
) {
667 htmlindexref
*hr
= snew(htmlindexref
);
671 hr
->referenced
= hr
->generated
= FALSE
;
672 hr
->section
= lastsect
;
676 lastsect
->file
->last_fragment_number
++);
677 hr
->fragment
= dupstr(buf
);
679 html_sanitise_fragment(&files
, hr
->section
->file
,
682 w
->private_data
= hr
;
684 tag
= index_findtag(idx
, w
->text
);
688 for (i
= 0; i
< tag
->nrefs
; i
++) {
689 indexentry
*entry
= tag
->refs
[i
];
690 htmlindex
*hi
= (htmlindex
*)entry
->backend_data
;
692 if (hi
->nrefs
>= hi
->refsize
) {
694 hi
->refs
= sresize(hi
->refs
, hi
->refsize
, word
*);
697 hi
->refs
[hi
->nrefs
++] = w
;
704 * Now we're ready to write out the actual HTML files.
708 * - we open that file and write its header
709 * - we run down the list of sections
710 * - for each section directly contained within that file, we
711 * output the section text
712 * - for each section which is not in the file but which has a
713 * parent that is, we output a contents entry for the
714 * section if appropriate
715 * - finally, we output the file trailer and close the file.
724 for (f
= files
.head
; f
; f
= f
->next
) {
727 enum LISTTYPE
{ NOLIST
, UL
, OL
, DL
};
728 enum ITEMTYPE
{ NOITEM
, LI
, DT
, DD
};
729 struct stackelement
{
730 struct stackelement
*next
;
731 enum LISTTYPE listtype
;
732 enum ITEMTYPE itemtype
;
735 #define listname(lt) ( (lt)==UL ? "ul" : (lt)==OL ? "ol" : "dl" )
736 #define itemname(lt) ( (lt)==LI ? "li" : (lt)==DT ? "dt" : "dd" )
738 ho
.fp
= fopen(f
->filename
, "w");
739 ho
.charset
= conf
.output_charset
;
740 ho
.restrict_charset
= conf
.restrict_charset
;
741 ho
.cstate
= charset_init_state
;
742 ho
.ver
= conf
.htmlver
;
743 ho
.state
= HO_NEUTRAL
;
744 ho
.contents_level
= 0;
747 switch (conf
.htmlver
) {
749 fprintf(ho
.fp
, "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD "
750 "HTML 3.2 Final//EN\">\n");
753 fprintf(ho
.fp
, "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML"
754 " 4.01//EN\"\n\"http://www.w3.org/TR/html4/"
758 fprintf(ho
.fp
, "<!DOCTYPE HTML PUBLIC \"ISO/IEC "
759 "15445:2000//DTD HTML//EN\">\n");
761 case XHTML_1_0_TRANSITIONAL
:
762 fprintf(ho
.fp
, "<?xml version=\"1.0\" encoding=\"%s\"?>\n",
763 charset_to_mimeenc(conf
.output_charset
));
764 fprintf(ho
.fp
, "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML"
765 " 1.0 Transitional//EN\"\n\"http://www.w3.org/TR/"
766 "xhtml1/DTD/xhtml1-transitional.dtd\">\n");
768 case XHTML_1_0_STRICT
:
769 fprintf(ho
.fp
, "<?xml version=\"1.0\" encoding=\"%s\"?>\n",
770 charset_to_mimeenc(conf
.output_charset
));
771 fprintf(ho
.fp
, "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML"
772 " 1.0 Strict//EN\"\n\"http://www.w3.org/TR/xhtml1/"
773 "DTD/xhtml1-strict.dtd\">\n");
777 element_open(&ho
, "html");
778 if (is_xhtml(conf
.htmlver
)) {
779 element_attr(&ho
, "xmlns", "http://www.w3.org/1999/xhtml");
783 element_open(&ho
, "head");
786 element_empty(&ho
, "meta");
787 element_attr(&ho
, "http-equiv", "content-type");
790 sprintf(buf
, "text/html; charset=%.150s",
791 charset_to_mimeenc(conf
.output_charset
));
792 element_attr(&ho
, "content", buf
);
797 element_empty(&ho
, "meta");
798 element_attr(&ho
, "name", "author");
799 element_attr_w(&ho
, "content", conf
.author
);
803 if (conf
.description
) {
804 element_empty(&ho
, "meta");
805 element_attr(&ho
, "name", "description");
806 element_attr_w(&ho
, "content", conf
.description
);
810 element_open(&ho
, "title");
811 if (f
->first
&& f
->first
->title
) {
812 html_words(&ho
, f
->first
->title
->words
, NOTHING
,
816 if (f
->last
!= f
->first
&& f
->last
->title
) {
817 html_text(&ho
, conf
.title_separator
);
818 html_words(&ho
, f
->last
->title
->words
, NOTHING
,
822 element_close(&ho
, "title");
826 html_raw(&ho
, conf
.head_end
);
828 element_close(&ho
, "head");
832 html_raw(&ho
, conf
.body_tag
);
834 element_open(&ho
, "body");
838 html_raw(&ho
, conf
.body_start
);
841 * Write out a nav bar. Special case: we don't do this
842 * if there is only one file.
