html.c

   1 /*      $Id: html.c,v 1.75 2009/10/31 06:10:57 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 <stdio.h>
  22 #include <stdarg.h>
  23 #include <stdint.h>
  24 #include <stdlib.h>
  25 #include <string.h>
  26 #include <unistd.h>
  27
  28 #include "out.h"
  29 #include "chars.h"
  30 #include "html.h"
  31 #include "main.h"
  32
  33 #define UNCONST(a)      ((void *)(uintptr_t)(const void *)(a))
  34
  35 #define DOCTYPE         "-//W3C//DTD HTML 4.01//EN"
  36 #define DTD             "http://www.w3.org/TR/html4/strict.dtd"
  37
  38 struct  htmldata {
  39         const char       *name;
  40         int               flags;
  41 #define HTML_CLRLINE     (1 << 0)
  42 #define HTML_NOSTACK     (1 << 1)
  43 };
  44
  45 static  const struct htmldata htmltags[TAG_MAX] = {
  46         {"html",        HTML_CLRLINE}, /* TAG_HTML */
  47         {"head",        HTML_CLRLINE}, /* TAG_HEAD */
  48         {"body",        HTML_CLRLINE}, /* TAG_BODY */
  49         {"meta",        HTML_CLRLINE | HTML_NOSTACK}, /* TAG_META */
  50         {"title",       HTML_CLRLINE}, /* TAG_TITLE */
  51         {"div",         HTML_CLRLINE}, /* TAG_DIV */
  52         {"h1",          0}, /* TAG_H1 */
  53         {"h2",          0}, /* TAG_H2 */
  54         {"p",           HTML_CLRLINE}, /* TAG_P */
  55         {"span",        0}, /* TAG_SPAN */
  56         {"link",        HTML_CLRLINE | HTML_NOSTACK}, /* TAG_LINK */
  57         {"br",          HTML_CLRLINE | HTML_NOSTACK}, /* TAG_LINK */
  58         {"a",           0}, /* TAG_A */
  59         {"table",       HTML_CLRLINE}, /* TAG_TABLE */
  60         {"col",         HTML_CLRLINE | HTML_NOSTACK}, /* 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         {"base",        HTML_CLRLINE | HTML_NOSTACK}, /* TAG_BASE */
  67 };
  68
  69 static  const char       *const htmlattrs[ATTR_MAX] = {
  70         "http-equiv",
  71         "content",
  72         "name",
  73         "rel",
  74         "href",
  75         "type",
  76         "media",
  77         "class",
  78         "style",
  79         "width",
  80         "valign",
  81         "target",
  82         "id",
  83         "summary",
  84 };
  85
  86 #ifdef __linux__
  87 extern  int               getsubopt(char **, char * const *, char **);
  88 #endif
  89
  90 void *
  91 html_alloc(char *outopts)
  92 {
  93         struct html     *h;
  94         const char      *toks[4];
  95         char            *v;
  96
  97         toks[0] = "style";
  98         toks[1] = "man";
  99         toks[2] = "includes";
 100         toks[3] = NULL;
 101
 102         h = calloc(1, sizeof(struct html));
 103         if (NULL == h) {
 104                 perror(NULL);
 105                 exit(EXIT_FAILURE);
 106         }
 107
 108         h->tags.head = NULL;
 109         h->ords.head = NULL;
 110         h->symtab = chars_init(CHARS_HTML);
 111
 112         while (outopts && *outopts)
 113                 switch (getsubopt(&outopts, UNCONST(toks), &v)) {
 114                 case (0):
 115                         h->style = v;
 116                         break;
 117                 case (1):
 118                         h->base_man = v;
 119                         break;
 120                 case (2):
 121                         h->base_includes = v;
 122                         break;
 123                 default:
 124                         break;
 125                 }
 126
 127         return(h);
 128 }
 129
 130
 131 void
 132 html_free(void *p)
 133 {
 134         struct tag      *tag;
 135         struct ord      *ord;
 136         struct html     *h;
 137
 138         h = (struct html *)p;
 139
 140         while ((ord = h->ords.head) != NULL) {
 141                 h->ords.head = ord->next;
 142                 free(ord);
 143         }
 144
 145         while ((tag = h->tags.head) != NULL) {
 146                 h->tags.head = tag->next;
 147                 free(tag);
 148         }
 149
 150         if (h->symtab)
 151                 chars_free(h->symtab);
 152
 153         free(h);
 154 }
 155
 156
 157 void
 158 print_gen_head(struct html *h)
 159 {
 160         struct htmlpair  tag[4];
 161
 162         tag[0].key = ATTR_HTTPEQUIV;
 163         tag[0].val = "Content-Type";
 164         tag[1].key = ATTR_CONTENT;
 165         tag[1].val = "text/html; charset=utf-8";
 166         print_otag(h, TAG_META, 2, tag);
 167
 168         tag[0].key = ATTR_NAME;
 169         tag[0].val = "resource-type";
 170         tag[1].key = ATTR_CONTENT;
 171         tag[1].val = "document";
 172         print_otag(h, TAG_META, 2, tag);
 173
 174         if (h->style) {
 175                 tag[0].key = ATTR_REL;
 176                 tag[0].val = "stylesheet";
 177                 tag[1].key = ATTR_HREF;
 178                 tag[1].val = h->style;
 179                 tag[2].key = ATTR_TYPE;
