78c73085 |
1 | /* |
2 | * HTML backend for Halibut |
3 | */ |
4 | |
5 | /* |
6 | * TODO: |
7 | * |
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.) |
13 | * |
14 | * - new configurability: |
15 | * * a few new things explicitly labelled as `FIXME: |
16 | * configurable' or similar. |
78c73085 |
17 | * * Some means of specifying the distinction between |
18 | * restrict-charset and output-charset. It seems to me that |
19 | * `html-charset' is output-charset, and that |
20 | * restrict-charset usually wants to be either output-charset |
21 | * or UTF-8 (the latter indicating that any Unicode character |
22 | * is fair game and it will be specified using &#foo; if it |
23 | * isn't in output-charset). However, since XHTML defaults to |
24 | * UTF-8 and it's fiddly to tell it otherwise, it's just |
25 | * possible that some user may need to set restrict-charset |
26 | * to their charset of choice while leaving _output_-charset |
27 | * at UTF-8. Figure out some configuration, and apply it. |
78c73085 |
28 | * |
3e82de8f |
29 | * - nonbreaking spaces. |
30 | * |
31 | * - free up all the data we have allocated while running this |
32 | * backend. |
78c73085 |
33 | */ |
34 | |
35 | #include <stdio.h> |
36 | #include <stdlib.h> |
37 | #include <assert.h> |
38 | #include <limits.h> |
39 | #include "halibut.h" |
40 | |
41 | #define is_heading_type(type) ( (type) == para_Title || \ |
42 | (type) == para_Chapter || \ |
43 | (type) == para_Appendix || \ |
44 | (type) == para_UnnumberedChapter || \ |
45 | (type) == para_Heading || \ |
46 | (type) == para_Subsect) |
47 | |
48 | #define heading_depth(p) ( (p)->type == para_Subsect ? (p)->aux + 1 : \ |
49 | (p)->type == para_Heading ? 1 : \ |
50 | (p)->type == para_Title ? -1 : 0 ) |
51 | |
52 | typedef struct { |
53 | int just_numbers; |
54 | wchar_t *number_suffix; |
55 | } sectlevel; |
56 | |
57 | typedef struct { |
58 | int nasect; |
59 | sectlevel achapter, *asect; |
60 | int *contents_depths; /* 0=main, 1=chapter, 2=sect etc */ |
61 | int ncdepths; |
62 | int address_section, visible_version_id; |
63 | int leaf_contains_contents, leaf_smallest_contents; |
64 | char *contents_filename; |
65 | char *index_filename; |
66 | char *template_filename; |
67 | char *single_filename; |
68 | char *template_fragment; |
69 | char *head_end, *body_start, *body_end, *addr_start, *addr_end; |
70 | char *body_tag, *nav_attr; |
71 | wchar_t *author, *description; |
72 | int restrict_charset, output_charset; |
73 | enum { |
27bdc5ab |
74 | HTML_3_2, HTML_4, ISO_HTML, |
78c73085 |
75 | XHTML_1_0_TRANSITIONAL, XHTML_1_0_STRICT |
76 | } htmlver; |
77 | wchar_t *lquote, *rquote; |
78 | int leaf_level; |
79 | } htmlconfig; |
80 | |
81 | #define contents_depth(conf, level) \ |
82 | ( (conf).ncdepths > (level) ? (conf).contents_depths[level] : (level)+2 ) |
83 | |
84 | #define is_xhtml(ver) ((ver) >= XHTML_1_0_TRANSITIONAL) |
85 | |
86 | typedef struct htmlfile htmlfile; |
87 | typedef struct htmlsect htmlsect; |
88 | |
89 | struct htmlfile { |
90 | htmlfile *next; |
91 | char *filename; |
92 | int last_fragment_number; |
93 | int min_heading_depth; |
94 | htmlsect *first, *last; /* first/last highest-level sections */ |
95 | }; |
96 | |
97 | struct htmlsect { |
98 | htmlsect *next, *parent; |
99 | htmlfile *file; |
100 | paragraph *title, *text; |
101 | enum { NORMAL, TOP, INDEX } type; |
102 | int contents_depth; |
103 | char *fragment; |
104 | }; |
105 | |
106 | typedef struct { |
107 | htmlfile *head, *tail; |
108 | htmlfile *single, *index; |
3e82de8f |
109 | tree234 *frags; |
78c73085 |
110 | } htmlfilelist; |
111 | |
112 | typedef struct { |
113 | htmlsect *head, *tail; |
114 | } htmlsectlist; |
115 | |
116 | typedef struct { |
3e82de8f |
117 | htmlfile *file; |
118 | char *fragment; |
119 | } htmlfragment; |
120 | |
121 | typedef struct { |
78c73085 |
122 | int nrefs, refsize; |
123 | word **refs; |
124 | } htmlindex; |
125 | |
126 | typedef struct { |
127 | htmlsect *section; |
128 | char *fragment; |
1b7bf715 |
129 | int generated, referenced; |
78c73085 |
130 | } htmlindexref; |
131 | |
132 | typedef struct { |
133 | /* |
134 | * This level deals with charset conversion, starting and |
135 | * ending tags, and writing to the file. It's the lexical |
136 | * level. |
137 | */ |
138 | FILE *fp; |
139 | int charset; |
140 | charset_state cstate; |
141 | int ver; |
142 | enum { |
143 | HO_NEUTRAL, HO_IN_TAG, HO_IN_EMPTY_TAG, HO_IN_TEXT |
144 | } state; |
145 | /* |
146 | * Stuff beyond here deals with the higher syntactic level: it |
147 | * tracks how many levels of <ul> are currently open when |
148 | * producing a contents list, for example. |
149 | */ |
150 | int contents_level; |
151 | } htmloutput; |
152 | |
3e82de8f |
153 | static int html_fragment_compare(void *av, void *bv) |
154 | { |
155 | htmlfragment *a = (htmlfragment *)av; |
156 | htmlfragment *b = (htmlfragment *)bv; |
157 | int cmp; |
158 | |
159 | if ((cmp = strcmp(a->file->filename, b->file->filename)) != 0) |
160 | return cmp; |
161 | else |
162 | return strcmp(a->fragment, b->fragment); |
163 | } |
164 | |
78c73085 |
165 | static void html_file_section(htmlconfig *cfg, htmlfilelist *files, |
166 | htmlsect *sect, int depth); |
167 | |
168 | static htmlfile *html_new_file(htmlfilelist *list, char *filename); |
169 | static htmlsect *html_new_sect(htmlsectlist *list, paragraph *title); |
170 | |
171 | /* Flags for html_words() flags parameter */ |
172 | #define NOTHING 0x00 |
173 | #define MARKUP 0x01 |
174 | #define LINKS 0x02 |
175 | #define INDEXENTS 0x04 |
176 | #define ALL 0x07 |
177 | static void html_words(htmloutput *ho, word *words, int flags, |
178 | htmlfile *file, keywordlist *keywords, htmlconfig *cfg); |
179 | static void html_codepara(htmloutput *ho, word *words); |
180 | |
181 | static void element_open(htmloutput *ho, char const *name); |
182 | static void element_close(htmloutput *ho, char const *name); |
183 | static void element_empty(htmloutput *ho, char const *name); |
184 | static void element_attr(htmloutput *ho, char const *name, char const *value); |
185 | static void element_attr_w(htmloutput *ho, char const *name, |
186 | wchar_t const *value); |
187 | static void html_text(htmloutput *ho, wchar_t const *str); |
188 | static void html_text_limit(htmloutput *ho, wchar_t const *str, int maxlen); |
189 | static void html_text_limit_internal(htmloutput *ho, wchar_t const *text, |
190 | int maxlen, int quote_quotes); |
191 | static void html_nl(htmloutput *ho); |
192 | static void html_raw(htmloutput *ho, char *text); |
193 | static void html_raw_as_attr(htmloutput *ho, char *text); |
194 | static void cleanup(htmloutput *ho); |
195 | |
196 | static void html_href(htmloutput *ho, htmlfile *thisfile, |
197 | htmlfile *targetfile, char *targetfrag); |
27bdc5ab |
198 | static void html_fragment(htmloutput *ho, char const *fragment); |
78c73085 |
199 | |
200 | static char *html_format(paragraph *p, char *template_string); |
3e82de8f |
201 | static char *html_sanitise_fragment(htmlfilelist *files, htmlfile *file, |
202 | char *text); |
78c73085 |
203 | |
204 | static void html_contents_entry(htmloutput *ho, int depth, htmlsect *s, |
205 | htmlfile *thisfile, keywordlist *keywords, |
206 | htmlconfig *cfg); |
207 | static void html_section_title(htmloutput *ho, htmlsect *s, |
208 | htmlfile *thisfile, keywordlist *keywords, |
23c9bbc2 |
209 | htmlconfig *cfg, int real); |
78c73085 |
210 | |
211 | static htmlconfig html_configure(paragraph *source) { |
212 | htmlconfig ret; |
213 | paragraph *p; |
214 | |
215 | /* |
216 | * Defaults. |
217 | */ |
218 | ret.leaf_level = 2; |
219 | ret.achapter.just_numbers = FALSE; |
220 | ret.achapter.number_suffix = L": "; |
221 | ret.nasect = 1; |
f1530049 |
222 | ret.asect = snewn(ret.nasect, sectlevel); |
78c73085 |
223 | ret.asect[0].just_numbers = TRUE; |
224 | ret.asect[0].number_suffix = L" "; |
225 | ret.ncdepths = 0; |
226 | ret.contents_depths = 0; |
