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 sect
->contents_depth
= 0;
544 html_file_section(&conf
, &files
, sect
, 0); /* peer of chapters */
545 sect
->fragment
= utoa_dup(conf
.index_text
, CS_ASCII
);
546 sect
->fragment
= html_sanitise_fragment(&files
, sect
->file
,
548 files
.index
= sect
->file
;
553 * Go through the keyword list and sort out fragment IDs for
554 * all the potentially referenced paragraphs which _aren't_
562 for (i
= 0; (kw
= index234(keywords
->keys
, i
)) != NULL
; i
++) {
563 paragraph
*q
, *p
= kw
->para
;
565 if (!is_heading_type(p
->type
)) {
569 * Find the paragraph's parent htmlsect, to
570 * determine which file it will end up in.
575 * Preamble paragraphs have no parent. So if we
576 * have a non-heading with no parent, it must
577 * be preamble, and therefore its parent
578 * htmlsect must be the preamble one.
581 sects
.head
->type
== TOP
);
584 parent
= (htmlsect
*)q
->private_data
;
587 * Now we can construct an htmlsect for this
588 * paragraph itself, taking care to put it in the
589 * list of non-sections rather than the list of
590 * sections (so that traverses of the `sects' list
591 * won't attempt to add it to the contents or
592 * anything weird like that).
594 sect
= html_new_sect(&nonsects
, p
);
595 sect
->file
= parent
->file
;
596 sect
->parent
= parent
;
597 p
->private_data
= sect
;
600 * Fragment IDs for these paragraphs will simply be
601 * `p' followed by an integer.
603 sect
->fragment
= snewn(40, char);
604 sprintf(sect
->fragment
, "p%d",
605 sect
->file
->last_fragment_number
++);
606 sect
->fragment
= html_sanitise_fragment(&files
, sect
->file
,
613 * Reset the fragment numbers in each file. I've just used them
614 * to generate `p' fragment IDs for non-section paragraphs
615 * (numbered list elements, bibliocited), and now I want to use
616 * them for `i' fragment IDs for index entries.
620 for (file
= files
.head
; file
; file
= file
->next
)
621 file
->last_fragment_number
= 0;
625 * Now sort out the index. This involves:
627 * - For each index term, we set up an htmlindex structure to
628 * store all the references to that term.
630 * - Then we make a pass over the actual document, finding
631 * every word_IndexRef; for each one, we actually figure out
632 * the HTML filename/fragment pair we will use to reference
633 * it, store that information in the private data field of
634 * the word_IndexRef itself (so we can recreate it when the
635 * time comes to output our HTML), and add a reference to it
636 * to the index term in question.
645 * Set up the htmlindex structures.
648 for (i
= 0; (entry
= index234(idx
->entries
, i
)) != NULL
; i
++) {
649 htmlindex
*hi
= snew(htmlindex
);
651 hi
->nrefs
= hi
->refsize
= 0;
654 entry
->backend_data
= hi
;
658 * Run over the document inventing fragments. Each fragment
659 * is of the form `i' followed by an integer.
661 lastsect
= sects
.head
; /* this is always the top section */
662 for (p
= sourceform
; p
; p
= p
->next
) {
663 if (is_heading_type(p
->type
) && p
->type
!= para_Title
)
664 lastsect
= (htmlsect
*)p
->private_data
;
666 for (w
= p
->words
; w
; w
= w
->next
)
667 if (w
->type
== word_IndexRef
) {
668 htmlindexref
*hr
= snew(htmlindexref
);
672 hr
->referenced
= hr
->generated
= FALSE
;
673 hr
->section
= lastsect
;
677 lastsect
->file
->last_fragment_number
++);
678 hr
->fragment
= dupstr(buf
);
680 html_sanitise_fragment(&files
, hr
->section
->file
,
683 w
->private_data
= hr
;
685 tag
= index_findtag(idx
, w
->text
);
689 for (i
= 0; i
< tag
->nrefs
; i
++) {
690 indexentry
*entry
= tag
->refs
[i
];
691 htmlindex
*hi
= (htmlindex
*)entry
->backend_data
;
693 if (hi
->nrefs
>= hi
->refsize
) {
695 hi
->refs
= sresize(hi
->refs
, hi
->refsize
, word
*);
698 hi
->refs
[hi
->nrefs
++] = w
;
705 * Now we're ready to write out the actual HTML files.
709 * - we open that file and write its header
710 * - we run down the list of sections
711 * - for each section directly contained within that file, we
712 * output the section text
713 * - for each section which is not in the file but which has a
714 * parent that is, we output a contents entry for the
715 * section if appropriate
716 * - finally, we output the file trailer and close the file.
