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.)
14 * - nonbreaking spaces.
16 * - free up all the data we have allocated while running this
26 #define is_heading_type(type) ( (type) == para_Title || \
27 (type) == para_Chapter || \
28 (type) == para_Appendix || \
29 (type) == para_UnnumberedChapter || \
30 (type) == para_Heading || \
31 (type) == para_Subsect)
33 #define heading_depth(p) ( (p)->type == para_Subsect ? (p)->aux + 1 : \
34 (p)->type == para_Heading ? 1 : \
35 (p)->type == para_Title ? -1 : 0 )
39 wchar_t *number_suffix
;
44 sectlevel achapter
, *asect
;
45 int *contents_depths
; /* 0=main, 1=chapter, 2=sect etc */
47 int address_section
, visible_version_id
;
48 int leaf_contains_contents
, leaf_smallest_contents
;
49 char *contents_filename
;
51 char *template_filename
;
52 char *single_filename
;
53 char *template_fragment
;
54 char *head_end
, *body_start
, *body_end
, *addr_start
, *addr_end
;
55 char *body_tag
, *nav_attr
;
56 wchar_t *author
, *description
;
57 wchar_t *index_text
, *contents_text
, *preamble_text
, *title_separator
;
58 wchar_t *nav_prev_text
, *nav_next_text
, *nav_separator
;
59 wchar_t *index_main_sep
, *index_multi_sep
;
60 wchar_t *pre_versionid
, *post_versionid
;
61 int restrict_charset
, output_charset
;
63 HTML_3_2
, HTML_4
, ISO_HTML
,
64 XHTML_1_0_TRANSITIONAL
, XHTML_1_0_STRICT
66 wchar_t *lquote
, *rquote
;
70 #define contents_depth(conf, level) \
71 ( (conf).ncdepths > (level) ? (conf).contents_depths[level] : (level)+2 )
73 #define is_xhtml(ver) ((ver) >= XHTML_1_0_TRANSITIONAL)
75 typedef struct htmlfile htmlfile
;
76 typedef struct htmlsect htmlsect
;
81 int last_fragment_number
;
82 int min_heading_depth
;
83 htmlsect
*first
, *last
; /* first/last highest-level sections */
87 htmlsect
*next
, *parent
;
89 paragraph
*title
, *text
;
90 enum { NORMAL
, TOP
, INDEX
} type
;
96 htmlfile
*head
, *tail
;
97 htmlfile
*single
, *index
;
102 htmlsect
*head
, *tail
;
118 int generated
, referenced
;
123 * This level deals with charset conversion, starting and
124 * ending tags, and writing to the file. It's the lexical
128 int charset
, restrict_charset
;
129 charset_state cstate
;
132 HO_NEUTRAL
, HO_IN_TAG
, HO_IN_EMPTY_TAG
, HO_IN_TEXT
135 * Stuff beyond here deals with the higher syntactic level: it
136 * tracks how many levels of <ul> are currently open when
137 * producing a contents list, for example.
142 static int html_fragment_compare(void *av
, void *bv
)
144 htmlfragment
*a
= (htmlfragment
*)av
;
145 htmlfragment
*b
= (htmlfragment
*)bv
;
148 if ((cmp
= strcmp(a
->file
->filename
, b
->file
->filename
)) != 0)
151 return strcmp(a
->fragment
, b
->fragment
);
154 static void html_file_section(htmlconfig
*cfg
, htmlfilelist
*files
,
155 htmlsect
*sect
, int depth
);
157 static htmlfile
*html_new_file(htmlfilelist
*list
, char *filename
);
158 static htmlsect
*html_new_sect(htmlsectlist
*list
, paragraph
*title
);
160 /* Flags for html_words() flags parameter */
164 #define INDEXENTS 0x04
166 static void html_words(htmloutput
*ho
, word
*words
, int flags
,
167 htmlfile
*file
, keywordlist
*keywords
, htmlconfig
*cfg
);
168 static void html_codepara(htmloutput
*ho
, word
*words
);
170 static void element_open(htmloutput
*ho
, char const *name
);
171 static void element_close(htmloutput
*ho
, char const *name
);
172 static void element_empty(htmloutput
*ho
, char const *name
);
173 static void element_attr(htmloutput
*ho
, char const *name
, char const *value
);
174 static void element_attr_w(htmloutput
*ho
, char const *name
,
175 wchar_t const *value
);
176 static void html_text(htmloutput
*ho
, wchar_t const *str
);
177 static void html_text_limit(htmloutput
*ho
, wchar_t const *str
, int maxlen
);
178 static void html_text_limit_internal(htmloutput
*ho
, wchar_t const *text
,
179 int maxlen
, int quote_quotes
);
180 static void html_nl(htmloutput
*ho
);
181 static void html_raw(htmloutput
*ho
, char *text
);
182 static void html_raw_as_attr(htmloutput
*ho
, char *text
);
183 static void cleanup(htmloutput
*ho
);
185 static void html_href(htmloutput
*ho
, htmlfile
*thisfile
,
186 htmlfile
*targetfile
, char *targetfrag
);
187 static void html_fragment(htmloutput
*ho
, char const *fragment
);
189 static char *html_format(paragraph
*p
, char *template_string
);
190 static char *html_sanitise_fragment(htmlfilelist
*files
, htmlfile
*file
,
193 static void html_contents_entry(htmloutput
*ho
, int depth
, htmlsect
*s
,
194 htmlfile
*thisfile
, keywordlist
*keywords
,
196 static void html_section_title(htmloutput
*ho
, htmlsect
*s
,
197 htmlfile
*thisfile
, keywordlist
*keywords
,
198 htmlconfig
*cfg
, int real
);
200 static htmlconfig
html_configure(paragraph
*source
) {
208 ret
.achapter
.just_numbers
= FALSE
;
209 ret
.achapter
.number_suffix
= L
": ";
211 ret
.asect
= snewn(ret
.nasect
, sectlevel
);
212 ret
.asect
[0].just_numbers
= TRUE
;
213 ret
.asect
[0].number_suffix
= L
" ";
215 ret
.contents_depths
= 0;
216 ret
.visible_version_id
= TRUE
;
217 ret
.address_section
= TRUE
;
218 ret
.leaf_contains_contents
= FALSE
;
219 ret
.leaf_smallest_contents
= 4;
220 ret
.single_filename
= dupstr("Manual.html");
221 ret
.contents_filename
= dupstr("Contents.html");
222 ret
.index_filename
= dupstr("IndexPage.html");
223 ret
.template_filename
= dupstr("%n.html");
224 ret
.template_fragment
= dupstr("%b");
225 ret
.head_end
= ret
.body_tag
= ret
.body_start
= ret
.body_end
=
226 ret
.addr_start
= ret
.addr_end
= ret
.nav_attr
= NULL
;
227 ret
.author
= ret
.description
= NULL
;
228 ret
.restrict_charset
= CS_UTF8
;
229 ret
.output_charset
= CS_ASCII
;
230 ret
.htmlver
= HTML_4
;
231 ret
.index_text
= L
"Index";
232 ret
.contents_text
= L
"Contents";
233 ret
.preamble_text
= L
"Preamble";
234 ret
.title_separator
= L
" - ";
235 ret
.nav_prev_text
= L
"Previous";
236 ret
.nav_next_text
= L
"Next";
237 ret
.nav_separator
= L
" | ";
238 ret
.index_main_sep
= L
": ";
239 ret
.index_multi_sep
= L
", ";
240 ret
.pre_versionid
= L
"[";
241 ret
.post_versionid
= L
"]";
243 * Default quote characters are Unicode matched single quotes,
244 * falling back to ordinary ASCII ".