844 if (files
.head
!= files
.tail
) {
845 element_open(&ho
, "p");
847 html_raw_as_attr(&ho
, conf
.nav_attr
);
850 element_open(&ho
, "a");
851 element_attr(&ho
, "href", prevf
->filename
);
853 html_text(&ho
, conf
.nav_prev_text
);
855 element_close(&ho
, "a");
857 html_text(&ho
, conf
.nav_separator
);
859 if (f
!= files
.head
) {
860 element_open(&ho
, "a");
861 element_attr(&ho
, "href", files
.head
->filename
);
863 html_text(&ho
, conf
.contents_text
);
865 element_close(&ho
, "a");
867 html_text(&ho
, conf
.nav_separator
);
870 if (f
!= files
.index
) {
871 element_open(&ho
, "a");
872 element_attr(&ho
, "href", files
.index
->filename
);
874 html_text(&ho
, conf
.index_text
);
875 if (f
!= files
.index
)
876 element_close(&ho
, "a");
879 html_text(&ho
, conf
.nav_separator
);
882 element_open(&ho
, "a");
883 element_attr(&ho
, "href", f
->next
->filename
);
885 html_text(&ho
, conf
.nav_next_text
);
887 element_close(&ho
, "a");
889 element_close(&ho
, "p");
895 * Write out a prefix TOC for the file.
897 * We start by going through the section list and
898 * collecting the sections which need to be added to
899 * the contents. On the way, we also test to see if
900 * this file is a leaf file (defined as one which
901 * contains all descendants of any section it
902 * contains), because this will play a part in our
903 * decision on whether or not to _output_ the TOC.
905 * Special case: we absolutely do not do this if we're
906 * in single-file mode.
908 if (files
.head
!= files
.tail
) {
909 int ntoc
= 0, tocsize
= 0;
910 htmlsect
**toc
= NULL
;
913 for (s
= sects
.head
; s
; s
= s
->next
) {
918 * Search up from this section until we find
919 * the highest-level one which belongs in this
924 for (ac
= s
; ac
; ac
= ac
->parent
) {
932 if (s
->file
!= f
&& a
!= NULL
)
936 if (adepth
<= a
->contents_depth
) {
937 if (ntoc
>= tocsize
) {
939 toc
= sresize(toc
, tocsize
, htmlsect
*);
946 if (leaf
&& conf
.leaf_contains_contents
&&
947 ntoc
>= conf
.leaf_smallest_contents
) {
950 for (i
= 0; i
< ntoc
; i
++) {
951 htmlsect
*s
= toc
[i
];
952 int hlevel
= (s
->type
== TOP ?
-1 :
953 s
->type
== INDEX ?
0 :
954 heading_depth(s
->title
))
955 - f
->min_heading_depth
+ 1;
958 html_contents_entry(&ho
, hlevel
, s
,
961 html_contents_entry(&ho
, 0, NULL
, f
, keywords
, &conf
);
966 * Now go through the document and output some real
970 for (s
= sects
.head
; s
; s
= s
->next
) {
973 * This section belongs in this file.
984 * Search up from this section until we find
985 * the highest-level one which belongs in this
990 for (ac
= s
; ac
; ac
= ac
->parent
) {
1000 * This section does not belong in this
1001 * file, but an ancestor of it does. Write
1002 * out a contents table entry, if the depth
1003 * doesn't exceed the maximum contents
1004 * depth for the ancestor section.
1006 if (adepth
<= a
->contents_depth
) {
1007 html_contents_entry(&ho
, adepth
, s
,
1008 f
, keywords
, &conf
);
1017 html_contents_entry(&ho
, 0, NULL
, f
, keywords
, &conf
);
1020 * Display the section heading.
1023 hlevel
= (s
->type
== TOP ?
-1 :
1024 s
->type
== INDEX ?
0 :
1025 heading_depth(s
->title
))
1026 - f
->min_heading_depth
+ 1;
1027 assert(hlevel
>= 1);
1028 /* HTML headings only go up to <h6> */
1032 htag
[1] = '0' + hlevel
;
1034 element_open(&ho
, htag
);
1037 * Provide anchor for cross-links to target.
