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 char *csname
= utoa_dup(uadv(k
), CS_ASCII
);
271 ret
.restrict_charset
= charset_from_localenc(csname
);
273 } else if (!ustricmp(k
, L
"html-output-charset")) {
274 char *csname
= utoa_dup(uadv(k
), CS_ASCII
);
275 ret
.output_charset
= charset_from_localenc(csname
);
277 } else if (!ustricmp(k
, L
"html-version")) {
278 wchar_t *vername
= uadv(k
);
279 static const struct {
283 {L
"html3.2", HTML_3_2
},
285 {L
"iso-html", ISO_HTML
},
286 {L
"xhtml1.0transitional", XHTML_1_0_TRANSITIONAL
},
287 {L
"xhtml1.0strict", XHTML_1_0_STRICT
}
291 for (i
= 0; i
< (int)lenof(versions
); i
++)
292 if (!ustricmp(versions
[i
].name
, vername
))
295 if (i
== lenof(versions
))
296 error(err_htmlver
, &p
->fpos
, vername
);
298 ret
.htmlver
= versions
[i
].ver
;
299 } else if (!ustricmp(k
, L
"html-single-filename")) {
300 sfree(ret
.single_filename
);
301 ret
.single_filename
= dupstr(adv(p
->origkeyword
));
302 } else if (!ustricmp(k
, L
"html-contents-filename")) {
303 sfree(ret
.contents_filename
);
304 ret
.contents_filename
= dupstr(adv(p
->origkeyword
));
305 } else if (!ustricmp(k
, L
"html-index-filename")) {
306 sfree(ret
.index_filename
);
307 ret
.index_filename
= dupstr(adv(p
->origkeyword
));
308 } else if (!ustricmp(k
, L
"html-template-filename")) {
309 sfree(ret
.template_filename
);
310 ret
.template_filename
= dupstr(adv(p
->origkeyword
));
311 } else if (!ustricmp(k
, L
"html-template-fragment")) {
312 sfree(ret
.template_fragment
);
313 ret
.template_fragment
= dupstr(adv(p
->origkeyword
));
314 } else if (!ustricmp(k
, L
"html-chapter-numeric")) {
315 ret
.achapter
.just_numbers
= utob(uadv(k
));
316 } else if (!ustricmp(k
, L
"html-chapter-suffix")) {
317 ret
.achapter
.number_suffix
= uadv(k
);
318 } else if (!ustricmp(k
, L
"html-leaf-level")) {
319 ret
.leaf_level
= utoi(uadv(k
));
320 } else if (!ustricmp(k
, L
"html-section-numeric")) {
321 wchar_t *q
= uadv(k
);
327 if (n
>= ret
.nasect
) {
329 ret
.asect
= sresize(ret
.asect
, n
+1, sectlevel
);
330 for (i
= ret
.nasect
; i
<= n
; i
++)
331 ret
.asect
[i
] = ret
.asect
[ret
.nasect
-1];
334 ret
.asect
[n
].just_numbers
= utob(q
);
335 } else if (!ustricmp(k
, L
"html-section-suffix")) {
336 wchar_t *q
= uadv(k
);
342 if (n
>= ret
.nasect
) {
344 ret
.asect
= sresize(ret
.asect
, n
+1, sectlevel
);
345 for (i
= ret
.nasect
; i
<= n
; i
++) {
346 ret
.asect
[i
] = ret
.asect
[ret
.nasect
-1];
350 ret
.asect
[n
].number_suffix
= q
;
351 } else if (!ustricmp(k
, L
"html-contents-depth") ||
352 !ustrnicmp(k
, L
"html-contents-depth-", 20)) {
354 * Relic of old implementation: this directive used
355 * to be written as \cfg{html-contents-depth-3}{2}
356 * rather than the usual Halibut convention of
357 * \cfg{html-contents-depth}{3}{2}. We therefore
360 wchar_t *q
= k
[19] ? k
+20 : uadv(k
);
366 if (n
>= ret
.ncdepths
) {
368 ret
.contents_depths
=
369 sresize(ret
.contents_depths
, n
+1, int);
370 for (i
= ret
.ncdepths
; i
<= n
; i
++) {
371 ret
.contents_depths
[i
] = i
+2;
375 ret
.contents_depths
[n
] = utoi(q
);
376 } else if (!ustricmp(k
, L
"html-head-end")) {
377 ret
.head_end
= adv(p
->origkeyword
);
378 } else if (!ustricmp(k
, L
"html-body-tag")) {
379 ret
.body_tag
= adv(p
->origkeyword
);
380 } else if (!ustricmp(k
, L
"html-body-start")) {
381 ret
.body_start
= adv(p
->origkeyword
);
382 } else if (!ustricmp(k
, L
"html-body-end")) {
383 ret
.body_end
= adv(p
->origkeyword
);
384 } else if (!ustricmp(k
, L
"html-address-start")) {
385 ret
.addr_start
= adv(p
->origkeyword
);
386 } else if (!ustricmp(k
, L
"html-address-end")) {
387 ret
.addr_end
= adv(p
->origkeyword
);
388 } else if (!ustricmp(k
, L
"html-navigation-attributes")) {
389 ret
.nav_attr
= adv(p
->origkeyword
);
390 } else if (!ustricmp(k
, L
"html-author")) {
391 ret
.author
= uadv(k
);
392 } else if (!ustricmp(k
, L
"html-description")) {
393 ret
.description
= uadv(k
);
394 } else if (!ustricmp(k
, L
"html-suppress-address")) {
395 ret
.address_section
= !utob(uadv(k
));
396 } else if (!ustricmp(k
, L
"html-versionid")) {
397 ret
.visible_version_id
= utob(uadv(k
));
398 } else if (!ustricmp(k
, L
"html-quotes")) {
399 if (*uadv(k
) && *uadv(uadv(k
))) {
400 ret
.lquote
= uadv(k
);
401 ret
.rquote
= uadv(ret
.lquote
);
403 } else if (!ustricmp(k
, L
"html-leaf-contains-contents")) {
404 ret
.leaf_contains_contents
= utob(uadv(k
));
405 } else if (!ustricmp(k
, L
"html-leaf-smallest-contents")) {
406 ret
.leaf_smallest_contents
= utoi(uadv(k
));
407 } else if (!ustricmp(k
, L
"html-index-text")) {
408 ret
.index_text
= uadv(k
);
409 } else if (!ustricmp(k
, L
"html-contents-text")) {
410 ret
.contents_text
= uadv(k
);
411 } else if (!ustricmp(k
, L
"html-preamble-text")) {
412 ret
.preamble_text
= uadv(k
);
413 } else if (!ustricmp(k
, L
"html-title-separator")) {
414 ret
.title_separator
= uadv(k
);
415 } else if (!ustricmp(k
, L
"html-nav-prev-text")) {
416 ret
.nav_prev_text
= uadv(k
);
417 } else if (!ustricmp(k
, L
"html-nav-next-text")) {
418 ret
.nav_next_text
= uadv(k
);
419 } else if (!ustricmp(k
, L
"html-nav-separator")) {
420 ret
.nav_separator
= uadv(k
);
421 } else if (!ustricmp(k
, L
"html-index-main-separator")) {
422 ret
.index_main_sep
= uadv(k
);
423 } else if (!ustricmp(k
, L
"html-index-multiple-separator")) {
424 ret
.index_multi_sep
= uadv(k
);
425 } else if (!ustricmp(k
, L
"html-pre-versionid")) {
426 ret
.pre_versionid
= uadv(k
);
427 } else if (!ustricmp(k
, L
"html-post-versionid")) {
428 ret
.post_versionid
= uadv(k
);
434 * Now process fallbacks on quote characters.
