html.c

   1 /*      $Id: html.c,v 1.73 2009/10/30 18:50:11 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 #include <sys/types.h>
  18
  19 #include <assert.h>
  20 #include <ctype.h>
  21 #include <err.h>
  22 #include <stdio.h>
  23 #include <stdarg.h>
  24 #include <stdint.h>
  25 #include <stdlib.h>
  26 #include <string.h>
  27 #include <unistd.h>
  28
  29 #include "out.h"
  30 #include "chars.h"
  31 #include "html.h"
  32 #include "main.h"
  33
  34 #define UNCONST(a)      ((void *)(uintptr_t)(const void *)(a))
  35
  36 #define DOCTYPE         "-//W3C//DTD HTML 4.01//EN"
  37 #define DTD             "http://www.w3.org/TR/html4/strict.dtd"
  38
  39 struct  htmldata {
  40         const char       *name;
  41         int               flags;
  42 #define HTML_CLRLINE     (1 << 0)
  43 #define HTML_NOSTACK     (1 << 1)
  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}, /* 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         {"p",           HTML_CLRLINE}, /* TAG_P */
  56         {"span",        0}, /* TAG_SPAN */
  57         {"link",        HTML_CLRLINE | HTML_NOSTACK}, /* TAG_LINK */
  58         {"br",          HTML_CLRLINE | HTML_NOSTACK}, /* TAG_LINK */
  59         {"a",           0}, /* TAG_A */
  60         {"table",       HTML_CLRLINE}, /* TAG_TABLE */
  61         {"col",         HTML_CLRLINE | HTML_NOSTACK}, /* 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         {"base",        HTML_CLRLINE | HTML_NOSTACK}, /* TAG_BASE */
  68 };
  69
  70 static  const char       *const htmlattrs[ATTR_MAX] = {
  71         "http-equiv",
  72         "content",
  73         "name",
  74         "rel",
  75         "href",
  76         "type",
  77         "media",
  78         "class",
  79         "style",
  80         "width",
  81         "valign",
  82         "target",
  83         "id",
  84         "summary",
  85 };
  86
  87 #ifdef __linux__
  88 extern  int               getsubopt(char **, char * const *, char **);
  89 #endif
  90
  91 void *
  92 html_alloc(char *outopts)
  93 {
  94         struct html     *h;
  95         const char      *toks[4];
  96         char            *v;
  97
  98         toks[0] = "style";
  99         toks[1] = "man";
 100         toks[2] = "includes";
 101         toks[3] = NULL;
 102
 103         h = calloc(1, sizeof(struct html));
 104         if (NULL == h) {
 105                 fprintf(stderr, "memory exhausted\n");
 106                 exit(EXIT_FAILURE);
 107         }
 108
 109         h->tags.head = NULL;
 110         h->ords.head = NULL;
 111         h->symtab = chars_init(CHARS_HTML);
 112
 113         while (outopts && *outopts)
 114                 switch (getsubopt(&outopts, UNCONST(toks), &v)) {
 115                 case (0):
 116                         h->style = v;
 117                         break;
 118                 case (1):
 119                         h->base_man = v;
 120                         break;
 121                 case (2):
 122                         h->base_includes = v;
 123                         break;
 124                 default:
 125                         break;
 126                 }
 127
 128         return(h);
 129 }
 130
 131
 132 void
 133 html_free(void *p)
 134 {
 135         struct tag      *tag;
 136         struct ord      *ord;
 137         struct html     *h;
 138
 139         h = (struct html *)p;
 140
 141         while ((ord = h->ords.head) != NULL) {
 142                 h->ords.head = ord->next;
 143                 free(ord);
 144         }
 145
 146         while ((tag = h->tags.head) != NULL) {
 147                 h->tags.head = tag->next;
 148                 free(tag);
 149         }
 150
 151         if (h->symtab)
 152                 chars_free(h->symtab);
 153
 154         free(h);
 155 }
 156
 157
 158 void
 159 print_gen_head(struct html *h)
 160 {
 161         struct htmlpair  tag[4];
 162
 163         tag[0].key = ATTR_HTTPEQUIV;
 164         tag[0].val = "Content-Type";
 165         tag[1].key = ATTR_CONTENT;
 166         tag[1].val = "text/html; charset=utf-8";
 167         print_otag(h, TAG_META, 2, tag);
 168
 169         tag[0].key = ATTR_NAME;
 170         tag[0].val = "resource-type";
 171         tag[1].key = ATTR_CONTENT;
 172         tag[1].val = "document";
 173         print_otag(h, TAG_META, 2, tag);
 174
 175         if (h->style) {
 176                 tag[0].key = ATTR_REL;
 177                 tag[0].val = "stylesheet";
 178                 tag[1].key = ATTR_HREF;
 179                 tag[1].val = h->style;
 180                 tag[2].key = ATTR_TYPE;
 181                 tag[2].val = "text/css";
 182                 tag[3].key = ATTR_MEDIA;
 183                 tag[3].val = "all";
 184                 print_otag(h, TAG_LINK, 4, tag);
 185         }
 186 }
 187
 188
