2 * HTML backend for Halibut
8 * - I'm never entirely convinced that having a fragment link to
9 * come in at the start of the real text in the file is
10 * sensible. Perhaps for the topmost section in the file, no
11 * fragment should be used? (Though it should probably still be
12 * _there_ even if unused.)
14 * - free up all the data we have allocated while running this
24 #define is_heading_type(type) ( (type) == para_Title || \
25 (type) == para_Chapter || \
26 (type) == para_Appendix || \
27 (type) == para_UnnumberedChapter || \
28 (type) == para_Heading || \
29 (type) == para_Subsect)
31 #define heading_depth(p) ( (p)->type == para_Subsect ? (p)->aux + 1 : \
32 (p)->type == para_Heading ? 1 : \
33 (p)->type == para_Title ? -1 : 0 )
37 wchar_t *number_suffix
;
42 sectlevel achapter
, *asect
;
43 int *contents_depths
; /* 0=main, 1=chapter, 2=sect etc */
45 int address_section
, visible_version_id
;
46 int leaf_contains_contents
, leaf_smallest_contents
;
47 char *contents_filename
;
49 char *template_filename
;
50 char *single_filename
;
51 char *template_fragment
;
52 char *head_end
, *body_start
, *body_end
, *addr_start
, *addr_end
;
53 char *body_tag
, *nav_attr
;
54 wchar_t *author
, *description
;
55 wchar_t *index_text
, *contents_text
, *preamble_text
, *title_separator
;
56 wchar_t *nav_prev_text
, *nav_next_text
, *nav_separator
;
57 wchar_t *index_main_sep
, *index_multi_sep
;
58 wchar_t *pre_versionid
, *post_versionid
;
59 int restrict_charset
, output_charset
;
61 HTML_3_2
, HTML_4
, ISO_HTML
,
62 XHTML_1_0_TRANSITIONAL
, XHTML_1_0_STRICT
64 wchar_t *lquote
, *rquote
;
68 #define contents_depth(conf, level) \
69 ( (conf).ncdepths > (level) ? (conf).contents_depths[level] : (level)+2 )
71 #define is_xhtml(ver) ((ver) >= XHTML_1_0_TRANSITIONAL)
73 typedef struct htmlfile htmlfile
;
74 typedef struct htmlsect htmlsect
;
79 int last_fragment_number
;
80 int min_heading_depth
;
81 htmlsect
*first
, *last
; /* first/last highest-level sections */
85 htmlsect
*next
, *parent
;
87 paragraph
*title
, *text
;
88 enum { NORMAL
, TOP
, INDEX
} type
;
94 htmlfile
*head
, *tail
;
95 htmlfile
*single
, *index
;
100 htmlsect
*head
, *tail
;
116 int generated
, referenced
;
121 * This level deals with charset conversion, starting and
122 * ending tags, and writing to the file. It's the lexical
126 int charset
, restrict_charset
;
127 charset_state cstate
;
130 HO_NEUTRAL
, HO_IN_TAG
, HO_IN_EMPTY_TAG
, HO_IN_TEXT
133 * Stuff beyond here deals with the higher syntactic level: it
134 * tracks how many levels of <ul> are currently open when
135 * producing a contents list, for example.
140 static int html_fragment_compare(void *av
, void *bv
)
142 htmlfragment
*a
= (htmlfragment
*)av
;
143 htmlfragment
*b
= (htmlfragment
*)bv
;
146 if ((cmp
= strcmp(a
->file
->filename
, b
->file
->filename
)) != 0)
149 return strcmp(a
->fragment
, b
->fragment
);
152 static void html_file_section(htmlconfig
*cfg
, htmlfilelist
*files
,
153 htmlsect
*sect
, int depth
);
155 static htmlfile
*html_new_file(htmlfilelist
*list
, char *filename
);
156 static htmlsect
*html_new_sect(htmlsectlist
*list
, paragraph
*title
);
158 /* Flags for html_words() flags parameter */
162 #define INDEXENTS 0x04
164 static void html_words(htmloutput
*ho
, word
*words
, int flags
,
165 htmlfile
*file
, keywordlist
*keywords
, htmlconfig
*cfg
);
166 static void html_codepara(htmloutput
*ho
, word
*words
);
168 static void element_open(htmloutput
*ho
, char const *name
);
169 static void element_close(htmloutput
*ho
, char const *name
);
170 static void element_empty(htmloutput
*ho
, char const *name
);
171 static void element_attr(htmloutput
*ho
, char const *name
, char const *value
);
172 static void element_attr_w(htmloutput
*ho
, char const *name
,
173 wchar_t const *value
);
174 static void html_text(htmloutput
*ho
, wchar_t const *str
);
175 static void html_text_nbsp(htmloutput
*ho
, wchar_t const *str
);
176 static void html_text_limit(htmloutput
*ho
, wchar_t const *str
, int maxlen
);
177 static void html_text_limit_internal(htmloutput
*ho
, wchar_t const *text
,
178 int maxlen
, int quote_quotes
, int nbsp
);
179 static void html_nl(htmloutput
*ho
);
180 static void html_raw(htmloutput
*ho
, char *text
);
181 static void html_raw_as_attr(htmloutput
*ho
, char *text
);
182 static void cleanup(htmloutput
*ho
);
184 static void html_href(htmloutput
*ho
, htmlfile
*thisfile
,
185 htmlfile
*targetfile
, char *targetfrag
);
186 static void html_fragment(htmloutput
*ho
, char const *fragment
);
188 static char *html_format(paragraph
*p
, char *template_string
);
189 static char *html_sanitise_fragment(htmlfilelist
*files
, htmlfile
*file
,
192 static void html_contents_entry(htmloutput
*ho
, int depth
, htmlsect
*s
,
193 htmlfile
*thisfile
, keywordlist
*keywords
,
195 static void html_section_title(htmloutput
*ho
, htmlsect
*s
,
196 htmlfile
*thisfile
, keywordlist
*keywords
,
197 htmlconfig
*cfg
, int real
);
199 static htmlconfig
html_configure(paragraph
*source
) {
207 ret
.achapter
.just_numbers
= FALSE
;
208 ret
.achapter
.number_suffix
= L
": ";
210 ret
.asect
= snewn(ret
.nasect
, sectlevel
);
211 ret
.asect
[0].just_numbers
= TRUE
;
212 ret
.asect
[0].number_suffix
= L
" ";
214 ret
.contents_depths
= 0;
215 ret
.visible_version_id
= TRUE
;
216 ret
.address_section
= TRUE
;
217 ret
.leaf_contains_contents
= FALSE
;
218 ret
.leaf_smallest_contents
= 4;
219 ret
.single_filename
= dupstr("Manual.html");
220 ret
.contents_filename
= dupstr("Contents.html");
221 ret
.index_filename
= dupstr("IndexPage.html");
222 ret
.template_filename
= dupstr("%n.html");
223 ret
.template_fragment
= dupstr("%b");
224 ret
.head_end
= ret
.body_tag
= ret
.body_start
= ret
.body_end
=
225 ret
.addr_start
= ret
.addr_end
= ret
.nav_attr
= NULL
;
226 ret
.author
= ret
.description
= NULL
;
227 ret
.restrict_charset
= CS_UTF8
;
228 ret
.output_charset
= CS_ASCII
;
229 ret
.htmlver
= HTML_4
;
230 ret
.index_text
= L
"Index";
231 ret
.contents_text
= L
"Contents";
232 ret
.preamble_text
= L
"Preamble";
233 ret
.title_separator
= L
" - ";
234 ret
.nav_prev_text
= L
"Previous";
235 ret
.nav_next_text
= L
"Next";
236 ret
.nav_separator
= L
" | ";
237 ret
.index_main_sep
= L
": ";
238 ret
.index_multi_sep
= L
", ";
239 ret
.pre_versionid
= L
"[";
240 ret
.post_versionid
= L
"]";
242 * Default quote characters are Unicode matched single quotes,
243 * falling back to ordinary ASCII ".