436 while (*uadv(ret
.rquote
) && *uadv(uadv(ret
.rquote
)) &&
437 (!cvt_ok(ret
.restrict_charset
, ret
.lquote
) ||
438 !cvt_ok(ret
.restrict_charset
, ret
.rquote
))) {
439 ret
.lquote
= uadv(ret
.rquote
);
440 ret
.rquote
= uadv(ret
.lquote
);
446 paragraph
*html_config_filename(char *filename
)
449 * If the user passes in a single filename as a parameter to
450 * the `--html' command-line option, then we should assume it
451 * to imply _two_ config directives:
452 * \cfg{html-single-filename}{whatever} and
453 * \cfg{html-leaf-level}{0}; the rationale being that the user
454 * wants their output _in that file_.
458 p
= cmdline_cfg_simple("html-single-filename", filename
, NULL
);
459 q
= cmdline_cfg_simple("html-leaf-level", "0", NULL
);
464 void html_backend(paragraph
*sourceform
, keywordlist
*keywords
,
465 indexdata
*idx
, void *unused
)
469 htmlfilelist files
= { NULL
, NULL
, NULL
, NULL
, NULL
};
470 htmlsectlist sects
= { NULL
, NULL
}, nonsects
= { NULL
, NULL
};
474 conf
= html_configure(sourceform
);
477 * We're going to make heavy use of paragraphs' private data
478 * fields in the forthcoming code. Clear them first, so we can
479 * reliably tell whether we have auxiliary data for a
480 * particular paragraph.
482 for (p
= sourceform
; p
; p
= p
->next
)
483 p
->private_data
= NULL
;
485 files
.frags
= newtree234(html_fragment_compare
);
488 * Start by figuring out into which file each piece of the
489 * document should be put. We'll do this by inventing an
490 * `htmlsect' structure and stashing it in the private_data
491 * field of each section paragraph; we also need one additional
492 * htmlsect for the document index, which won't show up in the
493 * source form but needs to be consistently mentioned in
496 * While we're here, we'll also invent the HTML fragment name
500 htmlsect
*topsect
, *sect
;
503 topsect
= html_new_sect(§s
, NULL
);
505 topsect
->title
= NULL
;
506 topsect
->text
= sourceform
;
507 topsect
->contents_depth
= contents_depth(conf
, 0);
508 html_file_section(&conf
, &files
, topsect
, -1);
509 topsect
->fragment
= NULL
;
511 for (p
= sourceform
; p
; p
= p
->next
)
512 if (is_heading_type(p
->type
)) {
513 d
= heading_depth(p
);
515 if (p
->type
== para_Title
) {
520 sect
= html_new_sect(§s
, p
);
521 sect
->text
= p
->next
;
523 sect
->contents_depth
= contents_depth(conf
, d
+1) - (d
+1);
526 sect
->parent
= (htmlsect
*)p
->parent
->private_data
;
527 assert(sect
->parent
!= NULL
);
529 sect
->parent
= topsect
;
530 p
->private_data
= sect
;
532 html_file_section(&conf
, &files
, sect
, d
);
534 sect
->fragment
= html_format(p
, conf
.template_fragment
);
535 sect
->fragment
= html_sanitise_fragment(&files
, sect
->file
,
540 sect
= html_new_sect(§s
, NULL
);
543 sect
->parent
= topsect
;
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
;
552 * Go through the keyword list and sort out fragment IDs for
553 * all the potentially referenced paragraphs which _aren't_
561 for (i
= 0; (kw
= index234(keywords
->keys
, i
)) != NULL
; i
++) {
562 paragraph
*q
, *p
= kw
->para
;
564 if (!is_heading_type(p
->type
)) {
568 * Find the paragraph's parent htmlsect, to
569 * determine which file it will end up in.
574 * Preamble paragraphs have no parent. So if we
575 * have a non-heading with no parent, it must
576 * be preamble, and therefore its parent
577 * htmlsect must be the preamble one.
580 sects
.head
->type
== TOP
);
583 parent
= (htmlsect
*)q
->private_data
;
586 * Now we can construct an htmlsect for this
587 * paragraph itself, taking care to put it in the
588 * list of non-sections rather than the list of
589 * sections (so that traverses of the `sects' list
590 * won't attempt to add it to the contents or
591 * anything weird like that).
593 sect
= html_new_sect(&nonsects
, p
);
594 sect
->file
= parent
->file
;
595 sect
->parent
= parent
;
596 p
->private_data
= sect
;
599 * Fragment IDs for these paragraphs will simply be
600 * `p' followed by an integer.
602 sect
->fragment
= snewn(40, char);
603 sprintf(sect
->fragment
, "p%d",
604 sect
->file
->last_fragment_number
++);
605 sect
->fragment
= html_sanitise_fragment(&files
, sect
->file
,
612 * Reset the fragment numbers in each file. I've just used them
613 * to generate `p' fragment IDs for non-section paragraphs
614 * (numbered list elements, bibliocited), and now I want to use
615 * them for `i' fragment IDs for index entries.