227 | ret.visible_version_id = TRUE; |
228 | ret.address_section = TRUE; |
229 | ret.leaf_contains_contents = FALSE; |
230 | ret.leaf_smallest_contents = 4; |
231 | ret.single_filename = dupstr("Manual.html"); |
232 | ret.contents_filename = dupstr("Contents.html"); |
233 | ret.index_filename = dupstr("IndexPage.html"); |
234 | ret.template_filename = dupstr("%n.html"); |
235 | ret.template_fragment = dupstr("%b"); |
236 | ret.head_end = ret.body_tag = ret.body_start = ret.body_end = |
237 | ret.addr_start = ret.addr_end = ret.nav_attr = NULL; |
238 | ret.author = ret.description = NULL; |
239 | ret.restrict_charset = CS_ASCII; |
240 | ret.output_charset = CS_ASCII; |
241 | ret.htmlver = HTML_4; |
242 | /* |
243 | * Default quote characters are Unicode matched single quotes, |
244 | * falling back to ordinary ASCII ". |
245 | */ |
246 | ret.lquote = L"\x2018\0\x2019\0\"\0\"\0\0"; |
247 | ret.rquote = uadv(ret.lquote); |
248 | |
249 | /* |
250 | * Two-pass configuration so that we can pick up global config |
251 | * (e.g. `quotes') before having it overridden by specific |
252 | * config (`html-quotes'), irrespective of the order in which |
253 | * they occur. |
254 | */ |
255 | for (p = source; p; p = p->next) { |
256 | if (p->type == para_Config) { |
257 | if (!ustricmp(p->keyword, L"quotes")) { |
258 | if (*uadv(p->keyword) && *uadv(uadv(p->keyword))) { |
259 | ret.lquote = uadv(p->keyword); |
260 | ret.rquote = uadv(ret.lquote); |
261 | } |
262 | } |
263 | } |
264 | } |
265 | |
266 | for (p = source; p; p = p->next) { |
267 | if (p->type == para_Config) { |
268 | wchar_t *k = p->keyword; |
269 | |
270 | if (!ustrnicmp(k, L"xhtml-", 6)) |
271 | k++; /* treat `xhtml-' and `html-' the same */ |
272 | |
273 | if (!ustricmp(k, L"html-charset")) { |
274 | char *csname = utoa_dup(uadv(k), CS_ASCII); |
275 | ret.restrict_charset = ret.output_charset = |
276 | charset_from_localenc(csname); |
277 | sfree(csname); |
27bdc5ab |
278 | } else if (!ustricmp(k, L"html-version")) { |
279 | wchar_t *vername = uadv(k); |
280 | static const struct { |
281 | const wchar_t *name; |
282 | int ver; |
283 | } versions[] = { |
284 | {L"html3.2", HTML_3_2}, |
285 | {L"html4", HTML_4}, |
286 | {L"iso-html", ISO_HTML}, |
287 | {L"xhtml1.0transitional", XHTML_1_0_TRANSITIONAL}, |
288 | {L"xhtml1.0strict", XHTML_1_0_STRICT} |
289 | }; |
290 | int i; |
291 | |
292 | for (i = 0; i < (int)lenof(versions); i++) |
293 | if (!ustricmp(versions[i].name, vername)) |
294 | break; |
295 | |
296 | if (i == lenof(versions)) |
297 | error(err_htmlver, &p->fpos, vername); |
298 | else |
299 | ret.htmlver = versions[i].ver; |
78c73085 |
300 | } else if (!ustricmp(k, L"html-single-filename")) { |
301 | sfree(ret.single_filename); |
302 | ret.single_filename = dupstr(adv(p->origkeyword)); |
303 | } else if (!ustricmp(k, L"html-contents-filename")) { |
304 | sfree(ret.contents_filename); |
305 | ret.contents_filename = dupstr(adv(p->origkeyword)); |
306 | } else if (!ustricmp(k, L"html-index-filename")) { |
307 | sfree(ret.index_filename); |
308 | ret.index_filename = dupstr(adv(p->origkeyword)); |
309 | } else if (!ustricmp(k, L"html-template-filename")) { |
310 | sfree(ret.template_filename); |
311 | ret.template_filename = dupstr(adv(p->origkeyword)); |
312 | } else if (!ustricmp(k, L"html-template-fragment")) { |
313 | sfree(ret.template_fragment); |
314 | ret.template_fragment = dupstr(adv(p->origkeyword)); |
315 | } else if (!ustricmp(k, L"html-chapter-numeric")) { |
316 | ret.achapter.just_numbers = utob(uadv(k)); |
317 | } else if (!ustricmp(k, L"html-chapter-suffix")) { |
318 | ret.achapter.number_suffix = uadv(k); |
319 | } else if (!ustricmp(k, L"html-leaf-level")) { |
320 | ret.leaf_level = utoi(uadv(k)); |
321 | } else if (!ustricmp(k, L"html-section-numeric")) { |
322 | wchar_t *q = uadv(k); |
323 | int n = 0; |
324 | if (uisdigit(*q)) { |
325 | n = utoi(q); |
326 | q = uadv(q); |
327 | } |
328 | if (n >= ret.nasect) { |
329 | int i; |
f1530049 |
330 | ret.asect = sresize(ret.asect, n+1, sectlevel); |
78c73085 |
331 | for (i = ret.nasect; i <= n; i++) |
332 | ret.asect[i] = ret.asect[ret.nasect-1]; |
333 | ret.nasect = n+1; |
334 | } |
335 | ret.asect[n].just_numbers = utob(q); |
336 | } else if (!ustricmp(k, L"html-section-suffix")) { |
337 | wchar_t *q = uadv(k); |
338 | int n = 0; |
339 | if (uisdigit(*q)) { |
340 | n = utoi(q); |
341 | q = uadv(q); |
342 | } |
343 | if (n >= ret.nasect) { |
344 | int i; |
f1530049 |
345 | ret.asect = sresize(ret.asect, n+1, sectlevel); |
78c73085 |
346 | for (i = ret.nasect; i <= n; i++) { |
347 | ret.asect[i] = ret.asect[ret.nasect-1]; |
348 | } |
349 | ret.nasect = n+1; |
350 | } |
351 | ret.asect[n].number_suffix = q; |
352 | } else if (!ustricmp(k, L"html-contents-depth") || |
353 | !ustrnicmp(k, L"html-contents-depth-", 20)) { |
354 | /* |
355 | * Relic of old implementation: this directive used |
356 | * to be written as \cfg{html-contents-depth-3}{2} |
357 | * rather than the usual Halibut convention of |
358 | * \cfg{html-contents-depth}{3}{2}. We therefore |
359 | * support both. |
360 | */ |
361 | wchar_t *q = k[19] ? k+20 : uadv(k); |
362 | int n = 0; |
363 | if (uisdigit(*q)) { |
364 | n = utoi(q); |
365 | q = uadv(q); |
366 | } |
367 | if (n >= ret.ncdepths) { |
368 | int i; |
f1530049 |
369 | ret.contents_depths = |
370 | sresize(ret.contents_depths, n+1, int); |
78c73085 |
371 | for (i = ret.ncdepths; i <= n; i++) { |
372 | ret.contents_depths[i] = i+2; |
373 | } |
374 | ret.ncdepths = n+1; |
375 | } |
376 | ret.contents_depths[n] = utoi(q); |
377 | } else if (!ustricmp(k, L"html-head-end")) { |
378 | ret.head_end = adv(p->origkeyword); |
379 | } else if (!ustricmp(k, L"html-body-tag")) { |
380 | ret.body_tag = adv(p->origkeyword); |
381 | } else if (!ustricmp(k, L"html-body-start")) { |
382 | ret.body_start = adv(p->origkeyword); |
383 | } else if (!ustricmp(k, L"html-body-end")) { |
384 | ret.body_end = adv(p->origkeyword); |
385 | } else if (!ustricmp(k, L"html-address-start")) { |
386 | ret.addr_start = adv(p->origkeyword); |
387 | } else if (!ustricmp(k, L"html-address-end")) { |
388 | ret.addr_end = adv(p->origkeyword); |
389 | } else if (!ustricmp(k, L"html-navigation-attributes")) { |
390 | ret.nav_attr = adv(p->origkeyword); |
391 | } else if (!ustricmp(k, L"html-author")) { |
392 | ret.author = uadv(k); |
393 | } else if (!ustricmp(k, L"html-description")) { |
394 | ret.description = uadv(k); |
395 | } else if (!ustricmp(k, L"html-suppress-address")) { |
396 | ret.address_section = !utob(uadv(k)); |
397 | } else if (!ustricmp(k, L"html-versionid")) { |
398 | ret.visible_version_id = utob(uadv(k)); |
399 | } else if (!ustricmp(k, L"html-quotes")) { |
400 | if (*uadv(k) && *uadv(uadv(k))) { |
401 | ret.lquote = uadv(k); |
402 | ret.rquote = uadv(ret.lquote); |
403 | } |
404 | } else if (!ustricmp(k, L"html-leaf-contains-contents")) { |
405 | ret.leaf_contains_contents = utob(uadv(k)); |
406 | } else if (!ustricmp(k, L"html-leaf-smallest-contents")) { |
407 | ret.leaf_smallest_contents = utoi(uadv(k)); |
408 | } |
409 | } |
410 | } |
411 | |
412 | /* |
413 | * Now process fallbacks on quote characters. |
414 | */ |
415 | while (*uadv(ret.rquote) && *uadv(uadv(ret.rquote)) && |
416 | (!cvt_ok(ret.restrict_charset, ret.lquote) || |
417 | !cvt_ok(ret.restrict_charset, ret.rquote))) { |
418 | ret.lquote = uadv(ret.rquote); |
419 | ret.rquote = uadv(ret.lquote); |
420 | } |
421 | |
422 | return ret; |
423 | } |
424 | |
425 | paragraph *html_config_filename(char *filename) |
426 | { |
427 | /* |
428 | * If the user passes in a single filename as a parameter to |
429 | * the `--html' command-line option, then we should assume it |
430 | * to imply _two_ config directives: |
431 | * \cfg{html-single-filename}{whatever} and |
432 | * \cfg{html-leaf-level}{0}; the rationale being that the user |
433 | * wants their output _in that file_. |
434 | */ |
435 | paragraph *p, *q; |