725 for (f
= files
.head
; f
; f
= f
->next
) {
728 enum LISTTYPE
{ NOLIST
, UL
, OL
, DL
};
729 enum ITEMTYPE
{ NOITEM
, LI
, DT
, DD
};
730 struct stackelement
{
731 struct stackelement
*next
;
732 enum LISTTYPE listtype
;
733 enum ITEMTYPE itemtype
;
736 #define listname(lt) ( (lt)==UL ? "ul" : (lt)==OL ? "ol" : "dl" )
737 #define itemname(lt) ( (lt)==LI ? "li" : (lt)==DT ? "dt" : "dd" )
739 ho
.fp
= fopen(f
->filename
, "w");
740 ho
.charset
= conf
.output_charset
;
741 ho
.restrict_charset
= conf
.restrict_charset
;
742 ho
.cstate
= charset_init_state
;
743 ho
.ver
= conf
.htmlver
;
744 ho
.state
= HO_NEUTRAL
;
745 ho
.contents_level
= 0;
748 switch (conf
.htmlver
) {
750 fprintf(ho
.fp
, "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD "
751 "HTML 3.2 Final//EN\">\n");
754 fprintf(ho
.fp
, "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML"
755 " 4.01//EN\"\n\"http://www.w3.org/TR/html4/"
759 fprintf(ho
.fp
, "<!DOCTYPE HTML PUBLIC \"ISO/IEC "
760 "15445:2000//DTD HTML//EN\">\n");
762 case XHTML_1_0_TRANSITIONAL
:
763 fprintf(ho
.fp
, "<?xml version=\"1.0\" encoding=\"%s\"?>\n",
764 charset_to_mimeenc(conf
.output_charset
));
765 fprintf(ho
.fp
, "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML"
766 " 1.0 Transitional//EN\"\n\"http://www.w3.org/TR/"
767 "xhtml1/DTD/xhtml1-transitional.dtd\">\n");
769 case XHTML_1_0_STRICT
:
770 fprintf(ho
.fp
, "<?xml version=\"1.0\" encoding=\"%s\"?>\n",
771 charset_to_mimeenc(conf
.output_charset
));
772 fprintf(ho
.fp
, "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML"
773 " 1.0 Strict//EN\"\n\"http://www.w3.org/TR/xhtml1/"
774 "DTD/xhtml1-strict.dtd\">\n");
778 element_open(&ho
, "html");
779 if (is_xhtml(conf
.htmlver
)) {
780 element_attr(&ho
, "xmlns", "http://www.w3.org/1999/xhtml");
784 element_open(&ho
, "head");
787 element_empty(&ho
, "meta");
788 element_attr(&ho
, "http-equiv", "content-type");
791 sprintf(buf
, "text/html; charset=%.150s",
792 charset_to_mimeenc(conf
.output_charset
));
793 element_attr(&ho
, "content", buf
);
798 element_empty(&ho
, "meta");
799 element_attr(&ho
, "name", "author");
800 element_attr_w(&ho
, "content", conf
.author
);
804 if (conf
.description
) {
805 element_empty(&ho
, "meta");
806 element_attr(&ho
, "name", "description");
807 element_attr_w(&ho
, "content", conf
.description
);
811 element_open(&ho
, "title");
812 if (f
->first
&& f
->first
->title
) {
813 html_words(&ho
, f
->first
->title
->words
, NOTHING
,
817 if (f
->last
!= f
->first
&& f
->last
->title
) {
818 html_text(&ho
, conf
.title_separator
);
819 html_words(&ho
, f
->last
->title
->words
, NOTHING
,
823 element_close(&ho
, "title");
827 html_raw(&ho
, conf
.head_end
);
830 * Add any <head> data defined in specific sections
831 * that go in this file. (This is mostly to allow <meta
832 * name="AppleTitle"> tags for Mac online help.)
834 for (s
= sects
.head
; s
; s
= s
->next
) {
835 if (s
->file
== f
&& s
->text
) {
837 p
&& (p
== s
->text
|| p
->type
== para_Title
||
838 !is_heading_type(p
->type
));
840 if (p
->type
== para_Config
) {
841 if (!ustricmp(p
->keyword
, L
"html-local-head")) {
842 html_raw(&ho
, adv(p
->origkeyword
));
849 element_close(&ho
, "head");
853 html_raw(&ho
, conf
.body_tag
);
855 element_open(&ho
, "body");
859 html_raw(&ho
, conf
.body_start
);
862 * Write out a nav bar. Special case: we don't do this
863 * if there is only one file.
865 if (files
.head
!= files
.tail
) {
866 element_open(&ho
, "p");
868 html_raw_as_attr(&ho
, conf
.nav_attr
);
871 element_open(&ho
, "a");
872 element_attr(&ho
, "href", prevf
->filename
);
874 html_text(&ho
, conf
.nav_prev_text
);
876 element_close(&ho
, "a");
878 html_text(&ho
, conf
.nav_separator
);
880 if (f
!= files
.head
) {
881 element_open(&ho
, "a");
882 element_attr(&ho
, "href", files
.head
->filename
);
884 html_text(&ho
, conf
.contents_text
);
886 element_close(&ho
, "a");
888 html_text(&ho
, conf
.nav_separator
);
891 if (f
!= files
.index
) {
892 element_open(&ho
, "a");
893 element_attr(&ho
, "href", files
.index
->filename
);
895 html_text(&ho
, conf
.index_text
);
896 if (f
!= files
.index
)
897 element_close(&ho
, "a");
900 html_text(&ho
, conf
.nav_separator
);
903 element_open(&ho
, "a");
904 element_attr(&ho
, "href", f
->next
->filename
);
906 html_text(&ho
, conf
.nav_next_text
);
908 element_close(&ho
, "a");
910 element_close(&ho
, "p");
916 * Write out a prefix TOC for the file.