619 for (file
= files
.head
; file
; file
= file
->next
)
620 file
->last_fragment_number
= 0;
624 * Now sort out the index. This involves:
626 * - For each index term, we set up an htmlindex structure to
627 * store all the references to that term.
629 * - Then we make a pass over the actual document, finding
630 * every word_IndexRef; for each one, we actually figure out
631 * the HTML filename/fragment pair we will use to reference
632 * it, store that information in the private data field of
633 * the word_IndexRef itself (so we can recreate it when the
634 * time comes to output our HTML), and add a reference to it
635 * to the index term in question.
644 * Set up the htmlindex structures.
647 for (i
= 0; (entry
= index234(idx
->entries
, i
)) != NULL
; i
++) {
648 htmlindex
*hi
= snew(htmlindex
);
650 hi
->nrefs
= hi
->refsize
= 0;
653 entry
->backend_data
= hi
;
657 * Run over the document inventing fragments. Each fragment
658 * is of the form `i' followed by an integer.
660 lastsect
= sects
.head
; /* this is always the top section */
661 for (p
= sourceform
; p
; p
= p
->next
) {
662 if (is_heading_type(p
->type
) && p
->type
!= para_Title
)
663 lastsect
= (htmlsect
*)p
->private_data
;
665 for (w
= p
->words
; w
; w
= w
->next
)
666 if (w
->type
== word_IndexRef
) {
667 htmlindexref
*hr
= snew(htmlindexref
);
671 hr
->referenced
= hr
->generated
= FALSE
;
672 hr
->section
= lastsect
;
676 lastsect
->file
->last_fragment_number
++);
677 hr
->fragment
= dupstr(buf
);
679 html_sanitise_fragment(&files
, hr
->section
->file
,
682 w
->private_data
= hr
;
684 tag
= index_findtag(idx
, w
->text
);
688 for (i
= 0; i
< tag
->nrefs
; i
++) {
689 indexentry
*entry
= tag
->refs
[i
];
690 htmlindex
*hi
= (htmlindex
*)entry
->backend_data
;
692 if (hi
->nrefs
>= hi
->refsize
) {
694 hi
->refs
= sresize(hi
->refs
, hi
->refsize
, word
*);
697 hi
->refs
[hi
->nrefs
++] = w
;
704 * Now we're ready to write out the actual HTML files.
708 * - we open that file and write its header
709 * - we run down the list of sections
710 * - for each section directly contained within that file, we
711 * output the section text
712 * - for each section which is not in the file but which has a
713 * parent that is, we output a contents entry for the
714 * section if appropriate
715 * - finally, we output the file trailer and close the file.
724 for (f
= files
.head
; f
; f
= f
->next
) {
727 enum LISTTYPE
{ NOLIST
, UL
, OL
, DL
};
728 enum ITEMTYPE
{ NOITEM
, LI
, DT
, DD
};
729 struct stackelement
{
730 struct stackelement
*next
;
731 enum LISTTYPE listtype
;
732 enum ITEMTYPE itemtype
;
735 #define listname(lt) ( (lt)==UL ? "ul" : (lt)==OL ? "ol" : "dl" )
736 #define itemname(lt) ( (lt)==LI ? "li" : (lt)==DT ? "dt" : "dd" )
738 ho
.fp
= fopen(f
->filename
, "w");
739 ho
.charset
= conf
.output_charset
;
740 ho
.restrict_charset
= conf
.restrict_charset
;
741 ho
.cstate
= charset_init_state
;
742 ho
.ver
= conf
.htmlver
;
743 ho
.state
= HO_NEUTRAL
;
744 ho
.contents_level
= 0;
747 switch (conf
.htmlver
) {
749 fprintf(ho
.fp
, "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD "
750 "HTML 3.2 Final//EN\">\n");
753 fprintf(ho
.fp
, "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML"
754 " 4.01//EN\"\n\"http://www.w3.org/TR/html4/"
758 fprintf(ho
.fp
, "<!DOCTYPE HTML PUBLIC \"ISO/IEC "
759 "15445:2000//DTD HTML//EN\">\n");
761 case XHTML_1_0_TRANSITIONAL
:
762 fprintf(ho
.fp
, "<?xml version=\"1.0\" encoding=\"%s\"?>\n",
763 charset_to_mimeenc(conf
.output_charset
));
764 fprintf(ho
.fp
, "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML"
765 " 1.0 Transitional//EN\"\n\"http://www.w3.org/TR/"
766 "xhtml1/DTD/xhtml1-transitional.dtd\">\n");
768 case XHTML_1_0_STRICT
:
769 fprintf(ho
.fp
, "<?xml version=\"1.0\" encoding=\"%s\"?>\n",
770 charset_to_mimeenc(conf
.output_charset
));
771 fprintf(ho
.fp
, "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML"
772 " 1.0 Strict//EN\"\n\"http://www.w3.org/TR/xhtml1/"
773 "DTD/xhtml1-strict.dtd\">\n");
777 element_open(&ho
, "html");
778 if (is_xhtml(conf
.htmlver
)) {
779 element_attr(&ho
, "xmlns", "http://www.w3.org/1999/xhtml");
783 element_open(&ho
, "head");
786 element_empty(&ho
, "meta");
787 element_attr(&ho
, "http-equiv", "content-type");
790 sprintf(buf
, "text/html; charset=%.150s",
791 charset_to_mimeenc(conf
.output_charset
));
792 element_attr(&ho
, "content", buf
);
797 element_empty(&ho
, "meta");
798 element_attr(&ho
, "name", "author");
799 element_attr_w(&ho
, "content", conf
.author
);
803 if (conf
.description
) {
804 element_empty(&ho
, "meta");
805 element_attr(&ho
, "name", "description");
806 element_attr_w(&ho
, "content", conf
.description
);
810 element_open(&ho
, "title");
811 if (f
->first
&& f
->first
->title
) {
812 html_words(&ho
, f
->first
->title
->words
, NOTHING
,
816 if (f
->last
!= f
->first
&& f
->last
->title
) {
817 html_text(&ho
, conf
.title_separator
);
818 html_words(&ho
, f
->last
->title
->words
, NOTHING
,
822 element_close(&ho
, "title");
826 html_raw(&ho
, conf
.head_end
);
828 element_close(&ho
, "head");
832 html_raw(&ho
, conf
.body_tag
);
834 element_open(&ho
, "body");
838 html_raw(&ho
, conf
.body_start
);
841 * Write out a nav bar. Special case: we don't do this
842 * if there is only one file.