1039 * (Also we'll have to do this separately in
1040 * other paragraph types - NumberedList and
1044 html_fragment(&ho
, s
->fragment
);
1046 html_section_title(&ho
, s
, f
, keywords
, &conf
, TRUE
);
1048 element_close(&ho
, htag
);
1051 * Now display the section text.
1054 stackhead
= snew(struct stackelement
);
1055 stackhead
->next
= NULL
;
1056 stackhead
->listtype
= NOLIST
;
1057 stackhead
->itemtype
= NOITEM
;
1059 for (p
= s
->text
;; p
= p
->next
) {
1060 enum LISTTYPE listtype
;
1061 struct stackelement
*se
;
1064 * Preliminary switch to figure out what
1065 * sort of list we expect to be inside at
1068 * Since p may still be NULL at this point,
1069 * I invent a harmless paragraph type for
1072 switch (p ? p
->type
: para_Normal
) {
1075 case para_Copyright
:
1076 case para_BiblioCited
:
1078 case para_QuotePush
:
1082 case para_UnnumberedChapter
:
1093 case para_NumberedList
:
1097 case para_DescribedThing
:
1098 case para_Description
:
1102 case para_LcontPush
:
1103 se
= snew(struct stackelement
);
1104 se
->next
= stackhead
;
1105 se
->listtype
= NOLIST
;
1106 se
->itemtype
= NOITEM
;
1110 default: /* some totally non-printing para */
1117 * Terminate the most recent list item, if
1118 * any. (We left this until after
1119 * processing LcontPush, since in that case
1120 * the list item won't want to be
1121 * terminated until after the corresponding
1124 if (stackhead
->itemtype
!= NOITEM
) {
1125 element_close(&ho
, itemname(stackhead
->itemtype
));
1128 stackhead
->itemtype
= NOITEM
;
1131 * Terminate the current list, if it's not
1132 * the one we want to be in.
1134 if (listtype
!= stackhead
->listtype
&&
1135 stackhead
->listtype
!= NOLIST
) {
1136 element_close(&ho
, listname(stackhead
->listtype
));
1141 * Leave the loop if our time has come.
1143 if (!p
|| (is_heading_type(p
->type
) &&
1144 p
->type
!= para_Title
))
1145 break; /* end of section text */
1148 * Start a fresh list if necessary.
1150 if (listtype
!= stackhead
->listtype
&&
1152 element_open(&ho
, listname(listtype
));
1154 stackhead
->listtype
= listtype
;
1158 element_empty(&ho
, "hr");
1161 html_codepara(&ho
, p
->words
);
1164 case para_Copyright
:
1165 element_open(&ho
, "p");
1167 html_words(&ho
, p
->words
, ALL
,
1168 f
, keywords
, &conf
);
1170 element_close(&ho
, "p");
1172 case para_BiblioCited
:
1173 element_open(&ho
, "p");
1174 if (p
->private_data
) {
1175 htmlsect
*s
= (htmlsect
*)p
->private_data
;
1176 html_fragment(&ho
, s
->fragment
);
1179 html_words(&ho
, p
->kwtext
, ALL
,
1180 f
, keywords
, &conf
);
1181 html_text(&ho
, L
" ");
1182 html_words(&ho
, p
->words
, ALL
,
1183 f
, keywords
, &conf
);
1185 element_close(&ho
, "p");
1188 case para_NumberedList
:
1189 element_open(&ho
, "li");
1190 if (p
->private_data
) {
1191 htmlsect
*s
= (htmlsect
*)p
->private_data
;
1192 html_fragment(&ho
, s
->fragment
);
1195 stackhead
->itemtype
= LI
;
1196 html_words(&ho
, p
->words
, ALL
,
1197 f
, keywords
, &conf
);
1199 case para_DescribedThing
:
1200 element_open(&ho
, "dt");
1202 stackhead
->itemtype
= DT
;
1203 html_words(&ho
, p
->words
, ALL
,
1204 f
, keywords
, &conf
);
1206 case para_Description
:
1207 element_open(&ho
, "dd");
1209 stackhead
->itemtype
= DD
;
1210 html_words(&ho
, p
->words
, ALL
,
1211 f
, keywords
, &conf
);
1214 case para_QuotePush
:
1215 element_open(&ho
, "blockquote");
1218 element_close(&ho
, "blockquote");
1223 stackhead
= stackhead
->next
;
1230 assert(stackhead
&& !stackhead
->next
);
1234 if (s
->type
== INDEX
) {
1239 * This section is the index. I'll just
1240 * render it as a single paragraph, with a
1241 * colon between the index term and the
1242 * references, and <br> in between each
1245 element_open(&ho
, "p");
1247 for (i
