html.c

   1 /*      $Id: html.c,v 1.109 2010/07/23 00:08:57 kristaps Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
   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 struct  htmldata {
  39         const char       *name;
  40         int               flags;
  41 #define HTML_CLRLINE     (1 << 0)
  42 #define HTML_NOSTACK     (1 << 1)
  43 #define HTML_AUTOCLOSE   (1 << 2) /* Tag has auto-closure. */
  44 };
  45
  46 static  const struct htmldata htmltags[TAG_MAX] = {
  47         {"html",        HTML_CLRLINE}, /* TAG_HTML */
  48         {"head",        HTML_CLRLINE}, /* TAG_HEAD */
  49         {"body",        HTML_CLRLINE}, /* TAG_BODY */
  50         {"meta",        HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_META */
  51         {"title",       HTML_CLRLINE}, /* TAG_TITLE */
  52         {"div",         HTML_CLRLINE}, /* TAG_DIV */
  53         {"h1",          0}, /* TAG_H1 */
  54         {"h2",          0}, /* TAG_H2 */
  55         {"span",        0}, /* TAG_SPAN */
  56         {"link",        HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_LINK */
  57         {"br",          HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_BR */
  58         {"a",           0}, /* TAG_A */
  59         {"table",       HTML_CLRLINE}, /* TAG_TABLE */
  60         {"col",         HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_COL */
  61         {"tr",          HTML_CLRLINE}, /* TAG_TR */
  62         {"td",          HTML_CLRLINE}, /* TAG_TD */
  63         {"li",          HTML_CLRLINE}, /* TAG_LI */
  64         {"ul",          HTML_CLRLINE}, /* TAG_UL */
  65         {"ol",          HTML_CLRLINE}, /* TAG_OL */
  66 };
  67
  68 static  const char      *const htmlfonts[HTMLFONT_MAX] = {
  69         "roman",
  70         "bold",
  71         "italic"
  72 };
  73
  74 static  const char      *const htmlattrs[ATTR_MAX] = {
  75         "http-equiv",
  76         "content",
  77         "name",
  78         "rel",
  79         "href",
  80         "type",
  81         "media",
  82         "class",
  83         "style",
  84         "width",
  85         "valign",
  86         "target",
  87         "id",
  88         "summary",
  89 };
  90
  91 static  void              print_spec(struct html *, enum roffdeco,
  92                                 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, enum roffdeco d, const char *p, size_t len)
 220 {
 221         int              cp;
 222         const char      *rhs;
 223         size_t           sz;
 224
 225         if ((cp = chars_spec2cp(h->symtab, p, len)) > 0) {
 226                 printf("&#%d;", cp);
 227                 return;
 228         } else if (-1 == cp && DECO_SSPECIAL == d) {
 229                 fwrite(p, 1, len, stdout);
 230                 return;
 231         } else if (-1 == cp)
 232                 return;
 233
 234         if (NULL != (rhs = chars_spec2str(h->symtab, p, len, &sz)))
 235                 fwrite(rhs, 1, sz, stdout);
 236 }
 237
 238
 239 static void
 240 print_res(struct html *h, const char *p, size_t len)
 241 {
 242         int              cp;
 243         const char      *rhs;
 244         size_t           sz;
 245
 246         if ((cp = chars_res2cp(h->symtab, p, len)) > 0) {
 247                 printf("&#%d;", cp);
 248                 return;
 249         } else if (-1 == cp)
 250                 return;
 251
 252         if (NULL != (rhs = chars_res2str(h->symtab, p, len, &sz)))
 253                 fwrite(rhs, 1, sz, stdout);
 254 }
 255
 256
 257 struct tag *
 258 print_ofont(struct html *h, enum htmlfont font)
 259 {
 260         struct htmlpair  tag;
 261
 262         h->metal = h->metac;
 263         h->metac = font;
 264
 265         /* FIXME: DECO_ROMAN should just close out preexisting. */
 266
 267         if (h->metaf && h->tags.head == h->metaf)
 268                 print_tagq(h, h->metaf);
 269
 270         PAIR_CLASS_INIT(&tag, htmlfonts[font]);
 271         h->metaf = print_otag(h, TAG_SPAN, 1, &tag);
 272         return(h->metaf);
 273 }
 274
 275
 276 static void
 277 print_metaf(struct html *h, enum roffdeco deco)
 278 {
 279         enum htmlfont    font;
 280
 281         switch (deco) {
 282         case (DECO_PREVIOUS):
 283                 font = h->metal;
 284                 break;
 285         case (DECO_ITALIC):
 286                 font = HTMLFONT_ITALIC;
 287                 break;
 288         case (DECO_BOLD):
 289                 font = HTMLFONT_BOLD;
 290                 break;
 291         case (DECO_ROMAN):
 292                 font = HTMLFONT_NONE;
 293                 break;
 294         default:
 295                 abort();
 296                 /* NOTREACHED */
 297         }
 298
 299         (void)print_ofont(h, font);
 300 }
 301
 302
 303 static int
 304 print_encode(struct html *h, const char *p, int norecurse)
 305 {
 306         size_t           sz;
 307         int              len, nospace;
 308         const char      *seq;
 309         enum roffdeco    deco;
 310         static const char rejs[6] = { '\\', '<', '>', '&', ASCII_HYPH, '\0' };
 311
 312         nospace = 0;
 313
 314         for (; *p; p++) {
 315                 sz = strcspn(p, rejs);
 316
 317                 fwrite(p, 1, sz, stdout);
 318                 p += /* LINTED */
 319                         sz;
 320
 321                 if ('<' == *p) {
 322                         printf("&lt;");
 323                         continue;
 324                 } else if ('>' == *p) {
 325                         printf("&gt;");
 326                         continue;
 327                 } else if ('&' == *p) {
 328                         printf("&amp;");
 329                         continue;
 330                 } else if (ASCII_HYPH == *p) {
 331                         /*
 332                          * Note: "soft hyphens" aren't graphically
 333                          * displayed when not breaking the text; we want
 334                          * them to be displayed.