844 if (files
.head
!= files
.tail
) {
845 element_open(&ho
, "p");
847 html_raw_as_attr(&ho
, conf
.nav_attr
);
850 element_open(&ho
, "a");
851 element_attr(&ho
, "href", prevf
->filename
);
853 html_text(&ho
, conf
.nav_prev_text
);
855 element_close(&ho
, "a");
857 html_text(&ho
, conf
.nav_separator
);
859 if (f
!= files
.head
) {
860 element_open(&ho
, "a");
861 element_attr(&ho
, "href", files
.head
->filename
);
863 html_text(&ho
, conf
.contents_text
);
865 element_close(&ho
, "a");
867 html_text(&ho
, conf
.nav_separator
);
869 if (f
!= files
.index
) {
870 element_open(&ho
, "a");
871 element_attr(&ho
, "href", files
.index
->filename
);
873 html_text(&ho
, conf
.index_text
);
874 if (f
!= files
.index
)
875 element_close(&ho
, "a");
877 html_text(&ho
, conf
.nav_separator
);
880 element_open(&ho
, "a");
881 element_attr(&ho
, "href", f
->next
->filename
);
883 html_text(&ho
, conf
.nav_next_text
);
885 element_close(&ho
, "a");
887 element_close(&ho
, "p");
893 * Write out a prefix TOC for the file.
895 * We start by going through the section list and
896 * collecting the sections which need to be added to
897 * the contents. On the way, we also test to see if
898 * this file is a leaf file (defined as one which
899 * contains all descendants of any section it
900 * contains), because this will play a part in our
901 * decision on whether or not to _output_ the TOC.
903 * Special case: we absolutely do not do this if we're
904 * in single-file mode.
906 if (files
.head
!= files
.tail
) {
907 int ntoc
= 0, tocsize
= 0;
908 htmlsect
**toc
= NULL
;
911 for (s
= sects
.head
; s
; s
= s
->next
) {
916 * Search up from this section until we find
917 * the highest-level one which belongs in this
922 for (ac
= s
; ac
; ac
= ac
->parent
) {
930 if (s
->file
!= f
&& a
!= NULL
)
934 if (adepth
<= a
->contents_depth
) {
935 if (ntoc
>= tocsize
) {
937 toc
= sresize(toc
, tocsize
, htmlsect
*);
944 if (leaf
&& conf
.leaf_contains_contents
&&
945 ntoc
>= conf
.leaf_smallest_contents
) {
948 for (i
= 0; i
< ntoc
; i
++) {
949 htmlsect
*s
= toc
[i
];
950 int hlevel
= (s
->type
== TOP ?
-1 :
951 s
->type
== INDEX ?
0 :
952 heading_depth(s
->title
))
953 - f
->min_heading_depth
+ 1;
956 html_contents_entry(&ho
, hlevel
, s
,
959 html_contents_entry(&ho
, 0, NULL
, f
, keywords
, &conf
);
964 * Now go through the document and output some real
968 for (s
= sects
.head
; s
; s
= s
->next
) {
971 * This section belongs in this file.
982 * Search up from this section until we find
983 * the highest-level one which belongs in this
988 for (ac
= s
; ac
; ac
= ac
->parent
) {
998 * This section does not belong in this
999 * file, but an ancestor of it does. Write
1000 * out a contents table entry, if the depth
1001 * doesn't exceed the maximum contents
1002 * depth for the ancestor section.
1004 if (adepth
<= a
->contents_depth
) {
1005 html_contents_entry(&ho
, adepth
, s
,
1006 f
, keywords
, &conf
);
1015 html_contents_entry(&ho
, 0, NULL
, f
, keywords
, &conf
);
1018 * Display the section heading.
1021 hlevel
= (s
->type
== TOP ?
-1 :
1022 s
->type
== INDEX ?
0 :
1023 heading_depth(s
->title
))
1024 - f
->min_heading_depth
+ 1;
1025 assert(hlevel
>= 1);
1026 /* HTML headings only go up to <h6> */
1030 htag
[1] = '0' + hlevel
;
1032 element_open(&ho
, htag
);
1035 * Provide anchor for cross-links to target.
1037 * (Also we'll have to do this separately in
1038 * other paragraph types - NumberedList and
1042 html_fragment(&ho
, s
->fragment
);
1044 html_section_title(&ho
, s
, f
, keywords
, &conf
, TRUE
);
1046 element_close(&ho
, htag
);
1049 * Now display the section text.
1052 stackhead
= snew(struct stackelement
);
1053 stackhead
->next
= NULL
;
1054 stackhead
->listtype
= NOLIST
;
1055 stackhead
->itemtype
= NOITEM
;
1057 for (p
= s
->text
;; p
= p
->next
) {
1058 enum LISTTYPE listtype
;
1059 struct stackelement
*se
;
1062 * Preliminary switch to figure out what
1063 * sort of list we expect to be inside at
1066 * Since p may still be NULL at this point,
1067 * I invent a harmless paragraph type for
1070 switch (p ? p
->type
: para_Normal
) {
1073 case para_Copyright
:
1074 case para_BiblioCited
:
1076 case para_QuotePush
:
1080 case para_UnnumberedChapter
:
1091 case para_NumberedList
:
1095 case para_DescribedThing
:
1096 case para_Description
:
1100 case para_LcontPush
:
1101 se
= snew(struct stackelement
);
1102 se
->next
= stackhead
;
1103 se
->listtype
= NOLIST
;
1104 se
->itemtype
= NOITEM
;
1108 default: /* some totally non-printing para */
1115 * Terminate the most recent list item, if
1116 * any. (We left this until after
1117 * processing LcontPush, since in that case
1118 * the list item won't want to be
1119 * terminated until after the corresponding
1122 if (stackhead
->itemtype
!= NOITEM
) {
1123 element_close(&ho
, itemname(stackhead
->itemtype
));
1126 stackhead
->itemtype
= NOITEM
;
1129 * Terminate the current list, if it's not
1130 * the one we want to be in.
1132 if (listtype
!= stackhead
->listtype
&&
1133 stackhead
->listtype
!= NOLIST
) {
1134 element_close(&ho
, listname(stackhead
->listtype
));
1139 * Leave the loop if our time has come.
1141 if (!p
|| (is_heading_type(p
->type
) &&
1142 p
->type
!= para_Title
))
1143 break; /* end of section text */
1146 * Start a fresh list if necessary.
