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.) |
78c73085 |
13 | */ |
14 | |
15 | #include <stdio.h> |
16 | #include <stdlib.h> |
17 | #include <assert.h> |
18 | #include <limits.h> |
19 | #include "halibut.h" |
20 | |
21 | #define is_heading_type(type) ( (type) == para_Title || \ |
22 | (type) == para_Chapter || \ |
23 | (type) == para_Appendix || \ |
24 | (type) == para_UnnumberedChapter || \ |
25 | (type) == para_Heading || \ |
26 | (type) == para_Subsect) |
27 | |
28 | #define heading_depth(p) ( (p)->type == para_Subsect ? (p)->aux + 1 : \ |
29 | (p)->type == para_Heading ? 1 : \ |
30 | (p)->type == para_Title ? -1 : 0 ) |
31 | |
32 | typedef struct { |
33 | int just_numbers; |
34 | wchar_t *number_suffix; |
35 | } sectlevel; |
36 | |
37 | typedef struct { |
38 | int nasect; |
39 | sectlevel achapter, *asect; |
40 | int *contents_depths; /* 0=main, 1=chapter, 2=sect etc */ |
41 | int ncdepths; |
42 | int address_section, visible_version_id; |
43 | int leaf_contains_contents, leaf_smallest_contents; |
44 | char *contents_filename; |
45 | char *index_filename; |
46 | char *template_filename; |
47 | char *single_filename; |
12f0ee84 |
48 | char **template_fragments; |
49 | int ntfragments; |
78c73085 |
50 | char *head_end, *body_start, *body_end, *addr_start, *addr_end; |
51 | char *body_tag, *nav_attr; |
52 | wchar_t *author, *description; |
56a99eb6 |
53 | wchar_t *index_text, *contents_text, *preamble_text, *title_separator; |
54 | wchar_t *nav_prev_text, *nav_next_text, *nav_separator; |
55 | wchar_t *index_main_sep, *index_multi_sep; |
56 | wchar_t *pre_versionid, *post_versionid; |
78c73085 |
57 | int restrict_charset, output_charset; |
58 | enum { |
27bdc5ab |
59 | HTML_3_2, HTML_4, ISO_HTML, |
78c73085 |
60 | XHTML_1_0_TRANSITIONAL, XHTML_1_0_STRICT |
61 | } htmlver; |
62 | wchar_t *lquote, *rquote; |
63 | int leaf_level; |
64 | } htmlconfig; |
65 | |
66 | #define contents_depth(conf, level) \ |
67 | ( (conf).ncdepths > (level) ? (conf).contents_depths[level] : (level)+2 ) |
68 | |
69 | #define is_xhtml(ver) ((ver) >= XHTML_1_0_TRANSITIONAL) |
70 | |
71 | typedef struct htmlfile htmlfile; |
72 | typedef struct htmlsect htmlsect; |
73 | |
74 | struct htmlfile { |
75 | htmlfile *next; |
76 | char *filename; |
77 | int last_fragment_number; |
78 | int min_heading_depth; |
79 | htmlsect *first, *last; /* first/last highest-level sections */ |
80 | }; |
81 | |
82 | struct htmlsect { |
83 | htmlsect *next, *parent; |
84 | htmlfile *file; |
85 | paragraph *title, *text; |
86 | enum { NORMAL, TOP, INDEX } type; |
87 | int contents_depth; |
12f0ee84 |
88 | char **fragments; |
78c73085 |
89 | }; |
90 | |
91 | typedef struct { |
92 | htmlfile *head, *tail; |
93 | htmlfile *single, *index; |
3e82de8f |
94 | tree234 *frags; |
78c73085 |
95 | } htmlfilelist; |
96 | |
97 | typedef struct { |
98 | htmlsect *head, *tail; |
99 | } htmlsectlist; |
100 | |
101 | typedef struct { |
3e82de8f |
102 | htmlfile *file; |
103 | char *fragment; |
104 | } htmlfragment; |
105 | |
106 | typedef struct { |
78c73085 |
107 | int nrefs, refsize; |
108 | word **refs; |
109 | } htmlindex; |
110 | |
111 | typedef struct { |
112 | htmlsect *section; |
113 | char *fragment; |
1b7bf715 |
114 | int generated, referenced; |
78c73085 |
115 | } htmlindexref; |
116 | |
117 | typedef struct { |
118 | /* |
119 | * This level deals with charset conversion, starting and |
120 | * ending tags, and writing to the file. It's the lexical |
121 | * level. |
122 | */ |
123 | FILE *fp; |
b7309494 |
124 | int charset, restrict_charset; |
78c73085 |
125 | charset_state cstate; |
126 | int ver; |
127 | enum { |
128 | HO_NEUTRAL, HO_IN_TAG, HO_IN_EMPTY_TAG, HO_IN_TEXT |
129 | } state; |
130 | /* |
131 | * Stuff beyond here deals with the higher syntactic level: it |
132 | * tracks how many levels of <ul> are currently open when |
133 | * producing a contents list, for example. |
134 | */ |
135 | int contents_level; |
136 | } htmloutput; |
137 | |
3e82de8f |
138 | static int html_fragment_compare(void *av, void *bv) |
139 | { |
140 | htmlfragment *a = (htmlfragment *)av; |
141 | htmlfragment *b = (htmlfragment *)bv; |
142 | int cmp; |
143 | |
144 | if ((cmp = strcmp(a->file->filename, b->file->filename)) != 0) |
145 | return cmp; |
146 | else |
147 | return strcmp(a->fragment, b->fragment); |
148 | } |
149 | |
78c73085 |
150 | static void html_file_section(htmlconfig *cfg, htmlfilelist *files, |
151 | htmlsect *sect, int depth); |
152 | |
153 | static htmlfile *html_new_file(htmlfilelist *list, char *filename); |
12f0ee84 |
154 | static htmlsect *html_new_sect(htmlsectlist *list, paragraph *title, |
155 | htmlconfig *cfg); |
78c73085 |
156 | |
157 | /* Flags for html_words() flags parameter */ |
158 | #define NOTHING 0x00 |
159 | #define MARKUP 0x01 |
160 | #define LINKS 0x02 |
161 | #define INDEXENTS 0x04 |
162 | #define ALL 0x07 |
163 | static void html_words(htmloutput *ho, word *words, int flags, |
164 | htmlfile *file, keywordlist *keywords, htmlconfig *cfg); |
165 | static void html_codepara(htmloutput *ho, word *words); |
166 | |
167 | static void element_open(htmloutput *ho, char const *name); |
168 | static void element_close(htmloutput *ho, char const *name); |
169 | static void element_empty(htmloutput *ho, char const *name); |
170 | static void element_attr(htmloutput *ho, char const *name, char const *value); |
171 | static void element_attr_w(htmloutput *ho, char const *name, |
172 | wchar_t const *value); |
173 | static void html_text(htmloutput *ho, wchar_t const *str); |
35b123a0 |
174 | static void html_text_nbsp(htmloutput *ho, wchar_t const *str); |
78c73085 |
175 | static void html_text_limit(htmloutput *ho, wchar_t const *str, int maxlen); |
176 | static void html_text_limit_internal(htmloutput *ho, wchar_t const *text, |
35b123a0 |
177 | int maxlen, int quote_quotes, int nbsp); |
78c73085 |
178 | static void html_nl(htmloutput *ho); |
179 | static void html_raw(htmloutput *ho, char *text); |
180 | static void html_raw_as_attr(htmloutput *ho, char *text); |
181 | static void cleanup(htmloutput *ho); |
182 | |
183 | static void html_href(htmloutput *ho, htmlfile *thisfile, |
184 | htmlfile *targetfile, char *targetfrag); |
27bdc5ab |
185 | static void html_fragment(htmloutput *ho, char const *fragment); |
78c73085 |
186 | |
187 | static char *html_format(paragraph *p, char *template_string); |
3e82de8f |
188 | static char *html_sanitise_fragment(htmlfilelist *files, htmlfile *file, |
189 | char *text); |
78c73085 |
190 | |
191 | static void html_contents_entry(htmloutput *ho, int depth, htmlsect *s, |
192 | htmlfile *thisfile, keywordlist *keywords, |
193 | htmlconfig *cfg); |
194 | static void html_section_title(htmloutput *ho, htmlsect *s, |
195 | htmlfile *thisfile, keywordlist *keywords, |
23c9bbc2 |
196 | htmlconfig *cfg, int real); |
78c73085 |
197 | |
198 | static htmlconfig html_configure(paragraph *source) { |
199 | htmlconfig ret; |
200 | paragraph *p; |
201 | |
202 | /* |
203 | * Defaults. |
204 | */ |
205 | ret.leaf_level = 2; |
206 | ret.achapter.just_numbers = FALSE; |
207 | ret.achapter.number_suffix = L": "; |
208 | ret.nasect = 1; |
f1530049 |
209 | ret.asect = snewn(ret.nasect, sectlevel); |
78c73085 |
210 | ret.asect[0].just_numbers = TRUE; |
211 | ret.asect[0].number_suffix = L" "; |
212 | ret.ncdepths = 0; |
213 | ret.contents_depths = 0; |
214 | ret.visible_version_id = TRUE; |
215 | ret.address_section = TRUE; |
216 | ret.leaf_contains_contents = FALSE; |
217 | ret.leaf_smallest_contents = 4; |
218 | ret.single_filename = dupstr("Manual.html"); |
219 | ret.contents_filename = dupstr("Contents.html"); |
220 | ret.index_filename = dupstr("IndexPage.html"); |
221 | ret.template_filename = dupstr("%n.html"); |
12f0ee84 |
222 | ret.ntfragments = 1; |
223 | ret.template_fragments = snewn(ret.ntfragments, char *); |
224 | ret.template_fragments[0] = dupstr("%b"); |
78c73085 |
225 | ret.head_end = ret.body_tag = ret.body_start = ret.body_end = |
226 | ret.addr_start = ret.addr_end = ret.nav_attr = NULL; |
227 | ret.author = ret.description = NULL; |
b7309494 |
228 | ret.restrict_charset = CS_UTF8; |
78c73085 |
229 | ret.output_charset = CS_ASCII; |
230 | ret.htmlver = HTML_4; |
56a99eb6 |
231 | ret.index_text = L"Index"; |
232 | ret.contents_text = L"Contents"; |
233 | ret.preamble_text = L"Preamble"; |
234 | ret.title_separator = L" - "; |
235 | ret.nav_prev_text = L"Previous"; |
236 | ret.nav_next_text = L"Next"; |
237 | ret.nav_separator = L" | "; |
238 | ret.index_main_sep = L": "; |
239 | ret.index_multi_sep = L", "; |
240 | ret.pre_versionid = L"["; |
241 | ret.post_versionid = L"]"; |
78c73085 |
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 | |
b7309494 |
273 | if (!ustricmp(k, L"html-restrict-charset")) { |
0960a3d8 |
274 | ret.restrict_charset = charset_from_ustr(&p->fpos, uadv(k)); |
b7309494 |