245 ret
.lquote
= L
"\x2018\0\x2019\0\"\0\"\0\0";
246 ret
.rquote
= uadv(ret
.lquote
);
249 * Two-pass configuration so that we can pick up global config
250 * (e.g. `quotes') before having it overridden by specific
251 * config (`html-quotes'), irrespective of the order in which
254 for (p
= source
; p
; p
= p
->next
) {
255 if (p
->type
== para_Config
) {
256 if (!ustricmp(p
->keyword
, L
"quotes")) {
257 if (*uadv(p
->keyword
) && *uadv(uadv(p
->keyword
))) {
258 ret
.lquote
= uadv(p
->keyword
);
259 ret
.rquote
= uadv(ret
.lquote
);
265 for (p
= source
; p
; p
= p
->next
) {
266 if (p
->type
== para_Config
) {
267 wchar_t *k
= p
->keyword
;
269 if (!ustrnicmp(k
, L
"xhtml-", 6))
270 k
++; /* treat `xhtml-' and `html-' the same */
272 if (!ustricmp(k
, L
"html-restrict-charset")) {
273 char *csname
= utoa_dup(uadv(k
), CS_ASCII
);
274 ret
.restrict_charset
= charset_from_localenc(csname
);
276 } else if (!ustricmp(k
, L
"html-output-charset")) {
277 char *csname
= utoa_dup(uadv(k
), CS_ASCII
);
278 ret
.output_charset
= charset_from_localenc(csname
);
280 } else if (!ustricmp(k
, L
"html-version")) {
281 wchar_t *vername
= uadv(k
);
282 static const struct {
286 {L
"html3.2", HTML_3_2
},
288 {L
"iso-html", ISO_HTML
},
289 {L
"xhtml1.0transitional", XHTML_1_0_TRANSITIONAL
},
290 {L
"xhtml1.0strict", XHTML_1_0_STRICT
}
294 for (i
= 0; i
< (int)lenof(versions
); i
++)
295 if (!ustricmp(versions
[i
].name
, vername
))
298 if (i
== lenof(versions
))
299 error(err_htmlver
, &p
->fpos
, vername
);
301 ret
.htmlver
= versions
[i
].ver
;
302 } else if (!ustricmp(k
, L
"html-single-filename")) {
303 sfree(ret
.single_filename
);
304 ret
.single_filename
= dupstr(adv(p
->origkeyword
));
305 } else if (!ustricmp(k
, L
"html-contents-filename")) {
306 sfree(ret
.contents_filename
);
307 ret
.contents_filename
= dupstr(adv(p
->origkeyword
));
308 } else if (!ustricmp(k
, L
"html-index-filename")) {
309 sfree(ret
.index_filename
);
310 ret
.index_filename
= dupstr(adv(p
->origkeyword
));
311 } else if (!ustricmp(k
, L
"html-template-filename")) {
312 sfree(ret
.template_filename
);
313 ret
.template_filename
= dupstr(adv(p
->origkeyword
));
314 } else if (!ustricmp(k
, L
"html-template-fragment")) {
315 sfree(ret
.template_fragment
);
316 ret
.template_fragment
= dupstr(adv(p
->origkeyword
));
317 } else if (!ustricmp(k
, L
"html-chapter-numeric")) {
318 ret
.achapter
.just_numbers
= utob(uadv(k
));
319 } else if (!ustricmp(k
, L
"html-chapter-suffix")) {
320 ret
.achapter
.number_suffix
= uadv(k
);
321 } else if (!ustricmp(k
, L
"html-leaf-level")) {
322 ret
.leaf_level
= utoi(uadv(k
));
323 } else if (!ustricmp(k
, L
"html-section-numeric")) {
324 wchar_t *q
= uadv(k
);
330 if (n
>= ret
.nasect
) {
332 ret
.asect
= sresize(ret
.asect
, n
+1, sectlevel
);
333 for (i
= ret
.nasect
; i
<= n
; i
++)
334 ret
.asect
[i
] = ret
.asect
[ret
.nasect
-1];
337 ret
.asect
[n
].just_numbers
= utob(q
);
338 } else if (!ustricmp(k
, L
"html-section-suffix")) {
339 wchar_t *q
= uadv(k
);
345 if (n
>= ret
.nasect
) {
347 ret
.asect
= sresize(ret
.asect
, n
+1, sectlevel
);
348 for (i
= ret
.nasect
; i
<= n
; i
++) {
349 ret
.asect
[i
] = ret
.asect
[ret
.nasect
-1];
353 ret
.asect
[n
].number_suffix
= q
;
354 } else if (!ustricmp(k
, L
"html-contents-depth") ||
355 !ustrnicmp(k
, L
"html-contents-depth-", 20)) {
357 * Relic of old implementation: this directive used
358 * to be written as \cfg{html-contents-depth-3}{2}
359 * rather than the usual Halibut convention of
360 * \cfg{html-contents-depth}{3}{2}. We therefore
363 wchar_t *q
= k
[19] ? k
+20 : uadv(k
);
369 if (n
>= ret
.ncdepths
) {
371 ret
.contents_depths
=
372 sresize(ret
.contents_depths
, n
+1, int);
373 for (i
= ret
.ncdepths
; i
<= n
; i
++) {
374 ret
.contents_depths
[i
] = i
+2;
378 ret
.contents_depths
[n
] = utoi(q
);
379 } else if (!ustricmp(k
, L
"html-head-end")) {
380 ret
.head_end
= adv(p
->origkeyword
);
381 } else if (!ustricmp(k
, L
"html-body-tag")) {
382 ret
.body_tag
= adv(p
->origkeyword
);
383 } else if (!ustricmp(k
, L
"html-body-start")) {
384 ret
.body_start
= adv(p
->origkeyword
);
385 } else if (!ustricmp(k
, L
"html-body-end")) {
386 ret
.body_end
= adv(p
->origkeyword
);
387 } else if (!ustricmp(k
, L
"html-address-start")) {
388 ret
.addr_start
= adv(p
->origkeyword
);
389 } else if (!ustricmp(k
, L
"html-address-end")) {
390 ret
.addr_end
= adv(p
->origkeyword
);
391 } else if (!ustricmp(k
, L
"html-navigation-attributes")) {
392 ret
.nav_attr
= adv(p
->origkeyword
);
393 } else if (!ustricmp(k
, L
"html-author")) {
394 ret
.author
= uadv(k
);
395 } else if (!ustricmp(k
, L
"html-description")) {
396 ret
.description
= uadv(k
);
397 } else if (!ustricmp(k
, L
"html-suppress-address")) {
398 ret
.address_section
= !utob(uadv(k
));
399 } else if (!ustricmp(k
, L
"html-versionid")) {
400 ret
.visible_version_id
= utob(uadv(k
));
401 } else if (!ustricmp(k
, L
"html-quotes")) {
402 if (*uadv(k
) && *uadv(uadv(k
))) {
403 ret
.lquote
= uadv(k
);
404 ret
.rquote
= uadv(ret
.lquote
);
406 } else if (!ustricmp(k
, L
"html-leaf-contains-contents")) {
407 ret
.leaf_contains_contents
= utob(uadv(k
));
408 } else if (!ustricmp(k
, L
"html-leaf-smallest-contents")) {
409 ret
.leaf_smallest_contents
= utoi(uadv(k
));
410 } else if (!ustricmp(k
, L
"html-index-text")) {
411 ret
.index_text
= uadv(k
);
412 } else if (!ustricmp(k
, L
"html-contents-text")) {
413 ret
.contents_text
= uadv(k
);
414 } else if (!ustricmp(k
, L
"html-preamble-text")) {
415 ret
.preamble_text
= uadv(k
);
416 } else if (!ustricmp(k
, L
"html-title-separator")) {
417 ret
.title_separator
= uadv(k
);
418 } else if (!ustricmp(k
, L
"html-nav-prev-text")) {
419 ret
.nav_prev_text
= uadv(k
);
420 } else if (!ustricmp(k
, L
"html-nav-next-text")) {
421 ret
.nav_next_text
= uadv(k
);
422 } else if (!ustricmp(k
, L
"html-nav-separator")) {
423 ret
.nav_separator
= uadv(k
);
424 } else if (!ustricmp(k
, L
"html-index-main-separator")) {
425 ret
.index_main_sep
= uadv(k
);
426 } else if (!ustricmp(k
, L
"html-index-multiple-separator")) {
427 ret
.index_multi_sep
= uadv(k
);
428 } else if (!ustricmp(k
, L
"html-pre-versionid")) {
429 ret
.pre_versionid
= uadv(k
);
430 } else if (!ustricmp(k
, L
"html-post-versionid")) {
431 ret
.post_versionid
= uadv(k
);
437 * Now process fallbacks on quote characters.
