html.c

   1 /*      $Id: html.c,v 1.100 2010/05/25 12:37:20 kristaps Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
   4  *
   5  * Permission to use, copy, modify, and distribute this software for any
   6  * purpose with or without fee is hereby granted, provided that the above
   7  * copyright notice and this permission notice appear in all copies.
   8  *
   9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  16  */
  17 #ifdef HAVE_CONFIG_H
  18 #include "config.h"
  19 #endif
  20
  21 #include <sys/types.h>
  22
  23 #include <assert.h>
  24 #include <ctype.h>
  25 #include <stdarg.h>
  26 #include <stdio.h>
  27 #include <stdint.h>
  28 #include <stdlib.h>
  29 #include <string.h>
  30 #include <unistd.h>
  31
  32 #include "mandoc.h"
  33 #include "out.h"
  34 #include "chars.h"
  35 #include "html.h"
  36 #include "main.h"
  37
  38 #define UNCONST(a)      ((void *)(uintptr_t)(const void *)(a))
  39
  40 struct  htmldata {
  41         const char       *name;
  42         int               flags;
  43 #define HTML_CLRLINE     (1 << 0)
  44 #define HTML_NOSTACK     (1 << 1)
  45 #define HTML_AUTOCLOSE   (1 << 2) /* Tag has auto-closure. */
  46 };
  47
  48 static  const struct htmldata htmltags[TAG_MAX] = {
  49         {"html",        HTML_CLRLINE}, /* TAG_HTML */
  50         {"head",        HTML_CLRLINE}, /* TAG_HEAD */
  51         {"body",        HTML_CLRLINE}, /* TAG_BODY */
  52         {"meta",        HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_META */
  53         {"title",       HTML_CLRLINE}, /* TAG_TITLE */
  54         {"div",         HTML_CLRLINE}, /* TAG_DIV */
  55         {"h1",          0}, /* TAG_H1 */
  56         {"h2",          0}, /* TAG_H2 */
  57         {"span",        0}, /* TAG_SPAN */
  58         {"link",        HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_LINK */
  59         {"br",          HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_BR */
  60         {"a",           0}, /* TAG_A */
  61         {"table",       HTML_CLRLINE}, /* TAG_TABLE */
  62         {"col",         HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_COL */
  63         {"tr",          HTML_CLRLINE}, /* TAG_TR */
  64         {"td",          HTML_CLRLINE}, /* TAG_TD */
  65         {"li",          HTML_CLRLINE}, /* TAG_LI */
  66         {"ul",          HTML_CLRLINE}, /* TAG_UL */
  67         {"ol",          HTML_CLRLINE}, /* TAG_OL */
  68 };
  69
  70 static  const char      *const htmlfonts[HTMLFONT_MAX] = {
  71         "roman",
  72         "bold",
  73         "italic"
  74 };
  75
  76 static  const char      *const htmlattrs[ATTR_MAX] = {
  77         "http-equiv",
  78         "content",
  79         "name",
  80         "rel",
  81         "href",
  82         "type",
  83         "media",
  84         "class",
  85         "style",
  86         "width",
  87         "valign",
  88         "target",
  89         "id",
  90         "summary",
  91 };
  92
  93 static  void              print_spec(struct html *, const char *, size_t);
  94 static  void              print_res(struct html *, const char *, size_t);
  95 static  void              print_ctag(struct html *, enum htmltag);
  96 static  void              print_doctype(struct html *);
  97 static  void              print_xmltype(struct html *);
  98 static  int               print_encode(struct html *, const char *, int);
  99 static  void              print_metaf(struct html *, enum roffdeco);
 100 static  void              print_attr(struct html *,
 101                                 const char *, const char *);
 102 static  void             *ml_alloc(char *, enum htmltype);
 103
 104
 105 static void *
 106 ml_alloc(char *outopts, enum htmltype type)
 107 {
 108         struct html     *h;
 109         const char      *toks[4];
 110         char            *v;
 111
 112         toks[0] = "style";
 113         toks[1] = "man";
 114         toks[2] = "includes";
 115         toks[3] = NULL;
 116
 117         h = calloc(1, sizeof(struct html));
 118         if (NULL == h) {
 119                 perror(NULL);
 120                 exit(EXIT_FAILURE);
 121         }
 122
 123         h->type = type;
 124         h->tags.head = NULL;
 125         h->ords.head = NULL;
 126         h->symtab = chars_init(CHARS_HTML);
 127
 128         while (outopts && *outopts)
 129                 switch (getsubopt(&outopts, UNCONST(toks), &v)) {
 130                 case (0):
 131                         h->style = v;
 132                         break;
 133                 case (1):
 134                         h->base_man = v;
 135                         break;
 136                 case (2):
 137                         h->base_includes = v;
 138                         break;
 139                 default:
 140                         break;
 141                 }
 142
 143         return(h);
 144 }
 145
 146 void *
 147 html_alloc(char *outopts)
 148 {
 149