275 | } else if (!ustricmp(k, L"html-output-charset")) { |
0960a3d8 |
276 | ret.output_charset = charset_from_ustr(&p->fpos, uadv(k)); |
27bdc5ab |
277 | } else if (!ustricmp(k, L"html-version")) { |
278 | wchar_t *vername = uadv(k); |
279 | static const struct { |
280 | const wchar_t *name; |
281 | int ver; |
282 | } versions[] = { |
283 | {L"html3.2", HTML_3_2}, |
284 | {L"html4", HTML_4}, |
285 | {L"iso-html", ISO_HTML}, |
286 | {L"xhtml1.0transitional", XHTML_1_0_TRANSITIONAL}, |
287 | {L"xhtml1.0strict", XHTML_1_0_STRICT} |
288 | }; |
289 | int i; |
290 | |
291 | for (i = 0; i < (int)lenof(versions); i++) |
292 | if (!ustricmp(versions[i].name, vername)) |
293 | break; |
294 | |
295 | if (i == lenof(versions)) |
296 | error(err_htmlver, &p->fpos, vername); |
297 | else |
298 | ret.htmlver = versions[i].ver; |
78c73085 |
299 | } else if (!ustricmp(k, L"html-single-filename")) { |
300 | sfree(ret.single_filename); |
301 | ret.single_filename = dupstr(adv(p->origkeyword)); |
302 | } else if (!ustricmp(k, L"html-contents-filename")) { |
303 | sfree(ret.contents_filename); |
304 | ret.contents_filename = dupstr(adv(p->origkeyword)); |
305 | } else if (!ustricmp(k, L"html-index-filename")) { |
306 | sfree(ret.index_filename); |
307 | ret.index_filename = dupstr(adv(p->origkeyword)); |
308 | } else if (!ustricmp(k, L"html-template-filename")) { |
309 | sfree(ret.template_filename); |
310 | ret.template_filename = dupstr(adv(p->origkeyword)); |
311 | } else if (!ustricmp(k, L"html-template-fragment")) { |
12f0ee84 |
312 | char *frag = adv(p->origkeyword); |
313 | if (*frag) { |
314 | while (ret.ntfragments--) |
315 | sfree(ret.template_fragments[ret.ntfragments]); |
316 | sfree(ret.template_fragments); |
317 | ret.template_fragments = NULL; |
318 | ret.ntfragments = 0; |
319 | while (*frag) { |
320 | ret.ntfragments++; |
321 | ret.template_fragments = |
322 | sresize(ret.template_fragments, |
323 | ret.ntfragments, char *); |
324 | ret.template_fragments[ret.ntfragments-1] = |
325 | dupstr(frag); |
326 | frag = adv(frag); |
327 | } |
328 | } else |
329 | error(err_cfginsufarg, &p->fpos, p->origkeyword, 1); |
78c73085 |
330 | } else if (!ustricmp(k, L"html-chapter-numeric")) { |
331 | ret.achapter.just_numbers = utob(uadv(k)); |
332 | } else if (!ustricmp(k, L"html-chapter-suffix")) { |
333 | ret.achapter.number_suffix = uadv(k); |
334 | } else if (!ustricmp(k, L"html-leaf-level")) { |
335 | ret.leaf_level = utoi(uadv(k)); |
336 | } else if (!ustricmp(k, L"html-section-numeric")) { |
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 | ret.nasect = n+1; |
349 | } |
350 | ret.asect[n].just_numbers = utob(q); |
351 | } else if (!ustricmp(k, L"html-section-suffix")) { |
352 | wchar_t *q = uadv(k); |
353 | int n = 0; |
354 | if (uisdigit(*q)) { |
355 | n = utoi(q); |
356 | q = uadv(q); |
357 | } |
358 | if (n >= ret.nasect) { |
359 | int i; |
f1530049 |
360 | ret.asect = sresize(ret.asect, n+1, sectlevel); |
78c73085 |
361 | for (i = ret.nasect; i <= n; i++) { |
362 | ret.asect[i] = ret.asect[ret.nasect-1]; |
363 | } |
364 | ret.nasect = n+1; |
365 | } |
366 | ret.asect[n].number_suffix = q; |
367 | } else if (!ustricmp(k, L"html-contents-depth") || |
368 | !ustrnicmp(k, L"html-contents-depth-", 20)) { |
369 | /* |
370 | * Relic of old implementation: this directive used |
371 | * to be written as \cfg{html-contents-depth-3}{2} |
372 | * rather than the usual Halibut convention of |
373 | * \cfg{html-contents-depth}{3}{2}. We therefore |
374 | * support both. |
375 | */ |
376 | wchar_t *q = k[19] ? k+20 : uadv(k); |
377 | int n = 0; |
378 | if (uisdigit(*q)) { |
379 | n = utoi(q); |
380 | q = uadv(q); |
381 | } |
382 | if (n >= ret.ncdepths) { |
383 | int i; |
f1530049 |
384 | ret.contents_depths = |
385 | sresize(ret.contents_depths, n+1, int); |
78c73085 |
386 | for (i = ret.ncdepths; i <= n; i++) { |
387 | ret.contents_depths[i] = i+2; |
388 | } |
389 | ret.ncdepths = n+1; |
390 | } |
391 | ret.contents_depths[n] = utoi(q); |
392 | } else if (!ustricmp(k, L"html-head-end")) { |
393 | ret.head_end = adv(p->origkeyword); |
394 | } else if (!ustricmp(k, L"html-body-tag")) { |
395 | ret.body_tag = adv(p->origkeyword); |
396 | } else if (!ustricmp(k, L"html-body-start")) { |
397 | ret.body_start = adv(p->origkeyword); |
398 | } else if (!ustricmp(k, L"html-body-end")) { |
399 | ret.body_end = adv(p->origkeyword); |
400 | } else if (!ustricmp(k, L"html-address-start")) { |
401 | ret.addr_start = adv(p->origkeyword); |
402 | } else if (!ustricmp(k, L"html-address-end")) { |
403 | ret.addr_end = adv(p->origkeyword); |
404 | } else if (!ustricmp(k, L"html-navigation-attributes")) { |
405 | ret.nav_attr = adv(p->origkeyword); |
406 | } else if (!ustricmp(k, L"html-author")) { |
407 | ret.author = uadv(k); |
408 | } else if (!ustricmp(k, L"html-description")) { |
409 | ret.description = uadv(k); |
410 | } else if (!ustricmp(k, L"html-suppress-address")) { |
411 | ret.address_section = !utob(uadv(k)); |
412 | } else if (!ustricmp(k, L"html-versionid")) { |
413 | ret.visible_version_id = utob(uadv(k)); |
414 | } else if (!ustricmp(k, L"html-quotes")) { |
415 | if (*uadv(k) && *uadv(uadv(k))) { |
416 | ret.lquote = uadv(k); |
417 | ret.rquote = uadv(ret.lquote); |
418 | } |
419 | } else if (!ustricmp(k, L"html-leaf-contains-contents")) { |
420 | ret.leaf_contains_contents = utob(uadv(k)); |
421 | } else if (!ustricmp(k, L"html-leaf-smallest-contents")) { |
422 | ret.leaf_smallest_contents = utoi(uadv(k)); |
75a96e91 |
423 | } else if (!ustricmp(k, L"html-index-text")) { |
424 | ret.index_text = uadv(k); |
425 | } else if (!ustricmp(k, L"html-contents-text")) { |
426 | ret.contents_text = uadv(k); |
427 | } else if (!ustricmp(k, L"html-preamble-text")) { |
428 | ret.preamble_text = uadv(k); |
429 | } else if (!ustricmp(k, L"html-title-separator")) { |
430 | ret.title_separator = uadv(k); |
431 | } else if (!ustricmp(k, L"html-nav-prev-text")) { |
432 | ret.nav_prev_text = uadv(k); |
433 | } else if (!ustricmp(k, L"html-nav-next-text")) { |
434 | ret.nav_next_text = uadv(k); |
435 | } else if (!ustricmp(k, L"html-nav-separator")) { |
436 | ret.nav_separator = uadv(k); |
437 | } else if (!ustricmp(k, L"html-index-main-separator")) { |
438 | ret.index_main_sep = uadv(k); |
439 | } else if (!ustricmp(k, L"html-index-multiple-separator")) { |
440 | ret.index_multi_sep = uadv(k); |
441 | } else if (!ustricmp(k, L"html-pre-versionid")) { |
442 | ret.pre_versionid = uadv(k); |
443 | } else if (!ustricmp(k, L"html-post-versionid")) { |
444 | ret.post_versionid = uadv(k); |
78c73085 |
445 | } |
446 | } |
447 | } |
448 | |
449 | /* |
450 | * Now process fallbacks on quote characters. |
451 | */ |
452 | while (*uadv(ret.rquote) && *uadv(uadv(ret.rquote)) && |
453 | (!cvt_ok(ret.restrict_charset, ret.lquote) || |
454 | !cvt_ok(ret.restrict_charset, ret.rquote))) { |
455 | ret.lquote = uadv(ret.rquote); |
456 | ret.rquote = uadv(ret.lquote); |
457 | } |
458 | |
459 | return ret; |
460 | } |
461 | |
462 | paragraph *html_config_filename(char *filename) |
463 | { |
464 | /* |
465 | * If the user passes in a single filename as a parameter to |
466 | * the `--html' command-line option, then we should assume it |
467 | * to imply _two_ config directives: |
468 | * \cfg{html-single-filename}{whatever} and |
469 | * \cfg{html-leaf-level}{0}; the rationale being that the user |
470 | * wants their output _in that file_. |
471 | */ |
472 | paragraph *p, *q; |
473 | |
474 | p = cmdline_cfg_simple("html-single-filename", filename, NULL); |
475 | q = cmdline_cfg_simple("html-leaf-level", "0", NULL); |
476 | p->next = q; |
477 | return p; |
478 | } |
479 | |
480 | void html_backend(paragraph *sourceform, keywordlist *keywords, |
529a6c83 |
481 | indexdata *idx, void *unused) |
482 | { |
78c73085 |
483 | paragraph *p; |
484 | htmlconfig conf; |
3e82de8f |
485 | htmlfilelist files = { NULL, NULL, NULL, NULL, NULL }; |
78c73085 |
486 | htmlsectlist sects = { NULL, NULL }, nonsects = { NULL, NULL }; |
03fcb340 |
487 | int has_index; |
78c73085 |
488 | |
489 | IGNORE(unused); |
490 | |
491 | conf = html_configure(sourceform); |
492 | |
493 | /* |
494 | * We're going to make heavy use of paragraphs' private data |
495 | * fields in the forthcoming code. Clear them first, so we can |
496 | * reliably tell whether we have auxiliary data for a |
497 | * particular paragraph. |
498 | */ |
499 | for (p = sourceform; p; p = p->next) |
500 | p->private_data = NULL; |
501 | |
3e82de8f |
502 | files.frags = newtree234(html_fragment_compare); |
503 | |
78c73085 |
504 | /* |
505 | * Start by figuring out into which file each piece of the |
506 | * document should be put. We'll do this by inventing an |
507 | * `htmlsect' structure and stashing it in the private_data |
508 | * field of each section paragraph; we also need one additional |
