usr.bin/mandoc/html.c

   1 /*      $Id: html.c,v 1.80 2009/11/02 06:22:44 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 <stdarg.h>
  22 #include <stdio.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         size_t           sz;
 193
 194         rhs = chars_a2ascii(h->symtab, p, (size_t)len, &sz);
 195
 196         if (NULL == rhs)
 197                 return;
 198         fwrite(rhs, 1, sz, stdout);
 199 }
 200
 201
 202 static void
 203 print_res(struct html *h, const char *p, int len)
 204 {
 205         const char      *rhs;
 206         size_t           sz;
 207
 208         rhs = chars_a2res(h->symtab, p, (size_t)len, &sz);
 209
 210         if (NULL == rhs)
 211                 return;
 212         fwrite(rhs, 1, sz, stdout);
 213 }
 214
 215
 216 static void
 217 print_escape(struct html *h, const char **p)
 218 {
 219         int              j, type;
 220         const char      *wp;
 221
 222         wp = *p;
 223         type = 1;
 224
 225         if (0 == *(++wp)) {
 226                 *p = wp;
 227                 return;
 228         }
 229
 230         if ('(' == *wp) {
 231                 wp++;
 232                 if (0 == *wp || 0 == *(wp + 1)) {
 233                         *p = 0 == *wp ? wp : wp + 1;
 234                         return;
 235                 }
 236
 237                 print_spec(h, wp, 2);
 238                 *p = ++wp;
 239                 return;
 240
 241         } else if ('*' == *wp) {
 242                 if (0 == *(++wp)) {
 243                         *p = wp;
 244                         return;
 245                 }
 246
 247                 switch (*wp) {
 248                 case ('('):
 249                         wp++;
 250                         if (0 == *wp || 0 == *(wp + 1)) {
 251                                 *p = 0 == *wp ? wp : wp + 1;
 252                                 return;
 253                         }
 254
 255                         print_res(h, wp, 2);
 256                         *p = ++wp;
 257                         return;
 258                 case ('['):
 259                         type = 0;
 260                         break;
 261                 default:
 262                         print_res(h, wp, 1);
 263                         *p = wp;
 264                         return;
 265                 }
 266
 267         } else if ('f' == *wp) {
 268                 if (0 == *(++wp)) {
 269                         *p = wp;
 270                         return;
 271                 }
 272
 273                 switch (*wp) {
 274                 case ('B'):
 275                         /* TODO */
 276                         break;
 277                 case ('I'):
 278                         /* TODO */
 279                         break;
 280                 case ('P'):
 281                         /* FALLTHROUGH */
 282                 case ('R'):
 283                         /* TODO */
 284                         break;
 285                 default:
 286                         break;
 287                 }
 288
 289                 *p = wp;
 290                 return;
 291
 292         } else if ('[' != *wp) {
 293                 print_spec(h, wp, 1);
 294                 *p = wp;
 295                 return;
 296         }
 297
 298         wp++;
 299         for (j = 0; *wp && ']' != *wp; wp++, j++)
 300                 /* Loop... */ ;
 301
 302         if (0 == *wp) {
 303                 *p = wp;
 304                 return;
 305         }
 306
 307         if (type)
 308                 print_spec(h, wp - j, j);
 309         else
 310                 print_res(h, wp - j, j);
 311
 312         *p = wp;
 313 }
 314
 315
 316 static void
 317 print_encode(struct html *h, const char *p)
 318 {
 319         size_t           sz;
 320
 321         for (; *p; p++) {
 322                 sz = strcspn(p, "\\<>&");
 323
 324                 fwrite(p, 1, sz, stdout);
 325                 p += /* LINTED */
 326                         sz;
 327
 328                 if ('\\' == *p) {
 329                         print_escape(h, &p);
 330                         continue;
 331                 } else if ('\0' == *p)
 332                         break;
 333
 334                 if ('<' == *p)
 335                         printf("&lt;");
 336                 else if ('>' == *p)
 337                         printf("&gt;");
 338                 else if ('&' == *p)
 339                         printf("&amp;");
 340         }
 341 }
 342
 343
 344 struct tag *
 345 print_otag(struct html *h, enum htmltag tag,
