html.c

   1 /*      $Id: html.c,v 1.99 2010/04/12 19:45:39 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 "out.h"
  33 #include "chars.h"
  34 #include "html.h"
  35 #include "main.h"
  36
  37 #define UNCONST(a)      ((void *)(uintptr_t)(const void *)(a))
  38
  39 struct  htmldata {
  40         const char       *name;
  41         int               flags;
  42 #define HTML_CLRLINE     (1 << 0)
  43 #define HTML_NOSTACK     (1 << 1)
  44 #define HTML_AUTOCLOSE   (1 << 2) /* Tag has auto-closure. */
  45 };
  46
  47 static  const struct htmldata htmltags[TAG_MAX] = {
  48         {"html",        HTML_CLRLINE}, /* TAG_HTML */
  49         {"head",        HTML_CLRLINE}, /* TAG_HEAD */
  50         {"body",        HTML_CLRLINE}, /* TAG_BODY */
  51         {"meta",        HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_META */
  52         {"title",       HTML_CLRLINE}, /* TAG_TITLE */
  53         {"div",         HTML_CLRLINE}, /* TAG_DIV */
  54         {"h1",          0}, /* TAG_H1 */
  55         {"h2",          0}, /* TAG_H2 */
  56         {"span",        0}, /* TAG_SPAN */
  57         {"link",        HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_LINK */
  58         {"br",          HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_BR */
  59         {"a",           0}, /* TAG_A */
  60         {"table",       HTML_CLRLINE}, /* TAG_TABLE */
  61         {"col",         HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_COL */
  62         {"tr",          HTML_CLRLINE}, /* TAG_TR */
  63         {"td",          HTML_CLRLINE}, /* TAG_TD */
  64         {"li",          HTML_CLRLINE}, /* TAG_LI */
  65         {"ul",          HTML_CLRLINE}, /* TAG_UL */
  66         {"ol",          HTML_CLRLINE}, /* TAG_OL */
  67 };
  68
  69 static  const char      *const htmlfonts[HTMLFONT_MAX] = {
  70         "roman",
  71         "bold",
  72         "italic"
  73 };
  74
  75 static  const char      *const htmlattrs[ATTR_MAX] = {
  76         "http-equiv",
  77         "content",
  78         "name",
  79         "rel",
  80         "href",
  81         "type",
  82         "media",
  83         "class",
  84         "style",
  85         "width",
  86         "valign",
  87         "target",
  88         "id",
  89         "summary",
  90 };
  91
  92 static  void              print_spec(struct html *, const char *, size_t);
  93 static  void              print_res(struct html *, const char *, size_t);
  94 static  void              print_ctag(struct html *, enum htmltag);
  95 static  void              print_doctype(struct html *);
  96 static  void              print_xmltype(struct html *);
  97 static  int               print_encode(struct html *, const char *, int);
  98 static  void              print_metaf(struct html *, enum roffdeco);
  99 static  void              print_attr(struct html *,
 100                                 const char *, const char *);
 101 static  void             *ml_alloc(char *, enum htmltype);
 102
 103
 104 static void *
 105 ml_alloc(char *outopts, enum htmltype type)
 106 {
 107         struct html     *h;
 108         const char      *toks[4];
 109         char            *v;
 110
 111         toks[0] = "style";
 112         toks[1] = "man";
 113         toks[2] = "includes";
 114         toks[3] = NULL;
 115
 116         h = calloc(1, sizeof(struct html));
 117         if (NULL == h) {
 118                 perror(NULL);
 119                 exit(EXIT_FAILURE);
 120         }
 121
 122         h->type = type;
 123         h->tags.head = NULL;
 124         h->ords.head = NULL;
 125         h->symtab = chars_init(CHARS_HTML);
 126
 127         while (outopts && *outopts)
 128                 switch (getsubopt(&outopts, UNCONST(toks), &v)) {
 129                 case (0):
 130                         h->style = v;
 131                         break;
 132                 case (1):
 133                         h->base_man = v;
 134                         break;
 135                 case (2):
 136                         h->base_includes = v;
 137                         break;
 138                 default:
 139                         break;
 140                 }
 141
 142         return(h);
 143 }
 144
 145 void *
 146 html_alloc(char *outopts)
 147 {
 148
 149         return(ml_alloc(outopts, HTML_HTML_4_01_STRICT));