 189 static void
 190 print_spec(struct html *h, const char *p, int len)
 191 {
 192         const char      *rhs;
 193         int              i;
 194         size_t           sz;
 195
 196         rhs = chars_a2ascii(h->symtab, p, (size_t)len, &sz);
 197
 198         if (NULL == rhs)
 199                 return;
 200         for (i = 0; i < (int)sz; i++)
 201                 putchar(rhs[i]);
 202 }
 203
 204
 205 static void
 206 print_res(struct html *h, const char *p, int len)
 207 {
 208         const char      *rhs;
 209         int              i;
 210         size_t           sz;
 211
 212         rhs = chars_a2res(h->symtab, p, (size_t)len, &sz);
 213
 214         if (NULL == rhs)
 215                 return;
 216         for (i = 0; i < (int)sz; i++)
 217                 putchar(rhs[i]);
 218 }
 219
 220
 221 static void
 222 print_escape(struct html *h, const char **p)
 223 {
 224         int              j, type;
 225         const char      *wp;
 226
 227         wp = *p;
 228         type = 1;
 229
 230         if (0 == *(++wp)) {
 231                 *p = wp;
 232                 return;
 233         }
 234
 235         if ('(' == *wp) {
 236                 wp++;
 237                 if (0 == *wp || 0 == *(wp + 1)) {
 238                         *p = 0 == *wp ? wp : wp + 1;
 239                         return;
 240                 }
 241
 242                 print_spec(h, wp, 2);
 243                 *p = ++wp;
 244                 return;
 245
 246         } else if ('*' == *wp) {
 247                 if (0 == *(++wp)) {
 248                         *p = wp;
 249                         return;
 250                 }
 251
 252                 switch (*wp) {
 253                 case ('('):
 254                         wp++;
 255                         if (0 == *wp || 0 == *(wp + 1)) {
 256                                 *p = 0 == *wp ? wp : wp + 1;
 257                                 return;
 258                         }
 259
 260                         print_res(h, wp, 2);
 261                         *p = ++wp;
 262                         return;
 263                 case ('['):
 264                         type = 0;
 265                         break;
 266                 default:
 267                         print_res(h, wp, 1);
 268                         *p = wp;
 269                         return;
 270                 }
 271
 272         } else if ('f' == *wp) {
 273                 if (0 == *(++wp)) {
 274                         *p = wp;
 275                         return;
 276                 }
 277
 278                 switch (*wp) {
 279                 case ('B'):
 280                         /* TODO */
 281                         break;
 282                 case ('I'):
 283                         /* TODO */
 284                         break;
 285                 case ('P'):
 286                         /* FALLTHROUGH */
 287                 case ('R'):
 288                         /* TODO */
 289                         break;
 290                 default:
 291                         break;
 292                 }
 293
 294                 *p = wp;
 295                 return;
 296
 297         } else if ('[' != *wp) {
 298                 print_spec(h, wp, 1);
 299                 *p = wp;
 300                 return;
 301         }
 302
 303         wp++;
 304         for (j = 0; *wp && ']' != *wp; wp++, j++)
 305                 /* Loop... */ ;
 306
 307         if (0 == *wp) {
 308                 *p = wp;
 309                 return;
 310         }
 311
 312         if (type)
 313                 print_spec(h, wp - j, j);
 314         else
 315                 print_res(h, wp - j, j);
 316
 317         *p = wp;
 318 }
 319
 320
 321 static void
 322 print_encode(struct html *h, const char *p)
 323 {
 324
 325         for (; *p; p++) {
 326                 if ('\\' == *p) {
 327                         print_escape(h, &p);
 328                         continue;
 329                 }
 330                 switch (*p) {
 331                 case ('<'):
 332                         printf("&lt;");
 333                         break;
 334                 case ('>'):
 335                         printf("&gt;");
 336                         break;
 337                 case ('&'):
 338                         printf("&amp;");
 339                         break;
 340                 default:
 341                         putchar(*p);
 342                         break;
 343                 }
 344         }
 345 }
 346
 347
 348 struct tag *
 349 print_otag(struct html *h, enum htmltag tag,
 350                 int sz, const struct htmlpair *p)
 351 {
 352         int              i;
 353         struct tag      *t;
 354
 355         if ( ! (HTML_NOSTACK & htmltags[tag].flags)) {