246 ret
.lquote
= L
"\x2018\0\x2019\0\"\0\"\0\0";
247 ret
.rquote
= uadv(ret
.lquote
);
250 * Two-pass configuration so that we can pick up global config
251 * (e.g. `quotes') before having it overridden by specific
252 * config (`html-quotes'), irrespective of the order in which
255 for (p
= source
; p
; p
= p
->next
) {
256 if (p
->type
== para_Config
) {
257 if (!ustricmp(p
->keyword
, L
"quotes")) {
258 if (*uadv(p
->keyword
) && *uadv(uadv(p
->keyword
))) {
259 ret
.lquote
= uadv(p
->keyword
);
260 ret
.rquote
= uadv(ret
.lquote
);
266 for (p
= source
; p
; p
= p
->next
) {
267 if (p
->type
== para_Config
) {
268 wchar_t *k
= p
->keyword
;
270 if (!ustrnicmp(k
, L
"xhtml-", 6))
271 k
++; /* treat `xhtml-' and `html-' the same */
273 if (!ustricmp(k
, L
"html-restrict-charset")) {
274 char *csname
= utoa_dup(uadv(k
), CS_ASCII
);
275 ret
.restrict_charset
= charset_from_localenc(csname
);
277 } else if (!ustricmp(k
, L
"html-output-charset")) {
278 char *csname
= utoa_dup(uadv(k
), CS_ASCII
);
279 ret
.output_charset
= charset_from_localenc(csname
);
281 } else if (!ustricmp(k
, L
"html-version")) {
282 wchar_t *vername
= uadv(k
);
283 static const struct {
287 {L
"html3.2", HTML_3_2
},
289 {L
"iso-html", ISO_HTML
},
290 {L
"xhtml1.0transitional", XHTML_1_0_TRANSITIONAL
},
291 {L
"xhtml1.0strict", XHTML_1_0_STRICT
}
295 for (i
= 0; i
< (int)lenof(versions
); i
++)
296 if (!ustricmp(versions
[i
].name
, vername
))
299 if (i
== lenof(versions
))
300 error(err_htmlver
, &p
->fpos
, vername
);
302 ret
.htmlver
= versions
[i
].ver
;
303 } else if (!ustricmp(k
, L
"html-single-filename")) {
304 sfree(ret
.single_filename
);
305 ret
.single_filename
= dupstr(adv(p
->origkeyword
));
306 } else if (!ustricmp(k
, L
"html-contents-filename")) {
307 sfree(ret
.contents_filename
);
308 ret
.contents_filename
= dupstr(adv(p
->origkeyword
));
309 } else if (!ustricmp(k
, L
"html-index-filename")) {
310 sfree(ret
.index_filename
);
311 ret
.index_filename
= dupstr(adv(p
->origkeyword
));
312 } else if (!ustricmp(k
, L
"html-template-filename")) {
313 sfree(ret
.template_filename
);
314 ret
.template_filename
= dupstr(adv(p
->origkeyword
));
315 } else if (!ustricmp(k
, L
"html-template-fragment")) {
316 sfree(ret
.template_fragment
);
317 ret
.template_fragment
= dupstr(adv(p
->origkeyword
));
318 } else if (!ustricmp(k
, L
"html-chapter-numeric")) {
319 ret
.achapter
.just_numbers
= utob(uadv(k
));
320 } else if (!ustricmp(k
, L
"html-chapter-suffix")) {
321 ret
.achapter
.number_suffix
= uadv(k
);
322 } else if (!ustricmp(k
, L
"html-leaf-level")) {
323 ret
.leaf_level
= utoi(uadv(k
));
324 } else if (!ustricmp(k
, L
"html-section-numeric")) {
325 wchar_t *q
= uadv(k
);
331 if (n
>= ret
.nasect
) {
333 ret
.asect
= sresize(ret
.asect
, n
+1, sectlevel
);
334 for (i
= ret
.nasect
; i
<= n
; i
++)
335 ret
.asect
[i
] = ret
.asect
[ret
.nasect
-1];
338 ret
.asect
[n
].just_numbers
= utob(q
);
339 } else if (!ustricmp(k
, L
"html-section-suffix")) {
340 wchar_t *q
= uadv(k
);
346 if (n
>= ret
.nasect
) {
348 ret
.asect
= sresize(ret
.asect
, n
+1, sectlevel
);
349 for (i
= ret
.nasect
; i
<= n
; i
++) {
350 ret
.asect
[i
] = ret
.asect
[ret
.nasect
-1];
354 ret
.asect
[n
].number_suffix
= q
;
355 } else if (!ustricmp(k
, L
"html-contents-depth") ||
356 !ustrnicmp(k
, L
"html-contents-depth-", 20)) {
358 * Relic of old implementation: this directive used
359 * to be written as \cfg{html-contents-depth-3}{2}
360 * rather than the usual Halibut convention of
361 * \cfg{html-contents-depth}{3}{2}. We therefore
364 wchar_t *q
= k
[19] ? k
+20 : uadv(k
);
370 if (n
>= ret
.ncdepths
) {
372 ret
.contents_depths
=
373 sresize(ret
.contents_depths
, n
+1, int);
374 for (i
= ret
.ncdepths
; i
<= n
; i
++) {
375 ret
.contents_depths
[i
] = i
+2;
379 ret
.contents_depths
[n
] = utoi(q
);
380 } else if (!ustricmp(k
, L
"html-head-end")) {
381 ret
.head_end
= adv(p
->origkeyword
);
382 } else if (!ustricmp(k
, L
"html-body-tag")) {
383 ret
.body_tag
= adv(p
->origkeyword
);
384 } else if (!ustricmp(k
, L
"html-body-start")) {
385 ret
.body_start
= adv(p
->origkeyword
);
386 } else if (!ustricmp(k
, L
"html-body-end")) {
387 ret
.body_end
= adv(p
->origkeyword
);
388 } else if (!ustricmp(k
, L
"html-address-start")) {
389 ret
.addr_start
= adv(p
->origkeyword
);
390 } else if (!ustricmp(k
, L
"html-address-end")) {
391 ret
.addr_end
= adv(p
->origkeyword
);
392 } else if (!ustricmp(k
, L
"html-navigation-attributes")) {
393 ret
.nav_attr
= adv(p
->origkeyword
);
394 } else if (!ustricmp(k
, L
"html-author")) {
395 ret
.author
= uadv(k
);
396 } else if (!ustricmp(k
, L
"html-description")) {
397 ret
.description
= uadv(k
);
398 } else if (!ustricmp(k
, L
"html-suppress-address")) {
399 ret
.address_section
= !utob(uadv(k
));
400 } else if (!ustricmp(k
, L
"html-versionid")) {
401 ret
.visible_version_id
= utob(uadv(k
));
402 } else if (!ustricmp(k
, L
"html-quotes")) {
403 if (*uadv(k
) && *uadv(uadv(k
))) {
404 ret
.lquote
= uadv(k
);
405 ret
.rquote
= uadv(ret
.lquote
);
407 } else if (!ustricmp(k
, L
"html-leaf-contains-contents")) {
408 ret
.leaf_contains_contents
= utob(uadv(k
));
409 } else if (!ustricmp(k
, L
"html-leaf-smallest-contents")) {
410 ret
.leaf_smallest_contents
= utoi(uadv(k
));
411 } else if (!ustricmp(k
, L
"html-index-text")) {
412 ret
.index_text
= uadv(k
);
413 } else if (!ustricmp(k
, L
"html-contents-text")) {
414 ret
.contents_text
= uadv(k
);
415 } else if (!ustricmp(k
, L
"html-preamble-text")) {
416 ret
.preamble_text
= uadv(k
);
417 } else if (!ustricmp(k
, L
"html-title-separator")) {
418 ret
.title_separator
= uadv(k
);
419 } else if (!ustricmp(k
, L
"html-nav-prev-text")) {
420 ret
.nav_prev_text
= uadv(k
);
421 } else if (!ustricmp(k
, L
"html-nav-next-text")) {
422 ret
.nav_next_text
= uadv(k
);
423 } else if (!ustricmp(k
, L
"html-nav-separator")) {
424 ret
.nav_separator
= uadv(k
);
425 } else if (!ustricmp(k
, L
"html-index-main-separator")) {
426 ret
.index_main_sep
= uadv(k
);
427 } else if (!ustricmp(k
, L
"html-index-multiple-separator")) {
428 ret
.index_multi_sep
= uadv(k
);
429 } else if (!ustricmp(k
, L
"html-pre-versionid")) {
430 ret
.pre_versionid
= uadv(k
);
431 } else if (!ustricmp(k
, L
"html-post-versionid")) {
432 ret
.post_versionid
= uadv(k
);
438 * Now process fallbacks on quote characters.