 346                 int sz, const struct htmlpair *p)
 347 {
 348         int              i;
 349         struct tag      *t;
 350
 351         if ( ! (HTML_NOSTACK & htmltags[tag].flags)) {
 352                 t = malloc(sizeof(struct tag));
 353                 if (NULL == t) {
 354                         perror(NULL);
 355                         exit(EXIT_FAILURE);
 356                 }
 357                 t->tag = tag;
 358                 t->next = h->tags.head;
 359                 h->tags.head = t;
 360         } else
 361                 t = NULL;
 362
 363         if ( ! (HTML_NOSPACE & h->flags))
 364                 if ( ! (HTML_CLRLINE & htmltags[tag].flags))
 365                         putchar(' ');
 366
 367         printf("<%s", htmltags[tag].name);
 368         for (i = 0; i < sz; i++) {
 369                 printf(" %s=\"", htmlattrs[p[i].key]);
 370                 assert(p->val);
 371                 print_encode(h, p[i].val);
 372                 putchar('\"');
 373         }
 374         putchar('>');
 375
 376         h->flags |= HTML_NOSPACE;
 377         if (HTML_CLRLINE & htmltags[tag].flags)
 378                 h->flags |= HTML_NEWLINE;
 379         else
 380                 h->flags &= ~HTML_NEWLINE;
 381
 382         return(t);
 383 }
 384
 385
 386 /* ARGSUSED */
 387 static void
 388 print_ctag(struct html *h, enum htmltag tag)
 389 {
 390
 391         printf("</%s>", htmltags[tag].name);
 392         if (HTML_CLRLINE & htmltags[tag].flags) {
 393                 h->flags |= HTML_NOSPACE;
 394                 h->flags |= HTML_NEWLINE;
 395                 putchar('\n');
 396         } else
 397                 h->flags &= ~HTML_NEWLINE;
 398 }
 399
 400
 401 /* ARGSUSED */
 402 void
 403 print_gen_doctype(struct html *h)
 404 {
 405
 406         printf("<!DOCTYPE HTML PUBLIC \"%s\" \"%s\">", DOCTYPE, DTD);
 407 }
 408
 409
 410 void
 411 print_text(struct html *h, const char *p)
 412 {
 413
 414         if (*p && 0 == *(p + 1))
 415                 switch (*p) {
 416                 case('.'):
 417                         /* FALLTHROUGH */
 418                 case(','):
 419                         /* FALLTHROUGH */
 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                         if ( ! (HTML_IGNDELIM & h->flags))
 434                                 h->flags |= HTML_NOSPACE;
 435                         break;
 436                 default:
 437                         break;
 438                 }
 439
 440         if ( ! (h->flags & HTML_NOSPACE))
 441                 putchar(' ');
 442
 443         h->flags &= ~HTML_NOSPACE;
 444         h->flags &= ~HTML_NEWLINE;
 445
 446         if (p)
 447                 print_encode(h, p);
 448
 449         if (*p && 0 == *(p + 1))
 450                 switch (*p) {
 451                 case('('):
 452                         /* FALLTHROUGH */
 453                 case('['):
 454                         /* FALLTHROUGH */
 455                 case('{'):
 456                         h->flags |= HTML_NOSPACE;
 457                         break;
 458                 default:
 459                         break;
 460                 }
 461 }
 462
 463
 464 void
 465 print_tagq(struct html *h, const struct tag *until)
 466 {
 467         struct tag      *tag;
 468
 469         while ((tag = h->tags.head) != NULL) {
 470                 print_ctag(h, tag->tag);
 471                 h->tags.head = tag->next;
 472                 free(tag);
 473                 if (until && tag == until)
 474                         return;
 475         }
 476 }
 477
 478
 479 void
 480 print_stagq(struct html *h, const struct tag *suntil)
 481 {
 482         struct tag      *tag;
 483
 484         while ((tag = h->tags.head) != NULL) {
 485                 if (suntil && tag == suntil)
 486                         return;
 487                 print_ctag(h, tag->tag);
 488                 h->tags.head = tag->next;
 489                 free(tag);
 490         }
 491 }
 492
 493
 494 void
 495 bufinit(struct html *h)
 496 {
 497
 498         h->buf[0] = '\0';
 499         h->buflen = 0;
 500 }
 501
 502
 503 void
 504 bufcat_style(struct html *h, const char *key, const char *val)
 505 {
 506
 507         bufcat(h, key);
 508         bufncat(h, ":", 1);
 509         bufcat(h, val);
 510         bufncat(h, ";", 1);
 511 }
 512
 513
 514 void
 515 bufcat(struct html *h, const char *p)
 516 {
 517
 518         bufncat(h, p, strlen(p));
 519 }
 520
 521
 522 void
 523 buffmt(struct html *h, const char *fmt, ...)