436 | |
437 | p = cmdline_cfg_simple("html-single-filename", filename, NULL); |
438 | q = cmdline_cfg_simple("html-leaf-level", "0", NULL); |
439 | p->next = q; |
440 | return p; |
441 | } |
442 | |
443 | void html_backend(paragraph *sourceform, keywordlist *keywords, |
444 | indexdata *idx, void *unused) { |
445 | paragraph *p; |
446 | htmlconfig conf; |
3e82de8f |
447 | htmlfilelist files = { NULL, NULL, NULL, NULL, NULL }; |
78c73085 |
448 | htmlsectlist sects = { NULL, NULL }, nonsects = { NULL, NULL }; |
449 | |
450 | IGNORE(unused); |
451 | |
452 | conf = html_configure(sourceform); |
453 | |
454 | /* |
455 | * We're going to make heavy use of paragraphs' private data |
456 | * fields in the forthcoming code. Clear them first, so we can |
457 | * reliably tell whether we have auxiliary data for a |
458 | * particular paragraph. |
459 | */ |
460 | for (p = sourceform; p; p = p->next) |
461 | p->private_data = NULL; |
462 | |
3e82de8f |
463 | files.frags = newtree234(html_fragment_compare); |
464 | |
78c73085 |
465 | /* |
466 | * Start by figuring out into which file each piece of the |
467 | * document should be put. We'll do this by inventing an |
468 | * `htmlsect' structure and stashing it in the private_data |
469 | * field of each section paragraph; we also need one additional |
470 | * htmlsect for the document index, which won't show up in the |
471 | * source form but needs to be consistently mentioned in |
472 | * contents links. |
473 | * |
474 | * While we're here, we'll also invent the HTML fragment name |
475 | * for each section. |
476 | */ |
477 | { |
478 | htmlsect *topsect, *sect; |
479 | int d; |
480 | |
481 | topsect = html_new_sect(§s, p); |
482 | topsect->type = TOP; |
483 | topsect->title = NULL; |
484 | topsect->text = sourceform; |
485 | topsect->contents_depth = contents_depth(conf, 0); |
486 | html_file_section(&conf, &files, topsect, -1); |
487 | topsect->fragment = NULL; |
488 | |
489 | for (p = sourceform; p; p = p->next) |
490 | if (is_heading_type(p->type)) { |
491 | d = heading_depth(p); |
492 | |
493 | if (p->type == para_Title) { |
494 | topsect->title = p; |
495 | continue; |
496 | } |
497 | |
498 | sect = html_new_sect(§s, p); |
499 | sect->text = p->next; |
500 | |
501 | sect->contents_depth = contents_depth(conf, d+1) - (d+1); |
502 | |
503 | if (p->parent) { |
504 | sect->parent = (htmlsect *)p->parent->private_data; |
505 | assert(sect->parent != NULL); |
506 | } else |
507 | sect->parent = topsect; |
508 | p->private_data = sect; |
509 | |
510 | html_file_section(&conf, &files, sect, d); |
511 | |
512 | sect->fragment = html_format(p, conf.template_fragment); |
3e82de8f |
513 | sect->fragment = html_sanitise_fragment(&files, sect->file, |
514 | sect->fragment); |
78c73085 |
515 | } |
516 | |
517 | /* And the index. */ |
518 | sect = html_new_sect(§s, NULL); |
78c73085 |
519 | sect->text = NULL; |
520 | sect->type = INDEX; |
521 | sect->parent = topsect; |
522 | html_file_section(&conf, &files, sect, 0); /* peer of chapters */ |
3e82de8f |
523 | sect->fragment = dupstr("Index"); /* FIXME: this _can't_ be right */ |
524 | sect->fragment = html_sanitise_fragment(&files, sect->file, |
525 | sect->fragment); |
78c73085 |
526 | files.index = sect->file; |
527 | } |
528 | |
529 | /* |
530 | * Go through the keyword list and sort out fragment IDs for |
531 | * all the potentially referenced paragraphs which _aren't_ |
532 | * headings. |
533 | */ |
534 | { |
535 | int i; |
536 | keyword *kw; |
537 | htmlsect *sect; |
538 | |
539 | for (i = 0; (kw = index234(keywords->keys, i)) != NULL; i++) { |
540 | paragraph *q, *p = kw->para; |
541 | |
542 | if (!is_heading_type(p->type)) { |
543 | htmlsect *parent; |
544 | |
545 | /* |
546 | * Find the paragraph's parent htmlsect, to |
547 | * determine which file it will end up in. |
548 | */ |
549 | q = p->parent; |
550 | if (!q) { |
551 | /* |
552 | * Preamble paragraphs have no parent. So if we |
553 | * have a non-heading with no parent, it must |
554 | * be preamble, and therefore its parent |
555 | * htmlsect must be the preamble one. |
556 | */ |
557 | assert(sects.head && |
558 | sects.head->type == TOP); |
559 | parent = sects.head; |
560 | } else |
561 | parent = (htmlsect *)q->private_data; |
562 | |
563 | /* |
564 | * Now we can construct an htmlsect for this |
565 | * paragraph itself, taking care to put it in the |
566 | * list of non-sections rather than the list of |
567 | * sections (so that traverses of the `sects' list |
568 | * won't attempt to add it to the contents or |
569 | * anything weird like that). |
570 | */ |
571 | sect = html_new_sect(&nonsects, p); |
572 | sect->file = parent->file; |
573 | sect->parent = parent; |
574 | p->private_data = sect; |
575 | |
576 | /* |
04781c84 |
577 | * Fragment IDs for these paragraphs will simply be |
578 | * `p' followed by an integer. |
78c73085 |
579 | */ |
f1530049 |
580 | sect->fragment = snewn(40, char); |
04781c84 |
581 | sprintf(sect->fragment, "p%d", |
582 | sect->file->last_fragment_number++); |
3e82de8f |
583 | sect->fragment = html_sanitise_fragment(&files, sect->file, |
584 | sect->fragment); |
78c73085 |
585 | } |
586 | } |
587 | } |
588 | |
589 | /* |
04781c84 |
590 | * Reset the fragment numbers in each file. I've just used them |
591 | * to generate `p' fragment IDs for non-section paragraphs |
592 | * (numbered list elements, bibliocited), and now I want to use |
593 | * them for `i' fragment IDs for index entries. |
594 | */ |
595 | { |
596 | htmlfile *file; |
597 | for (file = files.head; file; file = file->next) |
598 | file->last_fragment_number = 0; |
599 | } |
600 | |
601 | /* |
78c73085 |
602 | * Now sort out the index. This involves: |
603 | * |
604 | * - For each index term, we set up an htmlindex structure to |
605 | * store all the references to that term. |
606 | * |
607 | * - Then we make a pass over the actual document, finding |
608 | * every word_IndexRef; for each one, we actually figure out |
609 | * the HTML filename/fragment pair we will use to reference |
610 | * it, store that information in the private data field of |
611 | * the word_IndexRef itself (so we can recreate it when the |
612 | * time comes to output our HTML), and add a reference to it |
613 | * to the index term in question. |
614 | */ |
615 | { |
616 | int i; |
617 | indexentry *entry; |
618 | htmlsect *lastsect; |
619 | word *w; |
620 | |
621 | /* |
622 | * Set up the htmlindex structures. |
623 | */ |
624 | |
625 | for (i = 0; (entry = index234(idx->entries, i)) != NULL; i++) { |
f1530049 |
626 | htmlindex *hi = snew(htmlindex); |
78c73085 |
627 | |
628 | hi->nrefs = hi->refsize = 0; |
629 | hi->refs = NULL; |
630 | |
631 | entry->backend_data = hi; |
632 | } |
633 | |
634 | /* |
635 | * Run over the document inventing fragments. Each fragment |
636 | * is of the form `i' followed by an integer. |
78c73085 |
637 | */ |
638 | lastsect = NULL; |
639 | for (p = sourceform; p; p = p->next) { |
640 | if (is_heading_type(p->type)) |
641 | lastsect = (htmlsect *)p->private_data; |
642 | |
643 | for (w = p->words; w; w = w->next) |
644 | if (w->type == word_IndexRef) { |
f1530049 |
645 | htmlindexref *hr = snew(htmlindexref); |
78c73085 |
646 | indextag *tag; |
647 | int i; |
648 | |
1b7bf715 |
649 | hr->referenced = hr->generated = FALSE; |
78c73085 |
650 | hr->section = lastsect; |
78c73085 |
651 | { |
652 | char buf[40]; |
653 | sprintf(buf, "i%d", |
654 | lastsect->file->last_fragment_number++); |
655 | hr->fragment = dupstr(buf); |
3e82de8f |
656 | hr->fragment = |
657 | html_sanitise_fragment(&files, hr->section->file, |
658 | hr->fragment); |
78c73085 |
659 | } |
660 | w->private_data = hr; |
661 | |
662 | tag = index_findtag(idx, w->text); |
663 | if (!tag) |
664 | break; |
665 | |
666 | for (i = 0; i < tag->nrefs; i++) { |
667 | indexentry *entry = tag->refs[i]; |
668 | htmlindex *hi = (htmlindex *)entry->backend_data; |
669 | |
670 | if (hi->nrefs >= hi->refsize) { |