440 while (*uadv(ret
.rquote
) && *uadv(uadv(ret
.rquote
)) &&
441 (!cvt_ok(ret
.restrict_charset
, ret
.lquote
) ||
442 !cvt_ok(ret
.restrict_charset
, ret
.rquote
))) {
443 ret
.lquote
= uadv(ret
.rquote
);
444 ret
.rquote
= uadv(ret
.lquote
);
450 paragraph
*html_config_filename(char *filename
)
453 * If the user passes in a single filename as a parameter to
454 * the `--html' command-line option, then we should assume it
455 * to imply _two_ config directives:
456 * \cfg{html-single-filename}{whatever} and
457 * \cfg{html-leaf-level}{0}; the rationale being that the user
458 * wants their output _in that file_.
462 p
= cmdline_cfg_simple("html-single-filename", filename
, NULL
);
463 q
= cmdline_cfg_simple("html-leaf-level", "0", NULL
);
468 void html_backend(paragraph
*sourceform
, keywordlist
*keywords
,
469 indexdata
*idx
, void *unused
) {
472 htmlfilelist files
= { NULL
, NULL
, NULL
, NULL
, NULL
};
473 htmlsectlist sects
= { NULL
, NULL
}, nonsects
= { NULL
, NULL
};
477 conf
= html_configure(sourceform
);
480 * We're going to make heavy use of paragraphs' private data
481 * fields in the forthcoming code. Clear them first, so we can
482 * reliably tell whether we have auxiliary data for a
483 * particular paragraph.
485 for (p
= sourceform
; p
; p
= p
->next
)
486 p
->private_data
= NULL
;
488 files
.frags
= newtree234(html_fragment_compare
);
491 * Start by figuring out into which file each piece of the
492 * document should be put. We'll do this by inventing an
493 * `htmlsect' structure and stashing it in the private_data
494 * field of each section paragraph; we also need one additional
495 * htmlsect for the document index, which won't show up in the
496 * source form but needs to be consistently mentioned in
499 * While we're here, we'll also invent the HTML fragment name
503 htmlsect
*topsect
, *sect
;
506 topsect
= html_new_sect(§s
, NULL
);
508 topsect
->title
= NULL
;
509 topsect
->text
= sourceform
;
510 topsect
->contents_depth
= contents_depth(conf
, 0);
511 html_file_section(&conf
, &files
, topsect
, -1);
512 topsect
->fragment
= NULL
;
514 for (p
= sourceform
; p
; p
= p
->next
)
515 if (is_heading_type(p
->type
)) {
516 d
= heading_depth(p
);
518 if (p
->type
== para_Title
) {
523 sect
= html_new_sect(§s
, p
);
524 sect
->text
= p
->next
;
526 sect
->contents_depth
= contents_depth(conf
, d
+1) - (d
+1);
529 sect
->parent
= (htmlsect
*)p
->parent
->private_data
;
530 assert(sect
->parent
!= NULL
);
532 sect
->parent
= topsect
;
533 p
->private_data
= sect
;
535 html_file_section(&conf
, &files
, sect
, d
);
537 sect
->fragment
= html_format(p
, conf
.template_fragment
);
538 sect
->fragment
= html_sanitise_fragment(&files
, sect
->file
,
543 sect
= html_new_sect(§s
, NULL
);
546 sect
->parent
= topsect
;
547 html_file_section(&conf
, &files
, sect
, 0); /* peer of chapters */
548 sect
->fragment
= utoa_dup(conf
.index_text
, CS_ASCII
);
549 sect
->fragment
= html_sanitise_fragment(&files
, sect
->file
,
551 files
.index
= sect
->file
;
555 * Go through the keyword list and sort out fragment IDs for
556 * all the potentially referenced paragraphs which _aren't_
564 for (i
= 0; (kw
= index234(keywords
->keys
, i
)) != NULL
; i
++) {
565 paragraph
*q
, *p
= kw
->para
;
567 if (!is_heading_type(p
->type
)) {
571 * Find the paragraph's parent htmlsect, to
572 * determine which file it will end up in.
577 * Preamble paragraphs have no parent. So if we
578 * have a non-heading with no parent, it must
579 * be preamble, and therefore its parent
580 * htmlsect must be the preamble one.
583 sects
.head
->type
== TOP
);
586 parent
= (htmlsect
*)q
->private_data
;
589 * Now we can construct an htmlsect for this
590 * paragraph itself, taking care to put it in the
591 * list of non-sections rather than the list of
592 * sections (so that traverses of the `sects' list
593 * won't attempt to add it to the contents or
594 * anything weird like that).
596 sect
= html_new_sect(&nonsects
, p
);
597 sect
->file
= parent
->file
;
598 sect
->parent
= parent
;
599 p
->private_data
= sect
;
602 * Fragment IDs for these paragraphs will simply be
603 * `p' followed by an integer.
605 sect
->fragment
= snewn(40, char);
606 sprintf(sect
->fragment
, "p%d",
607 sect
->file
->last_fragment_number
++);
608 sect
->fragment
= html_sanitise_fragment(&files
, sect
->file
,
615 * Reset the fragment numbers in each file. I've just used them
616 * to generate `p' fragment IDs for non-section paragraphs
617 * (numbered list elements, bibliocited), and now I want to use
618 * them for `i' fragment IDs for index entries.