439 while (*uadv(ret
.rquote
) && *uadv(uadv(ret
.rquote
)) &&
440 (!cvt_ok(ret
.restrict_charset
, ret
.lquote
) ||
441 !cvt_ok(ret
.restrict_charset
, ret
.rquote
))) {
442 ret
.lquote
= uadv(ret
.rquote
);
443 ret
.rquote
= uadv(ret
.lquote
);
449 paragraph
*html_config_filename(char *filename
)
452 * If the user passes in a single filename as a parameter to
453 * the `--html' command-line option, then we should assume it
454 * to imply _two_ config directives:
455 * \cfg{html-single-filename}{whatever} and
456 * \cfg{html-leaf-level}{0}; the rationale being that the user
457 * wants their output _in that file_.
461 p
= cmdline_cfg_simple("html-single-filename", filename
, NULL
);
462 q
= cmdline_cfg_simple("html-leaf-level", "0", NULL
);
467 void html_backend(paragraph
*sourceform
, keywordlist
*keywords
,
468 indexdata
*idx
, void *unused
) {
471 htmlfilelist files
= { NULL
, NULL
, NULL
, NULL
, NULL
};
472 htmlsectlist sects
= { NULL
, NULL
}, nonsects
= { NULL
, NULL
};
476 conf
= html_configure(sourceform
);
479 * We're going to make heavy use of paragraphs' private data
480 * fields in the forthcoming code. Clear them first, so we can
481 * reliably tell whether we have auxiliary data for a
482 * particular paragraph.
484 for (p
= sourceform
; p
; p
= p
->next
)
485 p
->private_data
= NULL
;
487 files
.frags
= newtree234(html_fragment_compare
);
490 * Start by figuring out into which file each piece of the
491 * document should be put. We'll do this by inventing an
492 * `htmlsect' structure and stashing it in the private_data
493 * field of each section paragraph; we also need one additional
494 * htmlsect for the document index, which won't show up in the
495 * source form but needs to be consistently mentioned in
498 * While we're here, we'll also invent the HTML fragment name
502 htmlsect
*topsect
, *sect
;
505 topsect
= html_new_sect(§s
, NULL
);
507 topsect
->title
= NULL
;
508 topsect
->text
= sourceform
;
509 topsect
->contents_depth
= contents_depth(conf
, 0);
510 html_file_section(&conf
, &files
, topsect
, -1);
511 topsect
->fragment
= NULL
;
513 for (p
= sourceform
; p
; p
= p
->next
)
514 if (is_heading_type(p
->type
)) {
515 d
= heading_depth(p
);
517 if (p
->type
== para_Title
) {
522 sect
= html_new_sect(§s
, p
);
523 sect
->text
= p
->next
;
525 sect
->contents_depth
= contents_depth(conf
, d
+1) - (d
+1);
528 sect
->parent
= (htmlsect
*)p
->parent
->private_data
;
529 assert(sect
->parent
!= NULL
);
531 sect
->parent
= topsect
;
532 p
->private_data
= sect
;
534 html_file_section(&conf
, &files
, sect
, d
);
536 sect
->fragment
= html_format(p
, conf
.template_fragment
);
537 sect
->fragment
= html_sanitise_fragment(&files
, sect
->file
,
542 sect
= html_new_sect(§s
, NULL
);
545 sect
->parent
= topsect
;
546 html_file_section(&conf
, &files
, sect
, 0); /* peer of chapters */
547 sect
->fragment
= utoa_dup(conf
.index_text
, CS_ASCII
);
548 sect
->fragment
= html_sanitise_fragment(&files
, sect
->file
,
550 files
.index
= sect
->file
;
554 * Go through the keyword list and sort out fragment IDs for
555 * all the potentially referenced paragraphs which _aren't_
563 for (i
= 0; (kw
= index234(keywords
->keys
, i
)) != NULL
; i
++) {
564 paragraph
*q
, *p
= kw
->para
;
566 if (!is_heading_type(p
->type
)) {
570 * Find the paragraph's parent htmlsect, to
571 * determine which file it will end up in.
576 * Preamble paragraphs have no parent. So if we
577 * have a non-heading with no parent, it must
578 * be preamble, and therefore its parent
579 * htmlsect must be the preamble one.
582 sects
.head
->type
== TOP
);
585 parent
= (htmlsect
*)q
->private_data
;
588 * Now we can construct an htmlsect for this
589 * paragraph itself, taking care to put it in the
590 * list of non-sections rather than the list of
591 * sections (so that traverses of the `sects' list
592 * won't attempt to add it to the contents or
593 * anything weird like that).
595 sect
= html_new_sect(&nonsects
, p
);
596 sect
->file
= parent
->file
;
597 sect
->parent
= parent
;
598 p
->private_data
= sect
;
601 * Fragment IDs for these paragraphs will simply be
602 * `p' followed by an integer.
604 sect
->fragment
= snewn(40, char);
605 sprintf(sect
->fragment
, "p%d",
606 sect
->file
->last_fragment_number
++);
607 sect
->fragment
= html_sanitise_fragment(&files
, sect
->file
,
614 * Reset the fragment numbers in each file. I've just used them
615 * to generate `p' fragment IDs for non-section paragraphs
616 * (numbered list elements, bibliocited), and now I want to use
617 * them for `i' fragment IDs for index entries.
621 for (file
= files
.head
; file
; file
= file
->next
)
622 file
->last_fragment_number
= 0;
626 * Now sort out the index. This involves:
628 * - For each index term, we set up an htmlindex structure to
629 * store all the references to that term.