671 | hi->refsize += 32; |
f1530049 |
672 | hi->refs = sresize(hi->refs, hi->refsize, word *); |
78c73085 |
673 | } |
674 | |
675 | hi->refs[hi->nrefs++] = w; |
676 | } |
677 | } |
678 | } |
679 | } |
680 | |
681 | /* |
682 | * Now we're ready to write out the actual HTML files. |
683 | * |
684 | * For each file: |
685 | * |
686 | * - we open that file and write its header |
687 | * - we run down the list of sections |
688 | * - for each section directly contained within that file, we |
689 | * output the section text |
690 | * - for each section which is not in the file but which has a |
691 | * parent that is, we output a contents entry for the |
692 | * section if appropriate |
693 | * - finally, we output the file trailer and close the file. |
694 | */ |
695 | { |
696 | htmlfile *f, *prevf; |
697 | htmlsect *s; |
698 | paragraph *p; |
699 | |
700 | prevf = NULL; |
701 | |
702 | for (f = files.head; f; f = f->next) { |
703 | htmloutput ho; |
704 | int displaying; |
705 | enum LISTTYPE { NOLIST, UL, OL, DL }; |
706 | enum ITEMTYPE { NOITEM, LI, DT, DD }; |
707 | struct stackelement { |
708 | struct stackelement *next; |
709 | enum LISTTYPE listtype; |
710 | enum ITEMTYPE itemtype; |
711 | } *stackhead; |
712 | |
713 | #define listname(lt) ( (lt)==UL ? "ul" : (lt)==OL ? "ol" : "dl" ) |
714 | #define itemname(lt) ( (lt)==LI ? "li" : (lt)==DT ? "dt" : "dd" ) |
715 | |
716 | ho.fp = fopen(f->filename, "w"); |
717 | ho.charset = conf.output_charset; |
718 | ho.cstate = charset_init_state; |
719 | ho.ver = conf.htmlver; |
720 | ho.state = HO_NEUTRAL; |
721 | ho.contents_level = 0; |
722 | |
723 | /* <!DOCTYPE>. */ |
724 | switch (conf.htmlver) { |
725 | case HTML_3_2: |
726 | fprintf(ho.fp, "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD " |
727 | "HTML 3.2 Final//EN\">\n"); |
728 | break; |
729 | case HTML_4: |
730 | fprintf(ho.fp, "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML" |
731 | " 4.01//EN\"\n\"http://www.w3.org/TR/html4/" |
732 | "strict.dtd\">\n"); |
733 | break; |
27bdc5ab |
734 | case ISO_HTML: |
735 | fprintf(ho.fp, "<!DOCTYPE HTML PUBLIC \"ISO/IEC " |
736 | "15445:2000//DTD HTML//EN\">\n"); |
737 | break; |
78c73085 |
738 | case XHTML_1_0_TRANSITIONAL: |
27bdc5ab |
739 | fprintf(ho.fp, "<?xml version=\"1.0\" encoding=\"%s\"?>\n", |
740 | charset_to_mimeenc(conf.output_charset)); |
78c73085 |
741 | fprintf(ho.fp, "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML" |
742 | " 1.0 Transitional//EN\"\n\"http://www.w3.org/TR/" |
743 | "xhtml1/DTD/xhtml1-transitional.dtd\">\n"); |
744 | break; |
745 | case XHTML_1_0_STRICT: |
27bdc5ab |
746 | fprintf(ho.fp, "<?xml version=\"1.0\" encoding=\"%s\"?>\n", |
747 | charset_to_mimeenc(conf.output_charset)); |
78c73085 |
748 | fprintf(ho.fp, "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML" |
749 | " 1.0 Strict//EN\"\n\"http://www.w3.org/TR/xhtml1/" |
750 | "DTD/xhtml1-strict.dtd\">\n"); |
751 | break; |
752 | } |
753 | |
754 | element_open(&ho, "html"); |
755 | if (is_xhtml(conf.htmlver)) { |
756 | element_attr(&ho, "xmlns", "http://www.w3.org/1999/xhtml"); |
757 | } |
758 | html_nl(&ho); |
759 | |
760 | element_open(&ho, "head"); |
761 | html_nl(&ho); |
762 | |
763 | element_empty(&ho, "meta"); |
764 | element_attr(&ho, "http-equiv", "content-type"); |
765 | { |
766 | char buf[200]; |
767 | sprintf(buf, "text/html; charset=%.150s", |
768 | charset_to_mimeenc(conf.output_charset)); |
769 | element_attr(&ho, "content", buf); |
770 | } |
771 | html_nl(&ho); |
772 | |
773 | if (conf.author) { |
774 | element_empty(&ho, "meta"); |
775 | element_attr(&ho, "name", "author"); |
776 | element_attr_w(&ho, "content", conf.author); |
777 | html_nl(&ho); |
778 | } |
779 | |
780 | if (conf.description) { |
781 | element_empty(&ho, "meta"); |
782 | element_attr(&ho, "name", "description"); |
783 | element_attr_w(&ho, "content", conf.description); |
784 | html_nl(&ho); |
785 | } |
786 | |
787 | element_open(&ho, "title"); |
788 | if (f->first && f->first->title) { |
789 | html_words(&ho, f->first->title->words, NOTHING, |
790 | f, keywords, &conf); |
791 | |
792 | assert(f->last); |
793 | if (f->last != f->first && f->last->title) { |
794 | html_text(&ho, L" - "); /* FIXME: configurable? */ |
795 | html_words(&ho, f->last->title->words, NOTHING, |
796 | f, keywords, &conf); |
797 | } |
798 | } |
799 | element_close(&ho, "title"); |
800 | html_nl(&ho); |
801 | |
802 | if (conf.head_end) |
803 | html_raw(&ho, conf.head_end); |
804 | |
805 | element_close(&ho, "head"); |
806 | html_nl(&ho); |
807 | |
808 | /* FIXME: need to be able to specify replacement for this */ |
809 | if (conf.body_tag) |
810 | html_raw(&ho, conf.body_tag); |
811 | else |
812 | element_open(&ho, "body"); |
813 | html_nl(&ho); |
814 | |
815 | if (conf.body_start) |
816 | html_raw(&ho, conf.body_start); |
817 | |
818 | /* |
819 | * Write out a nav bar. Special case: we don't do this |
820 | * if there is only one file. |
821 | */ |
822 | if (files.head != files.tail) { |
823 | element_open(&ho, "p"); |
824 | if (conf.nav_attr) |
825 | html_raw_as_attr(&ho, conf.nav_attr); |
826 | |
827 | if (prevf) { |
828 | element_open(&ho, "a"); |
829 | element_attr(&ho, "href", prevf->filename); |
830 | } |
831 | html_text(&ho, L"Previous");/* FIXME: conf? */ |
832 | if (prevf) |
833 | element_close(&ho, "a"); |
834 | |
835 | html_text(&ho, L" | "); /* FIXME: conf? */ |
836 | |
837 | if (f != files.head) { |
838 | element_open(&ho, "a"); |
839 | element_attr(&ho, "href", files.head->filename); |
840 | } |
841 | html_text(&ho, L"Contents");/* FIXME: conf? */ |
842 | if (f != files.head) |
843 | element_close(&ho, "a"); |
844 | |
845 | html_text(&ho, L" | "); /* FIXME: conf? */ |
846 | |
847 | if (f != files.index) { |
848 | element_open(&ho, "a"); |
849 | element_attr(&ho, "href", files.index->filename); |
850 | } |
851 | html_text(&ho, L"Index");/* FIXME: conf? */ |
852 | if (f != files.index) |
853 | element_close(&ho, "a"); |
854 | |
855 | html_text(&ho, L" | "); /* FIXME: conf? */ |
856 | |
857 | if (f->next) { |
858 | element_open(&ho, "a"); |
859 | element_attr(&ho, "href", f->next->filename); |
860 | } |
861 | html_text(&ho, L"Next"); /* FIXME: conf? */ |
862 | if (f->next) |
863 | element_close(&ho, "a"); |
864 | |
865 | element_close(&ho, "p"); |
866 | html_nl(&ho); |
867 | } |
868 | prevf = f; |
869 | |
870 | /* |
871 | * Write out a prefix TOC for the file. |
872 | * |
873 | * We start by going through the section list and |
874 | * collecting the sections which need to be added to |
875 | * the contents. On the way, we also test to see if |
876 | * this file is a leaf file (defined as one which |
877 | * contains all descendants of any section it |
878 | * contains), because this will play a part in our |
879 | * decision on whether or not to _output_ the TOC. |
880 | * |
881 | * Special case: we absolutely do not do this if we're |
882 | * in single-file mode. |
883 | */ |
884 | if (files.head != files.tail) { |
885 | int ntoc = 0, tocsize = 0; |
886 | htmlsect **toc = NULL; |
887 | int leaf = TRUE; |
888 | |
889 | for (s = sects.head; s; s = s->next) { |
890 | htmlsect *a, *ac; |
891 | int depth, adepth; |
892 | |
893 | /* |
894 | * Search up from this section until we find |
895 | * the highest-level one which belongs in this |
896 | * file. |
897 | */ |
898 | depth = adepth = 0; |
899 | a = NULL; |
900 | for (ac = s; ac; ac = ac->parent) { |
901 | if (ac->file == f) { |
902 | a = ac; |
903 | adepth = depth; |
904 | } |
905 | depth++; |
906 | } |
907 | |
908 | if (s->file != f && a != NULL) |
909 | leaf = FALSE; |
910 | |
911 | if (a) { |
912 | if (adepth <= a->contents_depth) { |
913 | if (ntoc >= tocsize) { |
914 | tocsize += 64; |
f1530049 |
915 | toc = sresize(toc, tocsize, htmlsect *); |
78c73085 |
916 | } |
917 | toc[ntoc++] = s; |
918 | } |
919 | } |
920 | } |
921 | |
922 | if (leaf && conf.leaf_contains_contents && |
923 | ntoc >= conf.leaf_smallest_contents) { |