 150         return(ml_alloc(outopts, HTML_HTML_4_01_STRICT));
 151 }
 152
 153
 154 void *
 155 xhtml_alloc(char *outopts)
 156 {
 157
 158         return(ml_alloc(outopts, HTML_XHTML_1_0_STRICT));
 159 }
 160
 161
 162 void
 163 html_free(void *p)
 164 {
 165         struct tag      *tag;
 166         struct ord      *ord;
 167         struct html     *h;
 168
 169         h = (struct html *)p;
 170
 171         while ((ord = h->ords.head) != NULL) {
 172                 h->ords.head = ord->next;
 173                 free(ord);
 174         }
 175
 176         while ((tag = h->tags.head) != NULL) {
 177                 h->tags.head = tag->next;
 178                 free(tag);
 179         }
 180
 181         if (h->symtab)
 182                 chars_free(h->symtab);
 183
 184         free(h);
 185 }
 186
 187
 188 void
 189 print_gen_head(struct html *h)
 190 {
 191         struct htmlpair  tag[4];
 192
 193         tag[0].key = ATTR_HTTPEQUIV;
 194         tag[0].val = "Content-Type";
 195         tag[1].key = ATTR_CONTENT;
 196         tag[1].val = "text/html; charset=utf-8";
 197         print_otag(h, TAG_META, 2, tag);
 198
 199         tag[0].key = ATTR_NAME;
 200         tag[0].val = "resource-type";
 201         tag[1].key = ATTR_CONTENT;
 202         tag[1].val = "document";
 203         print_otag(h, TAG_META, 2, tag);
 204
 205         if (h->style) {
 206                 tag[0].key = ATTR_REL;
 207                 tag[0].val = "stylesheet";
 208                 tag[1].key = ATTR_HREF;
 209                 tag[1].val = h->style;
 210                 tag[2].key = ATTR_TYPE;
 211                 tag[2].val = "text/css";
 212                 tag[3].key = ATTR_MEDIA;
 213                 tag[3].val = "all";
 214                 print_otag(h, TAG_LINK, 4, tag);
 215         }
 216 }
 217
 218
 219 static void
 220 print_spec(struct html *h, const char *p, size_t len)
 221 {
 222         const char      *rhs;
 223         size_t           sz;
 224
 225         rhs = chars_a2ascii(h->symtab, p, len, &sz);
 226
 227         if (NULL == rhs)
 228                 return;
 229         fwrite(rhs, 1, sz, stdout);
 230 }
 231
 232
 233 static void
 234 print_res(struct html *h, const char *p, size_t len)
 235 {
 236         const char      *rhs;
 237         size_t           sz;
 238
 239         rhs = chars_a2res(h->symtab, p, len, &sz);
 240
 241         if (NULL == rhs)
 242                 return;
 243         fwrite(rhs, 1, sz, stdout);
 244 }
 245
 246
 247 struct tag *
 248 print_ofont(struct html *h, enum htmlfont font)
 249 {
 250         struct htmlpair  tag;
 251
 252         h->metal = h->metac;
 253         h->metac = font;
 254
 255         /* FIXME: DECO_ROMAN should just close out preexisting. */
 256
 257         if (h->metaf && h->tags.head == h->metaf)
 258                 print_tagq(h, h->metaf);
 259
 260         PAIR_CLASS_INIT(&tag, htmlfonts[font]);
 261         h->metaf = print_otag(h, TAG_SPAN, 1, &tag);
 262         return(h->metaf);
 263 }
 264
 265
 266 static void
 267 print_metaf(struct html *h, enum roffdeco deco)
 268 {
 269         enum htmlfont    font;
 270
 271         switch (deco) {
 272         case (DECO_PREVIOUS):
 273                 font = h->metal;
 274                 break;
 275         case (DECO_ITALIC):
 276                 font = HTMLFONT_ITALIC;
 277                 break;
 278         case (DECO_BOLD):
 279                 font = HTMLFONT_BOLD;
 280                 break;
 281         case (DECO_ROMAN):
 282                 font = HTMLFONT_NONE;
 283                 break;
 284         default:
 285                 abort();
 286                 /* NOTREACHED */
 287         }
 288
 289         (void)print_ofont(h, font);
 290 }
 291
 292
 293 static int
 294 print_encode(struct html *h, const char *p, int norecurse)
 295 {
 296         size_t           sz;
 297         int              len, nospace;
 298         const char      *seq;
 299         enum roffdeco    deco;
 300         static const char rejs[6] = { '\\', '<', '>', '&', ASCII_HYPH, '\0' };
 301
 302         nospace = 0;
 303
 304         for (; *p; p++) {
 305                 sz = strcspn(p, rejs);
 306
 307                 fwrite(p, 1, sz, stdout);
 308                 p += /* LINTED */
 309                         sz;
 310
 311                 if ('<' == *p) {
 312                         printf("&lt;");
 313                         continue;
 314                 } else if ('>' == *p) {
 315                         printf("&gt;");
 316                         continue;
 317                 } else if ('&' == *p) {
 318                         printf("&amp;");
 319                         continue;
 320                 } else if (ASCII_HYPH == *p) {
 321                         /*
 322                          * Note: "soft hyphens" aren't graphically
 323                          * displayed when not breaking the text; we want
 324                          * them to be displayed.