509 | * htmlsect for the document index, which won't show up in the |
510 | * source form but needs to be consistently mentioned in |
511 | * contents links. |
512 | * |
12f0ee84 |
513 | * While we're here, we'll also invent the HTML fragment name(s) |
78c73085 |
514 | * for each section. |
515 | */ |
516 | { |
517 | htmlsect *topsect, *sect; |
518 | int d; |
519 | |
12f0ee84 |
520 | topsect = html_new_sect(§s, NULL, &conf); |
78c73085 |
521 | topsect->type = TOP; |
522 | topsect->title = NULL; |
523 | topsect->text = sourceform; |
524 | topsect->contents_depth = contents_depth(conf, 0); |
525 | html_file_section(&conf, &files, topsect, -1); |
78c73085 |
526 | |
527 | for (p = sourceform; p; p = p->next) |
528 | if (is_heading_type(p->type)) { |
529 | d = heading_depth(p); |
530 | |
531 | if (p->type == para_Title) { |
532 | topsect->title = p; |
533 | continue; |
534 | } |
535 | |
12f0ee84 |
536 | sect = html_new_sect(§s, p, &conf); |
78c73085 |
537 | sect->text = p->next; |
538 | |
539 | sect->contents_depth = contents_depth(conf, d+1) - (d+1); |
540 | |
541 | if (p->parent) { |
542 | sect->parent = (htmlsect *)p->parent->private_data; |
543 | assert(sect->parent != NULL); |
544 | } else |
545 | sect->parent = topsect; |
546 | p->private_data = sect; |
547 | |
548 | html_file_section(&conf, &files, sect, d); |
549 | |
12f0ee84 |
550 | { |
551 | int i; |
552 | for (i=0; i < conf.ntfragments; i++) { |
553 | sect->fragments[i] = |
554 | html_format(p, conf.template_fragments[i]); |
555 | sect->fragments[i] = |
556 | html_sanitise_fragment(&files, sect->file, |
557 | sect->fragments[i]); |
558 | } |
559 | } |
78c73085 |
560 | } |
561 | |
03fcb340 |
562 | /* And the index, if we have one. */ |
563 | has_index = (count234(idx->entries) > 0); |
564 | if (has_index) { |
12f0ee84 |
565 | sect = html_new_sect(§s, NULL, &conf); |
03fcb340 |
566 | sect->text = NULL; |
567 | sect->type = INDEX; |
568 | sect->parent = topsect; |
cdb986cc |
569 | sect->contents_depth = 0; |
03fcb340 |
570 | html_file_section(&conf, &files, sect, 0); /* peer of chapters */ |
12f0ee84 |
571 | sect->fragments[0] = utoa_dup(conf.index_text, CS_ASCII); |
572 | sect->fragments[0] = html_sanitise_fragment(&files, sect->file, |
573 | sect->fragments[0]); |
03fcb340 |
574 | files.index = sect->file; |
575 | } |
78c73085 |
576 | } |
577 | |
578 | /* |
579 | * Go through the keyword list and sort out fragment IDs for |
580 | * all the potentially referenced paragraphs which _aren't_ |
581 | * headings. |
582 | */ |
583 | { |
584 | int i; |
585 | keyword *kw; |
586 | htmlsect *sect; |
587 | |
588 | for (i = 0; (kw = index234(keywords->keys, i)) != NULL; i++) { |
589 | paragraph *q, *p = kw->para; |
590 | |
591 | if (!is_heading_type(p->type)) { |
592 | htmlsect *parent; |
593 | |
594 | /* |
595 | * Find the paragraph's parent htmlsect, to |
596 | * determine which file it will end up in. |
597 | */ |
598 | q = p->parent; |
599 | if (!q) { |
600 | /* |
601 | * Preamble paragraphs have no parent. So if we |
602 | * have a non-heading with no parent, it must |
603 | * be preamble, and therefore its parent |
604 | * htmlsect must be the preamble one. |
605 | */ |
606 | assert(sects.head && |
607 | sects.head->type == TOP); |
608 | parent = sects.head; |
609 | } else |
610 | parent = (htmlsect *)q->private_data; |
611 | |
612 | /* |
613 | * Now we can construct an htmlsect for this |
614 | * paragraph itself, taking care to put it in the |
615 | * list of non-sections rather than the list of |
616 | * sections (so that traverses of the `sects' list |
617 | * won't attempt to add it to the contents or |
618 | * anything weird like that). |
619 | */ |
12f0ee84 |
620 | sect = html_new_sect(&nonsects, p, &conf); |
78c73085 |
621 | sect->file = parent->file; |
622 | sect->parent = parent; |
623 | p->private_data = sect; |
624 | |
625 | /* |
04781c84 |
626 | * Fragment IDs for these paragraphs will simply be |
627 | * `p' followed by an integer. |
78c73085 |
628 | */ |
12f0ee84 |
629 | sect->fragments[0] = snewn(40, char); |
630 | sprintf(sect->fragments[0], "p%d", |
04781c84 |
631 | sect->file->last_fragment_number++); |
12f0ee84 |
632 | sect->fragments[0] = html_sanitise_fragment(&files, sect->file, |
633 | sect->fragments[0]); |
78c73085 |
634 | } |
635 | } |
636 | } |
637 | |
638 | /* |
04781c84 |
639 | * Reset the fragment numbers in each file. I've just used them |
640 | * to generate `p' fragment IDs for non-section paragraphs |
641 | * (numbered list elements, bibliocited), and now I want to use |
642 | * them for `i' fragment IDs for index entries. |
643 | */ |
644 | { |
645 | htmlfile *file; |
646 | for (file = files.head; file; file = file->next) |
647 | file->last_fragment_number = 0; |
648 | } |
649 | |
650 | /* |
78c73085 |
651 | * Now sort out the index. This involves: |
652 | * |
653 | * - For each index term, we set up an htmlindex structure to |
654 | * store all the references to that term. |
655 | * |
656 | * - Then we make a pass over the actual document, finding |
657 | * every word_IndexRef; for each one, we actually figure out |
658 | * the HTML filename/fragment pair we will use to reference |
659 | * it, store that information in the private data field of |
660 | * the word_IndexRef itself (so we can recreate it when the |
661 | * time comes to output our HTML), and add a reference to it |
662 | * to the index term in question. |
663 | */ |
664 | { |
665 | int i; |
666 | indexentry *entry; |
667 | htmlsect *lastsect; |
668 | word *w; |
669 | |
670 | /* |
671 | * Set up the htmlindex structures. |
672 | */ |
673 | |
674 | for (i = 0; (entry = index234(idx->entries, i)) != NULL; i++) { |
f1530049 |
675 | htmlindex *hi = snew(htmlindex); |
78c73085 |
676 | |
677 | hi->nrefs = hi->refsize = 0; |
678 | hi->refs = NULL; |
679 | |
680 | entry->backend_data = hi; |
681 | } |
682 | |
683 | /* |
684 | * Run over the document inventing fragments. Each fragment |
685 | * is of the form `i' followed by an integer. |
78c73085 |
686 | */ |
56a99eb6 |
687 | lastsect = sects.head; /* this is always the top section */ |
78c73085 |
688 | for (p = sourceform; p; p = p->next) { |
56a99eb6 |
689 | if (is_heading_type(p->type) && p->type != para_Title) |
78c73085 |
690 | lastsect = (htmlsect *)p->private_data; |
691 | |
692 | for (w = p->words; w; w = w->next) |
693 | if (w->type == word_IndexRef) { |
f1530049 |
694 | htmlindexref *hr = snew(htmlindexref); |
78c73085 |
695 | indextag *tag; |
696 | int i; |
697 | |
1b7bf715 |
698 | hr->referenced = hr->generated = FALSE; |
78c73085 |
699 | hr->section = lastsect; |
78c73085 |
700 | { |
701 | char buf[40]; |
702 | sprintf(buf, "i%d", |
703 | lastsect->file->last_fragment_number++); |
704 | hr->fragment = dupstr(buf); |
3e82de8f |
705 | hr->fragment = |
706 | html_sanitise_fragment(&files, hr->section->file, |
707 | hr->fragment); |
78c73085 |
708 | } |
709 | w->private_data = hr; |
710 | |
711 | tag = index_findtag(idx, w->text); |
712 | if (!tag) |
713 | break; |
714 | |
715 | for (i = 0; i < tag->nrefs; i++) { |
716 | indexentry *entry = tag->refs[i]; |
717 | htmlindex *hi = (htmlindex *)entry->backend_data; |
718 | |
719 | if (hi->nrefs >= hi->refsize) { |
720 | hi->refsize += 32; |
f1530049 |
721 | hi->refs = sresize(hi->refs, hi->refsize, word *); |
78c73085 |
722 | } |
723 | |
724 | hi->refs[hi->nrefs++] = w; |
725 | } |
726 | } |
727 | } |
728 | } |
729 | |
730 | /* |
731 | * Now we're ready to write out the actual HTML files. |
732 | * |
733 | * For each file: |
734 | * |
735 | * - we open that file and write its header |
736 | * - we run down the list of sections |
737 | * - for each section directly contained within that file, we |
738 | * output the section text |
739 | * - for each section which is not in the file but which has a |
740 | * parent that is, we output a contents entry for the |
741 | * section if appropriate |
742 | * - finally, we output the file trailer and close the file. |
743 | */ |
744 | { |
745 | htmlfile *f, *prevf; |
746 | htmlsect *s; |
747 | paragraph *p; |
748 | |
749 | prevf = NULL; |
750 | |
751 | for (f = files.head; f; f = f->next) { |
752 | htmloutput ho; |
753 | int displaying; |
754 | enum LISTTYPE { NOLIST, UL, OL, DL }; |
755 | enum ITEMTYPE { NOITEM, LI, DT, DD }; |
756 | struct stackelement { |
757 | struct stackelement *next; |
758 | enum LISTTYPE listtype; |
759 | enum ITEMTYPE itemtype; |
760 | } *stackhead; |
761 | |
762 | #define listname(lt) ( (lt)==UL ? "ul" : (lt)==OL ? "ol" : "dl" ) |
763 | #define itemname(lt) ( (lt)==LI ? "li" : (lt)==DT ? "dt" : "dd" ) |
764 | |
765 | ho.fp = fopen(f->filename, "w"); |
766 | ho.charset = conf.output_charset; |
b7309494 |
767 | ho.restrict_charset = conf.restrict_charset; |
78c73085 |
768 | ho.cstate = charset_init_state; |
769 | ho.ver = conf.htmlver; |
770 | ho.state = HO_NEUTRAL; |
771 | ho.contents_level = 0; |
772 | |
773 | /* <!DOCTYPE>. */ |