 150 }
 151
 152
 153 void *
 154 xhtml_alloc(char *outopts)
 155 {
 156
 157         return(ml_alloc(outopts, HTML_XHTML_1_0_STRICT));
 158 }
 159
 160
 161 void
 162 html_free(void *p)
 163 {
 164         struct tag      *tag;
 165         struct ord      *ord;
 166         struct html     *h;
 167
 168         h = (struct html *)p;
 169
 170         while ((ord = h->ords.head) != NULL) {
 171                 h->ords.head = ord->next;
 172                 free(ord);
 173         }
 174
 175         while ((tag = h->tags.head) != NULL) {
 176                 h->tags.head = tag->next;
 177                 free(tag);
 178         }
 179
 180         if (h->symtab)
 181                 chars_free(h->symtab);
 182
 183         free(h);
 184 }
 185
 186
 187 void
 188 print_gen_head(struct html *h)
 189 {
 190         struct htmlpair  tag[4];
 191
 192         tag[0].key = ATTR_HTTPEQUIV;
 193         tag[0].val = "Content-Type";
 194         tag[1].key = ATTR_CONTENT;
 195         tag[1].val = "text/html; charset=utf-8";
 196         print_otag(h, TAG_META, 2, tag);
 197
 198         tag[0].key = ATTR_NAME;
 199         tag[0].val = "resource-type";
 200         tag[1].key = ATTR_CONTENT;
 201         tag[1].val = "document";
 202         print_otag(h, TAG_META, 2, tag);
 203
 204         if (h->style) {
 205                 tag[0].key = ATTR_REL;
 206                 tag[0].val = "stylesheet";
 207                 tag[1].key = ATTR_HREF;
 208                 tag[1].val = h->style;
 209                 tag[2].key = ATTR_TYPE;
 210                 tag[2].val = "text/css";
 211                 tag[3].key = ATTR_MEDIA;
 212                 tag[3].val = "all";
 213                 print_otag(h, TAG_LINK, 4, tag);
 214         }
 215 }
 216
 217
 218 static void
 219 print_spec(struct html *h, const char *p, size_t len)
 220 {
 221         const char      *rhs;
 222         size_t           sz;
 223
 224         rhs = chars_a2ascii(h->symtab, p, len, &sz);
 225
 226         if (NULL == rhs)
 227                 return;
 228         fwrite(rhs, 1, sz, stdout);
 229 }
 230
 231
 232 static void
 233 print_res(struct html *h, const char *p, size_t len)
 234 {
 235         const char      *rhs;
 236         size_t           sz;
 237
 238         rhs = chars_a2res(h->symtab, p, len, &sz);
 239
 240         if (NULL == rhs)
 241                 return;
 242         fwrite(rhs, 1, sz, stdout);
 243 }
 244
 245
 246 struct tag *
 247 print_ofont(struct html *h, enum htmlfont font)
 248 {
 249         struct htmlpair  tag;
 250
 251         h->metal = h->metac;
 252         h->metac = font;
 253
 254         /* FIXME: DECO_ROMAN should just close out preexisting. */
 255
 256         if (h->metaf && h->tags.head == h->metaf)
 257                 print_tagq(h, h->metaf);
 258
 259         PAIR_CLASS_INIT(&tag, htmlfonts[font]);
 260         h->metaf = print_otag(h, TAG_SPAN, 1, &tag);
 261         return(h->metaf);
 262 }
 263
 264
 265 static void
 266 print_metaf(struct html *h, enum roffdeco deco)
 267 {
 268         enum htmlfont    font;
 269
 270         switch (deco) {
 271         case (DECO_PREVIOUS):
 272                 font = h->metal;
 273                 break;
 274         case (DECO_ITALIC):
 275                 font = HTMLFONT_ITALIC;
 276                 break;
 277         case (DECO_BOLD):
 278                 font = HTMLFONT_BOLD;
 279                 break;
 280         case (DECO_ROMAN):
 281                 font = HTMLFONT_NONE;
 282                 break;
 283         default:
 284                 abort();
 285                 /* NOTREACHED */
 286         }
 287
 288         (void)print_ofont(h, font);
 289 }
 290
 291
 292 static int
 293 print_encode(struct html *h, const char *p, int norecurse)
 294 {
 295         size_t           sz;
 296         int              len, nospace;
 297         const char      *seq;
 298         enum roffdeco    deco;
 299
 300         nospace = 0;
 301
 302         for (; *p; p++) {
 303                 sz = strcspn(p, "\\<>&");
 304
 305                 fwrite(p, 1, sz, stdout);
 306                 p += /* LINTED */
 307                         sz;
 308
 309                 if ('<' == *p) {
 310                         printf("&lt;");
 311                         continue;
 312                 } else if ('>' == *p) {