 180                 tag[2].val = "text/css";
 181                 tag[3].key = ATTR_MEDIA;
 182                 tag[3].val = "all";
 183                 print_otag(h, TAG_LINK, 4, tag);
 184         }
 185 }
 186
 187
 188 static void
 189 print_spec(struct html *h, const char *p, int len)
 190 {
 191         const char      *rhs;
 192         int              i;
 193         size_t           sz;
 194
 195         rhs = chars_a2ascii(h->symtab, p, (size_t)len, &sz);
 196
 197         if (NULL == rhs)
 198                 return;
 199         for (i = 0; i < (int)sz; i++)
 200                 putchar(rhs[i]);
 201 }
 202
 203
 204 static void
 205 print_res(struct html *h, const char *p, int len)
 206 {
 207         const char      *rhs;
 208         int              i;
 209         size_t           sz;
 210
 211         rhs = chars_a2res(h->symtab, p, (size_t)len, &sz);
 212
 213         if (NULL == rhs)
 214                 return;
 215         for (i = 0; i < (int)sz; i++)
 216                 putchar(rhs[i]);
 217 }
 218
 219
 220 static void
 221 print_escape(struct html *h, const char **p)
 222 {
 223         int              j, type;
 224         const char      *wp;
 225
 226         wp = *p;
 227         type = 1;
 228
 229         if (0 == *(++wp)) {
 230                 *p = wp;
 231                 return;
 232         }
 233
 234         if ('(' == *wp) {
 235                 wp++;
 236                 if (0 == *wp || 0 == *(wp + 1)) {
 237                         *p = 0 == *wp ? wp : wp + 1;
 238                         return;
 239                 }
 240
 241                 print_spec(h, wp, 2);
 242                 *p = ++wp;
 243                 return;
 244
 245         } else if ('*' == *wp) {
 246                 if (0 == *(++wp)) {
 247                         *p = wp;
 248                         return;
 249                 }
 250
 251                 switch (*wp) {
 252                 case ('('):
 253                         wp++;
 254                         if (0 == *wp || 0 == *(wp + 1)) {
 255                                 *p = 0 == *wp ? wp : wp + 1;
 256                                 return;
 257                         }
 258
 259                         print_res(h, wp, 2);
 260                         *p = ++wp;
 261                         return;
 262                 case ('['):
 263                         type = 0;
 264                         break;
 265                 default:
 266                         print_res(h, wp, 1);
 267                         *p = wp;
 268                         return;
 269                 }
 270
 271         } else if ('f' == *wp) {
 272                 if (0 == *(++wp)) {
 273                         *p = wp;
 274                         return;
 275                 }
 276
 277                 switch (*wp) {
 278                 case ('B'):
 279                         /* TODO */
 280                         break;
 281                 case ('I'):
 282                         /* TODO */
 283                         break;
 284                 case ('P'):
 285                         /* FALLTHROUGH */
 286                 case ('R'):
 287                         /* TODO */
 288                         break;
 289                 default:
 290                         break;
 291                 }
 292
 293                 *p = wp;
 294                 return;
 295
 296         } else if ('[' != *wp) {
 297                 print_spec(h, wp, 1);
 298                 *p = wp;
 299                 return;
 300         }
 301
 302         wp++;
 303         for (j = 0; *wp && ']' != *wp; wp++, j++)
 304                 /* Loop... */ ;
 305
 306         if (0 == *wp) {
 307                 *p = wp;
 308                 return;
 309         }
 310
 311         if (type)
 312                 print_spec(h, wp - j, j);
 313         else
 314                 print_res(h, wp - j, j);
 315
 316         *p = wp;
 317 }
 318
 319
 320 static void
 321 print_encode(struct html *h, const char *p)
 322 {
 323
 324         for (; *p; p++) {
 325                 if ('\\' == *p) {
 326                         print_escape(h, &p);
 327                         continue;
 328                 }
 329                 switch (*p) {
 330                 case ('<'):
 331                         printf("&lt;");
 332                         break;
 333                 case ('>'):
 334                         printf("&gt;");
 335                         break;
 336                 case ('&'):
 337                         printf("&amp;");
 338                         break;
 339                 default:
 340                         putchar(*p);
 341                         break;
 342                 }
 343         }
 344 }
 345
 346
 347 struct tag *
 348 print_otag(struct html *h, enum htmltag tag,
 349                 int sz, const struct htmlpair *p)
 350 {
 351         int              i;