918 * We start by going through the section list and
919 * collecting the sections which need to be added to
920 * the contents. On the way, we also test to see if
921 * this file is a leaf file (defined as one which
922 * contains all descendants of any section it
923 * contains), because this will play a part in our
924 * decision on whether or not to _output_ the TOC.
926 * Special case: we absolutely do not do this if we're
927 * in single-file mode.
929 if (files
.head
!= files
.tail
) {
930 int ntoc
= 0, tocsize
= 0;
931 htmlsect
**toc
= NULL
;
934 for (s
= sects
.head
; s
; s
= s
->next
) {
939 * Search up from this section until we find
940 * the highest-level one which belongs in this
945 for (ac
= s
; ac
; ac
= ac
->parent
) {
953 if (s
->file
!= f
&& a
!= NULL
)
957 if (adepth
<= a
->contents_depth
) {
958 if (ntoc
>= tocsize
) {
960 toc
= sresize(toc
, tocsize
, htmlsect
*);
967 if (leaf
&& conf
.leaf_contains_contents
&&
968 ntoc
>= conf
.leaf_smallest_contents
) {
971 for (i
= 0; i
< ntoc
; i
++) {
972 htmlsect
*s
= toc
[i
];
973 int hlevel
= (s
->type
== TOP ?
-1 :
974 s
->type
== INDEX ?
0 :
975 heading_depth(s
->title
))
976 - f
->min_heading_depth
+ 1;
979 html_contents_entry(&ho
, hlevel
, s
,
982 html_contents_entry(&ho
, 0, NULL
, f
, keywords
, &conf
);
987 * Now go through the document and output some real
991 for (s
= sects
.head
; s
; s
= s
->next
) {
994 * This section belongs in this file.
1005 * Search up from this section until we find
1006 * the highest-level one which belongs in this
1011 for (ac
= s
; ac
; ac
= ac
->parent
) {
1012 if (ac
->file
== f
) {
1021 * This section does not belong in this
1022 * file, but an ancestor of it does. Write
1023 * out a contents table entry, if the depth
1024 * doesn't exceed the maximum contents
1025 * depth for the ancestor section.
1027 if (adepth
<= a
->contents_depth
) {
1028 html_contents_entry(&ho
, adepth
, s
,
1029 f
, keywords
, &conf
);
1038 html_contents_entry(&ho
, 0, NULL
, f
, keywords
, &conf
);
1041 * Display the section heading.
1044 hlevel
= (s
->type
== TOP ?
-1 :
1045 s
->type
== INDEX ?
0 :
1046 heading_depth(s
->title
))
1047 - f
->min_heading_depth
+ 1;
1048 assert(hlevel
>= 1);
1049 /* HTML headings only go up to <h6> */
1053 htag
[1] = '0' + hlevel
;
1055 element_open(&ho
, htag
);
1058 * Provide anchor for cross-links to target.
1060 * (Also we'll have to do this separately in
1061 * other paragraph types - NumberedList and
1065 html_fragment(&ho
, s
->fragment
);
1067 html_section_title(&ho
, s
, f
, keywords
, &conf
, TRUE
);
1069 element_close(&ho
, htag
);
1072 * Now display the section text.
1075 stackhead
= snew(struct stackelement
);
1076 stackhead
->next
= NULL
;
1077 stackhead
->listtype
= NOLIST
;
1078 stackhead
->itemtype
= NOITEM
;
1080 for (p
= s
->text
;; p
= p
->next
) {
1081 enum LISTTYPE listtype
;
1082 struct stackelement
*se
;
1085 * Preliminary switch to figure out what
1086 * sort of list we expect to be inside at
1089 * Since p may still be NULL at this point,
1090 * I invent a harmless paragraph type for
1093 switch (p ? p
->type
: para_Normal
) {
1096 case para_Copyright
:
1097 case para_BiblioCited
:
1099 case para_QuotePush
:
1103 case para_UnnumberedChapter
:
1114 case para_NumberedList
:
1118 case para_DescribedThing
:
1119 case para_Description
:
1123 case para_LcontPush
:
1124 se
= snew(struct stackelement
);
1125 se
->next
= stackhead
;
1126 se
->listtype
= NOLIST
;
1127 se
->itemtype
= NOITEM
;
1131 default: /* some totally non-printing para */
1138 * Terminate the most recent list item, if
1139 * any. (We left this until after
1140 * processing LcontPush, since in that case
1141 * the list item won't want to be
1142 * terminated until after the corresponding
1145 if (stackhead
->itemtype
!= NOITEM
) {
1146 element_close(&ho
, itemname(stackhead
->itemtype
));
1149 stackhead
->itemtype
= NOITEM
;
1152 * Terminate the current list, if it's not
1153 * the one we want to be in.
1155 if (listtype
!= stackhead
->listtype
&&
1156 stackhead
->listtype
!= NOLIST
) {
1157 element_close(&ho
, listname(stackhead
->listtype
));
1162 * Leave the loop if our time has come.