= 0; (entry
=
1248 index234(idx
->entries
, i
)) != NULL
; i
++) {
1249 htmlindex
*hi
= (htmlindex
*)entry
->backend_data
;
1253 element_empty(&ho
, "br");
1256 html_words(&ho
, entry
->text
, MARKUP
|LINKS
,
1257 f
, keywords
, &conf
);
1259 html_text(&ho
, conf
.index_main_sep
);
1261 for (j
= 0; j
< hi
->nrefs
; j
++) {
1263 (htmlindexref
*)hi
->refs
[j
]->private_data
;
1264 paragraph
*p
= hr
->section
->title
;
1267 html_text(&ho
, conf
.index_multi_sep
);
1269 html_href(&ho
, f
, hr
->section
->file
,
1271 hr
->referenced
= TRUE
;
1273 html_words(&ho
, p
->kwtext
, MARKUP
|LINKS
,
1274 f
, keywords
, &conf
);
1275 else if (p
&& p
->words
)
1276 html_words(&ho
, p
->words
, MARKUP
|LINKS
,
1277 f
, keywords
, &conf
);
1280 * If there is no title at all,
1281 * this must be because our
1282 * target section is the
1283 * preamble section and there
1284 * is no title. So we use the
1287 html_text(&ho
, conf
.preamble_text
);
1289 element_close(&ho
, "a");
1292 element_close(&ho
, "p");
1297 html_contents_entry(&ho
, 0, NULL
, f
, keywords
, &conf
);
1304 int done_version_ids
= FALSE
;
1306 element_empty(&ho
, "hr");
1309 html_raw(&ho
, conf
.body_end
);
1311 if (conf
.address_section
) {
1312 int started
= FALSE
;
1313 if (conf
.htmlver
== ISO_HTML
) {
1315 * The ISO-HTML validator complains if
1316 * there isn't a <div> tag surrounding the
1317 * <address> tag. I'm uncertain of why this
1318 * should be - there appears to be no
1319 * mention of this in the ISO-HTML spec,
1320 * suggesting that it doesn't represent a
1321 * change from HTML 4, but nonetheless the
1322 * HTML 4 validator doesn't seem to mind.
1324 element_open(&ho
, "div");
1326 element_open(&ho
, "address");
1327 if (conf
.addr_start
) {
1328 html_raw(&ho
, conf
.addr_start
);
1332 if (conf
.visible_version_id
) {
1333 for (p
= sourceform
; p
; p
= p
->next
)
1334 if (p
->type
== para_VersionID
) {
1336 element_empty(&ho
, "br");
1338 html_text(&ho
, conf
.pre_versionid
);
1339 html_words(&ho
, p
->words
, NOTHING
,
1340 f
, keywords
, &conf
);
1341 html_text(&ho
, conf
.post_versionid
);
1344 done_version_ids
= TRUE
;
1346 if (conf
.addr_end
) {
1348 element_empty(&ho
, "br");
1349 html_raw(&ho
, conf
.addr_end
);
1351 element_close(&ho
, "address");
1352 if (conf
.htmlver
== ISO_HTML
)
1353 element_close(&ho
, "div");
1356 if (!done_version_ids
) {
1358 * If the user didn't want the version IDs
1359 * visible, I think we still have a duty to put
1360 * them in an HTML comment.
1362 int started
= FALSE
;
1363 for (p
= sourceform
; p
; p
= p
->next
)
1364 if (p
->type
== para_VersionID
) {
1366 html_raw(&ho
, "<!-- version IDs:\n");
1369 html_words(&ho
, p
->words
, NOTHING
,
1370 f
, keywords
, &conf
);
1374 html_raw(&ho
, "-->\n");
1378 element_close(&ho
, "body");
1380 element_close(&ho
, "html");
1387 * Go through and check that no index fragments were referenced
1388 * without being generated, or indeed vice versa.
1390 * (When I actually get round to freeing everything, this can
1391 * probably be the freeing loop as well.)
1393 for (p
= sourceform
; p
; p
= p
->next
) {
1395 for (w
= p
->words
; w
; w
= w
->next
)
1396 if (w
->type
== word_IndexRef
) {
1397 htmlindexref
*hr
= (htmlindexref
*)w
->private_data
;
1399 assert(!hr
->referenced
== !hr
->generated
);
1404 * Free all the working data.
1407 sfree(conf
.single_filename
);
1408 sfree(conf
.contents_filename
);
1409 sfree(conf
.index_filename
);
1410 sfree(conf
.template_filename
);
1411 sfree(conf
.template_fragment
);
1414 while ( (frag
= (htmlfragment
*)delpos234(files
.frags
, 0)) != NULL
) {
1416 * frag->fragment is dynamically allocated, but will be
1417 * freed when we process the htmlsect structure which
1418 * it is attached to.