1148 if (listtype
!= stackhead
->listtype
&&
1150 element_open(&ho
, listname(listtype
));
1152 stackhead
->listtype
= listtype
;
1156 element_empty(&ho
, "hr");
1159 html_codepara(&ho
, p
->words
);
1162 case para_Copyright
:
1163 element_open(&ho
, "p");
1165 html_words(&ho
, p
->words
, ALL
,
1166 f
, keywords
, &conf
);
1168 element_close(&ho
, "p");
1170 case para_BiblioCited
:
1171 element_open(&ho
, "p");
1172 if (p
->private_data
) {
1173 htmlsect
*s
= (htmlsect
*)p
->private_data
;
1174 html_fragment(&ho
, s
->fragment
);
1177 html_words(&ho
, p
->kwtext
, ALL
,
1178 f
, keywords
, &conf
);
1179 html_text(&ho
, L
" ");
1180 html_words(&ho
, p
->words
, ALL
,
1181 f
, keywords
, &conf
);
1183 element_close(&ho
, "p");
1186 case para_NumberedList
:
1187 element_open(&ho
, "li");
1188 if (p
->private_data
) {
1189 htmlsect
*s
= (htmlsect
*)p
->private_data
;
1190 html_fragment(&ho
, s
->fragment
);
1193 stackhead
->itemtype
= LI
;
1194 html_words(&ho
, p
->words
, ALL
,
1195 f
, keywords
, &conf
);
1197 case para_DescribedThing
:
1198 element_open(&ho
, "dt");
1200 stackhead
->itemtype
= DT
;
1201 html_words(&ho
, p
->words
, ALL
,
1202 f
, keywords
, &conf
);
1204 case para_Description
:
1205 element_open(&ho
, "dd");
1207 stackhead
->itemtype
= DD
;
1208 html_words(&ho
, p
->words
, ALL
,
1209 f
, keywords
, &conf
);
1212 case para_QuotePush
:
1213 element_open(&ho
, "blockquote");
1216 element_close(&ho
, "blockquote");
1221 stackhead
= stackhead
->next
;
1228 assert(stackhead
&& !stackhead
->next
);
1232 if (s
->type
== INDEX
) {
1237 * This section is the index. I'll just
1238 * render it as a single paragraph, with a
1239 * colon between the index term and the
1240 * references, and <br> in between each
1243 element_open(&ho
, "p");
1245 for (i
= 0; (entry
=
1246 index234(idx
->entries
, i
)) != NULL
; i
++) {
1247 htmlindex
*hi
= (htmlindex
*)entry
->backend_data
;
1251 element_empty(&ho
, "br");
1254 html_words(&ho
, entry
->text
, MARKUP
|LINKS
,
1255 f
, keywords
, &conf
);
1257 html_text(&ho
, conf
.index_main_sep
);
1259 for (j
= 0; j
< hi
->nrefs
; j
++) {
1261 (htmlindexref
*)hi
->refs
[j
]->private_data
;
1262 paragraph
*p
= hr
->section
->title
;
1265 html_text(&ho
, conf
.index_multi_sep
);
1267 html_href(&ho
, f
, hr
->section
->file
,
1269 hr
->referenced
= TRUE
;
1271 html_words(&ho
, p
->kwtext
, MARKUP
|LINKS
,
1272 f
, keywords
, &conf
);
1273 else if (p
&& p
->words
)
1274 html_words(&ho
, p
->words
, MARKUP
|LINKS
,
1275 f
, keywords
, &conf
);
1278 * If there is no title at all,
1279 * this must be because our
1280 * target section is the
1281 * preamble section and there
1282 * is no title. So we use the
1285 html_text(&ho
, conf
.preamble_text
);
1287 element_close(&ho
, "a");
1290 element_close(&ho
, "p");
1295 html_contents_entry(&ho
, 0, NULL
, f
, keywords
, &conf
);
1302 int done_version_ids
= FALSE
;
1304 element_empty(&ho
, "hr");
1307 html_raw(&ho
, conf
.body_end
);
1309 if (conf
.address_section
) {
1310 int started
= FALSE
;
1311 if (conf
.htmlver
== ISO_HTML
) {
1313 * The ISO-HTML validator complains if
1314 * there isn't a <div> tag surrounding the
1315 * <address> tag. I'm uncertain of why this
1316 * should be - there appears to be no
1317 * mention of this in the ISO-HTML spec,
1318 * suggesting that it doesn't represent a
1319 * change from HTML 4, but nonetheless the
1320 * HTML 4 validator doesn't seem to mind.
1322 element_open(&ho
, "div");
1324 element_open(&ho
, "address");
1325 if (conf
.addr_start
) {
1326 html_raw(&ho
, conf
.addr_start
);
1330 if (conf
.visible_version_id
) {
1331 for (p
= sourceform
; p
; p
= p
->next
)
1332 if (p
->type
== para_VersionID
) {
1334 element_empty(&ho
, "br");
1336 html_text(&ho
, conf
.pre_versionid
);
1337 html_words(&ho
, p
->words
, NOTHING
,
1338 f
, keywords
, &conf
);
1339 html_text(&ho
, conf
.post_versionid
);
1342 done_version_ids
= TRUE
;
1344 if (conf
.addr_end
) {
1346 element_empty(&ho
, "br");
1347 html_raw(&ho
, conf
.addr_end
);
1349 element_close(&ho
, "address");
1350 if (conf
.htmlver
== ISO_HTML
)
1351 element_close(&ho
, "div");
1354 if (!done_version_ids
) {
1356 * If the user didn't want the version IDs
1357 * visible, I think we still have a duty to put
1358 * them in an HTML comment.
1360 int started
= FALSE
;
1361 for (p
= sourceform
; p
; p
= p
->next
)
1362 if (p
->type
== para_VersionID
) {
1364 html_raw(&ho
, "<!-- version IDs:\n");
1367 html_words(&ho
, p
->words
, NOTHING
,
1368 f
, keywords
, &conf
);
1372 html_raw(&ho
, "-->\n");
1376 element_close(&ho
, "body");
1378 element_close(&ho
, "html");
1385 * Go through and check that no index fragments were referenced
1386 * without being generated, or indeed vice versa.
1388 * (When I actually get round to freeing everything, this can
1389 * probably be the freeing loop as well.)