924 | int i; |
925 | |
926 | for (i = 0; i < ntoc; i++) { |
927 | htmlsect *s = toc[i]; |
928 | int hlevel = (s->type == TOP ? -1 : |
929 | s->type == INDEX ? 0 : |
930 | heading_depth(s->title)) |
931 | - f->min_heading_depth + 1; |
932 | |
933 | assert(hlevel >= 1); |
934 | html_contents_entry(&ho, hlevel, s, |
935 | f, keywords, &conf); |
936 | } |
937 | html_contents_entry(&ho, 0, NULL, f, keywords, &conf); |
938 | } |
939 | } |
940 | |
941 | /* |
942 | * Now go through the document and output some real |
943 | * text. |
944 | */ |
945 | displaying = FALSE; |
946 | for (s = sects.head; s; s = s->next) { |
947 | if (s->file == f) { |
948 | /* |
949 | * This section belongs in this file. |
950 | * Display it. |
951 | */ |
952 | displaying = TRUE; |
953 | } else { |
954 | htmlsect *a, *ac; |
955 | int depth, adepth; |
956 | |
957 | displaying = FALSE; |
958 | |
959 | /* |
960 | * Search up from this section until we find |
961 | * the highest-level one which belongs in this |
962 | * file. |
963 | */ |
964 | depth = adepth = 0; |
965 | a = NULL; |
966 | for (ac = s; ac; ac = ac->parent) { |
967 | if (ac->file == f) { |
968 | a = ac; |
969 | adepth = depth; |
970 | } |
971 | depth++; |
972 | } |
973 | |
974 | if (a != NULL) { |
975 | /* |
976 | * This section does not belong in this |
977 | * file, but an ancestor of it does. Write |
978 | * out a contents table entry, if the depth |
979 | * doesn't exceed the maximum contents |
980 | * depth for the ancestor section. |
981 | */ |
982 | if (adepth <= a->contents_depth) { |
983 | html_contents_entry(&ho, adepth, s, |
984 | f, keywords, &conf); |
985 | } |
986 | } |
987 | } |
988 | |
989 | if (displaying) { |
990 | int hlevel; |
991 | char htag[3]; |
992 | |
993 | html_contents_entry(&ho, 0, NULL, f, keywords, &conf); |
994 | |
995 | /* |
996 | * Display the section heading. |
997 | */ |
998 | |
999 | hlevel = (s->type == TOP ? -1 : |
1000 | s->type == INDEX ? 0 : |
1001 | heading_depth(s->title)) |
1002 | - f->min_heading_depth + 1; |
1003 | assert(hlevel >= 1); |
1004 | /* HTML headings only go up to <h6> */ |
1005 | if (hlevel > 6) |
1006 | hlevel = 6; |
1007 | htag[0] = 'h'; |
1008 | htag[1] = '0' + hlevel; |
1009 | htag[2] = '\0'; |
1010 | element_open(&ho, htag); |
1011 | |
1012 | /* |
1013 | * Provide anchor for cross-links to target. |
1014 | * |
78c73085 |
1015 | * (Also we'll have to do this separately in |
1016 | * other paragraph types - NumberedList and |
1017 | * BiblioCited.) |
1018 | */ |
27bdc5ab |
1019 | if (s->fragment) |
1020 | html_fragment(&ho, s->fragment); |
78c73085 |
1021 | |
23c9bbc2 |
1022 | html_section_title(&ho, s, f, keywords, &conf, TRUE); |
78c73085 |
1023 | |
1024 | element_close(&ho, htag); |
1025 | |
1026 | /* |
1027 | * Now display the section text. |
1028 | */ |
1029 | if (s->text) { |
f1530049 |
1030 | stackhead = snew(struct stackelement); |
78c73085 |
1031 | stackhead->next = NULL; |
1032 | stackhead->listtype = NOLIST; |
1033 | stackhead->itemtype = NOITEM; |
1034 | |
1035 | for (p = s->text;; p = p->next) { |
1036 | enum LISTTYPE listtype; |
1037 | struct stackelement *se; |
1038 | |
1039 | /* |
1040 | * Preliminary switch to figure out what |
1041 | * sort of list we expect to be inside at |
1042 | * this stage. |
1043 | * |
1044 | * Since p may still be NULL at this point, |
1045 | * I invent a harmless paragraph type for |
1046 | * it if it is. |
1047 | */ |
1048 | switch (p ? p->type : para_Normal) { |
1049 | case para_Rule: |
1050 | case para_Normal: |
1051 | case para_Copyright: |
1052 | case para_BiblioCited: |
1053 | case para_Code: |
1054 | case para_QuotePush: |
1055 | case para_QuotePop: |
1056 | case para_Chapter: |
1057 | case para_Appendix: |
1058 | case para_UnnumberedChapter: |
1059 | case para_Heading: |
1060 | case para_Subsect: |
1061 | case para_LcontPop: |
1062 | listtype = NOLIST; |
1063 | break; |
1064 | |
1065 | case para_Bullet: |
1066 | listtype = UL; |
1067 | break; |
1068 | |
1069 | case para_NumberedList: |
1070 | listtype = OL; |
1071 | break; |
1072 | |
1073 | case para_DescribedThing: |
1074 | case para_Description: |
1075 | listtype = DL; |
1076 | break; |
1077 | |
1078 | case para_LcontPush: |
f1530049 |
1079 | se = snew(struct stackelement); |
78c73085 |
1080 | se->next = stackhead; |
1081 | se->listtype = NOLIST; |
1082 | se->itemtype = NOITEM; |
1083 | stackhead = se; |
1084 | continue; |
1085 | |
1086 | default: /* some totally non-printing para */ |
1087 | continue; |
1088 | } |
1089 | |
1090 | html_nl(&ho); |
1091 | |
1092 | /* |
1093 | * Terminate the most recent list item, if |
1094 | * any. (We left this until after |
1095 | * processing LcontPush, since in that case |
1096 | * the list item won't want to be |
1097 | * terminated until after the corresponding |
1098 | * LcontPop.) |
1099 | */ |
1100 | if (stackhead->itemtype != NOITEM) { |
1101 | element_close(&ho, itemname(stackhead->itemtype)); |
1102 | html_nl(&ho); |
1103 | } |
1104 | stackhead->itemtype = NOITEM; |
1105 | |
1106 | /* |
1107 | * Terminate the current list, if it's not |
1108 | * the one we want to be in. |
1109 | */ |
1110 | if (listtype != stackhead->listtype && |
1111 | stackhead->listtype != NOLIST) { |
1112 | element_close(&ho, listname(stackhead->listtype)); |
1113 | html_nl(&ho); |
1114 | } |
1115 | |
1116 | /* |
1117 | * Leave the loop if our time has come. |
1118 | */ |
1119 | if (!p || (is_heading_type(p->type) && |
1120 | p->type != para_Title)) |
1121 | break; /* end of section text */ |
1122 | |
1123 | /* |
1124 | * Start a fresh list if necessary. |
1125 | */ |
1126 | if (listtype != stackhead->listtype && |
1127 | listtype != NOLIST) |
1128 | element_open(&ho, listname(listtype)); |
1129 | |
1130 | stackhead->listtype = listtype; |
1131 | |
1132 | switch (p->type) { |
1133 | case para_Rule: |
1134 | element_empty(&ho, "hr"); |
1135 | break; |
1136 | case para_Code: |
1137 | html_codepara(&ho, p->words); |
1138 | break; |
1139 | case para_Normal: |
1140 | case para_Copyright: |
1141 | element_open(&ho, "p"); |
1142 | html_nl(&ho); |
1143 | html_words(&ho, p->words, ALL, |
1144 | f, keywords, &conf); |
1145 | html_nl(&ho); |
1146 | element_close(&ho, "p"); |
1147 | break; |
1148 | case para_BiblioCited: |
1149 | element_open(&ho, "p"); |
1150 | if (p->private_data) { |
1151 | htmlsect *s = (htmlsect *)p->private_data; |
27bdc5ab |
1152 | html_fragment(&ho, s->fragment); |
78c73085 |
1153 | } |
1154 | html_nl(&ho); |
1155 | html_words(&ho, p->kwtext, ALL, |
1156 | f, keywords, &conf); |
1157 | html_text(&ho, L" "); |
1158 | html_words(&ho, p->words, ALL, |
1159 | f, keywords, &conf); |
1160 | html_nl(&ho); |
1161 | element_close(&ho, "p"); |
1162 | break; |
1163 | case para_Bullet: |
1164 | case para_NumberedList: |
1165 | element_open(&ho, "li"); |
1166 | if (p->private_data) { |
1167 | htmlsect *s = (htmlsect *)p->private_data; |
27bdc5ab |
1168 | html_fragment(&ho, s->fragment); |
78c73085 |
1169 | } |
1170 | html_nl(&ho); |
1171 | stackhead->itemtype = LI; |
1172 | html_words(&ho, p->words, ALL, |
1173 | f, keywords, &conf); |
1174 | break; |
1175 | case para_DescribedThing: |
1176 | element_open(&ho, "dt"); |
1177 | html_nl(&ho); |
1178 | stackhead->itemtype = DT; |
1179 | html_words(&ho, p->words, ALL, |
1180 | f, keywords, &conf); |
1181 | break; |
1182 | case para_Description: |
1183 | element_open(&ho, "dd"); |
1184 | html_nl(&ho); |
1185 | stackhead->itemtype = DD; |
1186 | html_words(&ho, p->words, ALL, |
1187 | f, keywords, &conf); |
1188 | break; |
1189 | |
1190 | case para_QuotePush: |
1191 | element_open(&ho, "blockquote"); |
1192 | break; |
1193 | case para_QuotePop: |
1194 | element_close(&ho, "blockquote"); |
1195 | break; |
1196 | |
1197 | case para_LcontPop: |
1198 | se = stackhead; |
1199 | stackhead = stackhead->next; |
1200 | assert(stackhead); |
1201 | sfree(se); |
1202 | break; |
1203 | } |
1204 | } |
1205 | |
1206 | assert(stackhead && !stackhead->next); |