 335                          */
 336                         /*printf("&#173;");*/
 337                         putchar('-');
 338                         continue;
 339                 } else if ('\0' == *p)
 340                         break;
 341
 342                 seq = ++p;
 343                 len = a2roffdeco(&deco, &seq, &sz);
 344
 345                 switch (deco) {
 346                 case (DECO_RESERVED):
 347                         print_res(h, seq, sz);
 348                         break;
 349                 case (DECO_SSPECIAL):
 350                         /* FALLTHROUGH */
 351                 case (DECO_SPECIAL):
 352                         print_spec(h, deco, seq, sz);
 353                         break;
 354                 case (DECO_PREVIOUS):
 355                         /* FALLTHROUGH */
 356                 case (DECO_BOLD):
 357                         /* FALLTHROUGH */
 358                 case (DECO_ITALIC):
 359                         /* FALLTHROUGH */
 360                 case (DECO_ROMAN):
 361                         if (norecurse)
 362                                 break;
 363                         print_metaf(h, deco);
 364                         break;
 365                 default:
 366                         break;
 367                 }
 368
 369                 p += len - 1;
 370
 371                 if (DECO_NOSPACE == deco && '\0' == *(p + 1))
 372                         nospace = 1;
 373         }
 374
 375         return(nospace);
 376 }
 377
 378
 379 static void
 380 print_attr(struct html *h, const char *key, const char *val)
 381 {
 382         printf(" %s=\"", key);
 383         (void)print_encode(h, val, 1);
 384         putchar('\"');
 385 }
 386
 387
 388 struct tag *
 389 print_otag(struct html *h, enum htmltag tag,
 390                 int sz, const struct htmlpair *p)
 391 {
 392         int              i;
 393         struct tag      *t;
 394
 395         /* Push this tags onto the stack of open scopes. */
 396
 397         if ( ! (HTML_NOSTACK & htmltags[tag].flags)) {
 398                 t = malloc(sizeof(struct tag));
 399                 if (NULL == t) {
 400                         perror(NULL);
 401                         exit(EXIT_FAILURE);
 402                 }
 403                 t->tag = tag;
 404                 t->next = h->tags.head;
 405                 h->tags.head = t;
 406         } else
 407                 t = NULL;
 408
 409         if ( ! (HTML_NOSPACE & h->flags))
 410                 if ( ! (HTML_CLRLINE & htmltags[tag].flags)) {
 411                         /* Manage keeps! */
 412                         if ( ! (HTML_KEEP & h->flags)) {
 413                                 if (HTML_PREKEEP & h->flags)
 414                                         h->flags |= HTML_KEEP;
 415                                 putchar(' ');
 416                         } else
 417                                 printf("&#160;");
 418                 }
 419
 420         if ( ! (h->flags & HTML_NONOSPACE))
 421                 h->flags &= ~HTML_NOSPACE;
 422
 423         /* Print out the tag name and attributes. */
 424
 425         printf("<%s", htmltags[tag].name);
 426         for (i = 0; i < sz; i++)
 427                 print_attr(h, htmlattrs[p[i].key], p[i].val);
 428
 429         /* Add non-overridable attributes. */
 430
 431         if (TAG_HTML == tag && HTML_XHTML_1_0_STRICT == h->type) {
 432                 print_attr(h, "xmlns", "http://www.w3.org/1999/xhtml");
 433                 print_attr(h, "xml:lang", "en");
 434                 print_attr(h, "lang", "en");
 435         }
 436
 437         /* Accomodate for XML "well-formed" singleton escaping. */
 438
 439         if (HTML_AUTOCLOSE & htmltags[tag].flags)
 440                 switch (h->type) {
 441                 case (HTML_XHTML_1_0_STRICT):
 442                         putchar('/');
 443                         break;
 444                 default:
 445                         break;
 446                 }
 447
 448         putchar('>');
 449
 450         h->flags |= HTML_NOSPACE;
 451         return(t);
 452 }
 453
 454
 455 static void
 456 print_ctag(struct html *h, enum htmltag tag)
 457 {
 458
 459         printf("</%s>", htmltags[tag].name);
 460         if (HTML_CLRLINE & htmltags[tag].flags) {
 461                 h->flags |= HTML_NOSPACE;
 462                 putchar('\n');
 463         }
 464 }
 465
 466
 467 void
 468 print_gen_decls(struct html *h)
 469 {
 470
 471         print_xmltype(h);
 472         print_doctype(h);
 473 }
 474
 475
 476 static void
 477 print_xmltype(struct html *h)
 478 {
 479