1391 for (p
= sourceform
; p
; p
= p
->next
) {
1393 for (w
= p
->words
; w
; w
= w
->next
)
1394 if (w
->type
== word_IndexRef
) {
1395 htmlindexref
*hr
= (htmlindexref
*)w
->private_data
;
1397 assert(!hr
->referenced
== !hr
->generated
);
1402 * Free all the working data.
1405 sfree(conf
.single_filename
);
1406 sfree(conf
.contents_filename
);
1407 sfree(conf
.index_filename
);
1408 sfree(conf
.template_filename
);
1409 sfree(conf
.template_fragment
);
1412 while ( (frag
= (htmlfragment
*)delpos234(files
.frags
, 0)) != NULL
) {
1414 * frag->fragment is dynamically allocated, but will be
1415 * freed when we process the htmlsect structure which
1416 * it is attached to.
1420 freetree234(files
.frags
);
1423 htmlsect
*sect
, *tmp
;
1427 sfree(sect
->fragment
);
1431 sect
= nonsects
.head
;
1434 sfree(sect
->fragment
);
1440 htmlfile
*file
, *tmp
;
1444 sfree(file
->filename
);
1452 for (i
= 0; (entry
= index234(idx
->entries
, i
)) != NULL
; i
++) {
1453 htmlindex
*hi
= (htmlindex
*)entry
->backend_data
;
1460 for (p
= sourceform
; p
; p
= p
->next
)
1461 for (w
= p
->words
; w
; w
= w
->next
)
1462 if (w
->type
== word_IndexRef
) {
1463 htmlindexref
*hr
= (htmlindexref
*)w
->private_data
;
1465 sfree(hr
->fragment
);
1471 static void html_file_section(htmlconfig
*cfg
, htmlfilelist
*files
,
1472 htmlsect
*sect
, int depth
)
1478 * `depth' is derived from the heading_depth() macro at the top
1479 * of this file, which counts title as -1, chapter as 0,
1480 * heading as 1 and subsection as 2. However, the semantics of
1481 * cfg->leaf_level are defined to count chapter as 1, heading
1482 * as 2 etc. So first I increment depth :-(
1486 if (cfg
->leaf_level
== 0) {
1488 * leaf_level==0 is a special case, in which everything is
1489 * put into a single file.
1492 files
->single
= html_new_file(files
, cfg
->single_filename
);
1494 file
= files
->single
;
1497 * If the depth of this section is at or above leaf_level,
1498 * we invent a fresh file and put this section at its head.
1499 * Otherwise, we put it in the same file as its parent
1502 if (ldepth
> cfg
->leaf_level
) {
1504 * We know that sect->parent cannot be NULL. The only
1505 * circumstance in which it can be is if sect is at
1506 * chapter or appendix level, i.e. ldepth==1; and if
1507 * that's the case, then we cannot have entered this
1508 * branch unless cfg->leaf_level==0, in which case we
1509 * would be in the single-file case above and not here
1512 assert(sect
->parent
);
1514 file
= sect
->parent
->file
;
1516 if (sect
->type
== TOP
) {
1517 file
= html_new_file(files
, cfg
->contents_filename
);
1518 } else if (sect
->type
== INDEX
) {
1519 file
= html_new_file(files
, cfg
->index_filename
);
1523 assert(ldepth
> 0 && sect
->title
);
1524 title
= html_format(sect
->title
, cfg
->template_filename
);
1525 file
= html_new_file(files
, title
);
1533 if (file
->min_heading_depth
> depth
) {
1535 * This heading is at a higher level than any heading we
1536 * have so far placed in this file; so we set the `first'
1539 file
->min_heading_depth
= depth
;
1543 if (file
->min_heading_depth
== depth
)
1547 static htmlfile
*html_new_file(htmlfilelist
*list
, char *filename
)
1549 htmlfile
*ret
= snew(htmlfile
);
1553 list
->tail
->next
= ret
;
1558 ret
->filename
= dupstr(filename
);
1559 ret
->last_fragment_number
= 0;
1560 ret
->min_heading_depth
= INT_MAX
;
1561 ret
->first
= ret
->last
= NULL
;
1566 static htmlsect
*html_new_sect(htmlsectlist
*list
, paragraph
*title
)
1568 htmlsect
*ret
= snew(htmlsect
);
1572 list
->tail
->next
= ret
;
1585 static void html_words(htmloutput
*ho
, word
*words
, int flags
,
1586 htmlfile
*file
, keywordlist
*keywords
, htmlconfig
*cfg
)
1592 for (w
= words
; w
; w
= w
->next
) switch (w
->type
) {
1593 case word_HyperLink
:
1594 if (flags
& LINKS
) {
1595 element_open(ho
, "a");
1596 c
= utoa_dup(w
->text
, CS_ASCII
);
1597 element_attr(ho
, "href", c
);
1601 case word_UpperXref
:
1602 case word_LowerXref
:
1603 if (flags
& LINKS
) {
1604 keyword
*kwl
= kw_lookup(keywords
, w
->text
);
1605 paragraph
*p
= kwl
->para
;
1606 htmlsect
*s
= (htmlsect
*)p
->private_data
;
1610 html_href(ho
, file
, s
->file
, s
->fragment
);
1616 element_close(ho
, "a");
1619 if (flags
& INDEXENTS
) {
1620 htmlindexref
*hr
= (htmlindexref
*)w
->private_data
;
1621 html_fragment(ho
, hr
->fragment
);
1622 hr
->generated
= TRUE
;
1629 case word_WhiteSpace
:
1630 case word_EmphSpace
:
1631 case word_CodeSpace
:
1632 case word_WkCodeSpace
:
1634 case word_EmphQuote
:
1635 case word_CodeQuote
:
1636 case word_WkCodeQuote
:
1637 style
= towordstyle(w
->type
);
1638 type
= removeattr(w
->type
);
1639 if (style
== word_Emph
&&
1640 (attraux(w
->aux
) == attr_First
||
1641 attraux(w
->aux
) == attr_Only
) &&
1643 element_open(ho
, "em");
1644 else if ((style
== word_Code
|| style
== word_WeakCode
) &&
1645 (attraux(w
->aux
) == attr_First
||
1646 attraux(w