 356                 t = malloc(sizeof(struct tag));
 357                 if (NULL == t) {
 358                         fprintf(stderr, "memory exhausted\n");
 359                         exit(EXIT_FAILURE);
 360                 }
 361                 t->tag = tag;
 362                 t->next = h->tags.head;
 363                 h->tags.head = t;
 364         } else
 365                 t = NULL;
 366
 367         if ( ! (HTML_NOSPACE & h->flags))
 368                 if ( ! (HTML_CLRLINE & htmltags[tag].flags))
 369                         printf(" ");
 370
 371         printf("<%s", htmltags[tag].name);
 372         for (i = 0; i < sz; i++) {
 373                 printf(" %s=\"", htmlattrs[p[i].key]);
 374                 assert(p->val);
 375                 print_encode(h, p[i].val);
 376                 printf("\"");
 377         }
 378         printf(">");
 379
 380         h->flags |= HTML_NOSPACE;
 381         if (HTML_CLRLINE & htmltags[tag].flags)
 382                 h->flags |= HTML_NEWLINE;
 383         else
 384                 h->flags &= ~HTML_NEWLINE;
 385
 386         return(t);
 387 }
 388
 389
 390 /* ARGSUSED */
 391 static void
 392 print_ctag(struct html *h, enum htmltag tag)
 393 {
 394
 395         printf("</%s>", htmltags[tag].name);
 396         if (HTML_CLRLINE & htmltags[tag].flags) {
 397                 h->flags |= HTML_NOSPACE;
 398                 h->flags |= HTML_NEWLINE;
 399                 printf("\n");
 400         } else
 401                 h->flags &= ~HTML_NEWLINE;
 402 }
 403
 404
 405 /* ARGSUSED */
 406 void
 407 print_gen_doctype(struct html *h)
 408 {
 409
 410         printf("<!DOCTYPE HTML PUBLIC \"%s\" \"%s\">", DOCTYPE, DTD);
 411 }
 412
 413
 414 void
 415 print_text(struct html *h, const char *p)
 416 {
 417
 418         if (*p && 0 == *(p + 1))
 419                 switch (*p) {
 420                 case('.'):
 421                         /* FALLTHROUGH */
 422                 case(','):
 423                         /* FALLTHROUGH */
 424                 case(';'):
 425                         /* FALLTHROUGH */
 426                 case(':'):
 427                         /* FALLTHROUGH */
 428                 case('?'):
 429                         /* FALLTHROUGH */
 430                 case('!'):
 431                         /* FALLTHROUGH */
 432                 case(')'):
 433                         /* FALLTHROUGH */
 434                 case(']'):
 435                         /* FALLTHROUGH */
 436                 case('}'):
 437                         if ( ! (HTML_IGNDELIM & h->flags))
 438                                 h->flags |= HTML_NOSPACE;
 439                         break;
 440                 default:
 441                         break;
 442                 }
 443
 444         if ( ! (h->flags & HTML_NOSPACE))
 445                 printf(" ");
 446
 447         h->flags &= ~HTML_NOSPACE;
 448         h->flags &= ~HTML_NEWLINE;
 449
 450         if (p)
 451                 print_encode(h, p);
 452
 453         if (*p && 0 == *(p + 1))
 454                 switch (*p) {
 455                 case('('):
 456                         /* FALLTHROUGH */
 457                 case('['):
 458                         /* FALLTHROUGH */
 459                 case('{'):
 460                         h->flags |= HTML_NOSPACE;
 461                         break;
 462                 default:
 463                         break;
 464                 }
 465 }
 466
 467
 468 void
 469 print_tagq(struct html *h, const struct tag *until)
 470 {
 471         struct tag      *tag;
 472
 473         while ((tag = h->tags.head) != NULL) {
 474                 print_ctag(h, tag->tag);
 475                 h->tags.head = tag->next;
 476                 free(tag);
 477                 if (until && tag == until)
 478                         return;
 479         }
 480 }
 481
 482
 483 void
 484 print_stagq(struct html *h, const struct tag *suntil)
 485 {
 486         struct tag      *tag;
 487
 488         while ((tag = h->tags.head) != NULL) {
 489                 if (suntil && tag == suntil)
 490                         return;
 491                 print_ctag(h, tag->tag);
 492                 h->tags.head = tag->next;
 493                 free(tag);
 494         }
 495 }
 496
 497
 498 void
 499 bufinit(struct html *h)
 500 {
 501
 502         h->buf[0] = '\0';
 503         h->buflen = 0;
 504 }
 505
 506
 507 void
 508 bufcat_style(struct html *h, const char *key, const char *val)
 509 {
 510
 511         bufcat(h, key);
 512         bufncat(h, ":", 1);
 513         bufcat(h, val);
 514         bufncat(h, ";", 1);
 515 }
 516
 517
 518 void
 519 bufcat(struct html *h, const char *p)
 520 {
 521
 522         bufncat(h, p, strlen(p));
 523 }
 524
 525
 526 void
 527 buffmt(struct html *h, const char *fmt, ...)