 325                          */
 326                         /*printf("&#173;");*/
 327                         putchar('-');
 328                         continue;
 329                 } else if ('\0' == *p)
 330                         break;
 331
 332                 seq = ++p;
 333                 len = a2roffdeco(&deco, &seq, &sz);
 334
 335                 switch (deco) {
 336                 case (DECO_RESERVED):
 337                         print_res(h, seq, sz);
 338                         break;
 339                 case (DECO_SPECIAL):
 340                         print_spec(h, seq, sz);
 341                         break;
 342                 case (DECO_PREVIOUS):
 343                         /* FALLTHROUGH */
 344                 case (DECO_BOLD):
 345                         /* FALLTHROUGH */
 346                 case (DECO_ITALIC):
 347                         /* FALLTHROUGH */
 348                 case (DECO_ROMAN):
 349                         if (norecurse)
 350                                 break;
 351                         print_metaf(h, deco);
 352                         break;
 353                 default:
 354                         break;
 355                 }
 356
 357                 p += len - 1;
 358
 359                 if (DECO_NOSPACE == deco && '\0' == *(p + 1))
 360                         nospace = 1;
 361         }
 362
 363         return(nospace);
 364 }
 365
 366
 367 static void
 368 print_attr(struct html *h, const char *key, const char *val)
 369 {
 370         printf(" %s=\"", key);
 371         (void)print_encode(h, val, 1);
 372         putchar('\"');
 373 }
 374
 375
 376 struct tag *
 377 print_otag(struct html *h, enum htmltag tag,
 378                 int sz, const struct htmlpair *p)
 379 {
 380         int              i;
 381         struct tag      *t;
 382
 383         /* Push this tags onto the stack of open scopes. */
 384
 385         if ( ! (HTML_NOSTACK & htmltags[tag].flags)) {
 386                 t = malloc(sizeof(struct tag));
 387                 if (NULL == t) {
 388                         perror(NULL);
 389                         exit(EXIT_FAILURE);
 390                 }
 391                 t->tag = tag;
 392                 t->next = h->tags.head;
 393                 h->tags.head = t;
 394         } else
 395                 t = NULL;
 396
 397         if ( ! (HTML_NOSPACE & h->flags))
 398                 if ( ! (HTML_CLRLINE & htmltags[tag].flags))
 399                         putchar(' ');
 400
 401         /* Print out the tag name and attributes. */
 402
 403         printf("<%s", htmltags[tag].name);
 404         for (i = 0; i < sz; i++)
 405                 print_attr(h, htmlattrs[p[i].key], p[i].val);
 406
 407         /* Add non-overridable attributes. */
 408
 409         if (TAG_HTML == tag && HTML_XHTML_1_0_STRICT == h->type) {
 410                 print_attr(h, "xmlns", "http://www.w3.org/1999/xhtml");
 411                 print_attr(h, "xml:lang", "en");
 412                 print_attr(h, "lang", "en");
 413         }
 414
 415         /* Accomodate for XML "well-formed" singleton escaping. */
 416
 417         if (HTML_AUTOCLOSE & htmltags[tag].flags)
 418                 switch (h->type) {
 419                 case (HTML_XHTML_1_0_STRICT):
 420                         putchar('/');
 421                         break;
 422                 default:
 423                         break;
 424                 }
 425
 426         putchar('>');
 427
 428         h->flags |= HTML_NOSPACE;
 429         return(t);
 430 }
 431
 432
 433 static void
 434 print_ctag(struct html *h, enum htmltag tag)
 435 {
 436
 437         printf("</%s>", htmltags[tag].name);
 438         if (HTML_CLRLINE & htmltags[tag].flags) {
 439                 h->flags |= HTML_NOSPACE;
 440                 putchar('\n');
 441         }
 442 }
 443
 444
 445 void
 446 print_gen_decls(struct html *h)
 447 {
 448
 449         print_xmltype(h);
 450         print_doctype(h);
 451 }
 452
 453
 454 static void
 455 print_xmltype(struct html *h)
 456 {
 457
 458         if (HTML_XHTML_1_0_STRICT == h->type)