->aux
) == attr_Only
) &&
1648 element_open(ho
, "code");
1650 if (type
== word_WhiteSpace
)
1651 html_text(ho
, L
" ");
1652 else if (type
== word_Quote
) {
1653 if (quoteaux(w
->aux
) == quote_Open
)
1654 html_text(ho
, cfg
->lquote
);
1656 html_text(ho
, cfg
->rquote
);
1658 if (!w
->alt
|| cvt_ok(ho
->restrict_charset
, w
->text
))
1659 html_text_nbsp(ho
, w
->text
);
1661 html_words(ho
, w
->alt
, flags
, file
, keywords
, cfg
);
1664 if (style
== word_Emph
&&
1665 (attraux(w
->aux
) == attr_Last
||
1666 attraux(w
->aux
) == attr_Only
) &&
1668 element_close(ho
, "em");
1669 else if ((style
== word_Code
|| style
== word_WeakCode
) &&
1670 (attraux(w
->aux
) == attr_Last
||
1671 attraux(w
->aux
) == attr_Only
) &&
1673 element_close(ho
, "code");
1679 static void html_codepara(htmloutput
*ho
, word
*words
)
1681 element_open(ho
, "pre");
1682 element_open(ho
, "code");
1683 for (; words
; words
= words
->next
) if (words
->type
== word_WeakCode
) {
1688 if (words
->next
&& words
->next
->type
== word_Emph
) {
1689 e
= words
->next
->text
;
1690 words
= words
->next
;
1694 while (e
&& *e
&& *t
) {
1698 for (n
= 0; t
[n
] && e
[n
] && e
[n
] == ec
; n
++);
1706 element_open(ho
, open_tag
);
1708 html_text_limit(ho
, t
, n
);
1711 element_close(ho
, open_tag
);
1719 element_close(ho
, "code");
1720 element_close(ho
, "pre");
1723 static void html_charset_cleanup(htmloutput
*ho
)
1728 bytes
= charset_from_unicode(NULL
, NULL
, outbuf
, lenof(outbuf
),
1729 ho
->charset
, &ho
->cstate
, NULL
);
1731 fwrite(outbuf
, 1, bytes
, ho
->fp
);
1734 static void return_mostly_to_neutral(htmloutput
*ho
)
1736 if (ho
->state
== HO_IN_EMPTY_TAG
&& is_xhtml(ho
->ver
)) {
1737 fprintf(ho
->fp
, " />");
1738 } else if (ho
->state
== HO_IN_EMPTY_TAG
|| ho
->state
== HO_IN_TAG
) {
1739 fprintf(ho
->fp
, ">");
1742 ho
->state
= HO_NEUTRAL
;
1745 static void return_to_neutral(htmloutput
*ho
)
1747 if (ho
->state
== HO_IN_TEXT
) {
1748 html_charset_cleanup(ho
);
1751 return_mostly_to_neutral(ho
);
1754 static void element_open(htmloutput
*ho
, char const *name
)
1756 return_to_neutral(ho
);
1757 fprintf(ho
->fp
, "<%s", name
);
1758 ho
->state
= HO_IN_TAG
;
1761 static void element_close(htmloutput
*ho
, char const *name
)
1763 return_to_neutral(ho
);
1764 fprintf(ho
->fp
, "</%s>", name
);
1765 ho
->state
= HO_NEUTRAL
;
1768 static void element_empty(htmloutput
*ho
, char const *name
)
1770 return_to_neutral(ho
);
1771 fprintf(ho
->fp
, "<%s", name
);
1772 ho
->state
= HO_IN_EMPTY_TAG
;
1775 static void html_nl(htmloutput
*ho
)
1777 return_to_neutral(ho
);
1778 fputc('\n', ho
->fp
);
1781 static void html_raw(htmloutput
*ho
, char *text
)
1783 return_to_neutral(ho
);
1784 fputs(text
, ho
->fp
);
1787 static void html_raw_as_attr(htmloutput
*ho
, char *text
)
1789 assert(ho
->state
== HO_IN_TAG
|| ho
->state
== HO_IN_EMPTY_TAG
);
1791 fputs(text
, ho
->fp
);
1794 static void element_attr(htmloutput
*ho
, char const *name
, char const *value
)
1796 html_charset_cleanup(ho
);
1797 assert(ho
->state
== HO_IN_TAG
|| ho
->state
== HO_IN_EMPTY_TAG
);
1798 fprintf(ho
->fp
, " %s=\"%s\"", name
, value
);
1801 static void element_attr_w(htmloutput
*ho
, char const *name
,
1802 wchar_t const *value
)
1804 html_charset_cleanup(ho
);
1805 fprintf(ho
->fp
, " %s=\"", name
);
1806 html_text_limit_internal(ho
, value
, 0, TRUE
, FALSE
);
1807 html_charset_cleanup(ho
);
1811 static void html_text(htmloutput
*ho
, wchar_t const *text
)
1813 return_mostly_to_neutral(ho
);
1814 html_text_limit_internal(ho
, text
, 0, FALSE
, FALSE
);
1817 static void html_text_nbsp(htmloutput
*ho
, wchar_t const *text
)
1819 return_mostly_to_neutral(ho
);
1820 html_text_limit_internal(ho
, text
, 0, FALSE
, TRUE
);
1823 static void html_text_limit(htmloutput
*ho
, wchar_t const *text
, int maxlen
)
1825 return_mostly_to_neutral(ho
);
1826 html_text_limit_internal(ho
, text
, maxlen
, FALSE
, FALSE
);
1829 static void html_text_limit_internal(htmloutput
*ho
, wchar_t const *text
,
1830 int maxlen
, int quote_quotes
, int nbsp
)
1832 int textlen
= ustrlen(text
);
1836 if (maxlen
> 0 && textlen
> maxlen
)
1839 while (textlen
> 0) {
1840 /* Scan ahead for characters we really can't display in HTML. */
1841 int lenbefore
, lenafter
;
1842 for (lenbefore
= 0; lenbefore
< textlen
; lenbefore
++)
1843 if (text
[lenbefore
] == L
'<' ||
1844 text
[lenbefore
] == L
'>' ||
1845 text
[lenbefore
] == L
'&' ||
1846 (text
[lenbefore
] == L
'"' && quote_quotes
) ||
1847 (text
[lenbefore
] == L
' ' && nbsp
))
1849 lenafter
= lenbefore
;
1850 bytes
= charset_from_unicode(&text
, &lenafter
, outbuf
, lenof(outbuf
),
1851 ho
->charset
, &ho
->cstate
, &err
);
1852 textlen
-= (lenbefore
- lenafter
);
1854 fwrite(outbuf
, 1, bytes
, ho
->fp
);
1857 * We have encountered a character that cannot be
1858 * displayed in the selected output charset. Therefore,
1859 * we use an HTML numeric entity reference.