774 | switch (conf.htmlver) { |
775 | case HTML_3_2: |
776 | fprintf(ho.fp, "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD " |
777 | "HTML 3.2 Final//EN\">\n"); |
778 | break; |
779 | case HTML_4: |
780 | fprintf(ho.fp, "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML" |
781 | " 4.01//EN\"\n\"http://www.w3.org/TR/html4/" |
782 | "strict.dtd\">\n"); |
783 | break; |
27bdc5ab |
784 | case ISO_HTML: |
785 | fprintf(ho.fp, "<!DOCTYPE HTML PUBLIC \"ISO/IEC " |
786 | "15445:2000//DTD HTML//EN\">\n"); |
787 | break; |
78c73085 |
788 | case XHTML_1_0_TRANSITIONAL: |
27bdc5ab |
789 | fprintf(ho.fp, "<?xml version=\"1.0\" encoding=\"%s\"?>\n", |
790 | charset_to_mimeenc(conf.output_charset)); |
78c73085 |
791 | fprintf(ho.fp, "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML" |
792 | " 1.0 Transitional//EN\"\n\"http://www.w3.org/TR/" |
793 | "xhtml1/DTD/xhtml1-transitional.dtd\">\n"); |
794 | break; |
795 | case XHTML_1_0_STRICT: |
27bdc5ab |
796 | fprintf(ho.fp, "<?xml version=\"1.0\" encoding=\"%s\"?>\n", |
797 | charset_to_mimeenc(conf.output_charset)); |
78c73085 |
798 | fprintf(ho.fp, "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML" |
799 | " 1.0 Strict//EN\"\n\"http://www.w3.org/TR/xhtml1/" |
800 | "DTD/xhtml1-strict.dtd\">\n"); |
801 | break; |
802 | } |
803 | |
804 | element_open(&ho, "html"); |
805 | if (is_xhtml(conf.htmlver)) { |
806 | element_attr(&ho, "xmlns", "http://www.w3.org/1999/xhtml"); |
807 | } |
808 | html_nl(&ho); |
809 | |
810 | element_open(&ho, "head"); |
811 | html_nl(&ho); |
812 | |
813 | element_empty(&ho, "meta"); |
814 | element_attr(&ho, "http-equiv", "content-type"); |
815 | { |
816 | char buf[200]; |
817 | sprintf(buf, "text/html; charset=%.150s", |
818 | charset_to_mimeenc(conf.output_charset)); |
819 | element_attr(&ho, "content", buf); |
820 | } |
821 | html_nl(&ho); |
822 | |
823 | if (conf.author) { |
824 | element_empty(&ho, "meta"); |
825 | element_attr(&ho, "name", "author"); |
826 | element_attr_w(&ho, "content", conf.author); |
827 | html_nl(&ho); |
828 | } |
829 | |
830 | if (conf.description) { |
831 | element_empty(&ho, "meta"); |
832 | element_attr(&ho, "name", "description"); |
833 | element_attr_w(&ho, "content", conf.description); |
834 | html_nl(&ho); |
835 | } |
836 | |
837 | element_open(&ho, "title"); |
838 | if (f->first && f->first->title) { |
839 | html_words(&ho, f->first->title->words, NOTHING, |
840 | f, keywords, &conf); |
841 | |
842 | assert(f->last); |
843 | if (f->last != f->first && f->last->title) { |
56a99eb6 |
844 | html_text(&ho, conf.title_separator); |
78c73085 |
845 | html_words(&ho, f->last->title->words, NOTHING, |
846 | f, keywords, &conf); |
847 | } |
848 | } |
849 | element_close(&ho, "title"); |
850 | html_nl(&ho); |
851 | |
852 | if (conf.head_end) |
853 | html_raw(&ho, conf.head_end); |
854 | |
9acfce4f |
855 | /* |
856 | * Add any <head> data defined in specific sections |
857 | * that go in this file. (This is mostly to allow <meta |
858 | * name="AppleTitle"> tags for Mac online help.) |
859 | */ |
860 | for (s = sects.head; s; s = s->next) { |
861 | if (s->file == f && s->text) { |
862 | for (p = s->text; |
6f0bdcde |
863 | p && (p == s->text || p->type == para_Title || |
864 | !is_heading_type(p->type)); |
9acfce4f |
865 | p = p->next) { |
866 | if (p->type == para_Config) { |
867 | if (!ustricmp(p->keyword, L"html-local-head")) { |
868 | html_raw(&ho, adv(p->origkeyword)); |
869 | } |
870 | } |
871 | } |
872 | } |
873 | } |
874 | |
78c73085 |
875 | element_close(&ho, "head"); |
876 | html_nl(&ho); |
877 | |
78c73085 |
878 | if (conf.body_tag) |
879 | html_raw(&ho, conf.body_tag); |
880 | else |
881 | element_open(&ho, "body"); |
882 | html_nl(&ho); |
883 | |
884 | if (conf.body_start) |
885 | html_raw(&ho, conf.body_start); |
886 | |
887 | /* |
888 | * Write out a nav bar. Special case: we don't do this |
889 | * if there is only one file. |
890 | */ |
891 | if (files.head != files.tail) { |
892 | element_open(&ho, "p"); |
893 | if (conf.nav_attr) |
894 | html_raw_as_attr(&ho, conf.nav_attr); |
895 | |
896 | if (prevf) { |
897 | element_open(&ho, "a"); |
898 | element_attr(&ho, "href", prevf->filename); |
899 | } |
56a99eb6 |
900 | html_text(&ho, conf.nav_prev_text); |
78c73085 |
901 | if (prevf) |
902 | element_close(&ho, "a"); |
903 | |
56a99eb6 |
904 | html_text(&ho, conf.nav_separator); |
78c73085 |
905 | |
906 | if (f != files.head) { |
907 | element_open(&ho, "a"); |
908 | element_attr(&ho, "href", files.head->filename); |
909 | } |
56a99eb6 |
910 | html_text(&ho, conf.contents_text); |
78c73085 |
911 | if (f != files.head) |
912 | element_close(&ho, "a"); |
913 | |
03fcb340 |
914 | if (has_index) { |
771d3da3 |
915 | html_text(&ho, conf.nav_separator); |
03fcb340 |
916 | if (f != files.index) { |
917 | element_open(&ho, "a"); |
918 | element_attr(&ho, "href", files.index->filename); |
919 | } |
920 | html_text(&ho, conf.index_text); |
921 | if (f != files.index) |
922 | element_close(&ho, "a"); |
78c73085 |
923 | } |
78c73085 |
924 | |
56a99eb6 |
925 | html_text(&ho, conf.nav_separator); |
78c73085 |
926 | |
927 | if (f->next) { |
928 | element_open(&ho, "a"); |
929 | element_attr(&ho, "href", f->next->filename); |
930 | } |
56a99eb6 |
931 | html_text(&ho, conf.nav_next_text); |
78c73085 |
932 | if (f->next) |
933 | element_close(&ho, "a"); |
934 | |
935 | element_close(&ho, "p"); |
936 | html_nl(&ho); |
937 | } |
938 | prevf = f; |
939 | |
940 | /* |
f5dee642 |
941 | * Write out a prefix TOC for the file (if a leaf file). |
78c73085 |
942 | * |
943 | * We start by going through the section list and |
944 | * collecting the sections which need to be added to |
945 | * the contents. On the way, we also test to see if |
946 | * this file is a leaf file (defined as one which |
947 | * contains all descendants of any section it |
948 | * contains), because this will play a part in our |
949 | * decision on whether or not to _output_ the TOC. |
950 | * |
951 | * Special case: we absolutely do not do this if we're |
952 | * in single-file mode. |
953 | */ |
954 | if (files.head != files.tail) { |
955 | int ntoc = 0, tocsize = 0; |
956 | htmlsect **toc = NULL; |
957 | int leaf = TRUE; |
958 | |
959 | for (s = sects.head; s; s = s->next) { |
960 | htmlsect *a, *ac; |
961 | int depth, adepth; |
962 | |
963 | /* |
964 | * Search up from this section until we find |
965 | * the highest-level one which belongs in this |
966 | * file. |
967 | */ |
968 | depth = adepth = 0; |
969 | a = NULL; |
970 | for (ac = s; ac; ac = ac->parent) { |
971 | if (ac->file == f) { |
972 | a = ac; |
973 | adepth = depth; |
974 | } |
975 | depth++; |
976 | } |
977 | |
978 | if (s->file != f && a != NULL) |
979 | leaf = FALSE; |
980 | |
981 | if (a) { |
982 | if (adepth <= a->contents_depth) { |
983 | if (ntoc >= tocsize) { |
984 | tocsize += 64; |
f1530049 |
985 | toc = sresize(toc, tocsize, htmlsect *); |
78c73085 |
986 | } |
987 | toc[ntoc++] = s; |
988 | } |
989 | } |
990 | } |
991 | |
992 | if (leaf && conf.leaf_contains_contents && |
993 | ntoc >= conf.leaf_smallest_contents) { |
994 | int i; |
995 | |
996 | for (i = 0; i < ntoc; i++) { |
997 | htmlsect *s = toc[i]; |
998 | int hlevel = (s->type == TOP ? -1 : |
999 | s->type == INDEX ? 0 : |
1000 | heading_depth(s->title)) |
1001 | - f->min_heading_depth + 1; |
1002 | |
1003 | assert(hlevel >= 1); |
1004 | html_contents_entry(&ho, hlevel, s, |
1005 | f, keywords, &conf); |
1006 | } |
1007 | html_contents_entry(&ho, 0, NULL, f, keywords, &conf); |
1008 | } |
1009 | } |
1010 | |
1011 | /* |
1012 | * Now go through the document and output some real |
1013 | * text. |
1014 | */ |
1015 | displaying = FALSE; |
1016 | for (s = sects.head; s; s = s->next) { |
1017 | if (s->file == f) { |
1018 | /* |
1019 | * This section belongs in this file. |
1020 | * Display it. |
1021 | */ |
1022 | displaying = TRUE; |
1023 | } else { |
f5dee642 |
1024 | /* |
1025 | * Doesn't belong in this file, but it may be |
1026 | * a descendant of a section which does, in |
1027 | * which case we should consider it for the |
1028 | * main TOC of this file (for non-leaf files). |
1029 | */ |
78c73085 |
1030 | htmlsect *a, *ac; |
1031 | int depth, adepth; |
1032 | |
1033 | displaying = FALSE; |
1034 | |
1035 | /* |
1036 | * Search up from this section until we find |
1037 | * the highest-level one which belongs in this |
1038 | * file. |
1039 | */ |
1040 | depth = adepth = 0; |
1041 | a = NULL; |
1042 | for (ac = s; ac; ac = ac->parent) { |
1043 | if (ac->file == f) { |
1044 | a = ac; |
1045 | adepth = depth; |
1046 | } |
1047 | depth++; |
1048 | } |
1049 | |
1050 | if (a != NULL) { |
1051 | /* |
1052 | * This section does not belong in this |
1053 | * file, but an ancestor of it does. Write |
1054 | * out a contents table entry, if the depth |
1055 | * doesn't exceed the maximum contents |