622 for (file
= files
.head
; file
; file
= file
->next
)
623 file
->last_fragment_number
= 0;
627 * Now sort out the index. This involves:
629 * - For each index term, we set up an htmlindex structure to
630 * store all the references to that term.
632 * - Then we make a pass over the actual document, finding
633 * every word_IndexRef; for each one, we actually figure out
634 * the HTML filename/fragment pair we will use to reference
635 * it, store that information in the private data field of
636 * the word_IndexRef itself (so we can recreate it when the
637 * time comes to output our HTML), and add a reference to it
638 * to the index term in question.
647 * Set up the htmlindex structures.
650 for (i
= 0; (entry
= index234(idx
->entries
, i
)) != NULL
; i
++) {
651 htmlindex
*hi
= snew(htmlindex
);
653 hi
->nrefs
= hi
->refsize
= 0;
656 entry
->backend_data
= hi
;
660 * Run over the document inventing fragments. Each fragment
661 * is of the form `i' followed by an integer.
663 lastsect
= sects
.head
; /* this is always the top section */
664 for (p
= sourceform
; p
; p
= p
->next
) {
665 if (is_heading_type(p
->type
) && p
->type
!= para_Title
)
666 lastsect
= (htmlsect
*)p
->private_data
;
668 for (w
= p
->words
; w
; w
= w
->next
)
669 if (w
->type
== word_IndexRef
) {
670 htmlindexref
*hr
= snew(htmlindexref
);
674 hr
->referenced
= hr
->generated
= FALSE
;
675 hr
->section
= lastsect
;
679 lastsect
->file
->last_fragment_number
++);
680 hr
->fragment
= dupstr(buf
);
682 html_sanitise_fragment(&files
, hr
->section
->file
,
685 w
->private_data
= hr
;
687 tag
= index_findtag(idx
, w
->text
);
691 for (i
= 0; i
< tag
->nrefs
; i
++) {
692 indexentry
*entry
= tag
->refs
[i
];
693 htmlindex
*hi
= (htmlindex
*)entry
->backend_data
;
695 if (hi
->nrefs
>= hi
->refsize
) {
697 hi
->refs
= sresize(hi
->refs
, hi
->refsize
, word
*);
700 hi
->refs
[hi
->nrefs
++] = w
;
707 * Now we're ready to write out the actual HTML files.
711 * - we open that file and write its header
712 * - we run down the list of sections
713 * - for each section directly contained within that file, we
714 * output the section text
715 * - for each section which is not in the file but which has a
716 * parent that is, we output a contents entry for the
717 * section if appropriate
718 * - finally, we output the file trailer and close the file.
727 for (f
= files
.head
; f
; f
= f
->next
) {
730 enum LISTTYPE
{ NOLIST
, UL
, OL
, DL
};
731 enum ITEMTYPE
{ NOITEM
, LI
, DT
, DD
};
732 struct stackelement
{
733 struct stackelement
*next
;
734 enum LISTTYPE listtype
;
735 enum ITEMTYPE itemtype
;
738 #define listname(lt) ( (lt)==UL ? "ul" : (lt)==OL ? "ol" : "dl" )
739 #define itemname(lt) ( (lt)==LI ? "li" : (lt)==DT ? "dt" : "dd" )
741 ho
.fp
= fopen(f
->filename
, "w");
742 ho
.charset
= conf
.output_charset
;
743 ho
.restrict_charset
= conf
.restrict_charset
;
744 ho
.cstate
= charset_init_state
;
745 ho
.ver
= conf
.htmlver
;
746 ho
.state
= HO_NEUTRAL
;
747 ho
.contents_level
= 0;
750 switch (conf
.htmlver
) {
752 fprintf(ho
.fp
, "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD "
753 "HTML 3.2 Final//EN\">\n");
756 fprintf(ho
.fp
, "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML"
757 " 4.01//EN\"\n\"http://www.w3.org/TR/html4/"
761 fprintf(ho
.fp
, "<!DOCTYPE HTML PUBLIC \"ISO/IEC "
762 "15445:2000//DTD HTML//EN\">\n");
764 case XHTML_1_0_TRANSITIONAL
:
765 fprintf(ho
.fp
, "<?xml version=\"1.0\" encoding=\"%s\"?>\n",
766 charset_to_mimeenc(conf
.output_charset
));
767 fprintf(ho
.fp
, "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML"
768 " 1.0 Transitional//EN\"\n\"http://www.w3.org/TR/"
769 "xhtml1/DTD/xhtml1-transitional.dtd\">\n");
771 case XHTML_1_0_STRICT
:
772 fprintf(ho
.fp
, "<?xml version=\"1.0\" encoding=\"%s\"?>\n",
773 charset_to_mimeenc(conf
.output_charset
));
774 fprintf(ho
.fp
, "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML"
775 " 1.0 Strict//EN\"\n\"http://www.w3.org/TR/xhtml1/"
776 "DTD/xhtml1-strict.dtd\">\n");
780 element_open(&ho
, "html");
781 if (is_xhtml(conf
.htmlver
)) {
782 element_attr(&ho
, "xmlns", "http://www.w3.org/1999/xhtml");
786 element_open(&ho
, "head");
789 element_empty(&ho
, "meta");
790 element_attr(&ho
, "http-equiv", "content-type");
793 sprintf(buf
, "text/html; charset=%.150s",
794 charset_to_mimeenc(conf
.output_charset
));
795 element_attr(&ho
, "content", buf
);
800 element_empty(&ho
, "meta");
801 element_attr(&ho
, "name", "author");
802 element_attr_w(&ho
, "content", conf
.author
);
806 if (conf
.description
) {
807 element_empty(&ho
, "meta");
808 element_attr(&ho
, "name", "description");
809 element_attr_w(&ho
, "content", conf
.description
);
813 element_open(&ho
, "title");
814 if (f
->first
&& f
->first
->title
) {
815 html_words(&ho
, f
->first
->title
->words
, NOTHING
,
819 if (f
->last
!= f
->first
&& f
->last
->title
) {
820 html_text(&ho
, conf
.title_separator
);
821 html_words(&ho
, f
->last
->title
->words
, NOTHING
,
825 element_close(&ho
, "title");
829 html_raw(&ho
, conf
.head_end
);
831 element_close(&ho
, "head");
835 html_raw(&ho
, conf
.body_tag
);
837 element_open(&ho
, "body");
841 html_raw(&ho
, conf
.body_start
);
844 * Write out a nav bar. Special case: we don't do this
845 * if there is only one file.
847 if (files
.head
!= files
.tail
) {
848 element_open(&ho
, "p");
850 html_raw_as_attr(&ho
, conf
.nav_attr
);
853 element_open(&ho
, "a");
854 element_attr(&ho
, "href", prevf
->filename
);
856 html_text(&ho
, conf
.nav_prev_text
);
858 element_close(&ho
, "a");
860 html_text(&ho
, conf
.nav_separator
);
862 if (f
!= files
.head
) {
863 element_open(&ho
, "a");
864 element_attr(&ho
, "href", files
.head
->filename
);
866 html_text(&ho
, conf
.contents_text
);
868 element_close(&ho
, "a");
870 html_text(&ho
, conf
.nav_separator
);
872 if (f
!= files
.index
) {
873 element_open(&ho
, "a");
874 element_attr(&ho
, "href", files
.index
->filename
);
876 html_text(&ho
, conf
.index_text
);
877 if (f
!= files
.index
)
878 element_close(&ho
, "a");
880 html_text(&ho
, conf
.nav_separator
);
883 element_open(&ho
, "a");
884 element_attr(&ho
, "href", f
->next
->filename
);
886 html_text(&ho
, conf
.nav_next_text
);
888 element_close(&ho
, "a");
890 element_close(&ho
, "p");
896 * Write out a prefix TOC for the file.