1422 freetree234(files
.frags
);
1425 htmlsect
*sect
, *tmp
;
1429 sfree(sect
->fragment
);
1433 sect
= nonsects
.head
;
1436 sfree(sect
->fragment
);
1442 htmlfile
*file
, *tmp
;
1446 sfree(file
->filename
);
1454 for (i
= 0; (entry
= index234(idx
->entries
, i
)) != NULL
; i
++) {
1455 htmlindex
*hi
= (htmlindex
*)entry
->backend_data
;
1462 for (p
= sourceform
; p
; p
= p
->next
)
1463 for (w
= p
->words
; w
; w
= w
->next
)
1464 if (w
->type
== word_IndexRef
) {
1465 htmlindexref
*hr
= (htmlindexref
*)w
->private_data
;
1467 sfree(hr
->fragment
);
1473 static void html_file_section(htmlconfig
*cfg
, htmlfilelist
*files
,
1474 htmlsect
*sect
, int depth
)
1480 * `depth' is derived from the heading_depth() macro at the top
1481 * of this file, which counts title as -1, chapter as 0,
1482 * heading as 1 and subsection as 2. However, the semantics of
1483 * cfg->leaf_level are defined to count chapter as 1, heading
1484 * as 2 etc. So first I increment depth :-(
1488 if (cfg
->leaf_level
== 0) {
1490 * leaf_level==0 is a special case, in which everything is
1491 * put into a single file.
1494 files
->single
= html_new_file(files
, cfg
->single_filename
);
1496 file
= files
->single
;
1499 * If the depth of this section is at or above leaf_level,
1500 * we invent a fresh file and put this section at its head.
1501 * Otherwise, we put it in the same file as its parent
1504 if (ldepth
> cfg
->leaf_level
) {
1506 * We know that sect->parent cannot be NULL. The only
1507 * circumstance in which it can be is if sect is at
1508 * chapter or appendix level, i.e. ldepth==1; and if
1509 * that's the case, then we cannot have entered this
1510 * branch unless cfg->leaf_level==0, in which case we
1511 * would be in the single-file case above and not here
1514 assert(sect
->parent
);
1516 file
= sect
->parent
->file
;
1518 if (sect
->type
== TOP
) {
1519 file
= html_new_file(files
, cfg
->contents_filename
);
1520 } else if (sect
->type
== INDEX
) {
1521 file
= html_new_file(files
, cfg
->index_filename
);
1525 assert(ldepth
> 0 && sect
->title
);
1526 title
= html_format(sect
->title
, cfg
->template_filename
);
1527 file
= html_new_file(files
, title
);
1535 if (file
->min_heading_depth
> depth
) {
1537 * This heading is at a higher level than any heading we
1538 * have so far placed in this file; so we set the `first'
1541 file
->min_heading_depth
= depth
;
1545 if (file
->min_heading_depth
== depth
)
1549 static htmlfile
*html_new_file(htmlfilelist
*list
, char *filename
)
1551 htmlfile
*ret
= snew(htmlfile
);
1555 list
->tail
->next
= ret
;
1560 ret
->filename
= dupstr(filename
);
1561 ret
->last_fragment_number
= 0;
1562 ret
->min_heading_depth
= INT_MAX
;
1563 ret
->first
= ret
->last
= NULL
;
1568 static htmlsect
*html_new_sect(htmlsectlist
*list
, paragraph
*title
)
1570 htmlsect
*ret
= snew(htmlsect
);
1574 list
->tail
->next
= ret
;
1587 static void html_words(htmloutput
*ho
, word
*words
, int flags
,
1588 htmlfile
*file
, keywordlist
*keywords
, htmlconfig
*cfg
)
1594 for (w
= words
; w
; w
= w
->next
) switch (w
->type
) {
1595 case word_HyperLink
:
1596 if (flags
& LINKS
) {
1597 element_open(ho
, "a");
1598 c
= utoa_dup(w
->text
, CS_ASCII
);
1599 element_attr(ho
, "href", c
);
1603 case word_UpperXref
:
1604 case word_LowerXref
:
1605 if (flags
& LINKS
) {
1606 keyword
*kwl
= kw_lookup(keywords
, w
->text
);
1607 paragraph
*p
= kwl
->para
;
1608 htmlsect
*s
= (htmlsect
*)p
->private_data
;
1612 html_href(ho