1164 if (!p
|| (is_heading_type(p
->type
) &&
1165 p
->type
!= para_Title
))
1166 break; /* end of section text */
1169 * Start a fresh list if necessary.
1171 if (listtype
!= stackhead
->listtype
&&
1173 element_open(&ho
, listname(listtype
));
1175 stackhead
->listtype
= listtype
;
1179 element_empty(&ho
, "hr");
1182 html_codepara(&ho
, p
->words
);
1185 case para_Copyright
:
1186 element_open(&ho
, "p");
1188 html_words(&ho
, p
->words
, ALL
,
1189 f
, keywords
, &conf
);
1191 element_close(&ho
, "p");
1193 case para_BiblioCited
:
1194 element_open(&ho
, "p");
1195 if (p
->private_data
) {
1196 htmlsect
*s
= (htmlsect
*)p
->private_data
;
1197 html_fragment(&ho
, s
->fragment
);
1200 html_words(&ho
, p
->kwtext
, ALL
,
1201 f
, keywords
, &conf
);
1202 html_text(&ho
, L
" ");
1203 html_words(&ho
, p
->words
, ALL
,
1204 f
, keywords
, &conf
);
1206 element_close(&ho
, "p");
1209 case para_NumberedList
:
1210 element_open(&ho
, "li");
1211 if (p
->private_data
) {
1212 htmlsect
*s
= (htmlsect
*)p
->private_data
;
1213 html_fragment(&ho
, s
->fragment
);
1216 stackhead
->itemtype
= LI
;
1217 html_words(&ho
, p
->words
, ALL
,
1218 f
, keywords
, &conf
);
1220 case para_DescribedThing
:
1221 element_open(&ho
, "dt");
1223 stackhead
->itemtype
= DT
;
1224 html_words(&ho
, p
->words
, ALL
,
1225 f
, keywords
, &conf
);
1227 case para_Description
:
1228 element_open(&ho
, "dd");
1230 stackhead
->itemtype
= DD
;
1231 html_words(&ho
, p
->words
, ALL
,
1232 f
, keywords
, &conf
);
1235 case para_QuotePush
:
1236 element_open(&ho
, "blockquote");
1239 element_close(&ho
, "blockquote");
1244 stackhead
= stackhead
->next
;
1251 assert(stackhead
&& !stackhead
->next
);
1255 if (s
->type
== INDEX
) {
1260 * This section is the index. I'll just
1261 * render it as a single paragraph, with a
1262 * colon between the index term and the
1263 * references, and <br> in between each
1266 element_open(&ho
, "p");
1268 for (i
= 0; (entry
=
1269 index234(idx
->entries
, i
)) != NULL
; i
++) {
1270 htmlindex
*hi
= (htmlindex
*)entry
->backend_data
;
1274 element_empty(&ho
, "br");
1277 html_words(&ho
, entry
->text
, MARKUP
|LINKS
,
1278 f
, keywords
, &conf
);
1280 html_text(&ho
, conf
.index_main_sep
);
1282 for (j
= 0; j
< hi
->nrefs
; j
++) {
1284 (htmlindexref
*)hi
->refs
[j
]->private_data
;
1285 paragraph
*p
= hr
->section
->title
;
1288 html_text(&ho
, conf
.index_multi_sep
);
1290 html_href(&ho
, f
, hr
->section
->file
,
1292 hr
->referenced
= TRUE
;
1294 html_words(&ho
, p
->kwtext
, MARKUP
|LINKS
,
1295 f
, keywords
, &conf
);
1296 else if (p
&& p
->words
)
1297 html_words(&ho
, p
->words
, MARKUP
|LINKS
,
1298 f
, keywords
, &conf
);
1301 * If there is no title at all,
1302 * this must be because our
1303 * target section is the
1304 * preamble section and there
1305 * is no title. So we use the
1308 html_text(&ho
, conf
.preamble_text
);
1310 element_close(&ho
, "a");
1313 element_close(&ho
, "p");
1318 html_contents_entry(&ho
, 0, NULL
, f
, keywords
, &conf
);
1325 int done_version_ids
= FALSE
;
1327 element_empty(&ho
, "hr");
1330 html_raw(&ho
, conf
.body_end
);
1332 if (conf
.address_section
) {
1333 int started
= FALSE
;
1334 if (conf
.htmlver
== ISO_HTML
) {
1336 * The ISO-HTML validator complains if
1337 * there isn't a <div> tag surrounding the
1338 * <address> tag. I'm uncertain of why this
1339 * should be - there appears to be no
1340 * mention of this in the ISO-HTML spec,
1341 * suggesting that it doesn't represent a
1342 * change from HTML 4, but nonetheless the
1343 * HTML 4 validator doesn't seem to mind.