 480         if (HTML_XHTML_1_0_STRICT == h->type)
 481                 printf("<?xml version=\"1.0\" encoding=\"UTF-8\"?>");
 482 }
 483
 484
 485 static void
 486 print_doctype(struct html *h)
 487 {
 488         const char      *doctype;
 489         const char      *dtd;
 490         const char      *name;
 491
 492         switch (h->type) {
 493         case (HTML_HTML_4_01_STRICT):
 494                 name = "HTML";
 495                 doctype = "-//W3C//DTD HTML 4.01//EN";
 496                 dtd = "http://www.w3.org/TR/html4/strict.dtd";
 497                 break;
 498         default:
 499                 name = "html";
 500                 doctype = "-//W3C//DTD XHTML 1.0 Strict//EN";
 501                 dtd = "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd";
 502                 break;
 503         }
 504
 505         printf("<!DOCTYPE %s PUBLIC \"%s\" \"%s\">\n",
 506                         name, doctype, dtd);
 507 }
 508
 509
 510 void
 511 print_text(struct html *h, const char *word)
 512 {
 513
 514         if (word[0] && '\0' == word[1])
 515                 switch (word[0]) {
 516                 case('.'):
 517                         /* FALLTHROUGH */
 518                 case(','):
 519                         /* FALLTHROUGH */
 520                 case(';'):
 521                         /* FALLTHROUGH */
 522                 case(':'):
 523                         /* FALLTHROUGH */
 524                 case('?'):
 525                         /* FALLTHROUGH */
 526                 case('!'):
 527                         /* FALLTHROUGH */
 528                 case(')'):
 529                         /* FALLTHROUGH */
 530                 case(']'):
 531                         if ( ! (HTML_IGNDELIM & h->flags))
 532                                 h->flags |= HTML_NOSPACE;
 533                         break;
 534                 default:
 535                         break;
 536                 }
 537
 538         if ( ! (HTML_NOSPACE & h->flags)) {
 539                 /* Manage keeps! */
 540                 if ( ! (HTML_KEEP & h->flags)) {
 541                         if (HTML_PREKEEP & h->flags)
 542                                 h->flags |= HTML_KEEP;
 543                         putchar(' ');
 544                 } else
 545                         printf("&#160;");
 546         }
 547
 548         assert(word);
 549         if ( ! print_encode(h, word, 0))
 550                 if ( ! (h->flags & HTML_NONOSPACE))
 551                         h->flags &= ~HTML_NOSPACE;
 552
 553         /*
 554          * Note that we don't process the pipe: the parser sees it as
 555          * punctuation, but we don't in terms of typography.
 556          */
 557         if (word[0] && '\0' == word[1])
 558                 switch (word[0]) {
 559                 case('('):
 560                         /* FALLTHROUGH */
 561                 case('['):
 562                         h->flags |= HTML_NOSPACE;
 563                         break;
 564                 default:
 565                         break;
 566                 }
 567 }
 568
 569
 570 void
 571 print_tagq(struct html *h, const struct tag *until)
 572 {
 573         struct tag      *tag;
 574
 575         while ((tag = h->tags.head) != NULL) {
 576                 if (tag == h->metaf)
 577                         h->metaf = NULL;
 578                 print_ctag(h, tag->tag);
 579                 h->tags.head = tag->next;
 580                 free(tag);
 581                 if (until && tag == until)
 582                         return;
 583         }
 584 }
 585
 586
 587 void
 588 print_stagq(struct html *h, const struct tag *suntil)
 589 {
 590         struct tag      *tag;
 591
 592         while ((tag = h->tags.head) != NULL) {
 593                 if (suntil && tag == suntil)
 594                         return;
 595                 if (tag == h->metaf)
 596                         h->metaf = NULL;
 597                 print_ctag(h, tag->tag);
 598                 h->tags.head = tag->next;
 599                 free(tag);
 600         }
 601 }
 602
 603
 604 void
 605 bufinit(struct html *h)
 606 {
 607
 608         h->buf[0] = '\0';
 609         h->buflen = 0;
 610 }
 611
 612
 613 void
 614 bufcat_style(struct html *h, const char *key, const char *val)
 615 {
 616
 617         bufcat(h, key);
 618         bufncat(h, ":", 1);
 619         bufcat(h, val);
 620         bufncat(h, ";", 1);
 621 }
 622
 623
 624 void
 625 bufcat(struct html *h, const char *p)
 626 {
 627
 628         bufncat(h, p, strlen(p));
 629 }
 630
 631
 632 void
 633 buffmt(struct html *h, const char *fmt, ...)