1056 | * depth for the ancestor section. |
1057 | */ |
1058 | if (adepth <= a->contents_depth) { |
1059 | html_contents_entry(&ho, adepth, s, |
1060 | f, keywords, &conf); |
1061 | } |
1062 | } |
1063 | } |
1064 | |
1065 | if (displaying) { |
1066 | int hlevel; |
1067 | char htag[3]; |
1068 | |
1069 | html_contents_entry(&ho, 0, NULL, f, keywords, &conf); |
1070 | |
1071 | /* |
1072 | * Display the section heading. |
1073 | */ |
1074 | |
1075 | hlevel = (s->type == TOP ? -1 : |
1076 | s->type == INDEX ? 0 : |
1077 | heading_depth(s->title)) |
1078 | - f->min_heading_depth + 1; |
1079 | assert(hlevel >= 1); |
1080 | /* HTML headings only go up to <h6> */ |
1081 | if (hlevel > 6) |
1082 | hlevel = 6; |
1083 | htag[0] = 'h'; |
1084 | htag[1] = '0' + hlevel; |
1085 | htag[2] = '\0'; |
1086 | element_open(&ho, htag); |
1087 | |
1088 | /* |
12f0ee84 |
1089 | * Provide anchor(s) for cross-links to target. |
78c73085 |
1090 | * |
78c73085 |
1091 | * (Also we'll have to do this separately in |
1092 | * other paragraph types - NumberedList and |
1093 | * BiblioCited.) |
1094 | */ |
12f0ee84 |
1095 | { |
1096 | int i; |
1097 | for (i=0; i < conf.ntfragments; i++) |
1098 | if (s->fragments[i]) |
1099 | html_fragment(&ho, s->fragments[i]); |
1100 | } |
78c73085 |
1101 | |
23c9bbc2 |
1102 | html_section_title(&ho, s, f, keywords, &conf, TRUE); |
78c73085 |
1103 | |
1104 | element_close(&ho, htag); |
1105 | |
1106 | /* |
1107 | * Now display the section text. |
1108 | */ |
1109 | if (s->text) { |
f1530049 |
1110 | stackhead = snew(struct stackelement); |
78c73085 |
1111 | stackhead->next = NULL; |
1112 | stackhead->listtype = NOLIST; |
1113 | stackhead->itemtype = NOITEM; |
1114 | |
1115 | for (p = s->text;; p = p->next) { |
1116 | enum LISTTYPE listtype; |
1117 | struct stackelement *se; |
1118 | |
1119 | /* |
1120 | * Preliminary switch to figure out what |
1121 | * sort of list we expect to be inside at |
1122 | * this stage. |
1123 | * |
1124 | * Since p may still be NULL at this point, |
1125 | * I invent a harmless paragraph type for |
1126 | * it if it is. |
1127 | */ |
1128 | switch (p ? p->type : para_Normal) { |
1129 | case para_Rule: |
1130 | case para_Normal: |
1131 | case para_Copyright: |
1132 | case para_BiblioCited: |
1133 | case para_Code: |
1134 | case para_QuotePush: |
1135 | case para_QuotePop: |
1136 | case para_Chapter: |
1137 | case para_Appendix: |
1138 | case para_UnnumberedChapter: |
1139 | case para_Heading: |
1140 | case para_Subsect: |
1141 | case para_LcontPop: |
1142 | listtype = NOLIST; |
1143 | break; |
1144 | |
1145 | case para_Bullet: |
1146 | listtype = UL; |
1147 | break; |
1148 | |
1149 | case para_NumberedList: |
1150 | listtype = OL; |
1151 | break; |
1152 | |
1153 | case para_DescribedThing: |
1154 | case para_Description: |
1155 | listtype = DL; |
1156 | break; |
1157 | |
1158 | case para_LcontPush: |
f1530049 |
1159 | se = snew(struct stackelement); |
78c73085 |
1160 | se->next = stackhead; |
1161 | se->listtype = NOLIST; |
1162 | se->itemtype = NOITEM; |
1163 | stackhead = se; |
1164 | continue; |
1165 | |
1166 | default: /* some totally non-printing para */ |
1167 | continue; |
1168 | } |
1169 | |
1170 | html_nl(&ho); |
1171 | |
1172 | /* |
1173 | * Terminate the most recent list item, if |
1174 | * any. (We left this until after |
1175 | * processing LcontPush, since in that case |
1176 | * the list item won't want to be |
1177 | * terminated until after the corresponding |
1178 | * LcontPop.) |
1179 | */ |
1180 | if (stackhead->itemtype != NOITEM) { |
1181 | element_close(&ho, itemname(stackhead->itemtype)); |
1182 | html_nl(&ho); |
1183 | } |
1184 | stackhead->itemtype = NOITEM; |
1185 | |
1186 | /* |
1187 | * Terminate the current list, if it's not |
1188 | * the one we want to be in. |
1189 | */ |
1190 | if (listtype != stackhead->listtype && |
1191 | stackhead->listtype != NOLIST) { |
1192 | element_close(&ho, listname(stackhead->listtype)); |
1193 | html_nl(&ho); |
1194 | } |
1195 | |
1196 | /* |
1197 | * Leave the loop if our time has come. |
1198 | */ |
1199 | if (!p || (is_heading_type(p->type) && |
1200 | p->type != para_Title)) |
1201 | break; /* end of section text */ |
1202 | |
1203 | /* |
1204 | * Start a fresh list if necessary. |
1205 | */ |
1206 | if (listtype != stackhead->listtype && |
1207 | listtype != NOLIST) |
1208 | element_open(&ho, listname(listtype)); |
1209 | |
1210 | stackhead->listtype = listtype; |
1211 | |
1212 | switch (p->type) { |
1213 | case para_Rule: |
1214 | element_empty(&ho, "hr"); |
1215 | break; |
1216 | case para_Code: |
1217 | html_codepara(&ho, p->words); |
1218 | break; |
1219 | case para_Normal: |
1220 | case para_Copyright: |
1221 | element_open(&ho, "p"); |
1222 | html_nl(&ho); |
1223 | html_words(&ho, p->words, ALL, |
1224 | f, keywords, &conf); |
1225 | html_nl(&ho); |
1226 | element_close(&ho, "p"); |
1227 | break; |
1228 | case para_BiblioCited: |
1229 | element_open(&ho, "p"); |
1230 | if (p->private_data) { |
1231 | htmlsect *s = (htmlsect *)p->private_data; |
12f0ee84 |
1232 | int i; |
1233 | for (i=0; i < conf.ntfragments; i++) |
1234 | if (s->fragments[i]) |
1235 | html_fragment(&ho, s->fragments[i]); |
78c73085 |
1236 | } |
1237 | html_nl(&ho); |
1238 | html_words(&ho, p->kwtext, ALL, |
1239 | f, keywords, &conf); |
1240 | html_text(&ho, L" "); |
1241 | html_words(&ho, p->words, ALL, |
1242 | f, keywords, &conf); |
1243 | html_nl(&ho); |
1244 | element_close(&ho, "p"); |
1245 | break; |
1246 | case para_Bullet: |
1247 | case para_NumberedList: |
1248 | element_open(&ho, "li"); |
1249 | if (p->private_data) { |
1250 | htmlsect *s = (htmlsect *)p->private_data; |
12f0ee84 |
1251 | int i; |
1252 | for (i=0; i < conf.ntfragments; i++) |
1253 | if (s->fragments[i]) |
1254 | html_fragment(&ho, s->fragments[i]); |
78c73085 |
1255 | } |
1256 | html_nl(&ho); |
1257 | stackhead->itemtype = LI; |
1258 | html_words(&ho, p->words, ALL, |
1259 | f, keywords, &conf); |
1260 | break; |
1261 | case para_DescribedThing: |
1262 | element_open(&ho, "dt"); |
1263 | html_nl(&ho); |
1264 | stackhead->itemtype = DT; |
1265 | html_words(&ho, p->words, ALL, |
1266 | f, keywords, &conf); |
1267 | break; |
1268 | case para_Description: |
1269 | element_open(&ho, "dd"); |
1270 | html_nl(&ho); |
1271 | stackhead->itemtype = DD; |
1272 | html_words(&ho, p->words, ALL, |
1273 | f, keywords, &conf); |
1274 | break; |
1275 | |
1276 | case para_QuotePush: |
1277 | element_open(&ho, "blockquote"); |
1278 | break; |
1279 | case para_QuotePop: |
1280 | element_close(&ho, "blockquote"); |
1281 | break; |
1282 | |
1283 | case para_LcontPop: |
1284 | se = stackhead; |
1285 | stackhead = stackhead->next; |
1286 | assert(stackhead); |
1287 | sfree(se); |
1288 | break; |
1289 | } |
1290 | } |
1291 | |
1292 | assert(stackhead && !stackhead->next); |
1293 | sfree(stackhead); |
1294 | } |
1295 | |
1296 | if (s->type == INDEX) { |
1297 | indexentry *entry; |
1298 | int i; |
1299 | |
1300 | /* |
1301 | * This section is the index. I'll just |
1302 | * render it as a single paragraph, with a |
1303 | * colon between the index term and the |
1304 | * references, and <br> in between each |
1305 | * entry. |
1306 | */ |
1307 | element_open(&ho, "p"); |
1308 | |
1309 | for (i = 0; (entry = |
1310 | index234(idx->entries, i)) != NULL; i++) { |
1311 | htmlindex *hi = (htmlindex *)entry->backend_data; |
1312 | int j; |
1313 | |
1314 | if (i > 0) |
1315 | element_empty(&ho, "br"); |
1316 | html_nl(&ho); |
1317 | |
1318 | html_words(&ho, entry->text, MARKUP|LINKS, |
1319 | f, keywords, &conf); |
1320 | |
56a99eb6 |
1321 | html_text(&ho, conf.index_main_sep); |
78c73085 |
1322 | |
1323 | for (j = 0; j < hi->nrefs; j++) { |
1324 | htmlindexref *hr = |
1325 | (htmlindexref *)hi->refs[j]->private_data; |
1326 | paragraph *p = hr->section->title; |
1327 | |
1328 | if (j > 0) |
56a99eb6 |
1329 | html_text(&ho, conf.index_multi_sep); |
78c73085 |
1330 | |
1331 | html_href(&ho, f, hr->section->file, |
1332 | hr->fragment); |
1b7bf715 |
1333 | hr->referenced = TRUE; |
78c73085 |
1334 | if (p && p->kwtext) |
1335 | html_words(&ho, p->kwtext, MARKUP|LINKS, |
1336 | f, keywords, &conf); |
1337 | else if (p && p->words) |
1338 | html_words(&ho, p->words, MARKUP|LINKS, |
1339 | f, keywords, &conf); |
56a99eb6 |
1340 | else { |
1341 | /* |
1342 | * If there is no title at all, |
1343 | * this must be because our |
1344 | * target section is the |
1345 | * preamble section and there |
1346 | * is no title. So we use the |
1347 | * preamble_text. |
1348 | */ |
1349 | html_text(&ho, conf.preamble_text); |
1350 | } |
78c73085 |
1351 | element_close(&ho, "a"); |
1352 | } |
1353 | } |
1354 | element_close(&ho, "p"); |
1355 | } |
1356 | } |
1357 | } |
1358 | |
1359 | html_contents_entry(&ho, 0, NULL, f, keywords, &conf); |
1360 | html_nl(&ho); |
1361 | |
1362 | { |
1363 | /* |
1364 | * Footer. |
1365 | */ |