898 * We start by going through the section list and
899 * collecting the sections which need to be added to
900 * the contents. On the way, we also test to see if
901 * this file is a leaf file (defined as one which
902 * contains all descendants of any section it
903 * contains), because this will play a part in our
904 * decision on whether or not to _output_ the TOC.
906 * Special case: we absolutely do not do this if we're
907 * in single-file mode.
909 if (files
.head
!= files
.tail
) {
910 int ntoc
= 0, tocsize
= 0;
911 htmlsect
**toc
= NULL
;
914 for (s
= sects
.head
; s
; s
= s
->next
) {
919 * Search up from this section until we find
920 * the highest-level one which belongs in this
925 for (ac
= s
; ac
; ac
= ac
->parent
) {
933 if (s
->file
!= f
&& a
!= NULL
)
937 if (adepth
<= a
->contents_depth
) {
938 if (ntoc
>= tocsize
) {
940 toc
= sresize(toc
, tocsize
, htmlsect
*);
947 if (leaf
&& conf
.leaf_contains_contents
&&
948 ntoc
>= conf
.leaf_smallest_contents
) {
951 for (i
= 0; i
< ntoc
; i
++) {
952 htmlsect
*s
= toc
[i
];
953 int hlevel
= (s
->type
== TOP ?
-1 :
954 s
->type
== INDEX ?
0 :
955 heading_depth(s
->title
))
956 - f
->min_heading_depth
+ 1;
959 html_contents_entry(&ho
, hlevel
, s
,
962 html_contents_entry(&ho
, 0, NULL
, f
, keywords
, &conf
);
967 * Now go through the document and output some real
971 for (s
= sects
.head
; s
; s
= s
->next
) {
974 * This section belongs in this file.
985 * Search up from this section until we find
986 * the highest-level one which belongs in this
991 for (ac
= s
; ac
; ac
= ac
->parent
) {
1001 * This section does not belong in this
1002 * file, but an ancestor of it does. Write
1003 * out a contents table entry, if the depth
1004 * doesn't exceed the maximum contents
1005 * depth for the ancestor section.
1007 if (adepth
<= a
->contents_depth
) {
1008 html_contents_entry(&ho
, adepth
, s
,
1009 f
, keywords
, &conf
);
1018 html_contents_entry(&ho
, 0, NULL
, f
, keywords
, &conf
);
1021 * Display the section heading.
1024 hlevel
= (s
->type
== TOP ?
-1 :
1025 s
->type
== INDEX ?
0 :
1026 heading_depth(s
->title
))
1027 - f
->min_heading_depth
+ 1;
1028 assert(hlevel
>= 1);
1029 /* HTML headings only go up to <h6> */
1033 htag
[1] = '0' + hlevel
;
1035 element_open(&ho
, htag
);
1038 * Provide anchor for cross-links to target.
1040 * (Also we'll have to do this separately in
1041 * other paragraph types - NumberedList and
1045 html_fragment(&ho
, s
->fragment
);
1047 html_section_title(&ho
, s
, f
, keywords
, &conf
, TRUE
);
1049 element_close(&ho
, htag
);
1052 * Now display the section text.
1055 stackhead
= snew(struct stackelement
);
1056 stackhead
->next
= NULL
;
1057 stackhead
->listtype
= NOLIST
;
1058 stackhead
->itemtype
= NOITEM
;
1060 for (p
= s
->text
;; p
= p
->next
) {
1061 enum LISTTYPE listtype
;
1062 struct stackelement
*se
;
1065 * Preliminary switch to figure out what
1066 * sort of list we expect to be inside at
1069 * Since p may still be NULL at this point,
1070 * I invent a harmless paragraph type for
1073 switch (p ? p
->type
: para_Normal
) {
1076 case para_Copyright
:
1077 case para_BiblioCited
:
1079 case para_QuotePush
:
1083 case para_UnnumberedChapter
:
1094 case para_NumberedList
:
1098 case para_DescribedThing
:
1099 case para_Description
:
1103 case para_LcontPush
:
1104 se
= snew(struct stackelement
);
1105 se
->next
= stackhead
;
1106 se
->listtype
= NOLIST
;
1107 se
->itemtype
= NOITEM
;
1111 default: /* some totally non-printing para */
1118 * Terminate the most recent list item, if
1119 * any. (We left this until after
1120 * processing LcontPush, since in that case
1121 * the list item won't want to be
1122 * terminated until after the corresponding
1125 if (stackhead
->itemtype
!= NOITEM
) {
1126 element_close(&ho
, itemname(stackhead
->itemtype
));
1129 stackhead
->itemtype
= NOITEM
;
1132 * Terminate the current list, if it's not
1133 * the one we want to be in.
1135 if (listtype
!= stackhead
->listtype
&&
1136 stackhead
->listtype
!= NOLIST
) {
1137 element_close(&ho
, listname(stackhead
->listtype
));
1142 * Leave the loop if our time has come.
1144 if (!p
|| (is_heading_type(p
->type
) &&
1145 p
->type
!= para_Title
))
1146 break; /* end of section text */
1149 * Start a fresh list if necessary.