 313                         printf("&gt;");
 314                         continue;
 315                 } else if ('&' == *p) {
 316                         printf("&amp;");
 317                         continue;
 318                 } else if ('\0' == *p)
 319                         break;
 320
 321                 seq = ++p;
 322                 len = a2roffdeco(&deco, &seq, &sz);
 323
 324                 switch (deco) {
 325                 case (DECO_RESERVED):
 326                         print_res(h, seq, sz);
 327                         break;
 328                 case (DECO_SPECIAL):
 329                         print_spec(h, seq, sz);
 330                         break;
 331                 case (DECO_PREVIOUS):
 332                         /* FALLTHROUGH */
 333                 case (DECO_BOLD):
 334                         /* FALLTHROUGH */
 335                 case (DECO_ITALIC):
 336                         /* FALLTHROUGH */
 337                 case (DECO_ROMAN):
 338                         if (norecurse)
 339                                 break;
 340                         print_metaf(h, deco);
 341                         break;
 342                 default:
 343                         break;
 344                 }
 345
 346                 p += len - 1;
 347
 348                 if (DECO_NOSPACE == deco && '\0' == *(p + 1))
 349                         nospace = 1;
 350         }
 351
 352         return(nospace);
 353 }
 354
 355
 356 static void
 357 print_attr(struct html *h, const char *key, const char *val)
 358 {
 359         printf(" %s=\"", key);
 360         (void)print_encode(h, val, 1);
 361         putchar('\"');
 362 }
 363
 364
 365 struct tag *
 366 print_otag(struct html *h, enum htmltag tag,
 367                 int sz, const struct htmlpair *p)
 368 {
 369         int              i;
 370         struct tag      *t;
 371
 372         /* Push this tags onto the stack of open scopes. */
 373
 374         if ( ! (HTML_NOSTACK & htmltags[tag].flags)) {
 375                 t = malloc(sizeof(struct tag));
 376                 if (NULL == t) {
 377                         perror(NULL);
 378                         exit(EXIT_FAILURE);
 379                 }
 380                 t->tag = tag;
 381                 t->next = h->tags.head;
 382                 h->tags.head = t;
 383         } else
 384                 t = NULL;
 385
 386         if ( ! (HTML_NOSPACE & h->flags))
 387                 if ( ! (HTML_CLRLINE & htmltags[tag].flags))
 388                         putchar(' ');
 389
 390         /* Print out the tag name and attributes. */
 391
 392         printf("<%s", htmltags[tag].name);
 393         for (i = 0; i < sz; i++)
 394                 print_attr(h, htmlattrs[p[i].key], p[i].val);
 395
 396         /* Add non-overridable attributes. */
 397
 398         if (TAG_HTML == tag && HTML_XHTML_1_0_STRICT == h->type) {
 399                 print_attr(h, "xmlns", "http://www.w3.org/1999/xhtml");
 400                 print_attr(h, "xml:lang", "en");
 401                 print_attr(h, "lang", "en");
 402         }
 403
 404         /* Accomodate for XML "well-formed" singleton escaping. */
 405
 406         if (HTML_AUTOCLOSE & htmltags[tag].flags)
 407                 switch (h->type) {
 408                 case (HTML_XHTML_1_0_STRICT):
 409                         putchar('/');
 410                         break;
 411                 default:
 412                         break;
 413                 }
 414
 415         putchar('>');
 416
 417         h->flags |= HTML_NOSPACE;
 418         return(t);
 419 }
 420
 421
 422 static void
 423 print_ctag(struct html *h, enum htmltag tag)
 424 {
 425
 426         printf("</%s>", htmltags[tag].name);
 427         if (HTML_CLRLINE & htmltags[tag].flags) {
 428                 h->flags |= HTML_NOSPACE;
 429                 putchar('\n');
 430         }
 431 }
 432
 433
 434 void
 435 print_gen_decls(struct html *h)
 436 {
 437
 438         print_xmltype(h);
 439         print_doctype(h);
 440 }
 441
 442
 443 static void
 444 print_xmltype(struct html *h)
 445 {
 446         const char      *decl;
 447
 448         switch (h->type) {
 449         case (HTML_XHTML_1_0_STRICT):
 450                 decl = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>";