1345 element_open(&ho
, "div");
1347 element_open(&ho
, "address");
1348 if (conf
.addr_start
) {
1349 html_raw(&ho
, conf
.addr_start
);
1353 if (conf
.visible_version_id
) {
1354 for (p
= sourceform
; p
; p
= p
->next
)
1355 if (p
->type
== para_VersionID
) {
1357 element_empty(&ho
, "br");
1359 html_text(&ho
, conf
.pre_versionid
);
1360 html_words(&ho
, p
->words
, NOTHING
,
1361 f
, keywords
, &conf
);
1362 html_text(&ho
, conf
.post_versionid
);
1365 done_version_ids
= TRUE
;
1367 if (conf
.addr_end
) {
1369 element_empty(&ho
, "br");
1370 html_raw(&ho
, conf
.addr_end
);
1372 element_close(&ho
, "address");
1373 if (conf
.htmlver
== ISO_HTML
)
1374 element_close(&ho
, "div");
1377 if (!done_version_ids
) {
1379 * If the user didn't want the version IDs
1380 * visible, I think we still have a duty to put
1381 * them in an HTML comment.
1383 int started
= FALSE
;
1384 for (p
= sourceform
; p
; p
= p
->next
)
1385 if (p
->type
== para_VersionID
) {
1387 html_raw(&ho
, "<!-- version IDs:\n");
1390 html_words(&ho
, p
->words
, NOTHING
,
1391 f
, keywords
, &conf
);
1395 html_raw(&ho
, "-->\n");
1399 element_close(&ho
, "body");
1401 element_close(&ho
, "html");
1408 * Go through and check that no index fragments were referenced
1409 * without being generated, or indeed vice versa.
1411 * (When I actually get round to freeing everything, this can
1412 * probably be the freeing loop as well.)
1414 for (p
= sourceform
; p
; p
= p
->next
) {
1416 for (w
= p
->words
; w
; w
= w
->next
)
1417 if (w
->type
== word_IndexRef
) {
1418 htmlindexref
*hr
= (htmlindexref
*)w
->private_data
;
1420 assert(!hr
->referenced
== !hr
->generated
);
1425 * Free all the working data.
1428 sfree(conf
.single_filename
);
1429 sfree(conf
.contents_filename
);
1430 sfree(conf
.index_filename
);
1431 sfree(conf
.template_filename
);
1432 sfree(conf
.template_fragment
);
1435 while ( (frag
= (htmlfragment
*)delpos234(files
.frags
, 0)) != NULL
) {
1437 * frag->fragment is dynamically allocated, but will be
1438 * freed when we process the htmlsect structure which
1439 * it is attached to.
1443 freetree234(files
.frags
);
1446 htmlsect
*sect
, *tmp
;
1450 sfree(sect
->fragment
);
1454 sect
= nonsects
.head
;
1457 sfree(sect
->fragment
);
1463 htmlfile
*file
, *tmp
;
1467 sfree(file
->filename
);
1475 for (i
= 0; (entry
= index234(idx
->entries
, i
)) != NULL
; i
++) {
1476 htmlindex
*hi
= (htmlindex
*)entry
->backend_data
;
1483 for (p
= sourceform
; p
; p
= p
->next
)
1484 for (w
= p
->words
; w
; w
= w
->next
)
1485 if (w
->type
== word_IndexRef
) {
1486 htmlindexref
*hr
= (htmlindexref
*)w
->private_data
;
1488 sfree(hr
->fragment
);
1494 static void html_file_section(htmlconfig
*cfg
, htmlfilelist
*files
,
1495 htmlsect
*sect
, int depth
)
1501 * `depth' is derived from the heading_depth() macro at the top
1502 * of this file, which counts title as -1, chapter as 0,
1503 * heading as 1 and subsection as 2. However, the semantics of
1504 * cfg->leaf_level are defined to count chapter as 1, heading
1505 * as 2 etc. So first I increment depth :-(
1509 if (cfg
->leaf_level
== 0) {
1511 * leaf_level==0 is a special case, in which everything is
1512 * put into a single file.
1515 files
->single
= html_new_file(files
, cfg
->single_filename
);
1517 file
= files
->single
;
1520 * If the depth of this section is at or above leaf_level,
1521 * we invent a fresh file and put this section at its head.