 528 {
 529         va_list          ap;
 530
 531         va_start(ap, fmt);
 532         (void)vsnprintf(h->buf + (int)h->buflen,
 533                         BUFSIZ - h->buflen - 1, fmt, ap);
 534         va_end(ap);
 535         h->buflen = strlen(h->buf);
 536 }
 537
 538
 539 void
 540 bufncat(struct html *h, const char *p, size_t sz)
 541 {
 542
 543         if (h->buflen + sz > BUFSIZ - 1)
 544                 sz = BUFSIZ - 1 - h->buflen;
 545
 546         (void)strncat(h->buf, p, sz);
 547         h->buflen += sz;
 548 }
 549
 550
 551 void
 552 buffmt_includes(struct html *h, const char *name)
 553 {
 554         const char      *p, *pp;
 555
 556         pp = h->base_includes;
 557
 558         while (NULL != (p = strchr(pp, '%'))) {
 559                 bufncat(h, pp, (size_t)(p - pp));
 560                 switch (*(p + 1)) {
 561                 case('I'):
 562                         bufcat(h, name);
 563                         break;
 564                 default:
 565                         bufncat(h, p, 2);
 566                         break;
 567                 }
 568                 pp = p + 2;
 569         }
 570         if (pp)
 571                 bufcat(h, pp);
 572 }
 573
 574
 575 void
 576 buffmt_man(struct html *h,
 577                 const char *name, const char *sec)
 578 {
 579         const char      *p, *pp;
 580
 581         pp = h->base_man;
 582
 583         /* LINTED */
 584         while (NULL != (p = strchr(pp, '%'))) {
 585                 bufncat(h, pp, (size_t)(p - pp));
 586                 switch (*(p + 1)) {
 587                 case('S'):
 588                         bufcat(h, sec ? sec : "1");
 589                         break;
 590                 case('N'):
 591                         buffmt(h, name);
 592                         break;
 593                 default:
 594                         bufncat(h, p, 2);
 595                         break;
 596                 }
 597                 pp = p + 2;
 598         }
 599         if (pp)
 600                 bufcat(h, pp);
 601 }
 602
 603
 604 void
 605 bufcat_su(struct html *h, const char *p, const struct roffsu *su)
 606 {
 607         double           v;
 608         const char      *u;
 609
 610         v = su->scale;
 611
 612         switch (su->unit) {
 613         case (SCALE_CM):
 614                 u = "cm";
 615                 break;
 616         case (SCALE_IN):
 617                 u = "in";
 618                 break;
 619         case (SCALE_PC):
 620                 u = "pc";
 621                 break;
 622         case (SCALE_PT):
 623                 u = "pt";
 624                 break;
 625         case (SCALE_EM):
 626                 u = "em";
 627                 break;
 628         case (SCALE_MM):
 629                 if (0 == (v /= 100))
 630                         v = 1;
 631                 u = "em";
 632                 break;
 633         case (SCALE_EN):
 634                 u = "ex";
 635                 break;
 636         case (SCALE_BU):
 637                 u = "ex";
 638                 break;
 639         case (SCALE_VS):
 640                 u = "em";
 641                 break;
 642         default:
 643                 u = "ex";
 644                 break;
 645         }
 646
 647         if (su->pt)
 648                 buffmt(h, "%s: %f%s;", p, v, u);
 649         else
 650                 /* LINTED */
 651                 buffmt(h, "%s: %d%s;", p, (int)v, u);
 652 }
 653
 654
 655 void
 656 html_idcat(char *dst, const char *src, int sz)
 657 {
 658         int              ssz;
 659
 660         assert(sz);
 661
 662         /* Cf. <http://www.w3.org/TR/html4/types.html#h-6.2>. */
 663
 664         for ( ; *dst != '\0' && sz; dst++, sz--)
 665                 /* Jump to end. */ ;
 666
 667         assert(sz > 2);
 668
 669         /* We can't start with a number (bah). */
 670
 671         *dst++ = 'x';
 672         *dst = '\0';
 673         sz--;
 674
 675         for ( ; *src != '\0' && sz > 1; src++) {
 676                 ssz = snprintf(dst, (size_t)sz, "%.2x", *src);
 677                 sz -= ssz;
 678                 dst += ssz;
 679         }
 680 }