 352         struct tag      *t;
 353
 354         if ( ! (HTML_NOSTACK & htmltags[tag].flags)) {
 355                 t = malloc(sizeof(struct tag));
 356                 if (NULL == t) {
 357                         perror(NULL);
 358                         exit(EXIT_FAILURE);
 359                 }
 360                 t->tag = tag;
 361                 t->next = h->tags.head;
 362                 h->tags.head = t;
 363         } else
 364                 t = NULL;
 365
 366         if ( ! (HTML_NOSPACE & h->flags))
 367                 if ( ! (HTML_CLRLINE & htmltags[tag].flags))
 368                         printf(" ");
 369
 370         printf("<%s", htmltags[tag].name);
 371         for (i = 0; i < sz; i++) {
 372                 printf(" %s=\"", htmlattrs[p[i].key]);
 373                 assert(p->val);
 374                 print_encode(h, p[i].val);
 375                 printf("\"");
 376         }
 377         printf(">");
 378
 379         h->flags |= HTML_NOSPACE;
 380         if (HTML_CLRLINE & htmltags[tag].flags)
 381                 h->flags |= HTML_NEWLINE;
 382         else
 383                 h->flags &= ~HTML_NEWLINE;
 384
 385         return(t);
 386 }
 387
 388
 389 /* ARGSUSED */
 390 static void
 391 print_ctag(struct html *h, enum htmltag tag)
 392 {
 393
 394         printf("</%s>", htmltags[tag].name);
 395         if (HTML_CLRLINE & htmltags[tag].flags) {
 396                 h->flags |= HTML_NOSPACE;
 397                 h->flags |= HTML_NEWLINE;
 398                 printf("\n");
 399         } else
 400                 h->flags &= ~HTML_NEWLINE;
 401 }
 402
 403
 404 /* ARGSUSED */
 405 void
 406 print_gen_doctype(struct html *h)
 407 {
 408
 409         printf("<!DOCTYPE HTML PUBLIC \"%s\" \"%s\">", DOCTYPE, DTD);
 410 }
 411
 412
 413 void
 414 print_text(struct html *h, const char *p)
 415 {
 416
 417         if (*p && 0 == *(p + 1))
 418                 switch (*p) {
 419                 case('.'):
 420                         /* FALLTHROUGH */
 421                 case(','):
 422                         /* FALLTHROUGH */
 423                 case(';'):
 424                         /* FALLTHROUGH */
 425                 case(':'):
 426                         /* FALLTHROUGH */
 427                 case('?'):
 428                         /* FALLTHROUGH */
 429                 case('!'):
 430                         /* FALLTHROUGH */
 431                 case(')'):
 432                         /* FALLTHROUGH */
 433                 case(']'):
 434                         /* FALLTHROUGH */
 435                 case('}'):
 436                         if ( ! (HTML_IGNDELIM & h->flags))
 437                                 h->flags |= HTML_NOSPACE;
 438                         break;
 439                 default:
 440                         break;
 441                 }
 442
 443         if ( ! (h->flags & HTML_NOSPACE))
 444                 printf(" ");
 445
 446         h->flags &= ~HTML_NOSPACE;
 447         h->flags &= ~HTML_NEWLINE;
 448
 449         if (p)
 450                 print_encode(h, p);
 451
 452         if (*p && 0 == *(p + 1))
 453                 switch (*p) {
 454                 case('('):
 455                         /* FALLTHROUGH */
 456                 case('['):
 457                         /* FALLTHROUGH */
 458                 case('{'):
 459                         h->flags |= HTML_NOSPACE;
 460                         break;
 461                 default:
 462                         break;
 463                 }
 464 }
 465
 466
 467 void
 468 print_tagq(struct html *h, const struct tag *until)
 469 {
 470         struct tag      *tag;
 471
 472         while ((tag = h->tags.head) != NULL) {
 473                 print_ctag(h, tag->tag);
 474                 h->tags.head = tag->next;
 475                 free(tag);
 476                 if (until && tag == until)
 477                         return;
 478         }
 479 }
 480
 481
 482 void
 483 print_stagq(struct html *h, const struct tag *suntil)
 484 {
 485         struct tag      *tag;
 486
 487         while ((tag = h->tags.head) != NULL) {
 488                 if (suntil && tag == suntil)
 489                         return;
 490                 print_ctag(h, tag->tag);
 491                 h->tags.head = tag->next;
 492                 free(tag);
 493         }
 494 }
 495
 496
 497 void
 498 bufinit(struct html *h)
 499 {
 500
 501         h->buf[0] = '\0';
 502         h->buflen = 0;
 503 }
 504
 505
 506 void
 507 bufcat_style(struct html *h, const char *key, const char *val)
 508 {
 509
 510         bufcat(h, key);
 511         bufncat(h, ":", 1);
 512         bufcat(h, val);
 513         bufncat(h, ";", 1);
 514 }
 515
 516
 517 void
 518 bufcat(struct html *h, const char *p)
 519 {
 520
 521         bufncat(h, p, strlen(p));
 522 }
 523
 524
 525 void
 526 buffmt(struct html *h, const char *fmt, ...)