 634 {
 635         va_list          ap;
 636
 637         va_start(ap, fmt);
 638         (void)vsnprintf(h->buf + (int)h->buflen,
 639                         BUFSIZ - h->buflen - 1, fmt, ap);
 640         va_end(ap);
 641         h->buflen = strlen(h->buf);
 642 }
 643
 644
 645 void
 646 bufncat(struct html *h, const char *p, size_t sz)
 647 {
 648
 649         if (h->buflen + sz > BUFSIZ - 1)
 650                 sz = BUFSIZ - 1 - h->buflen;
 651
 652         (void)strncat(h->buf, p, sz);
 653         h->buflen += sz;
 654 }
 655
 656
 657 void
 658 buffmt_includes(struct html *h, const char *name)
 659 {
 660         const char      *p, *pp;
 661
 662         pp = h->base_includes;
 663
 664         while (NULL != (p = strchr(pp, '%'))) {
 665                 bufncat(h, pp, (size_t)(p - pp));
 666                 switch (*(p + 1)) {
 667                 case('I'):
 668                         bufcat(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 buffmt_man(struct html *h,
 683                 const char *name, const char *sec)
 684 {
 685         const char      *p, *pp;
 686
 687         pp = h->base_man;
 688
 689         /* LINTED */
 690         while (NULL != (p = strchr(pp, '%'))) {
 691                 bufncat(h, pp, (size_t)(p - pp));
 692                 switch (*(p + 1)) {
 693                 case('S'):
 694                         bufcat(h, sec ? sec : "1");
 695                         break;
 696                 case('N'):
 697                         buffmt(h, name);
 698                         break;
 699                 default:
 700                         bufncat(h, p, 2);
 701                         break;
 702                 }
 703                 pp = p + 2;
 704         }
 705         if (pp)
 706                 bufcat(h, pp);
 707 }
 708
 709
 710 void
 711 bufcat_su(struct html *h, const char *p, const struct roffsu *su)
 712 {
 713         double           v;
 714         const char      *u;
 715
 716         v = su->scale;
 717
 718         switch (su->unit) {
 719         case (SCALE_CM):
 720                 u = "cm";
 721                 break;
 722         case (SCALE_IN):
 723                 u = "in";
 724                 break;
 725         case (SCALE_PC):
 726                 u = "pc";
 727                 break;
 728         case (SCALE_PT):
 729                 u = "pt";
 730                 break;
 731         case (SCALE_EM):
 732                 u = "em";
 733                 break;
 734         case (SCALE_MM):
 735                 if (0 == (v /= 100))
 736                         v = 1;
 737                 u = "em";
 738                 break;
 739         case (SCALE_EN):
 740                 u = "ex";
 741                 break;
 742         case (SCALE_BU):
 743                 u = "ex";
 744                 break;
 745         case (SCALE_VS):
 746                 u = "em";
 747                 break;
 748         default:
 749                 u = "ex";
 750                 break;
 751         }
 752
 753         /*
 754          * XXX: the CSS spec isn't clear as to which types accept
 755          * integer or real numbers, so we just make them all decimals.
 756          */
 757         buffmt(h, "%s: %.2f%s;", p, v, u);
 758 }
 759
 760
 761 void
 762 html_idcat(char *dst, const char *src, int sz)
 763 {
 764         int              ssz;
 765
 766         assert(sz);
 767
 768         /* Cf. <http://www.w3.org/TR/html4/types.html#h-6.2>. */
 769
 770         for ( ; *dst != '\0' && sz; dst++, sz--)
 771                 /* Jump to end. */ ;
 772
 773         assert(sz > 2);
 774
 775         /* We can't start with a number (bah). */
 776
 777         *dst++ = 'x';
 778         *dst = '\0';
 779         sz--;
 780
 781         for ( ; *src != '\0' && sz > 1; src++) {
 782                 ssz = snprintf(dst, (size_t)sz, "%.2x", *src);
 783                 sz -= ssz;
 784                 dst += ssz;
 785         }
 786 }