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