1151 if (listtype
!= stackhead
->listtype
&&
1153 element_open(&ho
, listname(listtype
));
1155 stackhead
->listtype
= listtype
;
1159 element_empty(&ho
, "hr");
1162 html_codepara(&ho
, p
->words
);
1165 case para_Copyright
:
1166 element_open(&ho
, "p");
1168 html_words(&ho
, p
->words
, ALL
,
1169 f
, keywords
, &conf
);
1171 element_close(&ho
, "p");
1173 case para_BiblioCited
:
1174 element_open(&ho
, "p");
1175 if (p
->private_data
) {
1176 htmlsect
*s
= (htmlsect
*)p
->private_data
;
1177 html_fragment(&ho
, s
->fragment
);
1180 html_words(&ho
, p
->kwtext
, ALL
,
1181 f
, keywords
, &conf
);
1182 html_text(&ho
, L
" ");
1183 html_words(&ho
, p
->words
, ALL
,
1184 f
, keywords
, &conf
);
1186 element_close(&ho
, "p");
1189 case para_NumberedList
:
1190 element_open(&ho
, "li");
1191 if (p
->private_data
) {
1192 htmlsect
*s
= (htmlsect
*)p
->private_data
;
1193 html_fragment(&ho
, s
->fragment
);
1196 stackhead
->itemtype
= LI
;
1197 html_words(&ho
, p
->words
, ALL
,
1198 f
, keywords
, &conf
);
1200 case para_DescribedThing
:
1201 element_open(&ho
, "dt");
1203 stackhead
->itemtype
= DT
;
1204 html_words(&ho
, p
->words
, ALL
,
1205 f
, keywords
, &conf
);
1207 case para_Description
:
1208 element_open(&ho
, "dd");
1210 stackhead
->itemtype
= DD
;
1211 html_words(&ho
, p
->words
, ALL
,
1212 f
, keywords
, &conf
);
1215 case para_QuotePush
:
1216 element_open(&ho
, "blockquote");
1219 element_close(&ho
, "blockquote");
1224 stackhead
= stackhead
->next
;
1231 assert(stackhead
&& !stackhead
->next
);
1235 if (s
->type
== INDEX
) {
1240 * This section is the index. I'll just
1241 * render it as a single paragraph, with a
1242 * colon between the index term and the
1243 * references, and <br> in between each
1246 element_open(&ho
, "p");
1248 for (i
= 0; (entry
=
1249 index234(idx
->entries
, i
)) != NULL
; i
++) {
1250 htmlindex
*hi
= (htmlindex
*)entry
->backend_data
;
1254 element_empty(&ho
, "br");
1257 html_words(&ho
, entry
->text
, MARKUP
|LINKS
,
1258 f
, keywords
, &conf
);
1260 html_text(&ho
, conf
.index_main_sep
);
1262 for (j
= 0; j
< hi
->nrefs
; j
++) {
1264 (htmlindexref
*)hi
->refs
[j
]->private_data
;
1265 paragraph
*p
= hr
->section
->title
;
1268 html_text(&ho
, conf
.index_multi_sep
);
1270 html_href(&ho
, f
, hr
->section
->file
,
1272 hr
->referenced
= TRUE
;
1274 html_words(&ho
, p
->kwtext
, MARKUP
|LINKS
,
1275 f
, keywords
, &conf
);
1276 else if (p
&& p
->words
)
1277 html_words(&ho
, p
->words
, MARKUP
|LINKS
,
1278 f
, keywords
, &conf
);
1281 * If there is no title at all,
1282 * this must be because our
1283 * target section is the
1284 * preamble section and there
1285 * is no title. So we use the
1288 html_text(&ho
, conf
.preamble_text
);
1290 element_close(&ho
, "a");
1293 element_close(&ho
, "p");
1298 html_contents_entry(&ho
, 0, NULL
, f
, keywords
, &conf
);
1305 int done_version_ids
= FALSE
;
1307 element_empty(&ho
, "hr");
1310 html_raw(&ho
, conf
.body_end
);
1312 if (conf
.address_section
) {
1313 int started
= FALSE
;
1314 if (conf
.htmlver
== ISO_HTML
) {
1316 * The ISO-HTML validator complains if
1317 * there isn't a <div> tag surrounding the
1318 * <address> tag. I'm uncertain of why this
1319 * should be - there appears to be no
1320 * mention of this in the ISO-HTML spec,
1321 * suggesting that it doesn't represent a
1322 * change from HTML 4, but nonetheless the
1323 * HTML 4 validator doesn't seem to mind.
1325 element_open(&ho
, "div");
1327 element_open(&ho
, "address");
1328 if (conf
.addr_start
) {
1329 html_raw(&ho
, conf
.addr_start
);
1333 if (conf
.visible_version_id
) {
1334 for (p
= sourceform
; p
; p
= p
->next
)
1335 if (p
->type
== para_VersionID
) {
1337 element_empty(&ho
, "br");
1339 html_text(&ho
, conf
.pre_versionid
);
1340 html_words(&ho
, p
->words
, NOTHING
,
1341 f
, keywords
, &conf
);
1342 html_text(&ho
, conf
.post_versionid
);
1345 done_version_ids
= TRUE
;
1347 if (conf
.addr_end
) {
1349 element_empty(&ho
, "br");
1350 html_raw(&ho
, conf
.addr_end
);
1352 element_close(&ho
, "address");
1353 if (conf
.htmlver
== ISO_HTML
)
1354 element_close(&ho
, "div");
1357 if (!done_version_ids
) {
1359 * If the user didn't want the version IDs
1360 * visible, I think we still have a duty to put
1361 * them in an HTML comment.
1363 int started
= FALSE
;
1364 for (p
= sourceform
; p
; p
= p
->next
)
1365 if (p
->type
== para_VersionID
) {
1367 html_raw(&ho
, "<!-- version IDs:\n");
1370 html_words(&ho
, p
->words
, NOTHING
,
1371 f
, keywords
, &conf
);
1375 html_raw(&ho
, "-->\n");
1379 element_close(&ho
, "body");
1381 element_close(&ho
, "html");
1388 * Go through and check that no index fragments were referenced
1389 * without being generated, or indeed vice versa.
1391 * (When I actually get round to freeing everything, this can
1392 * probably be the freeing loop as well.)
1394 for (p
= sourceform
; p
; p
= p
->next
) {
1396 for (w
= p
->words
; w
; w
= w
->next
)
1397 if (w
->type
== word_IndexRef
) {
1398 htmlindexref
*hr
= (htmlindexref
*)w
->private_data
;
1400 assert(!hr
->referenced
== !hr
->generated
);
1405 * FIXME: Free all the working data.
1409 static void html_file_section(htmlconfig
*cfg
, htmlfilelist
*files
,
1410 htmlsect
*sect
, int depth
)
1416 * `depth' is derived from the heading_depth() macro at the top
1417 * of this file, which counts title as -1, chapter as 0,
1418 * heading as 1 and subsection as 2. However, the semantics of
1419 * cfg->leaf_level are defined to count chapter as 1, heading
1420 * as 2 etc. So first I increment depth :-(
1424 if (cfg
->leaf_level
== 0) {
1426 * leaf_level==0 is a special case, in which everything is
1427 * put into a single file.
1430 files
->single
= html_new_file(files
, cfg
->single_filename
);
1432 file
= files
->single
;
1435 * If the depth of this section is at or above leaf_level,
1436 * we invent a fresh file and put this section at its head.