 527 {
 528         va_list          ap;
 529
 530         va_start(ap, fmt);
 531         (void)vsnprintf(h->buf + (int)h->buflen,
 532                         BUFSIZ - h->buflen - 1, fmt, ap);
 533         va_end(ap);
 534         h->buflen = strlen(h->buf);
 535 }
 536
 537
 538 void
 539 bufncat(struct html *h, const char *p, size_t sz)
 540 {
 541
 542         if (h->buflen + sz > BUFSIZ - 1)
 543                 sz = BUFSIZ - 1 - h->buflen;
 544
 545         (void)strncat(h->buf, p, sz);
 546         h->buflen += sz;
 547 }
 548
 549
 550 void
 551 buffmt_includes(struct html *h, const char *name)
 552 {
 553         const char      *p, *pp;
 554
 555         pp = h->base_includes;
 556
 557         while (NULL != (p = strchr(pp, '%'))) {
 558                 bufncat(h, pp, (size_t)(p - pp));
 559                 switch (*(p + 1)) {
 560                 case('I'):
 561                         bufcat(h, name);
 562                         break;
 563                 default:
 564                         bufncat(h, p, 2);
 565                         break;
 566                 }
 567                 pp = p + 2;
 568         }
 569         if (pp)
 570                 bufcat(h, pp);
 571 }
 572
 573
 574 void
 575 buffmt_man(struct html *h,
 576                 const char *name, const char *sec)
 577 {
 578         const char      *p, *pp;
 579
 580         pp = h->base_man;
 581
 582         /* LINTED */
 583         while (NULL != (p = strchr(pp, '%'))) {
 584                 bufncat(h, pp, (size_t)(p - pp));
 585                 switch (*(p + 1)) {
 586                 case('S'):
 587                         bufcat(h, sec ? sec : "1");
 588                         break;
 589                 case('N'):
 590                         buffmt(h, name);
 591                         break;
 592                 default:
 593                         bufncat(h, p, 2);
 594                         break;
 595                 }
 596                 pp = p + 2;
 597         }
 598         if (pp)
 599                 bufcat(h, pp);
 600 }
 601
 602
 603 void
 604 bufcat_su(struct html *h, const char *p, const struct roffsu *su)
 605 {
 606         double           v;
 607         const char      *u;
 608
 609         v = su->scale;
 610
 611         switch (su->unit) {
 612         case (SCALE_CM):
 613                 u = "cm";
 614                 break;
 615         case (SCALE_IN):
 616                 u = "in";
 617                 break;
 618         case (SCALE_PC):
 619                 u = "pc";
 620                 break;
 621         case (SCALE_PT):
 622                 u = "pt";
 623                 break;
 624         case (SCALE_EM):
 625                 u = "em";
 626                 break;
 627         case (SCALE_MM):
 628                 if (0 == (v /= 100))
 629                         v = 1;
 630                 u = "em";
 631                 break;
 632         case (SCALE_EN):
 633                 u = "ex";
 634                 break;
 635         case (SCALE_BU):
 636                 u = "ex";
 637                 break;
 638         case (SCALE_VS):
 639                 u = "em";
 640                 break;
 641         default:
 642                 u = "ex";
 643                 break;
 644         }
 645
 646         if (su->pt)
 647                 buffmt(h, "%s: %f%s;", p, v, u);
 648         else
 649                 /* LINTED */
 650                 buffmt(h, "%s: %d%s;", p, (int)v, u);
 651 }
 652
 653
 654 void
 655 html_idcat(char *dst, const char *src, int sz)
 656 {
 657         int              ssz;
 658
 659         assert(sz);
 660
 661         /* Cf. <http://www.w3.org/TR/html4/types.html#h-6.2>. */
 662
 663         for ( ; *dst != '\0' && sz; dst++, sz--)
 664                 /* Jump to end. */ ;
 665
 666         assert(sz > 2);
 667
 668         /* We can't start with a number (bah). */
 669
 670         *dst++ = 'x';
 671         *dst = '\0';
 672         sz--;
 673
 674         for ( ; *src != '\0' && sz > 1; src++) {
 675                 ssz = snprintf(dst, (size_t)sz, "%.2x", *src);
 676                 sz -= ssz;
 677                 dst += ssz;
 678         }
 679 }