1861 assert(textlen
> 0);
1862 fprintf(ho
->fp
, "&#%ld;", (long int)*text
);
1864 } else if (lenafter
== 0 && textlen
> 0) {
1866 * We have encountered a character which is special to
1870 fprintf(ho
->fp
, "<");
1871 else if (*text
== L
'>')
1872 fprintf(ho
->fp
, ">");
1873 else if (*text
== L
'&')
1874 fprintf(ho
->fp
, "&");
1875 else if (*text
== L
'"')
1876 fprintf(ho
->fp
, """);
1877 else if (*text
== L
' ') {
1879 fprintf(ho
->fp
, " ");
1881 assert(!"Can't happen");
1887 static void cleanup(htmloutput
*ho
)
1889 return_to_neutral(ho
);
1893 static void html_href(htmloutput
*ho
, htmlfile
*thisfile
,
1894 htmlfile
*targetfile
, char *targetfrag
)
1896 rdstringc rs
= { 0, 0, NULL
};
1899 if (targetfile
!= thisfile
)
1900 rdaddsc(&rs
, targetfile
->filename
);
1903 rdaddsc(&rs
, targetfrag
);
1907 element_open(ho
, "a");
1908 element_attr(ho
, "href", url
);
1912 static void html_fragment(htmloutput
*ho
, char const *fragment
)
1914 element_open(ho
, "a");
1915 element_attr(ho
, "name", fragment
);
1916 if (is_xhtml(ho
->ver
))
1917 element_attr(ho
, "id", fragment
);
1918 element_close(ho
, "a");
1921 static char *html_format(paragraph
*p
, char *template_string
)
1925 wchar_t *ws
, wsbuf
[2];
1926 rdstringc rs
= { 0, 0, NULL
};
1928 t
= template_string
;
1930 if (*t
== '%' && t
[1]) {
1944 if (p
->kwtext
&& fmt
== 'n')
1946 else if (p
->kwtext2
&& fmt
== 'b') {
1948 * HTML fragment names must start with a letter, so
1949 * simply `1.2.3' is not adequate. In this case I'm
1950 * going to cheat slightly by prepending the first
1951 * character of the first word of kwtext, so that
1952 * we get `C1' for chapter 1, `S2.3' for section
1955 if (p
->kwtext
&& p
->kwtext
->text
[0]) {
1958 wsbuf
[0] = p
->kwtext
->text
[0];
1961 } else if (p
->keyword
&& *p
->keyword
&& fmt
== 'k')
1967 c
= utoa_dup(ws
, CS_ASCII
);
1973 if (removeattr(w
->type
) == word_Normal
) {
1974 c
= utoa_dup(w
->text
, CS_ASCII
);
1985 return rdtrimc(&rs
);
1988 static char *html_sanitise_fragment(htmlfilelist
*files
, htmlfile
*file
,
1992 * The HTML 4 spec's strictest definition of fragment names (<a
1993 * name> and "id" attributes) says that they `must begin with a
1994 * letter and may be followed by any number of letters, digits,
1995 * hyphens, underscores, colons, and periods'.
1997 * So here we unceremoniously rip out any characters not
1998 * conforming to this limitation.
2000 char *p
= text
, *q
= text
;
2002 while (*p
&& !((*p
>='A' && *p
<='Z') || (*p
>='a' && *p
<='z')))
2004 if ((*q
++ = *p
++) != '\0') {
2006 if ((*p
>='A' && *p
<='Z') ||
2007 (*p
>='a' && *p
<='z') ||
2008 (*p
>='0' && *p
<='9') ||
2009 *p
=='-' || *p
=='_' || *p
==':' || *p
=='.')
2018 * Now we check for clashes with other fragment names, and
2019 * adjust this one if necessary by appending a hyphen followed
2023 htmlfragment
*frag
= snew(htmlfragment
);
2024 int len
= 0; /* >0 indicates we have resized */
2028 frag
->fragment
= text
;
2030 while (add234(files
->frags
, frag
) != frag
) {
2033 frag
->fragment
= text
= sresize(text
, len
+20, char);
2036 sprintf(text
+ len
, "-%d", ++suffix
);
2043 static void html_contents_entry(htmloutput
*ho
, int depth
, htmlsect
*s
,
2044 htmlfile
*thisfile
, keywordlist
*keywords
,
2047 while (ho
->contents_level
> depth
) {
2048 element_close(ho
, "ul");
2049 ho
->contents_level
--;
2052 while (ho
->contents_level
< depth
) {
2053 element_open(ho
, "ul");
2054 ho
->contents_level
++;
2060 element_open(ho
, "li");
2061 html_href(ho
, thisfile
, s
->file
, s
->fragment
);
2062 html_section_title(ho
, s
, thisfile
, keywords
, cfg
, FALSE
);
2063 element_close(ho
, "a");
2064 element_close(ho
, "li");
2067 static void html_section_title(htmloutput
*ho
, htmlsect
*s
, htmlfile
*thisfile
,
2068 keywordlist
*keywords
, htmlconfig
*cfg
,
2074 int depth
= heading_depth(s
->title
);
2078 else if (depth
== 0)
2079 sl
= &cfg
->achapter
;
2080 else if (depth
<= cfg
->nasect
)
2081 sl
= &cfg
->asect
[depth
-1];
2083 sl
= &cfg
->asect
[cfg
->nasect
-1];
2087 else if (sl
->just_numbers
)
2088 number
= s
->title
->kwtext2
;
2090 number
= s
->title
->kwtext
;
2093 html_words(ho
, number
, MARKUP
,
2094 thisfile
, keywords
, cfg
);
2095 html_text(ho
, sl
->number_suffix
);
2098 html_words(ho
, s
->title
->words
, real ? ALL
: MARKUP
,
2099 thisfile
, keywords
, cfg
);
2101 assert(s
->type
!= NORMAL
);
2103 * If we're printing the full document title for _real_ and
2104 * there isn't one, we don't want to print `Preamble' at
2105 * the top of what ought to just be some text. If we need
2106 * it in any other context such as TOCs, we need to print
2109 if (s
->type
== TOP
&& !real
)
2110 html_text(ho
, cfg
->preamble_text
);
2111 else if (s
->type
== INDEX
)
2112 html_text(ho
, cfg
->index_text
);