1207 | sfree(stackhead); |
1208 | } |
1209 | |
1210 | if (s->type == INDEX) { |
1211 | indexentry *entry; |
1212 | int i; |
1213 | |
1214 | /* |
1215 | * This section is the index. I'll just |
1216 | * render it as a single paragraph, with a |
1217 | * colon between the index term and the |
1218 | * references, and <br> in between each |
1219 | * entry. |
1220 | */ |
1221 | element_open(&ho, "p"); |
1222 | |
1223 | for (i = 0; (entry = |
1224 | index234(idx->entries, i)) != NULL; i++) { |
1225 | htmlindex *hi = (htmlindex *)entry->backend_data; |
1226 | int j; |
1227 | |
1228 | if (i > 0) |
1229 | element_empty(&ho, "br"); |
1230 | html_nl(&ho); |
1231 | |
1232 | html_words(&ho, entry->text, MARKUP|LINKS, |
1233 | f, keywords, &conf); |
1234 | |
1235 | html_text(&ho, L": ");/* FIXME: configurable */ |
1236 | |
1237 | for (j = 0; j < hi->nrefs; j++) { |
1238 | htmlindexref *hr = |
1239 | (htmlindexref *)hi->refs[j]->private_data; |
1240 | paragraph *p = hr->section->title; |
1241 | |
1242 | if (j > 0) |
1243 | html_text(&ho, L", "); /* FIXME: conf */ |
1244 | |
1245 | html_href(&ho, f, hr->section->file, |
1246 | hr->fragment); |
1b7bf715 |
1247 | hr->referenced = TRUE; |
78c73085 |
1248 | if (p && p->kwtext) |
1249 | html_words(&ho, p->kwtext, MARKUP|LINKS, |
1250 | f, keywords, &conf); |
1251 | else if (p && p->words) |
1252 | html_words(&ho, p->words, MARKUP|LINKS, |
1253 | f, keywords, &conf); |
1254 | else |
1255 | html_text(&ho, L"FIXME"); |
1256 | element_close(&ho, "a"); |
1257 | } |
1258 | } |
1259 | element_close(&ho, "p"); |
1260 | } |
1261 | } |
1262 | } |
1263 | |
1264 | html_contents_entry(&ho, 0, NULL, f, keywords, &conf); |
1265 | html_nl(&ho); |
1266 | |
1267 | { |
1268 | /* |
1269 | * Footer. |
1270 | */ |
1271 | int done_version_ids = FALSE; |
1272 | |
1273 | element_empty(&ho, "hr"); |
1274 | |
1275 | if (conf.body_end) |
1276 | html_raw(&ho, conf.body_end); |
1277 | |
1278 | if (conf.address_section) { |
27bdc5ab |
1279 | int started = FALSE; |
1280 | if (conf.htmlver == ISO_HTML) { |
1281 | /* |
1282 | * The ISO-HTML validator complains if |
1283 | * there isn't a <div> tag surrounding the |
1284 | * <address> tag. I'm uncertain of why this |
1285 | * should be - there appears to be no |
1286 | * mention of this in the ISO-HTML spec, |
1287 | * suggesting that it doesn't represent a |
1288 | * change from HTML 4, but nonetheless the |
1289 | * HTML 4 validator doesn't seem to mind. |
1290 | */ |
1291 | element_open(&ho, "div"); |
1292 | } |
78c73085 |
1293 | element_open(&ho, "address"); |
1294 | if (conf.addr_start) { |
1295 | html_raw(&ho, conf.addr_start); |
1296 | html_nl(&ho); |
27bdc5ab |
1297 | started = TRUE; |
78c73085 |
1298 | } |
1299 | if (conf.visible_version_id) { |
78c73085 |
1300 | for (p = sourceform; p; p = p->next) |
1301 | if (p->type == para_VersionID) { |
27bdc5ab |
1302 | if (started) |
78c73085 |
1303 | element_empty(&ho, "br"); |
1304 | html_nl(&ho); |
1305 | html_text(&ho, L"["); /* FIXME: conf? */ |
1306 | html_words(&ho, p->words, NOTHING, |
1307 | f, keywords, &conf); |
1308 | html_text(&ho, L"]"); /* FIXME: conf? */ |
1309 | started = TRUE; |
1310 | } |
78c73085 |
1311 | done_version_ids = TRUE; |
1312 | } |
27bdc5ab |
1313 | if (conf.addr_end) { |
1314 | if (started) |
1315 | element_empty(&ho, "br"); |
78c73085 |
1316 | html_raw(&ho, conf.addr_end); |
27bdc5ab |
1317 | } |
78c73085 |
1318 | element_close(&ho, "address"); |
27bdc5ab |
1319 | if (conf.htmlver == ISO_HTML) |
1320 | element_close(&ho, "div"); |
78c73085 |
1321 | } |
1322 | |
1323 | if (!done_version_ids) { |
1324 | /* |
1325 | * If the user didn't want the version IDs |
1326 | * visible, I think we still have a duty to put |
1327 | * them in an HTML comment. |
1328 | */ |
1329 | int started = FALSE; |
1330 | for (p = sourceform; p; p = p->next) |
1331 | if (p->type == para_VersionID) { |
1332 | if (!started) { |
1333 | html_raw(&ho, "<!-- version IDs:\n"); |
1334 | started = TRUE; |
1335 | } |
1336 | html_words(&ho, p->words, NOTHING, |
1337 | f, keywords, &conf); |
1338 | html_nl(&ho); |
1339 | } |
1340 | if (started) |
1341 | html_raw(&ho, "-->\n"); |
1342 | } |
1343 | } |
1344 | |
1345 | element_close(&ho, "body"); |
1346 | html_nl(&ho); |
1347 | element_close(&ho, "html"); |
1348 | html_nl(&ho); |
1349 | cleanup(&ho); |
1350 | } |
1351 | } |
1352 | |
1353 | /* |
1b7bf715 |
1354 | * Go through and check that no index fragments were referenced |
1355 | * without being generated, or indeed vice versa. |
1356 | * |
1357 | * (When I actually get round to freeing everything, this can |
1358 | * probably be the freeing loop as well.) |
1359 | */ |
1360 | for (p = sourceform; p; p = p->next) { |
1361 | word *w; |
1362 | for (w = p->words; w; w = w->next) |
1363 | if (w->type == word_IndexRef) { |
1364 | htmlindexref *hr = (htmlindexref *)w->private_data; |
1365 | |
1366 | assert(!hr->referenced == !hr->generated); |
1367 | } |
1368 | } |
1369 | |
1370 | /* |
3e82de8f |
1371 | * FIXME: Free all the working data. |
78c73085 |
1372 | */ |
1373 | } |
1374 | |
1375 | static void html_file_section(htmlconfig *cfg, htmlfilelist *files, |
1376 | htmlsect *sect, int depth) |
1377 | { |
1378 | htmlfile *file; |
1379 | int ldepth; |
1380 | |
1381 | /* |
1382 | * `depth' is derived from the heading_depth() macro at the top |
1383 | * of this file, which counts title as -1, chapter as 0, |
1384 | * heading as 1 and subsection as 2. However, the semantics of |
1385 | * cfg->leaf_level are defined to count chapter as 1, heading |
1386 | * as 2 etc. So first I increment depth :-( |
1387 | */ |
1388 | ldepth = depth + 1; |
1389 | |
1390 | if (cfg->leaf_level == 0) { |
1391 | /* |
1392 | * leaf_level==0 is a special case, in which everything is |
1393 | * put into a single file. |
1394 | */ |
1395 | if (!files->single) |
1396 | files->single = html_new_file(files, cfg->single_filename); |
1397 | |
1398 | file = files->single; |
1399 | } else { |
1400 | /* |
1401 | * If the depth of this section is at or above leaf_level, |
1402 | * we invent a fresh file and put this section at its head. |
1403 | * Otherwise, we put it in the same file as its parent |
1404 | * section. |
1405 | */ |
1406 | if (ldepth > cfg->leaf_level) { |
1407 | /* |
1408 | * We know that sect->parent cannot be NULL. The only |
1409 | * circumstance in which it can be is if sect is at |
1410 | * chapter or appendix level, i.e. ldepth==1; and if |
1411 | * that's the case, then we cannot have entered this |
1412 | * branch unless cfg->leaf_level==0, in which case we |
1413 | * would be in the single-file case above and not here |
1414 | * at all. |
1415 | */ |
1416 | assert(sect->parent); |
1417 | |
1418 | file = sect->parent->file; |
1419 | } else { |
1420 | if (sect->type == TOP) { |
1421 | file = html_new_file(files, cfg->contents_filename); |
1422 | } else if (sect->type == INDEX) { |
1423 | file = html_new_file(files, cfg->index_filename); |
1424 | } else { |
1425 | char *title; |
1426 | |
1427 | assert(ldepth > 0 && sect->title); |
1428 | title = html_format(sect->title, cfg->template_filename); |
1429 | file = html_new_file(files, title); |
1430 | sfree(title); |
1431 | } |
1432 | } |
1433 | } |
1434 | |
1435 | sect->file = file; |
1436 | |
1437 | if (file->min_heading_depth > depth) { |
1438 | /* |
1439 | * This heading is at a higher level than any heading we |
1440 | * have so far placed in this file; so we set the `first' |
1441 | * pointer. |
1442 | */ |
1443 | file->min_heading_depth = depth; |
1444 | file->first = sect; |
1445 | } |
1446 | |
1447 | if (file->min_heading_depth == depth) |
1448 | file->last = sect; |
1449 | } |
1450 | |
1451 | static htmlfile *html_new_file(htmlfilelist *list, char *filename) |
1452 | { |
f1530049 |
1453 | htmlfile *ret = snew(htmlfile); |
78c73085 |
1454 | |
1455 | ret->next = NULL; |
1456 | if (list->tail) |
1457 | list->tail->next = ret; |
1458 | else |
1459 | list->head = ret; |