1522 * Otherwise, we put it in the same file as its parent
1525 if (ldepth
> cfg
->leaf_level
) {
1527 * We know that sect->parent cannot be NULL. The only
1528 * circumstance in which it can be is if sect is at
1529 * chapter or appendix level, i.e. ldepth==1; and if
1530 * that's the case, then we cannot have entered this
1531 * branch unless cfg->leaf_level==0, in which case we
1532 * would be in the single-file case above and not here
1535 assert(sect
->parent
);
1537 file
= sect
->parent
->file
;
1539 if (sect
->type
== TOP
) {
1540 file
= html_new_file(files
, cfg
->contents_filename
);
1541 } else if (sect
->type
== INDEX
) {
1542 file
= html_new_file(files
, cfg
->index_filename
);
1546 assert(ldepth
> 0 && sect
->title
);
1547 title
= html_format(sect
->title
, cfg
->template_filename
);
1548 file
= html_new_file(files
, title
);
1556 if (file
->min_heading_depth
> depth
) {
1558 * This heading is at a higher level than any heading we
1559 * have so far placed in this file; so we set the `first'
1562 file
->min_heading_depth
= depth
;
1566 if (file
->min_heading_depth
== depth
)
1570 static htmlfile
*html_new_file(htmlfilelist
*list
, char *filename
)
1572 htmlfile
*ret
= snew(htmlfile
);
1576 list
->tail
->next
= ret
;
1581 ret
->filename
= dupstr(filename
);
1582 ret
->last_fragment_number
= 0;
1583 ret
->min_heading_depth
= INT_MAX
;
1584 ret
->first
= ret
->last
= NULL
;
1589 static htmlsect
*html_new_sect(htmlsectlist
*list
, paragraph
*title
)
1591 htmlsect
*ret
= snew(htmlsect
);
1595 list
->tail
->next
= ret
;
1608 static void html_words(htmloutput
*ho
, word
*words
, int flags
,
1609 htmlfile
*file
, keywordlist
*keywords
, htmlconfig
*cfg
)
1615 for (w
= words
; w
; w
= w
->next
) switch (w
->type
) {
1616 case word_HyperLink
:
1617 if (flags
& LINKS
) {
1618 element_open(ho
, "a");
1619 c
= utoa_dup(w
->text
, CS_ASCII
);
1620 element_attr(ho
, "href", c
);
1624 case word_UpperXref
:
1625 case word_LowerXref
:
1626 if (flags
& LINKS
) {
1627 keyword
*kwl
= kw_lookup(keywords
, w
->text
);
1628 paragraph
*p
= kwl
->para
;
1629 htmlsect
*s
= (htmlsect
*)p
->private_data
;
1633 html_href(ho
, file
, s
->file
, s
->fragment
);
1639 element_close(ho
, "a");
1642 if (flags
& INDEXENTS
) {
1643 htmlindexref
*hr
= (htmlindexref
*)w
->private_data
;
1644 html_fragment(ho
, hr
->fragment
);
1645 hr
->generated
= TRUE
;
1652 case word_WhiteSpace
:
1653 case word_EmphSpace
:
1654 case word_CodeSpace
:
1655 case word_WkCodeSpace
:
1657 case word_EmphQuote
:
1658 case word_CodeQuote
:
1659 case word_WkCodeQuote
:
1660 style
= towordstyle(w
->type
);
1661 type
= removeattr(w
->type
);
1662 if (style
== word_Emph
&&
1663 (attraux(w
->aux
) == attr_First
||
1664 attraux(w
->aux
) == attr_Only
) &&
1666 element_open(ho
, "em");
1667 else if ((style
== word_Code
|| style
== word_WeakCode
) &&
1668 (attraux(w
->aux
) == attr_First
||
1669 attraux(w
->aux
) == attr_Only
) &&
1671 element_open(ho
, "code");
1673 if (type
== word_WhiteSpace
)
1674 html_text(ho
, L
" ");
1675 else if (type
== word_Quote
) {
1676 if (quoteaux(w
->aux
) == quote_Open
)
1677 html_text(ho
, cfg
->lquote
);
1679 html_text(ho
, cfg
->rquote
);
1681 if (!w
->alt
|| cvt_ok(ho
->restrict_charset
, w
->text
))
1682 html_text_nbsp(ho
, w
->text
);
1684 html_words(ho
, w
->alt
, flags
, file
, keywords
, cfg
);
1687 if (style
== word_Emph
&&
1688 (attraux(w
->aux
) == attr_Last
||
1689 attraux(w
->aux
) == attr_Only
) &&
1691 element_close(ho
, "em");
1692 else if ((style
== word_Code
|| style
== word_WeakCode
) &&
1693 (attraux(w
->aux
) == attr_Last
||
1694 attraux(w
->aux
) == attr_Only
) &&
1696 element_close(ho
, "code");
1702 static void html_codepara(htmloutput
*ho
, word
*words
)
1704 element_open(ho
, "pre");
1705 element_open(ho
, "code");
1706 for (; words
; words
= words
->next
) if (words
->type
== word_WeakCode
) {
1711 if (words
->next
&& words
->next