1366 | int done_version_ids = FALSE; |
1367 | |
1368 | element_empty(&ho, "hr"); |
1369 | |
1370 | if (conf.body_end) |
1371 | html_raw(&ho, conf.body_end); |
1372 | |
1373 | if (conf.address_section) { |
27bdc5ab |
1374 | int started = FALSE; |
1375 | if (conf.htmlver == ISO_HTML) { |
1376 | /* |
1377 | * The ISO-HTML validator complains if |
1378 | * there isn't a <div> tag surrounding the |
1379 | * <address> tag. I'm uncertain of why this |
1380 | * should be - there appears to be no |
1381 | * mention of this in the ISO-HTML spec, |
1382 | * suggesting that it doesn't represent a |
1383 | * change from HTML 4, but nonetheless the |
1384 | * HTML 4 validator doesn't seem to mind. |
1385 | */ |
1386 | element_open(&ho, "div"); |
1387 | } |
78c73085 |
1388 | element_open(&ho, "address"); |
1389 | if (conf.addr_start) { |
1390 | html_raw(&ho, conf.addr_start); |
1391 | html_nl(&ho); |
27bdc5ab |
1392 | started = TRUE; |
78c73085 |
1393 | } |
1394 | if (conf.visible_version_id) { |
78c73085 |
1395 | for (p = sourceform; p; p = p->next) |
1396 | if (p->type == para_VersionID) { |
27bdc5ab |
1397 | if (started) |
78c73085 |
1398 | element_empty(&ho, "br"); |
1399 | html_nl(&ho); |
56a99eb6 |
1400 | html_text(&ho, conf.pre_versionid); |
78c73085 |
1401 | html_words(&ho, p->words, NOTHING, |
1402 | f, keywords, &conf); |
56a99eb6 |
1403 | html_text(&ho, conf.post_versionid); |
78c73085 |
1404 | started = TRUE; |
1405 | } |
78c73085 |
1406 | done_version_ids = TRUE; |
1407 | } |
27bdc5ab |
1408 | if (conf.addr_end) { |
1409 | if (started) |
1410 | element_empty(&ho, "br"); |
78c73085 |
1411 | html_raw(&ho, conf.addr_end); |
27bdc5ab |
1412 | } |
78c73085 |
1413 | element_close(&ho, "address"); |
27bdc5ab |
1414 | if (conf.htmlver == ISO_HTML) |
1415 | element_close(&ho, "div"); |
78c73085 |
1416 | } |
1417 | |
1418 | if (!done_version_ids) { |
1419 | /* |
1420 | * If the user didn't want the version IDs |
1421 | * visible, I think we still have a duty to put |
1422 | * them in an HTML comment. |
1423 | */ |
1424 | int started = FALSE; |
1425 | for (p = sourceform; p; p = p->next) |
1426 | if (p->type == para_VersionID) { |
1427 | if (!started) { |
1428 | html_raw(&ho, "<!-- version IDs:\n"); |
1429 | started = TRUE; |
1430 | } |
1431 | html_words(&ho, p->words, NOTHING, |
1432 | f, keywords, &conf); |
1433 | html_nl(&ho); |
1434 | } |
1435 | if (started) |
1436 | html_raw(&ho, "-->\n"); |
1437 | } |
1438 | } |
1439 | |
1440 | element_close(&ho, "body"); |
1441 | html_nl(&ho); |
1442 | element_close(&ho, "html"); |
1443 | html_nl(&ho); |
1444 | cleanup(&ho); |
1445 | } |
1446 | } |
1447 | |
1448 | /* |
1b7bf715 |
1449 | * Go through and check that no index fragments were referenced |
1450 | * without being generated, or indeed vice versa. |
1451 | * |
1452 | * (When I actually get round to freeing everything, this can |
1453 | * probably be the freeing loop as well.) |
1454 | */ |
1455 | for (p = sourceform; p; p = p->next) { |
1456 | word *w; |
1457 | for (w = p->words; w; w = w->next) |
1458 | if (w->type == word_IndexRef) { |
1459 | htmlindexref *hr = (htmlindexref *)w->private_data; |
1460 | |
1461 | assert(!hr->referenced == !hr->generated); |
1462 | } |
1463 | } |
1464 | |
1465 | /* |
529a6c83 |
1466 | * Free all the working data. |
78c73085 |
1467 | */ |
529a6c83 |
1468 | { |
1469 | htmlfragment *frag; |
1470 | while ( (frag = (htmlfragment *)delpos234(files.frags, 0)) != NULL ) { |
1471 | /* |
1472 | * frag->fragment is dynamically allocated, but will be |
1473 | * freed when we process the htmlsect structure which |
1474 | * it is attached to. |
1475 | */ |
1476 | sfree(frag); |
1477 | } |
1478 | freetree234(files.frags); |
1479 | } |
1480 | { |
1481 | htmlsect *sect, *tmp; |
1482 | sect = sects.head; |
1483 | while (sect) { |
12f0ee84 |
1484 | int i; |
529a6c83 |
1485 | tmp = sect->next; |
12f0ee84 |
1486 | for (i=0; i < conf.ntfragments; i++) |
1487 | sfree(sect->fragments[i]); |
1488 | sfree(sect->fragments); |
529a6c83 |
1489 | sfree(sect); |
1490 | sect = tmp; |
1491 | } |
1492 | sect = nonsects.head; |
1493 | while (sect) { |
12f0ee84 |
1494 | int i; |
529a6c83 |
1495 | tmp = sect->next; |
12f0ee84 |
1496 | for (i=0; i < conf.ntfragments; i++) |
1497 | sfree(sect->fragments[i]); |
1498 | sfree(sect->fragments); |
529a6c83 |
1499 | sfree(sect); |
1500 | sect = tmp; |
1501 | } |
1502 | } |
1503 | { |
1504 | htmlfile *file, *tmp; |
1505 | file = files.head; |
1506 | while (file) { |
1507 | tmp = file->next; |
1508 | sfree(file->filename); |
1509 | sfree(file); |
1510 | file = tmp; |
1511 | } |
1512 | } |
1513 | { |
1514 | int i; |
1515 | indexentry *entry; |
1516 | for (i = 0; (entry = index234(idx->entries, i)) != NULL; i++) { |
1517 | htmlindex *hi = (htmlindex *)entry->backend_data; |
1518 | sfree(hi); |
1519 | } |
1520 | } |
1521 | { |
1522 | paragraph *p; |
1523 | word *w; |
1524 | for (p = sourceform; p; p = p->next) |
1525 | for (w = p->words; w; w = w->next) |
1526 | if (w->type == word_IndexRef) { |
1527 | htmlindexref *hr = (htmlindexref *)w->private_data; |
1528 | assert(hr != NULL); |
1529 | sfree(hr->fragment); |
1530 | sfree(hr); |
1531 | } |
1532 | } |
12f0ee84 |
1533 | sfree(conf.asect); |
1534 | sfree(conf.single_filename); |
1535 | sfree(conf.contents_filename); |
1536 | sfree(conf.index_filename); |
1537 | sfree(conf.template_filename); |
1538 | while (conf.ntfragments--) |
1539 | sfree(conf.template_fragments[conf.ntfragments]); |
1540 | sfree(conf.template_fragments); |
78c73085 |
1541 | } |
1542 | |
1543 | static void html_file_section(htmlconfig *cfg, htmlfilelist *files, |
1544 | htmlsect *sect, int depth) |
1545 | { |
1546 | htmlfile *file; |
1547 | int ldepth; |
1548 | |
1549 | /* |
1550 | * `depth' is derived from the heading_depth() macro at the top |
1551 | * of this file, which counts title as -1, chapter as 0, |
1552 | * heading as 1 and subsection as 2. However, the semantics of |
1553 | * cfg->leaf_level are defined to count chapter as 1, heading |
1554 | * as 2 etc. So first I increment depth :-( |
1555 | */ |
1556 | ldepth = depth + 1; |
1557 | |
1558 | if (cfg->leaf_level == 0) { |
1559 | /* |
1560 | * leaf_level==0 is a special case, in which everything is |
1561 | * put into a single file. |
1562 | */ |
1563 | if (!files->single) |
1564 | files->single = html_new_file(files, cfg->single_filename); |
1565 | |
1566 | file = files->single; |
1567 | } else { |
1568 | /* |
1569 | * If the depth of this section is at or above leaf_level, |
1570 | * we invent a fresh file and put this section at its head. |
1571 | * Otherwise, we put it in the same file as its parent |
1572 | * section. |
1573 | */ |
1574 | if (ldepth > cfg->leaf_level) { |
1575 | /* |
1576 | * We know that sect->parent cannot be NULL. The only |
1577 | * circumstance in which it can be is if sect is at |
1578 | * chapter or appendix level, i.e. ldepth==1; and if |
1579 | * that's the case, then we cannot have entered this |
1580 | * branch unless cfg->leaf_level==0, in which case we |
1581 | * would be in the single-file case above and not here |
1582 | * at all. |
1583 | */ |
1584 | assert(sect->parent); |
1585 | |
1586 | file = sect->parent->file; |
1587 | } else { |
1588 | if (sect->type == TOP) { |
1589 | file = html_new_file(files, cfg->contents_filename); |
1590 | } else if (sect->type == INDEX) { |
1591 | file = html_new_file(files, cfg->index_filename); |
1592 | } else { |
1593 | char *title; |
1594 | |
1595 | assert(ldepth > 0 && sect->title); |
1596 | title = html_format(sect->title, cfg->template_filename); |
1597 | file = html_new_file(files, title); |
1598 | sfree(title); |
1599 | } |
1600 | } |
1601 | } |
1602 | |
1603 | sect->file = file; |
1604 | |
1605 | if (file->min_heading_depth > depth) { |
1606 | /* |
1607 | * This heading is at a higher level than any heading we |
1608 | * have so far placed in this file; so we set the `first' |
1609 | * pointer. |
1610 | */ |
1611 | file->min_heading_depth = depth; |
1612 | file->first = sect; |
1613 | } |
1614 | |
1615 | if (file->min_heading_depth == depth) |
1616 | file->last = sect; |
1617 | } |
1618 | |
1619 | static htmlfile *html_new_file(htmlfilelist *list, char *filename) |
1620 | { |
f1530049 |
1621 | htmlfile *ret = snew(htmlfile); |
78c73085 |
1622 | |
1623 | ret->next = NULL; |
1624 | if (list->tail) |
1625 | list->tail->next = ret; |
1626 | else |
1627 | list->head = ret; |
1628 | list->tail = ret; |
1629 | |
1630 | ret->filename = dupstr(filename); |
1631 | ret->last_fragment_number = 0; |
1632 | ret->min_heading_depth = INT_MAX; |
1633 | ret->first = ret->last = NULL; |
1634 | |
1635 | return ret; |
1636 | } |
1637 | |
12f0ee84 |
1638 | static htmlsect *html_new_sect(htmlsectlist *list, paragraph *title, |
1639 | htmlconfig *cfg) |
78c73085 |
1640 | { |
f1530049 |
1641 | htmlsect *ret = snew(htmlsect); |
78c73085 |
1642 | |