1437 * Otherwise, we put it in the same file as its parent
1440 if (ldepth
> cfg
->leaf_level
) {
1442 * We know that sect->parent cannot be NULL. The only
1443 * circumstance in which it can be is if sect is at
1444 * chapter or appendix level, i.e. ldepth==1; and if
1445 * that's the case, then we cannot have entered this
1446 * branch unless cfg->leaf_level==0, in which case we
1447 * would be in the single-file case above and not here
1450 assert(sect
->parent
);
1452 file
= sect
->parent
->file
;
1454 if (sect
->type
== TOP
) {
1455 file
= html_new_file(files
, cfg
->contents_filename
);
1456 } else if (sect
->type
== INDEX
) {
1457 file
= html_new_file(files
, cfg
->index_filename
);
1461 assert(ldepth
> 0 && sect
->title
);
1462 title
= html_format(sect
->title
, cfg
->template_filename
);
1463 file
= html_new_file(files
, title
);
1471 if (file
->min_heading_depth
> depth
) {
1473 * This heading is at a higher level than any heading we
1474 * have so far placed in this file; so we set the `first'
1477 file
->min_heading_depth
= depth
;
1481 if (file
->min_heading_depth
== depth
)
1485 static htmlfile
*html_new_file(htmlfilelist
*list
, char *filename
)
1487 htmlfile
*ret
= snew(htmlfile
);
1491 list
->tail
->next
= ret
;
1496 ret
->filename
= dupstr(filename
);
1497 ret
->last_fragment_number
= 0;
1498 ret
->min_heading_depth
= INT_MAX
;
1499 ret
->first
= ret
->last
= NULL
;
1504 static htmlsect
*html_new_sect(htmlsectlist
*list
, paragraph
*title
)
1506 htmlsect
*ret
= snew(htmlsect
);
1510 list
->tail
->next
= ret
;
1523 static void html_words(htmloutput
*ho
, word
*words
, int flags
,
1524 htmlfile
*file
, keywordlist
*keywords
, htmlconfig
*cfg
)
1530 for (w
= words
; w
; w
= w
->next
) switch (w
->type
) {
1531 case word_HyperLink
:
1532 if (flags
& LINKS
) {
1533 element_open(ho
, "a");
1534 c
= utoa_dup(w
->text
, CS_ASCII
);
1535 element_attr(ho
, "href", c
);
1539 case word_UpperXref
:
1540 case word_LowerXref
:
1541 if (flags
& LINKS
) {
1542 keyword
*kwl
= kw_lookup(keywords
, w
->text
);
1543 paragraph
*p
= kwl
->para
;
1544 htmlsect
*s
= (htmlsect
*)p
->private_data
;
1548 html_href(ho
, file
, s
->file
, s
->fragment
);
1554 element_close(ho
, "a");
1557 if (flags
& INDEXENTS
) {
1558 htmlindexref
*hr
= (htmlindexref
*)w
->private_data
;
1559 html_fragment(ho
, hr
->fragment
);
1560 hr
->generated
= TRUE
;
1567 case word_WhiteSpace
:
1568 case word_EmphSpace
:
1569 case word_CodeSpace
:
1570 case word_WkCodeSpace
:
1572 case word_EmphQuote
:
1573 case word_CodeQuote
:
1574 case word_WkCodeQuote
:
1575 style
= towordstyle(w
->type
);
1576 type
= removeattr(w
->type
);
1577 if (style
== word_Emph
&&
1578 (attraux(w
->aux
) == attr_First
||
1579 attraux(w
->aux
) == attr_Only
) &&
1581 element_open(ho
, "em");
1582 else if ((style
== word_Code
|| style
== word_WeakCode
) &&
1583 (attraux(w
->aux
) == attr_First
||
1584 attraux(w
->aux
) == attr_Only
) &&
1586 element_open(ho
, "code");
1588 if (type
== word_WhiteSpace
)
1589 html_text(ho
, L
" ");
1590 else if (type
== word_Quote
) {
1591 if (quoteaux(w
->aux
) == quote_Open
)
1592 html_text(ho
, cfg
->lquote
);
1594 html_text(ho
, cfg
->rquote
);
1596 if (cvt_ok(ho
->restrict_charset
, w
->text
) || !w
->alt
)
1597 html_text(ho
, w
->text
);
1599 html_words(ho
, w
->alt
, flags
, file
, keywords
, cfg
);
1602 if (style
== word_Emph
&&
1603 (attraux(w
->aux
) == attr_Last
||
1604 attraux(w
->aux
) == attr_Only
) &&
1606 element_close(ho
, "em");
1607 else if ((style
== word_Code
|| style
== word_WeakCode
) &&
1608 (attraux(w
->aux
) == attr_Last
||
1609 attraux(w
->aux
) == attr_Only
) &&
1611 element_close(ho
, "code");
1617 static void html_codepara(htmloutput
*ho
, word
*words
)
1619 element_open(ho
, "pre");
1620 element_open(ho
, "code");
1621 for (; words
; words
= words
->next
) if (words
->type
== word_WeakCode
) {
1626 if (words
->next
&& words
->next
->type
== word_Emph
) {
1627 e
= words
->next
->text
;
1628 words
= words
->next
;
1632 while (e
&& *e
&& *t
) {
1636 for (n
= 0; t
[n
] && e
[n
] && e
[n
] == ec
; n
++);
1644 element_open(ho
, open_tag
);
1646 html_text_limit(ho
, t
, n
);
1649 element_close(ho
, open_tag
);
1657 element_close(ho
, "code");
1658 element_close(ho
, "pre");
1661 static void html_charset_cleanup(htmloutput
*ho
)
1666 bytes
= charset_from_unicode(NULL
, NULL
, outbuf
, lenof(outbuf
),
1667 ho
->charset
, &ho
->cstate
, NULL
);
1669 fwrite(outbuf
, 1, bytes
, ho
->fp
);
1672 static void return_to_neutral(htmloutput
*ho
)
1674 if (ho
->state
== HO_IN_TEXT
) {
1675 html_charset_cleanup(ho
);
1676 } else if (ho
->state
== HO_IN_EMPTY_TAG
&& is_xhtml(ho
->ver
)) {
1677 fprintf(ho
->fp
, " />");
1678 } else if (ho
->state
== HO_IN_EMPTY_TAG
|| ho
->state
== HO_IN_TAG
) {
1679 fprintf(ho
->fp
, ">");
1682 ho
->state
= HO_NEUTRAL
;
1685 static void element_open(htmloutput
*ho
, char const *name
)
1687 return_to_neutral(ho
);
1688 fprintf(ho
->fp
, "<%s", name
);
1689 ho
->state
= HO_IN_TAG
;
1692 static void element_close(htmloutput
*ho
, char const *name
)
1694 return_to_neutral(ho
);
1695 fprintf(ho
->fp
, "</%s>", name
);
1696 ho
->state
= HO_NEUTRAL
;
1699 static void element_empty(htmloutput
*ho
, char const *name
)
1701 return_to_neutral(ho
);
1702 fprintf(ho
->fp
, "<%s", name
);
1703 ho
->state
= HO_IN_EMPTY_TAG
;
1706 static void html_nl(htmloutput
*ho
)
1708 return_to_neutral(ho
);
1709 fputc('\n', ho
->fp
);
1712 static void html_raw(htmloutput
*ho
, char *text
)
1714 return_to_neutral(ho
);
1715 fputs(text
, ho
->fp
);
1718 static void html_raw_as_attr(htmloutput
*ho
, char *text
)
1720 assert(ho
->state
== HO_IN_TAG
|| ho
->state
== HO_IN_EMPTY_TAG
);
1722 fputs(text
, ho
->fp
);
1725 static void element_attr(htmloutput
*ho
, char const *name
, char const *value
)
1727 html_charset_cleanup(ho
);
1728 assert(ho
->state
== HO_IN_TAG
|| ho
->state
== HO_IN_EMPTY_TAG
);
1729 fprintf(ho
->fp
, " %s=\"%s\"", name
, value
);
1732 static void element_attr_w(htmloutput
*ho
, char const *name
,
1733 wchar_t const *value
)
1735 html_charset_cleanup(ho
);
1736 fprintf(ho
->fp
, " %s=\"", name
);
1737 html_text_limit_internal(ho
, value
, 0, TRUE
);
1738 html_charset_cleanup(ho
);
1742 static void html_text(htmloutput
*ho
, wchar_t const *text
)
1744 html_text_limit(ho
, text
, 0);
1747 static void html_text_limit(htmloutput
*ho
, wchar_t const *text
, int maxlen
)
1749 return_to_neutral(ho
);
1750 html_text_limit_internal(ho
, text
, maxlen
, FALSE
);
1753 static void html_text_limit_internal(htmloutput
*ho
, wchar_t const *text
,
1754 int maxlen
, int quote_quotes
)
1756 int textlen
= ustrlen(text
);
1760 if (maxlen
> 0 && textlen
> maxlen
)
1763 while (textlen
> 0) {
1764 /* Scan ahead for characters we really can't display in HTML. */
1765 int lenbefore
, lenafter
;
1766 for (lenbefore
= 0; lenbefore
< textlen
; lenbefore
++)
1767 if (text
[lenbefore
] == L
'<' ||
1768 text
[lenbefore
] == L
'>' ||
1769 text
[lenbefore
] == L
'&' ||
1770 (text
[lenbefore
] == L
'"' && quote_quotes
))
1772 lenafter
= lenbefore
;
1773 bytes
= charset_from_unicode(&text
, &lenafter
, outbuf
, lenof(outbuf
),
1774 ho
->charset
, &ho
->cstate
, &err
);
1775 textlen
-= (lenbefore
- lenafter
);
1777 fwrite(outbuf
, 1, bytes
, ho
->fp
);
1780 * We have encountered a character that cannot be
1781 * displayed in the selected output charset. Therefore,
1782 * we use an HTML numeric entity reference.