, file
, s
->file
, s
->fragment
);
1618 element_close(ho
, "a");
1621 if (flags
& INDEXENTS
) {
1622 htmlindexref
*hr
= (htmlindexref
*)w
->private_data
;
1623 html_fragment(ho
, hr
->fragment
);
1624 hr
->generated
= TRUE
;
1631 case word_WhiteSpace
:
1632 case word_EmphSpace
:
1633 case word_CodeSpace
:
1634 case word_WkCodeSpace
:
1636 case word_EmphQuote
:
1637 case word_CodeQuote
:
1638 case word_WkCodeQuote
:
1639 style
= towordstyle(w
->type
);
1640 type
= removeattr(w
->type
);
1641 if (style
== word_Emph
&&
1642 (attraux(w
->aux
) == attr_First
||
1643 attraux(w
->aux
) == attr_Only
) &&
1645 element_open(ho
, "em");
1646 else if ((style
== word_Code
|| style
== word_WeakCode
) &&
1647 (attraux(w
->aux
) == attr_First
||
1648 attraux(w
->aux
) == attr_Only
) &&
1650 element_open(ho
, "code");
1652 if (type
== word_WhiteSpace
)
1653 html_text(ho
, L
" ");
1654 else if (type
== word_Quote
) {
1655 if (quoteaux(w
->aux
) == quote_Open
)
1656 html_text(ho
, cfg
->lquote
);
1658 html_text(ho
, cfg
->rquote
);
1660 if (!w
->alt
|| cvt_ok(ho
->restrict_charset
, w
->text
))
1661 html_text_nbsp(ho
, w
->text
);
1663 html_words(ho
, w
->alt
, flags
, file
, keywords
, cfg
);
1666 if (style
== word_Emph
&&
1667 (attraux(w
->aux
) == attr_Last
||
1668 attraux(w
->aux
) == attr_Only
) &&
1670 element_close(ho
, "em");
1671 else if ((style
== word_Code
|| style
== word_WeakCode
) &&
1672 (attraux(w
->aux
) == attr_Last
||
1673 attraux(w
->aux
) == attr_Only
) &&
1675 element_close(ho
, "code");
1681 static void html_codepara(htmloutput
*ho
, word
*words
)
1683 element_open(ho
, "pre");
1684 element_open(ho
, "code");
1685 for (; words
; words
= words
->next
) if (words
->type
== word_WeakCode
) {
1690 if (words
->next
&& words
->next
->type
== word_Emph
) {
1691 e
= words
->next
->text
;
1692 words
= words
->next
;
1696 while (e
&& *e
&& *t
) {
1700 for (n
= 0; t
[n
] && e
[n
] && e
[n
] == ec
; n
++);
1708 element_open(ho
, open_tag
);
1710 html_text_limit(ho
, t
, n
);
1713 element_close(ho
, open_tag
);
1721 element_close(ho
, "code");
1722 element_close(ho
, "pre");
1725 static void html_charset_cleanup(htmloutput
*ho
)
1730 bytes
= charset_from_unicode(NULL
, NULL
, outbuf
, lenof(outbuf
),
1731 ho
->charset
, &ho
->cstate
, NULL
);
1733 fwrite(outbuf
, 1, bytes
, ho
->fp
);
1736 static void return_mostly_to_neutral(htmloutput
*ho
)
1738 if (ho
->state
== HO_IN_EMPTY_TAG
&& is_xhtml(ho
->ver
)) {
1739 fprintf(ho
->fp
, " />");
1740 } else if (ho
->state
== HO_IN_EMPTY_TAG
|| ho
->state
== HO_IN_TAG
) {
1741 fprintf(ho
->fp
, ">");
1744 ho
->state
= HO_NEUTRAL
;
1747 static void return_to_neutral(htmloutput
*ho
)
1749 if (ho
->state
== HO_IN_TEXT
) {
1750 html_charset_cleanup(ho
);
1753 return_mostly_to_neutral(ho
);
1756 static void element_open(htmloutput
*ho
, char const *name
)
1758 return_to_neutral(ho
);
1759 fprintf(ho
->fp
, "<%s", name
);
1760 ho
->state
= HO_IN_TAG
;
1763 static void element_close(htmloutput
*ho
, char const *name
)
1765 return_to_neutral(ho
);
1766 fprintf(ho
->fp
, "</%s>", name
);
1767 ho
->state
= HO_NEUTRAL
;
1770 static void element_empty(htmloutput
*ho
, char const *name
)
1772 return_to_neutral(ho
);
1773 fprintf(ho
->fp
, "<%s", name
);
1774 ho
->state
= HO_IN_EMPTY_TAG
;
1777 static void html_nl(htmloutput
*ho
)
1779 return_to_neutral(ho
);
1780 fputc('\n', ho
->fp
);
1783 static void html_raw(htmloutput
*ho
, char *text
)
1785 return_to_neutral(ho
);
1786 fputs(text
, ho
->fp
);