 459                 printf("<?xml version=\"1.0\" encoding=\"UTF-8\"?>");
 460 }
 461
 462
 463 static void
 464 print_doctype(struct html *h)
 465 {
 466         const char      *doctype;
 467         const char      *dtd;
 468         const char      *name;
 469
 470         switch (h->type) {
 471         case (HTML_HTML_4_01_STRICT):
 472                 name = "HTML";
 473                 doctype = "-//W3C//DTD HTML 4.01//EN";
 474                 dtd = "http://www.w3.org/TR/html4/strict.dtd";
 475                 break;
 476         default:
 477                 name = "html";
 478                 doctype = "-//W3C//DTD XHTML 1.0 Strict//EN";
 479                 dtd = "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd";
 480                 break;
 481         }
 482
 483         printf("<!DOCTYPE %s PUBLIC \"%s\" \"%s\">\n",
 484                         name, doctype, dtd);
 485 }
 486
 487
 488 void
 489 print_text(struct html *h, const char *p)
 490 {
 491
 492         if (*p && 0 == *(p + 1))
 493                 switch (*p) {
 494                 case('.'):
 495                         /* FALLTHROUGH */
 496                 case(','):
 497                         /* FALLTHROUGH */
 498                 case(';'):
 499                         /* FALLTHROUGH */
 500                 case(':'):
 501                         /* FALLTHROUGH */
 502                 case('?'):
 503                         /* FALLTHROUGH */
 504                 case('!'):
 505                         /* FALLTHROUGH */
 506                 case(')'):
 507                         /* FALLTHROUGH */
 508                 case(']'):
 509                         if ( ! (HTML_IGNDELIM & h->flags))
 510                                 h->flags |= HTML_NOSPACE;
 511                         break;
 512                 default:
 513                         break;
 514                 }
 515
 516         if ( ! (h->flags & HTML_NOSPACE))
 517                 putchar(' ');
 518
 519         assert(p);
 520         if ( ! print_encode(h, p, 0))
 521                 h->flags &= ~HTML_NOSPACE;
 522
 523         /*
 524          * Note that we don't process the pipe: the parser sees it as
 525          * punctuation, but we don't in terms of typography.
 526          */
 527         if (*p && 0 == *(p + 1))
 528                 switch (*p) {
 529                 case('('):
 530                         /* FALLTHROUGH */
 531                 case('['):
 532                         h->flags |= HTML_NOSPACE;
 533                         break;
 534                 default:
 535                         break;
 536                 }
 537 }
 538
 539
 540 void
 541 print_tagq(struct html *h, const struct tag *until)
 542 {
 543         struct tag      *tag;
 544
 545         while ((tag = h->tags.head) != NULL) {
 546                 if (tag == h->metaf)
 547                         h->metaf = NULL;
 548                 print_ctag(h, tag->tag);
 549                 h->tags.head = tag->next;
 550                 free(tag);
 551                 if (until && tag == until)
 552                         return;
 553         }
 554 }
 555
 556
 557 void
 558 print_stagq(struct html *h, const struct tag *suntil)
 559 {
 560         struct tag      *tag;
 561
 562         while ((tag = h->tags.head) != NULL) {
 563                 if (suntil && tag == suntil)
 564                         return;
 565                 if (tag == h->metaf)
 566                         h->metaf = NULL;
 567                 print_ctag(h, tag->tag);
 568                 h->tags.head = tag->next;
 569                 free(tag);
 570         }
 571 }
 572
 573
 574 void
 575 bufinit(struct html *h)
 576 {
 577
 578         h->buf[0] = '\0';
 579         h->buflen = 0;
 580 }
 581
 582
 583 void
 584 bufcat_style(struct html *h, const char *key, const char *val)
 585 {
 586
 587         bufcat(h, key);
 588         bufncat(h, ":", 1);
 589         bufcat(h, val);
 590         bufncat(h, ";", 1);
 591 }
 592
 593
 594 void
 595 bufcat(struct html *h, const char *p)
 596 {
 597
 598         bufncat(h, p, strlen(p));
 599 }
 600
 601
 602 void
 603 buffmt(struct html *h, const char *fmt, ...)