1460 | list->tail = ret; |
1461 | |
1462 | ret->filename = dupstr(filename); |
1463 | ret->last_fragment_number = 0; |
1464 | ret->min_heading_depth = INT_MAX; |
1465 | ret->first = ret->last = NULL; |
1466 | |
1467 | return ret; |
1468 | } |
1469 | |
1470 | static htmlsect *html_new_sect(htmlsectlist *list, paragraph *title) |
1471 | { |
f1530049 |
1472 | htmlsect *ret = snew(htmlsect); |
78c73085 |
1473 | |
1474 | ret->next = NULL; |
1475 | if (list->tail) |
1476 | list->tail->next = ret; |
1477 | else |
1478 | list->head = ret; |
1479 | list->tail = ret; |
1480 | |
1481 | ret->title = title; |
1482 | ret->file = NULL; |
1483 | ret->parent = NULL; |
1484 | ret->type = NORMAL; |
1485 | |
1486 | return ret; |
1487 | } |
1488 | |
1489 | static void html_words(htmloutput *ho, word *words, int flags, |
1490 | htmlfile *file, keywordlist *keywords, htmlconfig *cfg) |
1491 | { |
1492 | word *w; |
1493 | char *c; |
1494 | int style, type; |
1495 | |
1496 | for (w = words; w; w = w->next) switch (w->type) { |
1497 | case word_HyperLink: |
1498 | if (flags & LINKS) { |
1499 | element_open(ho, "a"); |
1500 | c = utoa_dup(w->text, CS_ASCII); |
1501 | element_attr(ho, "href", c); |
1502 | sfree(c); |
1503 | } |
1504 | break; |
1505 | case word_UpperXref: |
1506 | case word_LowerXref: |
1507 | if (flags & LINKS) { |
1508 | keyword *kwl = kw_lookup(keywords, w->text); |
1509 | paragraph *p = kwl->para; |
1510 | htmlsect *s = (htmlsect *)p->private_data; |
1511 | |
1512 | assert(s); |
1513 | |
1514 | html_href(ho, file, s->file, s->fragment); |
1515 | } |
1516 | break; |
1517 | case word_HyperEnd: |
1518 | case word_XrefEnd: |
1519 | if (flags & LINKS) |
1520 | element_close(ho, "a"); |
1521 | break; |
1522 | case word_IndexRef: |
1523 | if (flags & INDEXENTS) { |
1524 | htmlindexref *hr = (htmlindexref *)w->private_data; |
27bdc5ab |
1525 | html_fragment(ho, hr->fragment); |
1b7bf715 |
1526 | hr->generated = TRUE; |
78c73085 |
1527 | } |
1528 | break; |
1529 | case word_Normal: |
1530 | case word_Emph: |
1531 | case word_Code: |
1532 | case word_WeakCode: |
1533 | case word_WhiteSpace: |
1534 | case word_EmphSpace: |
1535 | case word_CodeSpace: |
1536 | case word_WkCodeSpace: |
1537 | case word_Quote: |
1538 | case word_EmphQuote: |
1539 | case word_CodeQuote: |
1540 | case word_WkCodeQuote: |
1541 | style = towordstyle(w->type); |
1542 | type = removeattr(w->type); |
1543 | if (style == word_Emph && |
1544 | (attraux(w->aux) == attr_First || |
1545 | attraux(w->aux) == attr_Only) && |
1546 | (flags & MARKUP)) |
1547 | element_open(ho, "em"); |
1548 | else if ((style == word_Code || style == word_WeakCode) && |
1549 | (attraux(w->aux) == attr_First || |
1550 | attraux(w->aux) == attr_Only) && |
1551 | (flags & MARKUP)) |
1552 | element_open(ho, "code"); |
1553 | |
1554 | if (type == word_WhiteSpace) |
1555 | html_text(ho, L" "); |
1556 | else if (type == word_Quote) { |
1557 | if (quoteaux(w->aux) == quote_Open) |
1558 | html_text(ho, cfg->lquote); |
1559 | else |
1560 | html_text(ho, cfg->rquote); |
1561 | } else { |
1562 | if (cvt_ok(ho->charset, w->text) || !w->alt) |
1563 | html_text(ho, w->text); |
1564 | else |
1565 | html_words(ho, w->alt, flags, file, keywords, cfg); |
1566 | } |
1567 | |
1568 | if (style == word_Emph && |
1569 | (attraux(w->aux) == attr_Last || |
1570 | attraux(w->aux) == attr_Only) && |
1571 | (flags & MARKUP)) |
1572 | element_close(ho, "em"); |
1573 | else if ((style == word_Code || style == word_WeakCode) && |
1574 | (attraux(w->aux) == attr_Last || |
1575 | attraux(w->aux) == attr_Only) && |
1576 | (flags & MARKUP)) |
1577 | element_close(ho, "code"); |
1578 | |
1579 | break; |
1580 | } |
1581 | } |
1582 | |
1583 | static void html_codepara(htmloutput *ho, word *words) |
1584 | { |
1585 | element_open(ho, "pre"); |
1586 | element_open(ho, "code"); |
1587 | for (; words; words = words->next) if (words->type == word_WeakCode) { |
1588 | char *open_tag; |
1589 | wchar_t *t, *e; |
1590 | |
1591 | t = words->text; |
1592 | if (words->next && words->next->type == word_Emph) { |
1593 | e = words->next->text; |
1594 | words = words->next; |
1595 | } else |
1596 | e = NULL; |
1597 | |
1598 | while (e && *e && *t) { |
1599 | int n; |
1600 | int ec = *e; |
1601 | |
1602 | for (n = 0; t[n] && e[n] && e[n] == ec; n++); |
1603 | |
1604 | open_tag = NULL; |
1605 | if (ec == 'i') |
1606 | open_tag = "em"; |
1607 | else if (ec == 'b') |
1608 | open_tag = "b"; |
1609 | if (open_tag) |
1610 | element_open(ho, open_tag); |
1611 | |
1612 | html_text_limit(ho, t, n); |
1613 | |
1614 | if (open_tag) |
1615 | element_close(ho, open_tag); |
1616 | |
1617 | t += n; |
1618 | e += n; |
1619 | } |
1620 | html_text(ho, t); |
1621 | html_nl(ho); |
1622 | } |
1623 | element_close(ho, "code"); |
1624 | element_close(ho, "pre"); |
1625 | } |
1626 | |
1627 | static void html_charset_cleanup(htmloutput *ho) |
1628 | { |
1629 | char outbuf[256]; |
1630 | int bytes; |
1631 | |
1632 | bytes = charset_from_unicode(NULL, NULL, outbuf, lenof(outbuf), |
1633 | ho->charset, &ho->cstate, NULL); |
1634 | if (bytes > 0) |
1635 | fwrite(outbuf, 1, bytes, ho->fp); |
1636 | } |
1637 | |
1638 | static void return_to_neutral(htmloutput *ho) |
1639 | { |
1640 | if (ho->state == HO_IN_TEXT) { |
1641 | html_charset_cleanup(ho); |
1642 | } else if (ho->state == HO_IN_EMPTY_TAG && is_xhtml(ho->ver)) { |
1643 | fprintf(ho->fp, " />"); |
1644 | } else if (ho->state == HO_IN_EMPTY_TAG || ho->state == HO_IN_TAG) { |
1645 | fprintf(ho->fp, ">"); |
1646 | } |
1647 | |
1648 | ho->state = HO_NEUTRAL; |
1649 | } |
1650 | |
1651 | static void element_open(htmloutput *ho, char const *name) |
1652 | { |
1653 | return_to_neutral(ho); |
1654 | fprintf(ho->fp, "<%s", name); |
1655 | ho->state = HO_IN_TAG; |
1656 | } |
1657 | |
1658 | static void element_close(htmloutput *ho, char const *name) |
1659 | { |
1660 | return_to_neutral(ho); |
1661 | fprintf(ho->fp, "</%s>", name); |
1662 | ho->state = HO_NEUTRAL; |
1663 | } |
1664 | |
1665 | static void element_empty(htmloutput *ho, char const *name) |
1666 | { |
1667 | return_to_neutral(ho); |
1668 | fprintf(ho->fp, "<%s", name); |
1669 | ho->state = HO_IN_EMPTY_TAG; |
1670 | } |
1671 | |
1672 | static void html_nl(htmloutput *ho) |
1673 | { |
1674 | return_to_neutral(ho); |
1675 | fputc('\n', ho->fp); |
1676 | } |
1677 | |
1678 | static void html_raw(htmloutput *ho, char *text) |
1679 | { |
1680 | return_to_neutral(ho); |
1681 | fputs(text, ho->fp); |
1682 | } |
1683 | |
1684 | static void html_raw_as_attr(htmloutput *ho, char *text) |
1685 | { |
1686 | assert(ho->state == HO_IN_TAG || ho->state == HO_IN_EMPTY_TAG); |
1687 | fputc(' ', ho->fp); |
1688 | fputs(text, ho->fp); |
1689 | } |
1690 | |
1691 | static void element_attr(htmloutput *ho, char const *name, char const *value) |
1692 | { |
1693 | html_charset_cleanup(ho); |
1694 | assert(ho->state == HO_IN_TAG || ho->state == HO_IN_EMPTY_TAG); |
1695 | fprintf(ho->fp, " %s=\"%s\"", name, value); |
1696 | } |
1697 | |
1698 | static void element_attr_w(htmloutput *ho, char const *name, |
1699 | wchar_t const *value) |
1700 | { |
1701 | html_charset_cleanup(ho); |
1702 | fprintf(ho->fp, " %s=\"", name); |
1703 | html_text_limit_internal(ho, value, 0, TRUE); |
1704 | html_charset_cleanup(ho); |
1705 | fputc('"', ho->fp); |
1706 | } |
1707 | |
1708 | static void html_text(htmloutput *ho, wchar_t const *text) |
1709 | { |
1710 | html_text_limit(ho, text, 0); |
1711 | } |
1712 | |
1713 | static void html_text_limit(htmloutput *ho, wchar_t const *text, int maxlen) |
1714 | { |
1715 | return_to_neutral(ho); |
1716 | html_text_limit_internal(ho, text, maxlen, FALSE); |
1717 | } |
1718 | |
1719 | static void html_text_limit_internal(htmloutput *ho, wchar_t const *text, |
1720 | int maxlen, int quote_quotes) |
1721 | { |
1722 | int textlen = ustrlen(text); |
1723 | char outbuf[256]; |
1724 | int bytes, err; |
1725 | |
1726 | if (maxlen > 0 && textlen > maxlen) |