1789 static void html_raw_as_attr(htmloutput
*ho
, char *text
)
1791 assert(ho
->state
== HO_IN_TAG
|| ho
->state
== HO_IN_EMPTY_TAG
);
1793 fputs(text
, ho
->fp
);
1796 static void element_attr(htmloutput
*ho
, char const *name
, char const *value
)
1798 html_charset_cleanup(ho
);
1799 assert(ho
->state
== HO_IN_TAG
|| ho
->state
== HO_IN_EMPTY_TAG
);
1800 fprintf(ho
->fp
, " %s=\"%s\"", name
, value
);
1803 static void element_attr_w(htmloutput
*ho
, char const *name
,
1804 wchar_t const *value
)
1806 html_charset_cleanup(ho
);
1807 fprintf(ho
->fp
, " %s=\"", name
);
1808 html_text_limit_internal(ho
, value
, 0, TRUE
, FALSE
);
1809 html_charset_cleanup(ho
);
1813 static void html_text(htmloutput
*ho
, wchar_t const *text
)
1815 return_mostly_to_neutral(ho
);
1816 html_text_limit_internal(ho
, text
, 0, FALSE
, FALSE
);
1819 static void html_text_nbsp(htmloutput
*ho
, wchar_t const *text
)
1821 return_mostly_to_neutral(ho
);
1822 html_text_limit_internal(ho
, text
, 0, FALSE
, TRUE
);
1825 static void html_text_limit(htmloutput
*ho
, wchar_t const *text
, int maxlen
)
1827 return_mostly_to_neutral(ho
);
1828 html_text_limit_internal(ho
, text
, maxlen
, FALSE
, FALSE
);
1831 static void html_text_limit_internal(htmloutput
*ho
, wchar_t const *text
,
1832 int maxlen
, int quote_quotes
, int nbsp
)
1834 int textlen
= ustrlen(text
);
1838 if (maxlen
> 0 && textlen
> maxlen
)
1841 while (textlen
> 0) {
1842 /* Scan ahead for characters we really can't display in HTML. */
1843 int lenbefore
, lenafter
;
1844 for (lenbefore
= 0; lenbefore
< textlen
; lenbefore
++)
1845 if (text
[lenbefore
] == L
'<' ||
1846 text
[lenbefore
] == L
'>' ||
1847 text
[lenbefore
] == L
'&' ||
1848 (text
[lenbefore
] == L
'"' && quote_quotes
) ||
1849 (text
[lenbefore
] == L
' ' && nbsp
))
1851 lenafter
= lenbefore
;
1852 bytes
= charset_from_unicode(&text
, &lenafter
, outbuf
, lenof(outbuf
),
1853 ho
->charset
, &ho
->cstate
, &err
);
1854 textlen
-= (lenbefore
- lenafter
);
1856 fwrite(outbuf
, 1, bytes
, ho
->fp
);
1859 * We have encountered a character that cannot be
1860 * displayed in the selected output charset. Therefore,
1861 * we use an HTML numeric entity reference.
1863 assert(textlen
> 0);
1864 fprintf(ho
->fp
, "&#%ld;", (long int)*text
);
1866 } else if (lenafter
== 0 && textlen
> 0) {
1868 * We have encountered a character which is special to
1872 fprintf(ho
->fp
, "<");
1873 else if (*text
== L
'>')
1874 fprintf(ho
->fp
, ">");
1875 else if (*text
== L
'&')
1876 fprintf(ho
->fp
, "&");
1877 else if (*text
== L
'"')
1878 fprintf(ho
->fp
, """);
1879 else if (*text
== L
' ') {
1881 fprintf(ho
->fp
, " ");
1883 assert(!"Can't happen");
1889 static void cleanup(htmloutput
*ho
)
1891 return_to_neutral(ho
);
1895 static void html_href(htmloutput
*ho
, htmlfile
*thisfile
,
1896 htmlfile
*targetfile
, char *targetfrag
)
1898 rdstringc rs
= { 0, 0, NULL
};
1901 if (targetfile
!= thisfile
)
1902 rdaddsc(&rs
, targetfile
->filename
);
1905 rdaddsc(&rs
, targetfrag
);
1909 element_open(ho
, "a");
1910 element_attr(ho
, "href", url
);
1914 static void html_fragment(htmloutput
*ho
, char const *fragment
)
1916 element_open(ho
, "a");
1917 element_attr(ho
, "name", fragment
);
1918 if (is_xhtml(ho
->ver
))
1919 element_attr(ho
, "id", fragment
);
1920 element_close(ho
, "a");
1923 static char *html_format(paragraph
*p
, char *template_string
)
1927 wchar_t *ws
, wsbuf
[2];
1928 rdstringc rs
= { 0, 0, NULL
};
1930 t
= template_string
;