1643 | ret->next = NULL; |
1644 | if (list->tail) |
1645 | list->tail->next = ret; |
1646 | else |
1647 | list->head = ret; |
1648 | list->tail = ret; |
1649 | |
1650 | ret->title = title; |
1651 | ret->file = NULL; |
1652 | ret->parent = NULL; |
1653 | ret->type = NORMAL; |
1654 | |
12f0ee84 |
1655 | ret->fragments = snewn(cfg->ntfragments, char *); |
1656 | { |
1657 | int i; |
1658 | for (i=0; i < cfg->ntfragments; i++) |
1659 | ret->fragments[i] = NULL; |
1660 | } |
1661 | |
78c73085 |
1662 | return ret; |
1663 | } |
1664 | |
1665 | static void html_words(htmloutput *ho, word *words, int flags, |
1666 | htmlfile *file, keywordlist *keywords, htmlconfig *cfg) |
1667 | { |
1668 | word *w; |
1669 | char *c; |
1670 | int style, type; |
1671 | |
1672 | for (w = words; w; w = w->next) switch (w->type) { |
1673 | case word_HyperLink: |
1674 | if (flags & LINKS) { |
1675 | element_open(ho, "a"); |
1676 | c = utoa_dup(w->text, CS_ASCII); |
1677 | element_attr(ho, "href", c); |
1678 | sfree(c); |
1679 | } |
1680 | break; |
1681 | case word_UpperXref: |
1682 | case word_LowerXref: |
1683 | if (flags & LINKS) { |
1684 | keyword *kwl = kw_lookup(keywords, w->text); |
2b0986a3 |
1685 | paragraph *p; |
1686 | htmlsect *s; |
1687 | |
1688 | assert(kwl); |
1689 | p = kwl->para; |
1690 | s = (htmlsect *)p->private_data; |
78c73085 |
1691 | |
1692 | assert(s); |
1693 | |
12f0ee84 |
1694 | html_href(ho, file, s->file, s->fragments[0]); |
78c73085 |
1695 | } |
1696 | break; |
1697 | case word_HyperEnd: |
1698 | case word_XrefEnd: |
1699 | if (flags & LINKS) |
1700 | element_close(ho, "a"); |
1701 | break; |
1702 | case word_IndexRef: |
1703 | if (flags & INDEXENTS) { |
1704 | htmlindexref *hr = (htmlindexref *)w->private_data; |
27bdc5ab |
1705 | html_fragment(ho, hr->fragment); |
1b7bf715 |
1706 | hr->generated = TRUE; |
78c73085 |
1707 | } |
1708 | break; |
1709 | case word_Normal: |
1710 | case word_Emph: |
1711 | case word_Code: |
1712 | case word_WeakCode: |
1713 | case word_WhiteSpace: |
1714 | case word_EmphSpace: |
1715 | case word_CodeSpace: |
1716 | case word_WkCodeSpace: |
1717 | case word_Quote: |
1718 | case word_EmphQuote: |
1719 | case word_CodeQuote: |
1720 | case word_WkCodeQuote: |
1721 | style = towordstyle(w->type); |
1722 | type = removeattr(w->type); |
1723 | if (style == word_Emph && |
1724 | (attraux(w->aux) == attr_First || |
1725 | attraux(w->aux) == attr_Only) && |
1726 | (flags & MARKUP)) |
1727 | element_open(ho, "em"); |
1728 | else if ((style == word_Code || style == word_WeakCode) && |
1729 | (attraux(w->aux) == attr_First || |
1730 | attraux(w->aux) == attr_Only) && |
1731 | (flags & MARKUP)) |
1732 | element_open(ho, "code"); |
1733 | |
1734 | if (type == word_WhiteSpace) |
1735 | html_text(ho, L" "); |
1736 | else if (type == word_Quote) { |
1737 | if (quoteaux(w->aux) == quote_Open) |
1738 | html_text(ho, cfg->lquote); |
1739 | else |
1740 | html_text(ho, cfg->rquote); |
1741 | } else { |
35b123a0 |
1742 | if (!w->alt || cvt_ok(ho->restrict_charset, w->text)) |
1743 | html_text_nbsp(ho, w->text); |
78c73085 |
1744 | else |
1745 | html_words(ho, w->alt, flags, file, keywords, cfg); |
1746 | } |
1747 | |
1748 | if (style == word_Emph && |
1749 | (attraux(w->aux) == attr_Last || |
1750 | attraux(w->aux) == attr_Only) && |
1751 | (flags & MARKUP)) |
1752 | element_close(ho, "em"); |
1753 | else if ((style == word_Code || style == word_WeakCode) && |
1754 | (attraux(w->aux) == attr_Last || |
1755 | attraux(w->aux) == attr_Only) && |
1756 | (flags & MARKUP)) |
1757 | element_close(ho, "code"); |
1758 | |
1759 | break; |
1760 | } |
1761 | } |
1762 | |
1763 | static void html_codepara(htmloutput *ho, word *words) |
1764 | { |
1765 | element_open(ho, "pre"); |
1766 | element_open(ho, "code"); |
1767 | for (; words; words = words->next) if (words->type == word_WeakCode) { |
1768 | char *open_tag; |
1769 | wchar_t *t, *e; |
1770 | |
1771 | t = words->text; |
1772 | if (words->next && words->next->type == word_Emph) { |
1773 | e = words->next->text; |
1774 | words = words->next; |
1775 | } else |
1776 | e = NULL; |
1777 | |
1778 | while (e && *e && *t) { |
1779 | int n; |
1780 | int ec = *e; |
1781 | |
1782 | for (n = 0; t[n] && e[n] && e[n] == ec; n++); |
1783 | |
1784 | open_tag = NULL; |
1785 | if (ec == 'i') |
1786 | open_tag = "em"; |
1787 | else if (ec == 'b') |
1788 | open_tag = "b"; |
1789 | if (open_tag) |
1790 | element_open(ho, open_tag); |
1791 | |
1792 | html_text_limit(ho, t, n); |
1793 | |
1794 | if (open_tag) |
1795 | element_close(ho, open_tag); |
1796 | |
1797 | t += n; |
1798 | e += n; |
1799 | } |
1800 | html_text(ho, t); |
1801 | html_nl(ho); |
1802 | } |
1803 | element_close(ho, "code"); |
1804 | element_close(ho, "pre"); |
1805 | } |
1806 | |
1807 | static void html_charset_cleanup(htmloutput *ho) |
1808 | { |
1809 | char outbuf[256]; |
1810 | int bytes; |
1811 | |
1812 | bytes = charset_from_unicode(NULL, NULL, outbuf, lenof(outbuf), |
1813 | ho->charset, &ho->cstate, NULL); |
1814 | if (bytes > 0) |
1815 | fwrite(outbuf, 1, bytes, ho->fp); |
1816 | } |
1817 | |
35b123a0 |
1818 | static void return_mostly_to_neutral(htmloutput *ho) |
78c73085 |
1819 | { |
35b123a0 |
1820 | if (ho->state == HO_IN_EMPTY_TAG && is_xhtml(ho->ver)) { |
78c73085 |
1821 | fprintf(ho->fp, " />"); |
1822 | } else if (ho->state == HO_IN_EMPTY_TAG || ho->state == HO_IN_TAG) { |
1823 | fprintf(ho->fp, ">"); |
1824 | } |
1825 | |
1826 | ho->state = HO_NEUTRAL; |
1827 | } |
1828 | |
35b123a0 |
1829 | static void return_to_neutral(htmloutput *ho) |
1830 | { |
1831 | if (ho->state == HO_IN_TEXT) { |
1832 | html_charset_cleanup(ho); |
1833 | } |
1834 | |
1835 | return_mostly_to_neutral(ho); |
1836 | } |
1837 | |
78c73085 |
1838 | static void element_open(htmloutput *ho, char const *name) |
1839 | { |
1840 | return_to_neutral(ho); |
1841 | fprintf(ho->fp, "<%s", name); |
1842 | ho->state = HO_IN_TAG; |
1843 | } |
1844 | |
1845 | static void element_close(htmloutput *ho, char const *name) |
1846 | { |
1847 | return_to_neutral(ho); |
1848 | fprintf(ho->fp, "</%s>", name); |
1849 | ho->state = HO_NEUTRAL; |
1850 | } |
1851 | |
1852 | static void element_empty(htmloutput *ho, char const *name) |
1853 | { |
1854 | return_to_neutral(ho); |
1855 | fprintf(ho->fp, "<%s", name); |
1856 | ho->state = HO_IN_EMPTY_TAG; |
1857 | } |
1858 | |
1859 | static void html_nl(htmloutput *ho) |
1860 | { |
1861 | return_to_neutral(ho); |
1862 | fputc('\n', ho->fp); |
1863 | } |
1864 | |
1865 | static void html_raw(htmloutput *ho, char *text) |
1866 | { |
1867 | return_to_neutral(ho); |
1868 | fputs(text, ho->fp); |
1869 | } |
1870 | |
1871 | static void html_raw_as_attr(htmloutput *ho, char *text) |
1872 | { |
1873 | assert(ho->state == HO_IN_TAG || ho->state == HO_IN_EMPTY_TAG); |
1874 | fputc(' ', ho->fp); |
1875 | fputs(text, ho->fp); |
1876 | } |
1877 | |
1878 | static void element_attr(htmloutput *ho, char const *name, char const *value) |
1879 | { |
1880 | html_charset_cleanup(ho); |
1881 | assert(ho->state == HO_IN_TAG || ho->state == HO_IN_EMPTY_TAG); |
1882 | fprintf(ho->fp, " %s=\"%s\"", name, value); |
1883 | } |
1884 | |
1885 | static void element_attr_w(htmloutput *ho, char const *name, |
1886 | wchar_t const *value) |
1887 | { |
1888 | html_charset_cleanup(ho); |
1889 | fprintf(ho->fp, " %s=\"", name); |
35b123a0 |
1890 | html_text_limit_internal(ho, value, 0, TRUE, FALSE); |
78c73085 |
1891 | html_charset_cleanup(ho); |
1892 | fputc('"', ho->fp); |
1893 | } |
1894 | |
1895 | static void html_text(htmloutput *ho, wchar_t const *text) |
1896 | { |
35b123a0 |
1897 | return_mostly_to_neutral(ho); |
1898 | html_text_limit_internal(ho, text, 0, FALSE, FALSE); |
1899 | } |
1900 | |
1901 | static void html_text_nbsp(htmloutput *ho, wchar_t const *text) |
1902 | { |
1903 | return_mostly_to_neutral(ho); |
1904 | html_text_limit_internal(ho, text, 0, FALSE, TRUE); |
78c73085 |
1905 | } |
1906 | |
1907 | static void html_text_limit(htmloutput *ho, wchar_t const *text, int maxlen) |
1908 | { |
35b123a0 |
1909 | return_mostly_to_neutral(ho); |
1910 | html_text_limit_internal(ho, text, maxlen, FALSE, FALSE); |
78c73085 |
1911 | } |
1912 | |
1913 | static void html_text_limit_internal(htmloutput *ho, wchar_t const *text, |
35b123a0 |
1914 | int maxlen, int quote_quotes, int nbsp) |
78c73085 |
1915 | { |
1916 | int textlen = ustrlen(text); |
1917 | char outbuf[256]; |
1918 | int bytes, err; |
1919 | |
1920 | if (maxlen > 0 && textlen > maxlen) |
1921 | textlen = maxlen; |
1922 | |
1923 | while (textlen > 0) { |
1924 | /* Scan ahead for characters we really can't display in HTML. */ |
1925 | int lenbefore, lenafter; |
1926 | for (lenbefore = 0; lenbefore < textlen; lenbefore++) |
1927 | if (text[lenbefore] == L'<' || |
1928 | text[lenbefore] == L'>' || |
1929 | text[lenbefore] == L'&' || |
35b123a0 |
1930 | (text[lenbefore] == L'"' && quote_quotes) || |