 451                 break;
 452         default:
 453                 decl = NULL;
 454                 break;
 455         }
 456
 457         if (NULL == decl)
 458                 return;
 459
 460         printf("%s\n", decl);
 461 }
 462
 463
 464 static void
 465 print_doctype(struct html *h)
 466 {
 467         const char      *doctype;
 468         const char      *dtd;
 469         const char      *name;
 470
 471         switch (h->type) {
 472         case (HTML_HTML_4_01_STRICT):
 473                 name = "HTML";
 474                 doctype = "-//W3C//DTD HTML 4.01//EN";
 475                 dtd = "http://www.w3.org/TR/html4/strict.dtd";
 476                 break;
 477         default:
 478                 name = "html";
 479                 doctype = "-//W3C//DTD XHTML 1.0 Strict//EN";
 480                 dtd = "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd";
 481                 break;
 482         }
 483
 484         printf("<!DOCTYPE %s PUBLIC \"%s\" \"%s\">\n",
 485                         name, doctype, dtd);
 486 }
 487
 488
 489 void
 490 print_text(struct html *h, const char *p)
 491 {
 492
 493         if (*p && 0 == *(p + 1))
 494                 switch (*p) {
 495                 case('.'):
 496                         /* FALLTHROUGH */
 497                 case(','):
 498                         /* FALLTHROUGH */
 499                 case(';'):
 500                         /* FALLTHROUGH */
 501                 case(':'):
 502                         /* FALLTHROUGH */
 503                 case('?'):
 504                         /* FALLTHROUGH */
 505                 case('!'):
 506                         /* FALLTHROUGH */
 507                 case(')'):
 508                         /* FALLTHROUGH */
 509                 case(']'):
 510                         if ( ! (HTML_IGNDELIM & h->flags))
 511                                 h->flags |= HTML_NOSPACE;
 512                         break;
 513                 default:
 514                         break;
 515                 }
 516
 517         if ( ! (h->flags & HTML_NOSPACE))
 518                 putchar(' ');
 519
 520         assert(p);
 521         if ( ! print_encode(h, p, 0))
 522                 h->flags &= ~HTML_NOSPACE;
 523
 524         /*
 525          * Note that we don't process the pipe: the parser sees it as
 526          * punctuation, but we don't in terms of typography.
 527          */
 528         if (*p && 0 == *(p + 1))
 529                 switch (*p) {
 530                 case('('):
 531                         /* FALLTHROUGH */
 532                 case('['):
 533                         h->flags |= HTML_NOSPACE;
 534                         break;
 535                 default:
 536                         break;
 537                 }
 538 }
 539
 540
 541 void
 542 print_tagq(struct html *h, const struct tag *until)
 543 {
 544         struct tag      *tag;
 545
 546         while ((tag = h->tags.head) != NULL) {
 547                 if (tag == h->metaf)
 548                         h->metaf = NULL;
 549                 print_ctag(h, tag->tag);
 550                 h->tags.head = tag->next;
 551                 free(tag);
 552                 if (until && tag == until)
 553                         return;
 554         }
 555 }
 556
 557
 558 void
 559 print_stagq(struct html *h, const struct tag *suntil)
 560 {
 561         struct tag      *tag;
 562
 563         while ((tag = h->tags.head) != NULL) {
 564                 if (suntil && tag == suntil)
 565                         return;
 566                 if (tag == h->metaf)
 567                         h->metaf = NULL;
 568                 print_ctag(h, tag->tag);
 569                 h->tags.head = tag->next;
 570                 free(tag);
 571         }
 572 }
 573
 574
 575 void
 576 bufinit(struct html *h)
 577 {
 578
 579         h->buf[0] = '\0';
 580         h->buflen = 0;
 581 }
 582
 583
 584 void
 585 bufcat_style(struct html *h, const char *key, const char *val)
 586 {
 587
 588         bufcat(h, key);
 589         bufncat(h, ":", 1);
 590         bufcat(h, val);
 591         bufncat(h, ";", 1);
 592 }
 593
 594
 595 void
 596 bufcat(struct html *h, const char *p)
 597 {
 598
 599         bufncat(h, p, strlen(p));
 600 }
 601
 602
 603 void
 604 buffmt(struct html *h, const char *fmt, ...)