1932 if (*t
== '%' && t
[1]) {
1946 if (p
->kwtext
&& fmt
== 'n')
1948 else if (p
->kwtext2
&& fmt
== 'b') {
1950 * HTML fragment names must start with a letter, so
1951 * simply `1.2.3' is not adequate. In this case I'm
1952 * going to cheat slightly by prepending the first
1953 * character of the first word of kwtext, so that
1954 * we get `C1' for chapter 1, `S2.3' for section
1957 if (p
->kwtext
&& p
->kwtext
->text
[0]) {
1960 wsbuf
[0] = p
->kwtext
->text
[0];
1963 } else if (p
->keyword
&& *p
->keyword
&& fmt
== 'k')
1969 c
= utoa_dup(ws
, CS_ASCII
);
1975 if (removeattr(w
->type
) == word_Normal
) {
1976 c
= utoa_dup(w
->text
, CS_ASCII
);
1987 return rdtrimc(&rs
);
1990 static char *html_sanitise_fragment(htmlfilelist
*files
, htmlfile
*file
,
1994 * The HTML 4 spec's strictest definition of fragment names (<a
1995 * name> and "id" attributes) says that they `must begin with a
1996 * letter and may be followed by any number of letters, digits,
1997 * hyphens, underscores, colons, and periods'.
1999 * So here we unceremoniously rip out any characters not
2000 * conforming to this limitation.
2002 char *p
= text
, *q
= text
;
2004 while (*p
&& !((*p
>='A' && *p
<='Z') || (*p
>='a' && *p
<='z')))
2006 if ((*q
++ = *p
++) != '\0') {
2008 if ((*p
>='A' && *p
<='Z') ||
2009 (*p
>='a' && *p
<='z') ||
2010 (*p
>='0' && *p
<='9') ||
2011 *p
=='-' || *p
=='_' || *p
==':' || *p
=='.')
2020 * Now we check for clashes with other fragment names, and
2021 * adjust this one if necessary by appending a hyphen followed
2025 htmlfragment
*frag
= snew(htmlfragment
);
2026 int len
= 0; /* >0 indicates we have resized */
2030 frag
->fragment
= text
;
2032 while (add234(files
->frags
, frag
) != frag
) {
2035 frag
->fragment
= text
= sresize(text
, len
+20, char);
2038 sprintf(text
+ len
, "-%d", ++suffix
);
2045 static void html_contents_entry(htmloutput
*ho
, int depth
, htmlsect
*s
,
2046 htmlfile
*thisfile
, keywordlist
*keywords
,
2049 while (ho
->contents_level
> depth
) {
2050 element_close(ho
, "ul");
2051 ho
->contents_level
--;
2054 while (ho
->contents_level
< depth
) {
2055 element_open(ho
, "ul");
2056 ho
->contents_level
++;
2062 element_open(ho
, "li");
2063 html_href(ho
, thisfile
, s
->file
, s
->fragment
);
2064 html_section_title(ho
, s
, thisfile
, keywords
, cfg
, FALSE
);
2065 element_close(ho
, "a");
2066 element_close(ho
, "li");
2069 static void html_section_title(htmloutput
*ho
, htmlsect
*s
, htmlfile
*thisfile
,
2070 keywordlist
*keywords
, htmlconfig
*cfg
,
2076 int depth
= heading_depth(s
->title
);
2080 else if (depth
== 0)
2081 sl
= &cfg
->achapter
;
2082 else if (depth
<= cfg
->nasect
)
2083 sl
= &cfg
->asect
[depth
-1];
2085 sl
= &cfg
->asect
[cfg
->nasect
-1];
2089 else if (sl
->just_numbers
)
2090 number
= s
->title
->kwtext2
;
2092 number
= s
->title
->kwtext
;
2095 html_words(ho
, number
, MARKUP
,
2096 thisfile
, keywords
, cfg
);
2097 html_text(ho
, sl
->number_suffix
);
2100 html_words(ho
, s
->title
->words
, real ? ALL
: MARKUP
,
2101 thisfile
, keywords
, cfg
);
2103 assert(s
->type
!= NORMAL
);
2105 * If we're printing the full document title for _real_ and
2106 * there isn't one, we don't want to print `Preamble' at
2107 * the top of what ought to just be some text. If we need
2108 * it in any other context such as TOCs, we need to print
2111 if (s
->type
== TOP
&& !real
)
2112 html_text(ho
, cfg
->preamble_text
);
2113 else if (s
->type
== INDEX
)
2114 html_text(ho
, cfg
->index_text
);