1931 | (text[lenbefore] == L' ' && nbsp)) |
78c73085 |
1932 | break; |
1933 | lenafter = lenbefore; |
1934 | bytes = charset_from_unicode(&text, &lenafter, outbuf, lenof(outbuf), |
1935 | ho->charset, &ho->cstate, &err); |
1936 | textlen -= (lenbefore - lenafter); |
1937 | if (bytes > 0) |
1938 | fwrite(outbuf, 1, bytes, ho->fp); |
1939 | if (err) { |
1940 | /* |
1941 | * We have encountered a character that cannot be |
1942 | * displayed in the selected output charset. Therefore, |
1943 | * we use an HTML numeric entity reference. |
1944 | */ |
1945 | assert(textlen > 0); |
1946 | fprintf(ho->fp, "&#%ld;", (long int)*text); |
1947 | text++, textlen--; |
1948 | } else if (lenafter == 0 && textlen > 0) { |
1949 | /* |
1950 | * We have encountered a character which is special to |
1951 | * HTML. |
1952 | */ |
1953 | if (*text == L'<') |
1954 | fprintf(ho->fp, "<"); |
1955 | else if (*text == L'>') |
1956 | fprintf(ho->fp, ">"); |
1957 | else if (*text == L'&') |
1958 | fprintf(ho->fp, "&"); |
1959 | else if (*text == L'"') |
1960 | fprintf(ho->fp, """); |
35b123a0 |
1961 | else if (*text == L' ') { |
1962 | assert(nbsp); |
1963 | fprintf(ho->fp, " "); |
1964 | } else |
78c73085 |
1965 | assert(!"Can't happen"); |
1966 | text++, textlen--; |
1967 | } |
1968 | } |
1969 | } |
1970 | |
1971 | static void cleanup(htmloutput *ho) |
1972 | { |
1973 | return_to_neutral(ho); |
1974 | fclose(ho->fp); |
1975 | } |
1976 | |
1977 | static void html_href(htmloutput *ho, htmlfile *thisfile, |
1978 | htmlfile *targetfile, char *targetfrag) |
1979 | { |
1980 | rdstringc rs = { 0, 0, NULL }; |
1981 | char *url; |
1982 | |
1983 | if (targetfile != thisfile) |
1984 | rdaddsc(&rs, targetfile->filename); |
1985 | if (targetfrag) { |
1986 | rdaddc(&rs, '#'); |
1987 | rdaddsc(&rs, targetfrag); |
1988 | } |
1989 | url = rs.text; |
1990 | |
1991 | element_open(ho, "a"); |
1992 | element_attr(ho, "href", url); |
1993 | sfree(url); |
1994 | } |
1995 | |
27bdc5ab |
1996 | static void html_fragment(htmloutput *ho, char const *fragment) |
1997 | { |
1998 | element_open(ho, "a"); |
1999 | element_attr(ho, "name", fragment); |
2000 | if (is_xhtml(ho->ver)) |
2001 | element_attr(ho, "id", fragment); |
2002 | element_close(ho, "a"); |
2003 | } |
2004 | |
78c73085 |
2005 | static char *html_format(paragraph *p, char *template_string) |
2006 | { |
2007 | char *c, *t; |
2008 | word *w; |
2009 | wchar_t *ws, wsbuf[2]; |
2010 | rdstringc rs = { 0, 0, NULL }; |
2011 | |
2012 | t = template_string; |
2013 | while (*t) { |
2014 | if (*t == '%' && t[1]) { |
2015 | int fmt; |
2016 | |
2017 | t++; |
2018 | fmt = *t++; |
2019 | |
2020 | if (fmt == '%') { |
2021 | rdaddc(&rs, fmt); |
2022 | continue; |
2023 | } |
2024 | |
2025 | w = NULL; |
2026 | ws = NULL; |
2027 | |
2028 | if (p->kwtext && fmt == 'n') |
2029 | w = p->kwtext; |
2030 | else if (p->kwtext2 && fmt == 'b') { |
2031 | /* |
2032 | * HTML fragment names must start with a letter, so |
2033 | * simply `1.2.3' is not adequate. In this case I'm |
2034 | * going to cheat slightly by prepending the first |
2035 | * character of the first word of kwtext, so that |
2036 | * we get `C1' for chapter 1, `S2.3' for section |
2037 | * 2.3 etc. |
2038 | */ |
2039 | if (p->kwtext && p->kwtext->text[0]) { |
2040 | ws = wsbuf; |
2041 | wsbuf[1] = '\0'; |
2042 | wsbuf[0] = p->kwtext->text[0]; |
2043 | } |
2044 | w = p->kwtext2; |
2045 | } else if (p->keyword && *p->keyword && fmt == 'k') |
2046 | ws = p->keyword; |
2047 | else |
46c7302f |
2048 | /* %N comes here; also failure cases of other fmts */ |
78c73085 |
2049 | w = p->words; |
2050 | |
2051 | if (ws) { |
2052 | c = utoa_dup(ws, CS_ASCII); |
2053 | rdaddsc(&rs,c); |
2054 | sfree(c); |
2055 | } |
2056 | |
2057 | while (w) { |
2058 | if (removeattr(w->type) == word_Normal) { |
2059 | c = utoa_dup(w->text, CS_ASCII); |
2060 | rdaddsc(&rs,c); |
2061 | sfree(c); |
2062 | } |
2063 | w = w->next; |
2064 | } |
2065 | } else { |
2066 | rdaddc(&rs, *t++); |
2067 | } |
2068 | } |
2069 | |
2070 | return rdtrimc(&rs); |
2071 | } |
2072 | |
3e82de8f |
2073 | static char *html_sanitise_fragment(htmlfilelist *files, htmlfile *file, |
2074 | char *text) |
78c73085 |
2075 | { |
2076 | /* |
2077 | * The HTML 4 spec's strictest definition of fragment names (<a |
2078 | * name> and "id" attributes) says that they `must begin with a |
2079 | * letter and may be followed by any number of letters, digits, |
2080 | * hyphens, underscores, colons, and periods'. |
2081 | * |
2082 | * So here we unceremoniously rip out any characters not |
2083 | * conforming to this limitation. |
2084 | */ |
2085 | char *p = text, *q = text; |
2086 | |
2087 | while (*p && !((*p>='A' && *p<='Z') || (*p>='a' && *p<='z'))) |
2088 | p++; |
3e82de8f |
2089 | if ((*q++ = *p++) != '\0') { |
2090 | while (*p) { |
2091 | if ((*p>='A' && *p<='Z') || |
2092 | (*p>='a' && *p<='z') || |
2093 | (*p>='0' && *p<='9') || |
2094 | *p=='-' || *p=='_' || *p==':' || *p=='.') |
2095 | *q++ = *p; |
2096 | p++; |
2097 | } |
2098 | |
2099 | *q = '\0'; |
2100 | } |
2101 | |
12f0ee84 |
2102 | /* If there's nothing left, make something valid up */ |
2103 | if (!*text) { |
0bac815b |
2104 | static const char anonfrag[] = "anon"; |
12f0ee84 |
2105 | text = sresize(text, lenof(anonfrag), char); |
2106 | strcpy(text, anonfrag); |
2107 | } |
2108 | |
3e82de8f |
2109 | /* |
2110 | * Now we check for clashes with other fragment names, and |
2111 | * adjust this one if necessary by appending a hyphen followed |
2112 | * by a number. |
2113 | */ |
2114 | { |
2115 | htmlfragment *frag = snew(htmlfragment); |
2116 | int len = 0; /* >0 indicates we have resized */ |
2117 | int suffix = 1; |
2118 | |
2119 | frag->file = file; |
2120 | frag->fragment = text; |
2121 | |
2122 | while (add234(files->frags, frag) != frag) { |
2123 | if (!len) { |
2124 | len = strlen(text); |
2125 | frag->fragment = text = sresize(text, len+20, char); |
2126 | } |
2127 | |
2128 | sprintf(text + len, "-%d", ++suffix); |
2129 | } |
78c73085 |
2130 | } |
2131 | |
3e82de8f |
2132 | return text; |
78c73085 |
2133 | } |
2134 | |
2135 | static void html_contents_entry(htmloutput *ho, int depth, htmlsect *s, |
2136 | htmlfile *thisfile, keywordlist *keywords, |
2137 | htmlconfig *cfg) |
2138 | { |
f5dee642 |
2139 | if (ho->contents_level >= depth && ho->contents_level > 0) { |
2140 | element_close(ho, "li"); |
2141 | html_nl(ho); |
2142 | } |
2143 | |
78c73085 |
2144 | while (ho->contents_level > depth) { |
2145 | element_close(ho, "ul"); |
2146 | ho->contents_level--; |
f5dee642 |
2147 | if (ho->contents_level > 0) { |
2148 | element_close(ho, "li"); |
2149 | } |
2150 | html_nl(ho); |
78c73085 |
2151 | } |
2152 | |
2153 | while (ho->contents_level < depth) { |
f5dee642 |
2154 | html_nl(ho); |
78c73085 |
2155 | element_open(ho, "ul"); |
f5dee642 |
2156 | html_nl(ho); |
78c73085 |
2157 | ho->contents_level++; |
2158 | } |
2159 | |
2160 | if (!s) |
2161 | return; |
2162 | |
2163 | element_open(ho, "li"); |
12f0ee84 |
2164 | html_href(ho, thisfile, s->file, s->fragments[0]); |
23c9bbc2 |
2165 | html_section_title(ho, s, thisfile, keywords, cfg, FALSE); |
78c73085 |
2166 | element_close(ho, "a"); |
f5dee642 |
2167 | /* <li> will be closed by a later invocation */ |
78c73085 |
2168 | } |
2169 | |
2170 | static void html_section_title(htmloutput *ho, htmlsect *s, htmlfile *thisfile, |
23c9bbc2 |
2171 | keywordlist *keywords, htmlconfig *cfg, |
2172 | int real) |
78c73085 |
2173 | { |
2174 | if (s->title) { |
2175 | sectlevel *sl; |
2176 | word *number; |
2177 | int depth = heading_depth(s->title); |
2178 | |
2179 | if (depth < 0) |
2180 | sl = NULL; |
2181 | else if (depth == 0) |
2182 | sl = &cfg->achapter; |
2183 | else if (depth <= cfg->nasect) |
2184 | sl = &cfg->asect[depth-1]; |
2185 | else |
2186 | sl = &cfg->asect[cfg->nasect-1]; |
2187 | |
2188 | if (!sl) |
2189 | number = NULL; |
2190 | else if (sl->just_numbers) |
2191 | number = s->title->kwtext2; |
2192 | else |
2193 | number = s->title->kwtext; |
2194 | |
2195 | if (number) { |
2196 | html_words(ho, number, MARKUP, |
2197 | thisfile, keywords, cfg); |
2198 | html_text(ho, sl->number_suffix); |
2199 | } |
2200 | |
23c9bbc2 |
2201 | html_words(ho, s->title->words, real ? ALL : MARKUP, |
78c73085 |
2202 | thisfile, keywords, cfg); |
2203 | } else { |
2204 | assert(s->type != NORMAL); |
56a99eb6 |
2205 | /* |
2206 | * If we're printing the full document title for _real_ and |
2207 | * there isn't one, we don't want to print `Preamble' at |
2208 | * the top of what ought to just be some text. If we need |
2209 | * it in any other context such as TOCs, we need to print |
2210 | * `Preamble'. |
2211 | */ |
2212 | if (s->type == TOP && !real) |
2213 | html_text(ho, cfg->preamble_text); |
78c73085 |
2214 | else if (s->type == INDEX) |
56a99eb6 |
2215 | html_text(ho, cfg->index_text); |
78c73085 |
2216 | } |
2217 | } |