631 * - Then we make a pass over the actual document, finding
632 * every word_IndexRef; for each one, we actually figure out
633 * the HTML filename/fragment pair we will use to reference
634 * it, store that information in the private data field of
635 * the word_IndexRef itself (so we can recreate it when the
636 * time comes to output our HTML), and add a reference to it
637 * to the index term in question.
646 * Set up the htmlindex structures.
649 for (i
= 0; (entry
= index234(idx
->entries
, i
)) != NULL
; i
++) {
650 htmlindex
*hi
= snew(htmlindex
);
652 hi
->nrefs
= hi
->refsize
= 0;
655 entry
->backend_data
= hi
;
659 * Run over the document inventing fragments. Each fragment
660 * is of the form `i' followed by an integer.
662 lastsect
= sects
.head
; /* this is always the top section */
663 for (p
= sourceform
; p
; p
= p
->next
) {
664 if (is_heading_type(p
->type
) && p
->type
!= para_Title
)
665 lastsect
= (htmlsect
*)p
->private_data
;
667 for (w
= p
->words
; w
; w
= w
->next
)
668 if (w
->type
== word_IndexRef
) {
669 htmlindexref
*hr
= snew(htmlindexref
);
673 hr
->referenced
= hr
->generated
= FALSE
;
674 hr
->section
= lastsect
;
678 lastsect
->file
->last_fragment_number
++);
679 hr
->fragment
= dupstr(buf
);
681 html_sanitise_fragment(&files
, hr
->section
->file
,
684 w
->private_data
= hr
;
686 tag
= index_findtag(idx
, w
->text
);
690 for (i
= 0; i
< tag
->nrefs
; i
++) {
691 indexentry
*entry
= tag
->refs
[i
];
692 htmlindex
*hi
= (htmlindex
*)entry
->backend_data
;
694 if (hi
->nrefs
>= hi
->refsize
) {
696 hi
->refs
= sresize(hi
->refs
, hi
->refsize
, word
*);
699 hi
->refs
[hi
->nrefs
++] = w
;
706 * Now we're ready to write out the actual HTML files.
710 * - we open that file and write its header
711 * - we run down the list of sections
712 * - for each section directly contained within that file, we
713 * output the section text
714 * - for each section which is not in the file but which has a
715 * parent that is, we output a contents entry for the
716 * section if appropriate
717 * - finally, we output the file trailer and close the file.
726 for (f
= files
.head
; f
; f
= f
->next
) {
729 enum LISTTYPE
{ NOLIST
, UL
, OL
, DL
};
730 enum ITEMTYPE
{ NOITEM
, LI
, DT
, DD
};
731 struct stackelement
{
732 struct stackelement
*next
;
733 enum LISTTYPE listtype
;
734 enum ITEMTYPE itemtype
;
737 #define listname(lt) ( (lt)==UL ? "ul" : (lt)==OL ? "ol" : "dl" )
738 #define itemname(lt) ( (lt)==LI ? "li" : (lt)==DT ? "dt" : "dd" )
740 ho
.fp
= fopen(f
->filename
, "w");
741 ho
.charset
= conf
.output_charset
;
742 ho
.restrict_charset
= conf
.restrict_charset
;
743 ho
.cstate
= charset_init_state
;
744 ho
.ver
= conf
.htmlver
;
745 ho
.state
= HO_NEUTRAL
;
746 ho
.contents_level
= 0;
749 switch (conf
.htmlver
) {
751 fprintf(ho
.fp
, "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD "
752 "HTML 3.2 Final//EN\">\n");
755 fprintf(ho
.fp
, "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML"
756 " 4.01//EN\"\n\"http://www.w3.org/TR/html4/"
760 fprintf(ho
.fp
, "<!DOCTYPE HTML PUBLIC \"ISO/IEC "
761 "15445:2000//DTD HTML//EN\">\n");
763 case XHTML_1_0_TRANSITIONAL
:
764 fprintf(ho
.fp
, "<?xml version=\"1.0\" encoding=\"%s\"?>\n",
765 charset_to_mimeenc(conf
.output_charset
));
766 fprintf(ho
.fp
, "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML"
767 " 1.0 Transitional//EN\"\n\"http://www.w3.org/TR/"
768 "xhtml1/DTD/xhtml1-transitional.dtd\">\n");
770 case XHTML_1_0_STRICT
:
771 fprintf(ho
.fp
, "<?xml version=\"1.0\" encoding=\"%s\"?>\n",
772 charset_to_mimeenc(conf
.output_charset
));
773 fprintf(ho
.fp
, "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML"
774 " 1.0 Strict//EN\"\n\"http://www.w3.org/TR/xhtml1/"
775 "DTD/xhtml1-strict.dtd\">\n");
779 element_open(&ho
, "html");
780 if (is_xhtml(conf
.htmlver
)) {
781 element_attr(&ho
, "xmlns", "http://www.w3.org/1999/xhtml");
785 element_open(&ho
, "head");
788 element_empty(&ho
, "meta");
789 element_attr(&ho
, "http-equiv", "content-type");
792 sprintf(buf
, "text/html; charset=%.150s",
793 charset_to_mimeenc(conf
.output_charset
));
794 element_attr(&ho
, "content", buf
);
799 element_empty(&ho
, "meta");
800 element_attr(&ho
, "name", "author");
801 element_attr_w(&ho
, "content", conf
.author
);
805 if (conf
.description
) {
806 element_empty(&ho
, "meta");
807 element_attr(&ho
, "name", "description");
808 element_attr_w(&ho
, "content", conf
.description
);
812 element_open(&ho
, "title");
813 if (f
->first
&& f
->first
->title
) {
814 html_words(&ho
, f
->first
->title
->words
, NOTHING
,
818 if (f
->last
!= f
->first
&& f
->last
->title
) {
819 html_text(&ho
, conf
.title_separator
);
820 html_words(&ho
, f
->last
->title
->words
, NOTHING
,
824 element_close(&ho
, "title");
828 html_raw(&ho
, conf
.head_end
);
830 element_close(&ho
, "head");
834 html_raw(&ho
, conf
.body_tag
);
836 element_open(&ho
, "body");
840 html_raw(&ho
, conf
.body_start
);
843 * Write out a nav bar. Special case: we don't do this
844 * if there is only one file.
846 if (files
.head
!= files
.tail
) {
847 element_open(&ho
, "p");
849 html_raw_as_attr(&ho
, conf
.nav_attr
);
852 element_open(&ho
, "a");
853 element_attr(&ho
, "href", prevf
->filename
);
855 html_text(&ho
, conf
.nav_prev_text
);
857 element_close(&ho
, "a");
859 html_text(&ho
, conf
.nav_separator
);
861 if (f
!= files
.head
) {
862 element_open(&ho
, "a");
863 element_attr(&ho
, "href", files
.head
->filename
);
865 html_text(&ho
, conf
.contents_text
);
867 element_close(&ho
, "a");
869 html_text(&ho
, conf
.nav_separator
);
871 if (f
!= files
.index
) {
872 element_open(&ho
, "a");
873 element_attr(&ho
, "href", files
.index
->filename
);
875 html_text(&ho
, conf
.index_text
);
876 if (f
!= files
.index
)
877 element_close(&ho
, "a");
879 html_text(&ho
, conf
.nav_separator
);
882 element_open(&ho
, "a");
883 element_attr(&ho
, "href", f
->next
->filename
);
885 html_text(&ho
, conf
.nav_next_text
);
887 element_close(&ho
, "a");
889 element_close(&ho
, "p");
895 * Write out a prefix TOC for the file.