1784 assert(textlen
> 0);
1785 fprintf(ho
->fp
, "&#%ld;", (long int)*text
);
1787 } else if (lenafter
== 0 && textlen
> 0) {
1789 * We have encountered a character which is special to
1793 fprintf(ho
->fp
, "<");
1794 else if (*text
== L
'>')
1795 fprintf(ho
->fp
, ">");
1796 else if (*text
== L
'&')
1797 fprintf(ho
->fp
, "&");
1798 else if (*text
== L
'"')
1799 fprintf(ho
->fp
, """);
1801 assert(!"Can't happen");
1807 static void cleanup(htmloutput
*ho
)
1809 return_to_neutral(ho
);
1813 static void html_href(htmloutput
*ho
, htmlfile
*thisfile
,
1814 htmlfile
*targetfile
, char *targetfrag
)
1816 rdstringc rs
= { 0, 0, NULL
};
1819 if (targetfile
!= thisfile
)
1820 rdaddsc(&rs
, targetfile
->filename
);
1823 rdaddsc(&rs
, targetfrag
);
1827 element_open(ho
, "a");
1828 element_attr(ho
, "href", url
);
1832 static void html_fragment(htmloutput
*ho
, char const *fragment
)
1834 element_open(ho
, "a");
1835 element_attr(ho
, "name", fragment
);
1836 if (is_xhtml(ho
->ver
))
1837 element_attr(ho
, "id", fragment
);
1838 element_close(ho
, "a");
1841 static char *html_format(paragraph
*p
, char *template_string
)
1845 wchar_t *ws
, wsbuf
[2];
1846 rdstringc rs
= { 0, 0, NULL
};
1848 t
= template_string
;
1850 if (*t
== '%' && t
[1]) {
1864 if (p
->kwtext
&& fmt
== 'n')
1866 else if (p
->kwtext2
&& fmt
== 'b') {
1868 * HTML fragment names must start with a letter, so
1869 * simply `1.2.3' is not adequate. In this case I'm
1870 * going to cheat slightly by prepending the first
1871 * character of the first word of kwtext, so that
1872 * we get `C1' for chapter 1, `S2.3' for section
1875 if (p
->kwtext
&& p
->kwtext
->text
[0]) {
1878 wsbuf
[0] = p
->kwtext
->text
[0];
1881 } else if (p
->keyword
&& *p
->keyword
&& fmt
== 'k')
1887 c
= utoa_dup(ws
, CS_ASCII
);
1893 if (removeattr(w
->type
) == word_Normal
) {
1894 c
= utoa_dup(w
->text
, CS_ASCII
);
1905 return rdtrimc(&rs
);
1908 static char *html_sanitise_fragment(htmlfilelist
*files
, htmlfile
*file
,
1912 * The HTML 4 spec's strictest definition of fragment names (<a
1913 * name> and "id" attributes) says that they `must begin with a
1914 * letter and may be followed by any number of letters, digits,
1915 * hyphens, underscores, colons, and periods'.
1917 * So here we unceremoniously rip out any characters not
1918 * conforming to this limitation.
1920 char *p
= text
, *q
= text
;
1922 while (*p
&& !((*p
>='A' && *p
<='Z') || (*p
>='a' && *p
<='z')))
1924 if ((*q
++ = *p
++) != '\0') {
1926 if ((*p
>='A' && *p
<='Z') ||
1927 (*p
>='a' && *p
<='z') ||
1928 (*p
>='0' && *p
<='9') ||
1929 *p
=='-' || *p
=='_' || *p
==':' || *p
=='.')
1938 * Now we check for clashes with other fragment names, and
1939 * adjust this one if necessary by appending a hyphen followed
1943 htmlfragment
*frag
= snew(htmlfragment
);
1944 int len
= 0; /* >0 indicates we have resized */
1948 frag
->fragment
= text
;
1950 while (add234(files
->frags
, frag
) != frag
) {
1953 frag
->fragment
= text
= sresize(text
, len
+20, char);
1956 sprintf(text
+ len
, "-%d", ++suffix
);
1963 static void html_contents_entry(htmloutput
*ho
, int depth
, htmlsect
*s
,
1964 htmlfile
*thisfile
, keywordlist
*keywords
,
1967 while (ho
->contents_level
> depth
) {
1968 element_close(ho
, "ul");
1969 ho
->contents_level
--;
1972 while (ho
->contents_level
< depth
) {
1973 element_open(ho
, "ul");
1974 ho
->contents_level
++;
1980 element_open(ho
, "li");
1981 html_href(ho
, thisfile
, s
->file
, s
->fragment
);
1982 html_section_title(ho
, s
, thisfile
, keywords
, cfg
, FALSE
);
1983 element_close(ho
, "a");
1984 element_close(ho
, "li");
1987 static void html_section_title(htmloutput
*ho
, htmlsect
*s
, htmlfile
*thisfile
,
1988 keywordlist
*keywords
, htmlconfig
*cfg
,
1994 int depth
= heading_depth(s
->title
);
1998 else if (depth
== 0)
1999 sl
= &cfg
->achapter
;
2000 else if (depth
<= cfg
->nasect
)
2001 sl
= &cfg
->asect
[depth
-1];
2003 sl
= &cfg
->asect
[cfg
->nasect
-1];
2007 else if (sl
->just_numbers
)
2008 number
= s
->title
->kwtext2
;
2010 number
= s
->title
->kwtext
;
2013 html_words(ho
, number
, MARKUP
,
2014 thisfile
, keywords
, cfg
);
2015 html_text(ho
, sl
->number_suffix
);
2018 html_words(ho
, s
->title
->words
, real ? ALL
: MARKUP
,
2019 thisfile
, keywords
, cfg
);
2021 assert(s
->type
!= NORMAL
);
2023 * If we're printing the full document title for _real_ and
2024 * there isn't one, we don't want to print `Preamble' at
2025 * the top of what ought to just be some text. If we need
2026 * it in any other context such as TOCs, we need to print
2029 if (s
->type
== TOP
&& !real
)
2030 html_text(ho
, cfg
->preamble_text
);
2031 else if (s
->type
== INDEX
)
2032 html_text(ho
, cfg
->index_text
);