 605 {
 606         va_list          ap;
 607
 608         va_start(ap, fmt);
 609         (void)vsnprintf(h->buf + (int)h->buflen,
 610                         BUFSIZ - h->buflen - 1, fmt, ap);
 611         va_end(ap);
 612         h->buflen = strlen(h->buf);
 613 }
 614
 615
 616 void
 617 bufncat(struct html *h, const char *p, size_t sz)
 618 {
 619
 620         if (h->buflen + sz > BUFSIZ - 1)
 621                 sz = BUFSIZ - 1 - h->buflen;
 622
 623         (void)strncat(h->buf, p, sz);
 624         h->buflen += sz;
 625 }
 626
 627
 628 void
 629 buffmt_includes(struct html *h, const char *name)
 630 {
 631         const char      *p, *pp;
 632
 633         pp = h->base_includes;
 634
 635         while (NULL != (p = strchr(pp, '%'))) {
 636                 bufncat(h, pp, (size_t)(p - pp));
 637                 switch (*(p + 1)) {
 638                 case('I'):
 639                         bufcat(h, name);
 640                         break;
 641                 default:
 642                         bufncat(h, p, 2);
 643                         break;
 644                 }
 645                 pp = p + 2;
 646         }
 647         if (pp)
 648                 bufcat(h, pp);
 649 }
 650
 651
 652 void
 653 buffmt_man(struct html *h,
 654                 const char *name, const char *sec)
 655 {
 656         const char      *p, *pp;
 657
 658         pp = h->base_man;
 659
 660         /* LINTED */
 661         while (NULL != (p = strchr(pp, '%'))) {
 662                 bufncat(h, pp, (size_t)(p - pp));
 663                 switch (*(p + 1)) {
 664                 case('S'):
 665                         bufcat(h, sec ? sec : "1");
 666                         break;
 667                 case('N'):
 668                         buffmt(h, name);
 669                         break;
 670                 default:
 671                         bufncat(h, p, 2);
 672                         break;
 673                 }
 674                 pp = p + 2;
 675         }
 676         if (pp)
 677                 bufcat(h, pp);
 678 }
 679
 680
 681 void
 682 bufcat_su(struct html *h, const char *p, const struct roffsu *su)
 683 {
 684         double           v;
 685         const char      *u;
 686
 687         v = su->scale;
 688
 689         switch (su->unit) {
 690         case (SCALE_CM):
 691                 u = "cm";
 692                 break;
 693         case (SCALE_IN):
 694                 u = "in";
 695                 break;
 696         case (SCALE_PC):
 697                 u = "pc";
 698                 break;
 699         case (SCALE_PT):
 700                 u = "pt";
 701                 break;
 702         case (SCALE_EM):
 703                 u = "em";
 704                 break;
 705         case (SCALE_MM):
 706                 if (0 == (v /= 100))
 707                         v = 1;
 708                 u = "em";
 709                 break;
 710         case (SCALE_EN):
 711                 u = "ex";
 712                 break;
 713         case (SCALE_BU):
 714                 u = "ex";
 715                 break;
 716         case (SCALE_VS):
 717                 u = "em";
 718                 break;
 719         default:
 720                 u = "ex";
 721                 break;
 722         }
 723
 724         if (su->pt)
 725                 buffmt(h, "%s: %f%s;", p, v, u);
 726         else
 727                 /* LINTED */
 728                 buffmt(h, "%s: %d%s;", p, (int)v, u);
 729 }
 730
 731
 732 void
 733 html_idcat(char *dst, const char *src, int sz)
 734 {
 735         int              ssz;
 736
 737         assert(sz);
 738
 739         /* Cf. <http://www.w3.org/TR/html4/types.html#h-6.2>. */
 740
 741         for ( ; *dst != '\0' && sz; dst++, sz--)
 742                 /* Jump to end. */ ;
 743
 744         assert(sz > 2);
 745
 746         /* We can't start with a number (bah). */
 747
 748         *dst++ = 'x';
 749         *dst = '\0';
 750         sz--;
 751
 752         for ( ; *src != '\0' && sz > 1; src++) {
 753                 ssz = snprintf(dst, (size_t)sz, "%.2x", *src);
 754                 sz -= ssz;
 755                 dst += ssz;
 756         }
 757 }