897 * We start by going through the section list and
898 * collecting the sections which need to be added to
899 * the contents. On the way, we also test to see if
900 * this file is a leaf file (defined as one which
901 * contains all descendants of any section it
902 * contains), because this will play a part in our
903 * decision on whether or not to _output_ the TOC.
905 * Special case: we absolutely do not do this if we're
906 * in single-file mode.
908 if (files
.head
!= files
.tail
) {
909 int ntoc
= 0, tocsize
= 0;
910 htmlsect
**toc
= NULL
;
913 for (s
= sects
.head
; s
; s
= s
->next
) {
918 * Search up from this section until we find
919 * the highest-level one which belongs in this
924 for (ac
= s
; ac
; ac
= ac
->parent
) {
932 if (s
->file
!= f
&& a
!= NULL
)
936 if (adepth
<= a
->contents_depth
) {
937 if (ntoc
>= tocsize
) {
939 toc
= sresize(toc
, tocsize
, htmlsect
*);
946 if (leaf
&& conf
.leaf_contains_contents
&&
947 ntoc
>= conf
.leaf_smallest_contents
) {
950 for (i
= 0; i
< ntoc
; i
++) {
951 htmlsect
*s
= toc
[i
];
952 int hlevel
= (s
->type
== TOP ?
-1 :
953 s
->type
== INDEX ?
0 :
954 heading_depth(s
->title
))
955 - f
->min_heading_depth
+ 1;
958 html_contents_entry(&ho
, hlevel
, s
,
961 html_contents_entry(&ho
, 0, NULL
, f
, keywords
, &conf
);
966 * Now go through the document and output some real
970 for (s
= sects
.head
; s
; s
= s
->next
) {
973 * This section belongs in this file.
984 * Search up from this section until we find
985 * the highest-level one which belongs in this
990 for (ac
= s
; ac
; ac
= ac
->parent
) {
1000 * This section does not belong in this
1001 * file, but an ancestor of it does. Write
1002 * out a contents table entry, if the depth
1003 * doesn't exceed the maximum contents
1004 * depth for the ancestor section.
1006 if (adepth
<= a
->contents_depth
) {
1007 html_contents_entry(&ho
, adepth
, s
,
1008 f
, keywords
, &conf
);
1017 html_contents_entry(&ho
, 0, NULL
, f
, keywords
, &conf
);
1020 * Display the section heading.
1023 hlevel
= (s
->type
== TOP ?
-1 :
1024 s
->type
== INDEX ?
0 :
1025 heading_depth(s
->title
))
1026 - f
->min_heading_depth
+ 1;
1027 assert(hlevel
>= 1);
1028 /* HTML headings only go up to <h6> */
1032 htag
[1] = '0' + hlevel
;
1034 element_open(&ho
, htag
);
1037 * Provide anchor for cross-links to target.
1039 * (Also we'll have to do this separately in
1040 * other paragraph types - NumberedList and
1044 html_fragment(&ho
, s
->fragment
);
1046 html_section_title(&ho
, s
, f
, keywords
, &conf
, TRUE
);
1048 element_close(&ho
, htag
);
1051 * Now display the section text.
1054 stackhead
= snew(struct stackelement
);
1055 stackhead
->next
= NULL
;
1056 stackhead
->listtype
= NOLIST
;
1057 stackhead
->itemtype
= NOITEM
;
1059 for (p
= s
->text
;; p
= p
->next
) {
1060 enum LISTTYPE listtype
;
1061 struct stackelement
*se
;
1064 * Preliminary switch to figure out what
1065 * sort of list we expect to be inside at
1068 * Since p may still be NULL at this point,
1069 * I invent a harmless paragraph type for
1072 switch (p ? p
->type
: para_Normal
) {
1075 case para_Copyright
:
1076 case para_BiblioCited
:
1078 case para_QuotePush
:
1082 case para_UnnumberedChapter
:
1093 case para_NumberedList
:
1097 case para_DescribedThing
:
1098 case para_Description
:
1102 case para_LcontPush
:
1103 se
= snew(struct stackelement
);
1104 se
->next
= stackhead
;
1105 se
->listtype
= NOLIST
;
1106 se
->itemtype
= NOITEM
;
1110 default: /* some totally non-printing para */
1117 * Terminate the most recent list item, if
1118 * any. (We left this until after
1119 * processing LcontPush, since in that case
1120 * the list item won't want to be
1121 * terminated until after the corresponding
1124 if (stackhead
->itemtype
!= NOITEM
) {
1125 element_close(&ho
, itemname(stackhead
->itemtype
));
1128 stackhead
->itemtype
= NOITEM
;
1131 * Terminate the current list, if it's not
1132 * the one we want to be in.
1134 if (listtype
!= stackhead
->listtype
&&
1135 stackhead
->listtype
!= NOLIST
) {
1136 element_close(&ho
, listname(stackhead
->listtype
));
1141 * Leave the loop if our time has come.
1143 if (!p
|| (is_heading_type(p
->type
) &&
1144 p
->type
!= para_Title
))
1145 break; /* end of section text */
1148 * Start a fresh list if necessary.