1727 | textlen = maxlen; |
1728 | |
1729 | while (textlen > 0) { |
1730 | /* Scan ahead for characters we really can't display in HTML. */ |
1731 | int lenbefore, lenafter; |
1732 | for (lenbefore = 0; lenbefore < textlen; lenbefore++) |
1733 | if (text[lenbefore] == L'<' || |
1734 | text[lenbefore] == L'>' || |
1735 | text[lenbefore] == L'&' || |
1736 | (text[lenbefore] == L'"' && quote_quotes)) |
1737 | break; |
1738 | lenafter = lenbefore; |
1739 | bytes = charset_from_unicode(&text, &lenafter, outbuf, lenof(outbuf), |
1740 | ho->charset, &ho->cstate, &err); |
1741 | textlen -= (lenbefore - lenafter); |
1742 | if (bytes > 0) |
1743 | fwrite(outbuf, 1, bytes, ho->fp); |
1744 | if (err) { |
1745 | /* |
1746 | * We have encountered a character that cannot be |
1747 | * displayed in the selected output charset. Therefore, |
1748 | * we use an HTML numeric entity reference. |
1749 | */ |
1750 | assert(textlen > 0); |
1751 | fprintf(ho->fp, "&#%ld;", (long int)*text); |
1752 | text++, textlen--; |
1753 | } else if (lenafter == 0 && textlen > 0) { |
1754 | /* |
1755 | * We have encountered a character which is special to |
1756 | * HTML. |
1757 | */ |
1758 | if (*text == L'<') |
1759 | fprintf(ho->fp, "<"); |
1760 | else if (*text == L'>') |
1761 | fprintf(ho->fp, ">"); |
1762 | else if (*text == L'&') |
1763 | fprintf(ho->fp, "&"); |
1764 | else if (*text == L'"') |
1765 | fprintf(ho->fp, """); |
1766 | else |
1767 | assert(!"Can't happen"); |
1768 | text++, textlen--; |
1769 | } |
1770 | } |
1771 | } |
1772 | |
1773 | static void cleanup(htmloutput *ho) |
1774 | { |
1775 | return_to_neutral(ho); |
1776 | fclose(ho->fp); |
1777 | } |
1778 | |
1779 | static void html_href(htmloutput *ho, htmlfile *thisfile, |
1780 | htmlfile *targetfile, char *targetfrag) |
1781 | { |
1782 | rdstringc rs = { 0, 0, NULL }; |
1783 | char *url; |
1784 | |
1785 | if (targetfile != thisfile) |
1786 | rdaddsc(&rs, targetfile->filename); |
1787 | if (targetfrag) { |
1788 | rdaddc(&rs, '#'); |
1789 | rdaddsc(&rs, targetfrag); |
1790 | } |
1791 | url = rs.text; |
1792 | |
1793 | element_open(ho, "a"); |
1794 | element_attr(ho, "href", url); |
1795 | sfree(url); |
1796 | } |
1797 | |
27bdc5ab |
1798 | static void html_fragment(htmloutput *ho, char const *fragment) |
1799 | { |
1800 | element_open(ho, "a"); |
1801 | element_attr(ho, "name", fragment); |
1802 | if (is_xhtml(ho->ver)) |
1803 | element_attr(ho, "id", fragment); |
1804 | element_close(ho, "a"); |
1805 | } |
1806 | |
78c73085 |
1807 | static char *html_format(paragraph *p, char *template_string) |
1808 | { |
1809 | char *c, *t; |
1810 | word *w; |
1811 | wchar_t *ws, wsbuf[2]; |
1812 | rdstringc rs = { 0, 0, NULL }; |
1813 | |
1814 | t = template_string; |
1815 | while (*t) { |
1816 | if (*t == '%' && t[1]) { |
1817 | int fmt; |
1818 | |
1819 | t++; |
1820 | fmt = *t++; |
1821 | |
1822 | if (fmt == '%') { |
1823 | rdaddc(&rs, fmt); |
1824 | continue; |
1825 | } |
1826 | |
1827 | w = NULL; |
1828 | ws = NULL; |
1829 | |
1830 | if (p->kwtext && fmt == 'n') |
1831 | w = p->kwtext; |
1832 | else if (p->kwtext2 && fmt == 'b') { |
1833 | /* |
1834 | * HTML fragment names must start with a letter, so |
1835 | * simply `1.2.3' is not adequate. In this case I'm |
1836 | * going to cheat slightly by prepending the first |
1837 | * character of the first word of kwtext, so that |
1838 | * we get `C1' for chapter 1, `S2.3' for section |
1839 | * 2.3 etc. |
1840 | */ |
1841 | if (p->kwtext && p->kwtext->text[0]) { |
1842 | ws = wsbuf; |
1843 | wsbuf[1] = '\0'; |
1844 | wsbuf[0] = p->kwtext->text[0]; |
1845 | } |
1846 | w = p->kwtext2; |
1847 | } else if (p->keyword && *p->keyword && fmt == 'k') |
1848 | ws = p->keyword; |
1849 | else |
1850 | w = p->words; |
1851 | |
1852 | if (ws) { |
1853 | c = utoa_dup(ws, CS_ASCII); |
1854 | rdaddsc(&rs,c); |
1855 | sfree(c); |
1856 | } |
1857 | |
1858 | while (w) { |
1859 | if (removeattr(w->type) == word_Normal) { |
1860 | c = utoa_dup(w->text, CS_ASCII); |
1861 | rdaddsc(&rs,c); |
1862 | sfree(c); |
1863 | } |
1864 | w = w->next; |
1865 | } |
1866 | } else { |
1867 | rdaddc(&rs, *t++); |
1868 | } |
1869 | } |
1870 | |
1871 | return rdtrimc(&rs); |
1872 | } |
1873 | |
3e82de8f |
1874 | static char *html_sanitise_fragment(htmlfilelist *files, htmlfile *file, |
1875 | char *text) |
78c73085 |
1876 | { |
1877 | /* |
1878 | * The HTML 4 spec's strictest definition of fragment names (<a |
1879 | * name> and "id" attributes) says that they `must begin with a |
1880 | * letter and may be followed by any number of letters, digits, |
1881 | * hyphens, underscores, colons, and periods'. |
1882 | * |
1883 | * So here we unceremoniously rip out any characters not |
1884 | * conforming to this limitation. |
1885 | */ |
1886 | char *p = text, *q = text; |
1887 | |
1888 | while (*p && !((*p>='A' && *p<='Z') || (*p>='a' && *p<='z'))) |
1889 | p++; |
3e82de8f |
1890 | if ((*q++ = *p++) != '\0') { |
1891 | while (*p) { |
1892 | if ((*p>='A' && *p<='Z') || |
1893 | (*p>='a' && *p<='z') || |
1894 | (*p>='0' && *p<='9') || |
1895 | *p=='-' || *p=='_' || *p==':' || *p=='.') |
1896 | *q++ = *p; |
1897 | p++; |
1898 | } |
1899 | |
1900 | *q = '\0'; |
1901 | } |
1902 | |
1903 | /* |
1904 | * Now we check for clashes with other fragment names, and |
1905 | * adjust this one if necessary by appending a hyphen followed |
1906 | * by a number. |
1907 | */ |
1908 | { |
1909 | htmlfragment *frag = snew(htmlfragment); |
1910 | int len = 0; /* >0 indicates we have resized */ |
1911 | int suffix = 1; |
1912 | |
1913 | frag->file = file; |
1914 | frag->fragment = text; |
1915 | |
1916 | while (add234(files->frags, frag) != frag) { |
1917 | if (!len) { |
1918 | len = strlen(text); |
1919 | frag->fragment = text = sresize(text, len+20, char); |
1920 | } |
1921 | |
1922 | sprintf(text + len, "-%d", ++suffix); |
1923 | } |
78c73085 |
1924 | } |
1925 | |
3e82de8f |
1926 | return text; |
78c73085 |
1927 | } |
1928 | |
1929 | static void html_contents_entry(htmloutput *ho, int depth, htmlsect *s, |
1930 | htmlfile *thisfile, keywordlist *keywords, |
1931 | htmlconfig *cfg) |
1932 | { |
1933 | while (ho->contents_level > depth) { |
1934 | element_close(ho, "ul"); |
1935 | ho->contents_level--; |
1936 | } |
1937 | |
1938 | while (ho->contents_level < depth) { |
1939 | element_open(ho, "ul"); |
1940 | ho->contents_level++; |
1941 | } |
1942 | |
1943 | if (!s) |
1944 | return; |
1945 | |
1946 | element_open(ho, "li"); |
1947 | html_href(ho, thisfile, s->file, s->fragment); |
23c9bbc2 |
1948 | html_section_title(ho, s, thisfile, keywords, cfg, FALSE); |
78c73085 |
1949 | element_close(ho, "a"); |
1950 | element_close(ho, "li"); |
1951 | } |
1952 | |
1953 | static void html_section_title(htmloutput *ho, htmlsect *s, htmlfile *thisfile, |
23c9bbc2 |
1954 | keywordlist *keywords, htmlconfig *cfg, |
1955 | int real) |
78c73085 |
1956 | { |
1957 | if (s->title) { |
1958 | sectlevel *sl; |
1959 | word *number; |
1960 | int depth = heading_depth(s->title); |
1961 | |
1962 | if (depth < 0) |
1963 | sl = NULL; |
1964 | else if (depth == 0) |
1965 | sl = &cfg->achapter; |
1966 | else if (depth <= cfg->nasect) |
1967 | sl = &cfg->asect[depth-1]; |
1968 | else |
1969 | sl = &cfg->asect[cfg->nasect-1]; |
1970 | |
1971 | if (!sl) |
1972 | number = NULL; |
1973 | else if (sl->just_numbers) |
1974 | number = s->title->kwtext2; |
1975 | else |
1976 | number = s->title->kwtext; |
1977 | |
1978 | if (number) { |
1979 | html_words(ho, number, MARKUP, |
1980 | thisfile, keywords, cfg); |
1981 | html_text(ho, sl->number_suffix); |
1982 | } |
1983 | |
23c9bbc2 |
1984 | html_words(ho, s->title->words, real ? ALL : MARKUP, |
78c73085 |
1985 | thisfile, keywords, cfg); |
1986 | } else { |
1987 | assert(s->type != NORMAL); |
1988 | if (s->type == TOP) |
1989 | html_text(ho, L"Preamble");/* FIXME: configure */ |
1990 | else if (s->type == INDEX) |
1991 | html_text(ho, L"Index");/* FIXME: configure */ |
1992 | } |
1993 | } |