->type
== word_Emph
) {
1712 e
= words
->next
->text
;
1713 words
= words
->next
;
1717 while (e
&& *e
&& *t
) {
1721 for (n
= 0; t
[n
] && e
[n
] && e
[n
] == ec
; n
++);
1729 element_open(ho
, open_tag
);
1731 html_text_limit(ho
, t
, n
);
1734 element_close(ho
, open_tag
);
1742 element_close(ho
, "code");
1743 element_close(ho
, "pre");
1746 static void html_charset_cleanup(htmloutput
*ho
)
1751 bytes
= charset_from_unicode(NULL
, NULL
, outbuf
, lenof(outbuf
),
1752 ho
->charset
, &ho
->cstate
, NULL
);
1754 fwrite(outbuf
, 1, bytes
, ho
->fp
);
1757 static void return_mostly_to_neutral(htmloutput
*ho
)
1759 if (ho
->state
== HO_IN_EMPTY_TAG
&& is_xhtml(ho
->ver
)) {
1760 fprintf(ho
->fp
, " />");
1761 } else if (ho
->state
== HO_IN_EMPTY_TAG
|| ho
->state
== HO_IN_TAG
) {
1762 fprintf(ho
->fp
, ">");
1765 ho
->state
= HO_NEUTRAL
;
1768 static void return_to_neutral(htmloutput
*ho
)
1770 if (ho
->state
== HO_IN_TEXT
) {
1771 html_charset_cleanup(ho
);
1774 return_mostly_to_neutral(ho
);
1777 static void element_open(htmloutput
*ho
, char const *name
)
1779 return_to_neutral(ho
);
1780 fprintf(ho
->fp
, "<%s", name
);
1781 ho
->state
= HO_IN_TAG
;
1784 static void element_close(htmloutput
*ho
, char const *name
)
1786 return_to_neutral(ho
);
1787 fprintf(ho
->fp
, "</%s>", name
);
1788 ho
->state
= HO_NEUTRAL
;
1791 static void element_empty(htmloutput
*ho
, char const *name
)
1793 return_to_neutral(ho
);
1794 fprintf(ho
->fp
, "<%s", name
);
1795 ho
->state
= HO_IN_EMPTY_TAG
;
1798 static void html_nl(htmloutput
*ho
)
1800 return_to_neutral(ho
);
1801 fputc('\n', ho
->fp
);
1804 static void html_raw(htmloutput
*ho
, char *text
)
1806 return_to_neutral(ho
);
1807 fputs(text
, ho
->fp
);
1810 static void html_raw_as_attr(htmloutput
*ho
, char *text
)
1812 assert(ho
->state
== HO_IN_TAG
|| ho
->state
== HO_IN_EMPTY_TAG
);
1814 fputs(text
, ho
->fp
);
1817 static void element_attr(htmloutput
*ho
, char const *name
, char const *value
)
1819 html_charset_cleanup(ho
);
1820 assert(ho
->state
== HO_IN_TAG
|| ho
->state
== HO_IN_EMPTY_TAG
);
1821 fprintf(ho
->fp
, " %s=\"%s\"", name
, value
);
1824 static void element_attr_w(htmloutput
*ho
, char const *name
,
1825 wchar_t const *value
)
1827 html_charset_cleanup(ho
);
1828 fprintf(ho
->fp
, " %s=\"", name
);
1829 html_text_limit_internal(ho
, value
, 0, TRUE
, FALSE
);
1830 html_charset_cleanup(ho
);
1834 static void html_text(htmloutput
*ho
, wchar_t const *text
)
1836 return_mostly_to_neutral(ho
);
1837 html_text_limit_internal(ho
, text
, 0, FALSE
, FALSE
);
1840 static void html_text_nbsp(htmloutput
*ho
, wchar_t const *text
)
1842 return_mostly_to_neutral(ho
);
1843 html_text_limit_internal(ho
, text
, 0, FALSE
, TRUE
);
1846 static void html_text_limit(htmloutput
*ho
, wchar_t const *text
, int maxlen
)
1848 return_mostly_to_neutral(ho
);
1849 html_text_limit_internal(ho
, text
, maxlen
, FALSE
, FALSE
);
1852 static void html_text_limit_internal(htmloutput
*ho
, wchar_t const *text
,
1853 int maxlen
, int quote_quotes
, int nbsp
)
1855 int textlen
= ustrlen(text
);
1859 if (maxlen
> 0 && textlen
> maxlen
)
1862 while (textlen
> 0) {
1863 /* Scan ahead for characters we really can't display in HTML. */
1864 int lenbefore
, lenafter
;
1865 for (lenbefore
= 0; lenbefore
< textlen
; lenbefore
++)
1866 if (text
[lenbefore
] == L
'<' ||
1867 text
[lenbefore
] == L
'>' ||
1868 text
[lenbefore
] == L
'&' ||
1869 (text
[lenbefore
] == L
'"' && quote_quotes
) ||
1870 (text
[lenbefore
] == L
' ' && nbsp
))
1872 lenafter
= lenbefore
;
1873 bytes
= charset_from_unicode(&text
, &lenafter
, outbuf
, lenof(outbuf
),
1874 ho
->charset
, &ho
->cstate
, &err
);
1875 textlen
-= (lenbefore
- lenafter
);
1877 fwrite(outbuf
, 1, bytes
, ho
->fp
);
1880 * We have encountered a character that cannot be
1881 * displayed in the selected output charset. Therefore,
1882 * we use an HTML numeric entity reference.