1150 if (listtype
!= stackhead
->listtype
&&
1152 element_open(&ho
, listname(listtype
));
1154 stackhead
->listtype
= listtype
;
1158 element_empty(&ho
, "hr");
1161 html_codepara(&ho
, p
->words
);
1164 case para_Copyright
:
1165 element_open(&ho
, "p");
1167 html_words(&ho
, p
->words
, ALL
,
1168 f
, keywords
, &conf
);
1170 element_close(&ho
, "p");
1172 case para_BiblioCited
:
1173 element_open(&ho
, "p");
1174 if (p
->private_data
) {
1175 htmlsect
*s
= (htmlsect
*)p
->private_data
;
1176 html_fragment(&ho
, s
->fragment
);
1179 html_words(&ho
, p
->kwtext
, ALL
,
1180 f
, keywords
, &conf
);
1181 html_text(&ho
, L
" ");
1182 html_words(&ho
, p
->words
, ALL
,
1183 f
, keywords
, &conf
);
1185 element_close(&ho
, "p");
1188 case para_NumberedList
:
1189 element_open(&ho
, "li");
1190 if (p
->private_data
) {
1191 htmlsect
*s
= (htmlsect
*)p
->private_data
;
1192 html_fragment(&ho
, s
->fragment
);
1195 stackhead
->itemtype
= LI
;
1196 html_words(&ho
, p
->words
, ALL
,
1197 f
, keywords
, &conf
);
1199 case para_DescribedThing
:
1200 element_open(&ho
, "dt");
1202 stackhead
->itemtype
= DT
;
1203 html_words(&ho
, p
->words
, ALL
,
1204 f
, keywords
, &conf
);
1206 case para_Description
:
1207 element_open(&ho
, "dd");
1209 stackhead
->itemtype
= DD
;
1210 html_words(&ho
, p
->words
, ALL
,
1211 f
, keywords
, &conf
);
1214 case para_QuotePush
:
1215 element_open(&ho
, "blockquote");
1218 element_close(&ho
, "blockquote");
1223 stackhead
= stackhead
->next
;
1230 assert(stackhead
&& !stackhead
->next
);
1234 if (s
->type
== INDEX
) {
1239 * This section is the index. I'll just
1240 * render it as a single paragraph, with a
1241 * colon between the index term and the
1242 * references, and <br> in between each
1245 element_open(&ho
, "p");
1247 for (i
= 0; (entry
=
1248 index234(idx
->entries
, i
)) != NULL
; i
++) {
1249 htmlindex
*hi
= (htmlindex
*)entry
->backend_data
;
1253 element_empty(&ho
, "br");
1256 html_words(&ho
, entry
->text
, MARKUP
|LINKS
,
1257 f
, keywords
, &conf
);
1259 html_text(&ho
, conf
.index_main_sep
);
1261 for (j
= 0; j
< hi
->nrefs
; j
++) {
1263 (htmlindexref
*)hi
->refs
[j
]->private_data
;
1264 paragraph
*p
= hr
->section
->title
;
1267 html_text(&ho
, conf
.index_multi_sep
);
1269 html_href(&ho
, f
, hr
->section
->file
,
1271 hr
->referenced
= TRUE
;
1273 html_words(&ho
, p
->kwtext
, MARKUP
|LINKS
,
1274 f
, keywords
, &conf
);
1275 else if (p
&& p
->words
)
1276 html_words(&ho
, p
->words
, MARKUP
|LINKS
,
1277 f
, keywords
, &conf
);
1280 * If there is no title at all,
1281 * this must be because our
1282 * target section is the
1283 * preamble section and there
1284 * is no title. So we use the
1287 html_text(&ho
, conf
.preamble_text
);
1289 element_close(&ho
, "a");
1292 element_close(&ho
, "p");
1297 html_contents_entry(&ho
, 0, NULL
, f
, keywords
, &conf
);
1304 int done_version_ids
= FALSE
;
1306 element_empty(&ho
, "hr");
1309 html_raw(&ho
, conf
.body_end
);
1311 if (conf
.address_section
) {
1312 int started
= FALSE
;
1313 if (conf
.htmlver
== ISO_HTML
) {
1315 * The ISO-HTML validator complains if
1316 * there isn't a <div> tag surrounding the
1317 * <address> tag. I'm uncertain of why this
1318 * should be - there appears to be no
1319 * mention of this in the ISO-HTML spec,
1320 * suggesting that it doesn't represent a
1321 * change from HTML 4, but nonetheless the
1322 * HTML 4 validator doesn't seem to mind.
1324 element_open(&ho
, "div");
1326 element_open(&ho
, "address");
1327 if (conf
.addr_start
) {
1328 html_raw(&ho
, conf
.addr_start
);
1332 if (conf
.visible_version_id
) {
1333 for (p
= sourceform
; p
; p
= p
->next
)
1334 if (p
->type
== para_VersionID
) {
1336 element_empty(&ho
, "br");
1338 html_text(&ho
, conf
.pre_versionid
);
1339 html_words(&ho
, p
->words
, NOTHING
,
1340 f
, keywords
, &conf
);
1341 html_text(&ho
, conf
.post_versionid
);
1344 done_version_ids
= TRUE
;
1346 if (conf
.addr_end
) {
1348 element_empty(&ho
, "br");
1349 html_raw(&ho
, conf
.addr_end
);
1351 element_close(&ho
, "address");
1352 if (conf
.htmlver
== ISO_HTML
)
1353 element_close(&ho
, "div");
1356 if (!done_version_ids
) {
1358 * If the user didn't want the version IDs
1359 * visible, I think we still have a duty to put
1360 * them in an HTML comment.
1362 int started
= FALSE
;
1363 for (p
= sourceform
; p
; p
= p
->next
)
1364 if (p
->type
== para_VersionID
) {
1366 html_raw(&ho
, "<!-- version IDs:\n");
1369 html_words(&ho
, p
->words
, NOTHING
,
1370 f
, keywords
, &conf
);
1374 html_raw(&ho
, "-->\n");
1378 element_close(&ho
, "body");
1380 element_close(&ho
, "html");
1387 * Go through and check that no index fragments were referenced
1388 * without being generated, or indeed vice versa.
1390 * (When I actually get round to freeing everything, this can
1391 * probably be the freeing loop as well.)
1393 for (p
= sourceform
; p
; p
= p
->next
) {
1395 for (w
= p
->words
; w
; w
= w
->next
)
1396 if (w
->type
== word_IndexRef
) {
1397 htmlindexref
*hr
= (htmlindexref
*)w
->private_data
;
1399 assert(!hr
->referenced
== !hr
->generated
);
1404 * FIXME: Free all the working data.
1408 static void html_file_section(htmlconfig
*cfg
, htmlfilelist
*files
,
1409 htmlsect
*sect
, int depth
)
1415 * `depth' is derived from the heading_depth() macro at the top
1416 * of this file, which counts title as -1, chapter as 0,
1417 * heading as 1 and subsection as 2. However, the semantics of
1418 * cfg->leaf_level are defined to count chapter as 1, heading
1419 * as 2 etc. So first I increment depth :-(
1423 if (cfg
->leaf_level
== 0) {
1425 * leaf_level==0 is a special case, in which everything is
1426 * put into a single file.
1429 files
->single
= html_new_file(files
, cfg
->single_filename
);
1431 file
= files
->single
;
1434 * If the depth of this section is at or above leaf_level,
1435 * we invent a fresh file and put this section at its head.