 524 {
 525         va_list          ap;
 526
 527         va_start(ap, fmt);
 528         (void)vsnprintf(h->buf + (int)h->buflen,
 529                         BUFSIZ - h->buflen - 1, fmt, ap);
 530         va_end(ap);
 531         h->buflen = strlen(h->buf);
 532 }
 533
 534
 535 void
 536 bufncat(struct html *h, const char *p, size_t sz)
 537 {
 538
 539         if (h->buflen + sz > BUFSIZ - 1)
 540                 sz = BUFSIZ - 1 - h->buflen;
 541
 542         (void)strncat(h->buf, p, sz);
 543         h->buflen += sz;
 544 }
 545
 546
 547 void
 548 buffmt_includes(struct html *h, const char *name)
 549 {
 550         const char      *p, *pp;
 551
 552         pp = h->base_includes;
 553
 554         while (NULL != (p = strchr(pp, '%'))) {
 555                 bufncat(h, pp, (size_t)(p - pp));
 556                 switch (*(p + 1)) {
 557                 case('I'):
 558                         bufcat(h, name);
 559                         break;
 560                 default:
 561                         bufncat(h, p, 2);
 562                         break;
 563                 }
 564                 pp = p + 2;
 565         }
 566         if (pp)
 567                 bufcat(h, pp);
 568 }
 569
 570
 571 void
 572 buffmt_man(struct html *h,
 573                 const char *name, const char *sec)
 574 {
 575         const char      *p, *pp;
 576
 577         pp = h->base_man;
 578
 579         /* LINTED */
 580         while (NULL != (p = strchr(pp, '%'))) {
 581                 bufncat(h, pp, (size_t)(p - pp));
 582                 switch (*(p + 1)) {
 583                 case('S'):
 584                         bufcat(h, sec ? sec : "1");
 585                         break;
 586                 case('N'):
 587                         buffmt(h, name);
 588                         break;
 589                 default:
 590                         bufncat(h, p, 2);
 591                         break;
 592                 }
 593                 pp = p + 2;
 594         }
 595         if (pp)
 596                 bufcat(h, pp);
 597 }
 598
 599
 600 void
 601 bufcat_su(struct html *h, const char *p, const struct roffsu *su)
 602 {
 603         double           v;
 604         const char      *u;
 605
 606         v = su->scale;
 607
 608         switch (su->unit) {
 609         case (SCALE_CM):
 610                 u = "cm";
 611                 break;
 612         case (SCALE_IN):
 613                 u = "in";
 614                 break;
 615         case (SCALE_PC):
 616                 u = "pc";
 617                 break;
 618         case (SCALE_PT):
 619                 u = "pt";
 620                 break;
 621         case (SCALE_EM):
 622                 u = "em";
 623                 break;
 624         case (SCALE_MM):
 625                 if (0 == (v /= 100))
 626                         v = 1;
 627                 u = "em";
 628                 break;
 629         case (SCALE_EN):
 630                 u = "ex";
 631                 break;
 632         case (SCALE_BU):
 633                 u = "ex";
 634                 break;
 635         case (SCALE_VS):
 636                 u = "em";
 637                 break;
 638         default:
 639                 u = "ex";
 640                 break;
 641         }
 642
 643         if (su->pt)
 644                 buffmt(h, "%s: %f%s;", p, v, u);
 645         else
 646                 /* LINTED */
 647                 buffmt(h, "%s: %d%s;", p, (int)v, u);
 648 }
 649
 650
 651 void
 652 html_idcat(char *dst, const char *src, int sz)
 653 {
 654         int              ssz;
 655
 656         assert(sz);
 657
 658         /* Cf. <http://www.w3.org/TR/html4/types.html#h-6.2>. */
 659
 660         for ( ; *dst != '\0' && sz; dst++, sz--)
 661                 /* Jump to end. */ ;
 662
 663         assert(sz > 2);
 664
 665         /* We can't start with a number (bah). */
 666
 667         *dst++ = 'x';
 668         *dst = '\0';
 669         sz--;
 670
 671         for ( ; *src != '\0' && sz > 1; src++) {
 672                 ssz = snprintf(dst, (size_t)sz, "%.2x", *src);
 673                 sz -= ssz;
 674                 dst += ssz;
 675         }
 676 }