 604 {
 605         va_list          ap;
 606
 607         va_start(ap, fmt);
 608         (void)vsnprintf(h->buf + (int)h->buflen,
 609                         BUFSIZ - h->buflen - 1, fmt, ap);
 610         va_end(ap);
 611         h->buflen = strlen(h->buf);
 612 }
 613
 614
 615 void
 616 bufncat(struct html *h, const char *p, size_t sz)
 617 {
 618
 619         if (h->buflen + sz > BUFSIZ - 1)
 620                 sz = BUFSIZ - 1 - h->buflen;
 621
 622         (void)strncat(h->buf, p, sz);
 623         h->buflen += sz;
 624 }
 625
 626
 627 void
 628 buffmt_includes(struct html *h, const char *name)
 629 {
 630         const char      *p, *pp;
 631
 632         pp = h->base_includes;
 633
 634         while (NULL != (p = strchr(pp, '%'))) {
 635                 bufncat(h, pp, (size_t)(p - pp));
 636                 switch (*(p + 1)) {
 637                 case('I'):
 638                         bufcat(h, name);
 639                         break;
 640                 default:
 641                         bufncat(h, p, 2);
 642                         break;
 643                 }
 644                 pp = p + 2;
 645         }
 646         if (pp)
 647                 bufcat(h, pp);
 648 }
 649
 650
 651 void
 652 buffmt_man(struct html *h,
 653                 const char *name, const char *sec)
 654 {
 655         const char      *p, *pp;
 656
 657         pp = h->base_man;
 658
 659         /* LINTED */
 660         while (NULL != (p = strchr(pp, '%'))) {
 661                 bufncat(h, pp, (size_t)(p - pp));
 662                 switch (*(p + 1)) {
 663                 case('S'):
 664                         bufcat(h, sec ? sec : "1");
 665                         break;
 666                 case('N'):
 667                         buffmt(h, name);
 668                         break;
 669                 default:
 670                         bufncat(h, p, 2);
 671                         break;
 672                 }
 673                 pp = p + 2;
 674         }
 675         if (pp)
 676                 bufcat(h, pp);
 677 }
 678
 679
 680 void
 681 bufcat_su(struct html *h, const char *p, const struct roffsu *su)
 682 {
 683         double           v;
 684         const char      *u;
 685
 686         v = su->scale;
 687
 688         switch (su->unit) {
 689         case (SCALE_CM):
 690                 u = "cm";
 691                 break;
 692         case (SCALE_IN):
 693                 u = "in";
 694                 break;
 695         case (SCALE_PC):
 696                 u = "pc";
 697                 break;
 698         case (SCALE_PT):
 699                 u = "pt";
 700                 break;
 701         case (SCALE_EM):
 702                 u = "em";
 703                 break;
 704         case (SCALE_MM):
 705                 if (0 == (v /= 100))
 706                         v = 1;
 707                 u = "em";
 708                 break;
 709         case (SCALE_EN):
 710                 u = "ex";
 711                 break;
 712         case (SCALE_BU):
 713                 u = "ex";
 714                 break;
 715         case (SCALE_VS):
 716                 u = "em";
 717                 break;
 718         default:
 719                 u = "ex";
 720                 break;
 721         }
 722
 723         if (su->pt)
 724                 buffmt(h, "%s: %f%s;", p, v, u);
 725         else
 726                 /* LINTED */
 727                 buffmt(h, "%s: %d%s;", p, (int)v, u);
 728 }
 729
 730
 731 void
 732 html_idcat(char *dst, const char *src, int sz)
 733 {
 734         int              ssz;
 735
 736         assert(sz);
 737
 738         /* Cf. <http://www.w3.org/TR/html4/types.html#h-6.2>. */
 739
 740         for ( ; *dst != '\0' && sz; dst++, sz--)
 741                 /* Jump to end. */ ;
 742
 743         assert(sz > 2);
 744
 745         /* We can't start with a number (bah). */
 746
 747         *dst++ = 'x';
 748         *dst = '\0';
 749         sz--;
 750
 751         for ( ; *src != '\0' && sz > 1; src++) {
 752                 ssz = snprintf(dst, (size_t)sz, "%.2x", *src);
 753                 sz -= ssz;
 754                 dst += ssz;
 755         }
 756 }