1436 * Otherwise, we put it in the same file as its parent
1439 if (ldepth
> cfg
->leaf_level
) {
1441 * We know that sect->parent cannot be NULL. The only
1442 * circumstance in which it can be is if sect is at
1443 * chapter or appendix level, i.e. ldepth==1; and if
1444 * that's the case, then we cannot have entered this
1445 * branch unless cfg->leaf_level==0, in which case we
1446 * would be in the single-file case above and not here
1449 assert(sect
->parent
);
1451 file
= sect
->parent
->file
;
1453 if (sect
->type
== TOP
) {
1454 file
= html_new_file(files
, cfg
->contents_filename
);
1455 } else if (sect
->type
== INDEX
) {
1456 file
= html_new_file(files
, cfg
->index_filename
);
1460 assert(ldepth
> 0 && sect
->title
);
1461 title
= html_format(sect
->title
, cfg
->template_filename
);
1462 file
= html_new_file(files
, title
);
1470 if (file
->min_heading_depth
> depth
) {
1472 * This heading is at a higher level than any heading we
1473 * have so far placed in this file; so we set the `first'
1476 file
->min_heading_depth
= depth
;
1480 if (file
->min_heading_depth
== depth
)
1484 static htmlfile
*html_new_file(htmlfilelist
*list
, char *filename
)
1486 htmlfile
*ret
= snew(htmlfile
);
1490 list
->tail
->next
= ret
;
1495 ret
->filename
= dupstr(filename
);
1496 ret
->last_fragment_number
= 0;
1497 ret
->min_heading_depth
= INT_MAX
;
1498 ret
->first
= ret
->last
= NULL
;
1503 static htmlsect
*html_new_sect(htmlsectlist
*list
, paragraph
*title
)
1505 htmlsect
*ret
= snew(htmlsect
);
1509 list
->tail
->next
= ret
;
1522 static void html_words(htmloutput
*ho
, word
*words
, int flags
,
1523 htmlfile
*file
, keywordlist
*keywords
, htmlconfig
*cfg
)
1529 for (w
= words
; w
; w
= w
->next
) switch (w
->type
) {
1530 case word_HyperLink
:
1531 if (flags
& LINKS
) {
1532 element_open(ho
, "a");
1533 c
= utoa_dup(w
->text
, CS_ASCII
);
1534 element_attr(ho
, "href", c
);
1538 case word_UpperXref
:
1539 case word_LowerXref
:
1540 if (flags
& LINKS
) {
1541 keyword
*kwl
= kw_lookup(keywords
, w
->text
);
1542 paragraph
*p
= kwl
->para
;
1543 htmlsect
*s
= (htmlsect
*)p
->private_data
;
1547 html_href(ho
, file
, s
->file
, s
->fragment
);
1553 element_close(ho
, "a");
1556 if (flags
& INDEXENTS
) {
1557 htmlindexref
*hr
= (htmlindexref
*)w
->private_data
;
1558 html_fragment(ho
, hr
->fragment
);
1559 hr
->generated
= TRUE
;
1566 case word_WhiteSpace
:
1567 case word_EmphSpace
:
1568 case word_CodeSpace
:
1569 case word_WkCodeSpace
:
1571 case word_EmphQuote
:
1572 case word_CodeQuote
:
1573 case word_WkCodeQuote
:
1574 style
= towordstyle(w
->type
);
1575 type
= removeattr(w
->type
);
1576 if (style
== word_Emph
&&
1577 (attraux(w
->aux
) == attr_First
||
1578 attraux(w
->aux
) == attr_Only
) &&
1580 element_open(ho
, "em");
1581 else if ((style
== word_Code
|| style
== word_WeakCode
) &&
1582 (attraux(w
->aux
) == attr_First
||
1583 attraux(w
->aux
) == attr_Only
) &&
1585 element_open(ho
, "code");
1587 if (type
== word_WhiteSpace
)
1588 html_text(ho
, L
" ");
1589 else if (type
== word_Quote
) {
1590 if (quoteaux(w
->aux
) == quote_Open
)
1591 html_text(ho
, cfg
->lquote
);
1593 html_text(ho
, cfg
->rquote
);
1595 if (!w
->alt
|| cvt_ok(ho
->restrict_charset
, w
->text
))
1596 html_text_nbsp(ho
, w
->text
);
1598 html_words(ho
, w
->alt
, flags
, file
, keywords
, cfg
);
1601 if (style
== word_Emph
&&
1602 (attraux(w
->aux
) == attr_Last
||
1603 attraux(w
->aux
) == attr_Only
) &&
1605 element_close(ho
, "em");
1606 else if ((style
== word_Code
|| style
== word_WeakCode
) &&
1607 (attraux(w
->aux
) == attr_Last
||
1608 attraux(w
->aux
) == attr_Only
) &&
1610 element_close(ho
, "code");
1616 static void html_codepara(htmloutput
*ho
, word
*words
)
1618 element_open(ho
, "pre");
1619 element_open(ho
, "code");
1620 for (; words
; words
= words
->next
) if (words
->type
== word_WeakCode
) {
1625 if (words
->next
&& words
->next
->type
== word_Emph
) {
1626 e
= words
->next
->text
;
1627 words
= words
->next
;
1631 while (e
&& *e
&& *t
) {
1635 for (n
= 0; t
[n
] && e
[n
] && e
[n
] == ec
; n
++);
1643 element_open(ho
, open_tag
);
1645 html_text_limit(ho
, t
, n
);
1648 element_close(ho
, open_tag
);
1656 element_close(ho
, "code");
1657 element_close(ho
, "pre");
1660 static void html_charset_cleanup(htmloutput
*ho
)
1665 bytes
= charset_from_unicode(NULL
, NULL
, outbuf
, lenof(outbuf
),
1666 ho
->charset
, &ho
->cstate
, NULL
);
1668 fwrite(outbuf
, 1, bytes
, ho
->fp
);
1671 static void return_mostly_to_neutral(htmloutput
*ho
)
1673 if (ho
->state
== HO_IN_EMPTY_TAG
&& is_xhtml(ho
->ver
)) {
1674 fprintf(ho
->fp
, " />");
1675 } else if (ho
->state
== HO_IN_EMPTY_TAG
|| ho
->state
== HO_IN_TAG
) {
1676 fprintf(ho
->fp
, ">");
1679 ho
->state
= HO_NEUTRAL
;
1682 static void return_to_neutral(htmloutput
*ho
)
1684 if (ho
->state
== HO_IN_TEXT
) {
1685 html_charset_cleanup(ho
);
1688 return_mostly_to_neutral(ho
);
1691 static void element_open(htmloutput
*ho
, char const *name
)
1693 return_to_neutral(ho
);
1694 fprintf(ho
->fp
, "<%s", name
);
1695 ho
->state
= HO_IN_TAG
;
1698 static void element_close(htmloutput
*ho
, char const *name
)
1700 return_to_neutral(ho
);
1701 fprintf(ho
->fp
, "</%s>", name
);
1702 ho
->state
= HO_NEUTRAL
;
1705 static void element_empty(htmloutput
*ho
, char const *name
)
1707 return_to_neutral(ho
);
1708 fprintf(ho
->fp
, "<%s", name
);
1709 ho
->state
= HO_IN_EMPTY_TAG
;
1712 static void html_nl(htmloutput
*ho
)
1714 return_to_neutral(ho
);
1715 fputc('\n', ho
->fp
);
1718 static void html_raw(htmloutput
*ho
, char *text
)
1720 return_to_neutral(ho
);
1721 fputs(text
, ho
->fp
);
1724 static void html_raw_as_attr(htmloutput
*ho
, char *text
)
1726 assert(ho
->state
== HO_IN_TAG
|| ho
->state
== HO_IN_EMPTY_TAG
);
1728 fputs(text
, ho
->fp
);
1731 static void element_attr(htmloutput
*ho
, char const *name
, char const *value
)
1733 html_charset_cleanup(ho
);
1734 assert(ho
->state
== HO_IN_TAG
|| ho
->state
== HO_IN_EMPTY_TAG
);
1735 fprintf(ho
->fp
, " %s=\"%s\"", name
, value
);
1738 static void element_attr_w(htmloutput
*ho
, char const *name
,
1739 wchar_t const *value
)
1741 html_charset_cleanup(ho
);
1742 fprintf(ho
->fp
, " %s=\"", name
);
1743 html_text_limit_internal(ho
, value
, 0, TRUE
, FALSE
);
1744 html_charset_cleanup(ho
);
1748 static void html_text(htmloutput
*ho
, wchar_t const *text
)
1750 return_mostly_to_neutral(ho
);
1751 html_text_limit_internal(ho
, text
, 0, FALSE
, FALSE
);
1754 static void html_text_nbsp(htmloutput
*ho
, wchar_t const *text
)
1756 return_mostly_to_neutral(ho
);
1757 html_text_limit_internal(ho
, text
, 0, FALSE
, TRUE
);
1760 static void html_text_limit(htmloutput
*ho
, wchar_t const *text
, int maxlen
)
1762 return_mostly_to_neutral(ho
);
1763 html_text_limit_internal(ho
, text
, maxlen
, FALSE
, FALSE
);
1766 static void html_text_limit_internal(htmloutput
*ho
, wchar_t const *text
,
1767 int maxlen
, int quote_quotes
, int nbsp
)
1769 int textlen
= ustrlen(text
);
1773 if (maxlen
> 0 && textlen
> maxlen
)
1776 while (textlen
> 0) {
1777 /* Scan ahead for characters we really can't display in HTML. */
1778 int lenbefore
, lenafter
;
1779 for (lenbefore
= 0; lenbefore
< textlen
; lenbefore
++)
1780 if (text
[lenbefore
] == L
'<' ||
1781 text
[lenbefore
] == L
'>' ||
1782 text
[lenbefore
] == L
'&' ||
1783 (text
[lenbefore
] == L
'"' && quote_quotes
) ||
1784 (text
[lenbefore
] == L
' ' && nbsp
))
1786 lenafter
= lenbefore
;
1787 bytes
= charset_from_unicode(&text
, &lenafter
, outbuf
, lenof(outbuf
),
1788 ho
->charset
, &ho
->cstate
, &err
);
1789 textlen
-= (lenbefore
- lenafter
);
1791 fwrite(outbuf
, 1, bytes
, ho
->fp
);
1794 * We have encountered a character that cannot be
1795 * displayed in the selected output charset. Therefore,
1796 * we use an HTML numeric entity reference.