1884 assert(textlen
> 0);
1885 fprintf(ho
->fp
, "&#%ld;", (long int)*text
);
1887 } else if (lenafter
== 0 && textlen
> 0) {
1889 * We have encountered a character which is special to
1893 fprintf(ho
->fp
, "<");
1894 else if (*text
== L
'>')
1895 fprintf(ho
->fp
, ">");
1896 else if (*text
== L
'&')
1897 fprintf(ho
->fp
, "&");
1898 else if (*text
== L
'"')
1899 fprintf(ho
->fp
, """);
1900 else if (*text
== L
' ') {
1902 fprintf(ho
->fp
, " ");
1904 assert(!"Can't happen");
1910 static void cleanup(htmloutput
*ho
)
1912 return_to_neutral(ho
);
1916 static void html_href(htmloutput
*ho
, htmlfile
*thisfile
,
1917 htmlfile
*targetfile
, char *targetfrag
)
1919 rdstringc rs
= { 0, 0, NULL
};
1922 if (targetfile
!= thisfile
)
1923 rdaddsc(&rs
, targetfile
->filename
);
1926 rdaddsc(&rs
, targetfrag
);
1930 element_open(ho
, "a");
1931 element_attr(ho
, "href", url
);
1935 static void html_fragment(htmloutput
*ho
, char const *fragment
)
1937 element_open(ho
, "a");
1938 element_attr(ho
, "name", fragment
);
1939 if (is_xhtml(ho
->ver
))
1940 element_attr(ho
, "id", fragment
);
1941 element_close(ho
, "a");
1944 static char *html_format(paragraph
*p
, char *template_string
)
1948 wchar_t *ws
, wsbuf
[2];
1949 rdstringc rs
= { 0, 0, NULL
};
1951 t
= template_string
;
1953 if (*t
== '%' && t
[1]) {
1967 if (p
->kwtext
&& fmt
== 'n')
1969 else if (p
->kwtext2
&& fmt
== 'b') {
1971 * HTML fragment names must start with a letter, so
1972 * simply `1.2.3' is not adequate. In this case I'm
1973 * going to cheat slightly by prepending the first
1974 * character of the first word of kwtext, so that
1975 * we get `C1' for chapter 1, `S2.3' for section
1978 if (p
->kwtext
&& p
->kwtext
->text
[0]) {
1981 wsbuf
[0] = p
->kwtext
->text
[0];
1984 } else if (p
->keyword
&& *p
->keyword
&& fmt
== 'k')
1990 c
= utoa_dup(ws
, CS_ASCII
);
1996 if (removeattr(w
->type
) == word_Normal
) {
1997 c
= utoa_dup(w
->text
, CS_ASCII
);
2008 return rdtrimc(&rs
);
2011 static char *html_sanitise_fragment(htmlfilelist
*files
, htmlfile
*file
,
2015 * The HTML 4 spec's strictest definition of fragment names (<a
2016 * name> and "id" attributes) says that they `must begin with a
2017 * letter and may be followed by any number of letters, digits,
2018 * hyphens, underscores, colons, and periods'.
2020 * So here we unceremoniously rip out any characters not
2021 * conforming to this limitation.
2023 char *p
= text
, *q
= text
;
2025 while (*p
&& !((*p
>='A' && *p
<='Z') || (*p
>='a' && *p
<='z')))
2027 if ((*q
++ = *p
++) != '\0') {
2029 if ((*p
>='A' && *p
<='Z') ||
2030 (*p
>='a' && *p
<='z') ||
2031 (*p
>='0' && *p
<='9') ||
2032 *p
=='-' || *p
=='_' || *p
==':' || *p
=='.')
2041 * Now we check for clashes with other fragment names, and
2042 * adjust this one if necessary by appending a hyphen followed
2046 htmlfragment
*frag
= snew(htmlfragment
);
2047 int len
= 0; /* >0 indicates we have resized */
2051 frag
->fragment
= text
;
2053 while (add234(files
->frags
, frag
) != frag
) {
2056 frag
->fragment
= text
= sresize(text
, len
+20, char);
2059 sprintf(text
+ len
, "-%d", ++suffix
);
2066 static void html_contents_entry(htmloutput
*ho
, int depth
, htmlsect
*s
,
2067 htmlfile
*thisfile
, keywordlist
*keywords
,
2070 while (ho
->contents_level
> depth
) {
2071 element_close(ho
, "ul");
2072 ho
->contents_level
--;
2075 while (ho
->contents_level
< depth
) {
2076 element_open(ho
, "ul");
2077 ho
->contents_level
++;
2083 element_open(ho
, "li");
2084 html_href(ho
, thisfile
, s
->file
, s
->fragment
);
2085 html_section_title(ho
, s
, thisfile
, keywords
, cfg
, FALSE
);
2086 element_close(ho
, "a");
2087 element_close(ho
, "li");
2090 static void html_section_title(htmloutput
*ho
, htmlsect
*s
, htmlfile
*thisfile
,
2091 keywordlist
*keywords
, htmlconfig
*cfg
,
2097 int depth
= heading_depth(s
->title
);
2101 else if (depth
== 0)
2102 sl
= &cfg
->achapter
;
2103 else if (depth
<= cfg
->nasect
)
2104 sl
= &cfg
->asect
[depth
-1];
2106 sl
= &cfg
->asect
[cfg
->nasect
-1];
2110 else if (sl
->just_numbers
)
2111 number
= s
->title
->kwtext2
;
2113 number
= s
->title
->kwtext
;
2116 html_words(ho
, number
, MARKUP
,
2117 thisfile
, keywords
, cfg
);
2118 html_text(ho
, sl
->number_suffix
);
2121 html_words(ho
, s
->title
->words
, real ? ALL
: MARKUP
,
2122 thisfile
, keywords
, cfg
);
2124 assert(s
->type
!= NORMAL
);
2126 * If we're printing the full document title for _real_ and
2127 * there isn't one, we don't want to print `Preamble' at
2128 * the top of what ought to just be some text. If we need
2129 * it in any other context such as TOCs, we need to print
2132 if (s
->type
== TOP
&& !real
)
2133 html_text(ho
, cfg
->preamble_text
);
2134 else if (s
->type
== INDEX
)
2135 html_text(ho
, cfg
->index_text
);