1798 assert(textlen
> 0);
1799 fprintf(ho
->fp
, "&#%ld;", (long int)*text
);
1801 } else if (lenafter
== 0 && textlen
> 0) {
1803 * We have encountered a character which is special to
1807 fprintf(ho
->fp
, "<");
1808 else if (*text
== L
'>')
1809 fprintf(ho
->fp
, ">");
1810 else if (*text
== L
'&')
1811 fprintf(ho
->fp
, "&");
1812 else if (*text
== L
'"')
1813 fprintf(ho
->fp
, """);
1814 else if (*text
== L
' ') {
1816 fprintf(ho
->fp
, " ");
1818 assert(!"Can't happen");
1824 static void cleanup(htmloutput
*ho
)
1826 return_to_neutral(ho
);
1830 static void html_href(htmloutput
*ho
, htmlfile
*thisfile
,
1831 htmlfile
*targetfile
, char *targetfrag
)
1833 rdstringc rs
= { 0, 0, NULL
};
1836 if (targetfile
!= thisfile
)
1837 rdaddsc(&rs
, targetfile
->filename
);
1840 rdaddsc(&rs
, targetfrag
);
1844 element_open(ho
, "a");
1845 element_attr(ho
, "href", url
);
1849 static void html_fragment(htmloutput
*ho
, char const *fragment
)
1851 element_open(ho
, "a");
1852 element_attr(ho
, "name", fragment
);
1853 if (is_xhtml(ho
->ver
))
1854 element_attr(ho
, "id", fragment
);
1855 element_close(ho
, "a");
1858 static char *html_format(paragraph
*p
, char *template_string
)
1862 wchar_t *ws
, wsbuf
[2];
1863 rdstringc rs
= { 0, 0, NULL
};
1865 t
= template_string
;
1867 if (*t
== '%' && t
[1]) {
1881 if (p
->kwtext
&& fmt
== 'n')
1883 else if (p
->kwtext2
&& fmt
== 'b') {
1885 * HTML fragment names must start with a letter, so
1886 * simply `1.2.3' is not adequate. In this case I'm
1887 * going to cheat slightly by prepending the first
1888 * character of the first word of kwtext, so that
1889 * we get `C1' for chapter 1, `S2.3' for section
1892 if (p
->kwtext
&& p
->kwtext
->text
[0]) {
1895 wsbuf
[0] = p
->kwtext
->text
[0];
1898 } else if (p
->keyword
&& *p
->keyword
&& fmt
== 'k')
1904 c
= utoa_dup(ws
, CS_ASCII
);
1910 if (removeattr(w
->type
) == word_Normal
) {
1911 c
= utoa_dup(w
->text
, CS_ASCII
);
1922 return rdtrimc(&rs
);
1925 static char *html_sanitise_fragment(htmlfilelist
*files
, htmlfile
*file
,
1929 * The HTML 4 spec's strictest definition of fragment names (<a
1930 * name> and "id" attributes) says that they `must begin with a
1931 * letter and may be followed by any number of letters, digits,
1932 * hyphens, underscores, colons, and periods'.
1934 * So here we unceremoniously rip out any characters not
1935 * conforming to this limitation.
1937 char *p
= text
, *q
= text
;
1939 while (*p
&& !((*p
>='A' && *p
<='Z') || (*p
>='a' && *p
<='z')))
1941 if ((*q
++ = *p
++) != '\0') {
1943 if ((*p
>='A' && *p
<='Z') ||
1944 (*p
>='a' && *p
<='z') ||
1945 (*p
>='0' && *p
<='9') ||
1946 *p
=='-' || *p
=='_' || *p
==':' || *p
=='.')
1955 * Now we check for clashes with other fragment names, and
1956 * adjust this one if necessary by appending a hyphen followed
1960 htmlfragment
*frag
= snew(htmlfragment
);
1961 int len
= 0; /* >0 indicates we have resized */
1965 frag
->fragment
= text
;
1967 while (add234(files
->frags
, frag
) != frag
) {
1970 frag
->fragment
= text
= sresize(text
, len
+20, char);
1973 sprintf(text
+ len
, "-%d", ++suffix
);
1980 static void html_contents_entry(htmloutput
*ho
, int depth
, htmlsect
*s
,
1981 htmlfile
*thisfile
, keywordlist
*keywords
,
1984 while (ho
->contents_level
> depth
) {
1985 element_close(ho
, "ul");
1986 ho
->contents_level
--;
1989 while (ho
->contents_level
< depth
) {
1990 element_open(ho
, "ul");
1991 ho
->contents_level
++;
1997 element_open(ho
, "li");
1998 html_href(ho
, thisfile
, s
->file
, s
->fragment
);
1999 html_section_title(ho
, s
, thisfile
, keywords
, cfg
, FALSE
);
2000 element_close(ho
, "a");
2001 element_close(ho
, "li");
2004 static void html_section_title(htmloutput
*ho
, htmlsect
*s
, htmlfile
*thisfile
,
2005 keywordlist
*keywords
, htmlconfig
*cfg
,
2011 int depth
= heading_depth(s
->title
);
2015 else if (depth
== 0)
2016 sl
= &cfg
->achapter
;
2017 else if (depth
<= cfg
->nasect
)
2018 sl
= &cfg
->asect
[depth
-1];
2020 sl
= &cfg
->asect
[cfg
->nasect
-1];
2024 else if (sl
->just_numbers
)
2025 number
= s
->title
->kwtext2
;
2027 number
= s
->title
->kwtext
;
2030 html_words(ho
, number
, MARKUP
,
2031 thisfile
, keywords
, cfg
);
2032 html_text(ho
, sl
->number_suffix
);
2035 html_words(ho
, s
->title
->words
, real ? ALL
: MARKUP
,
2036 thisfile
, keywords
, cfg
);
2038 assert(s
->type
!= NORMAL
);
2040 * If we're printing the full document title for _real_ and
2041 * there isn't one, we don't want to print `Preamble' at
2042 * the top of what ought to just be some text. If we need
2043 * it in any other context such as TOCs, we need to print
2046 if (s
->type
== TOP
&& !real
)
2047 html_text(ho
, cfg
->preamble_text
);
2048 else if (s
->type
== INDEX
)
2049 html_text(ho
, cfg
->index_text
);