usr.bin/mandoc/html.c

   1 /*      $Id: html.c,v 1.2 2009/10/27 21:40:07 schwarze 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 <err.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 };
  84
  85 void *
  86 html_alloc(char *outopts)
  87 {
  88         struct html     *h;
  89         const char      *toks[4];
  90         char            *v;
  91
  92         toks[0] = "style";
  93         toks[1] = "man";
  94         toks[2] = "includes";
  95         toks[3] = NULL;
  96
  97         if (NULL == (h = calloc(1, sizeof(struct html))))
  98                 return(NULL);
  99
 100         h->tags.head = NULL;
 101         h->ords.head = NULL;
 102
 103         if (NULL == (h->symtab = chars_init(CHARS_HTML))) {
 104                 free(h);
 105                 return(NULL);
 106         }
 107
 108         while (outopts && *outopts)
 109                 switch (getsubopt(&outopts, UNCONST(toks), &v)) {
 110                 case (0):
 111                         h->style = v;
 112                         break;
 113                 case (1):
 114                         h->base_man = v;
 115                         break;
 116                 case (2):
 117                         h->base_includes = v;
 118                         break;
 119                 default:
 120                         break;
 121                 }
 122
 123         return(h);
 124 }
 125
 126
 127 void
 128 html_free(void *p)
 129 {
 130         struct tag      *tag;
 131         struct ord      *ord;
 132         struct html     *h;
 133
 134         h = (struct html *)p;
 135
 136         while ((ord = h->ords.head) != NULL) {
 137                 h->ords.head = ord->next;
 138                 free(ord);
 139         }
 140
 141         while ((tag = h->tags.head) != NULL) {
 142                 h->tags.head = tag->next;
 143                 free(tag);
 144         }
 145
 146         if (h->symtab)
 147                 chars_free(h->symtab);
 148
 149         free(h);
 150 }
 151
 152
 153 void
 154 print_gen_head(struct html *h)
 155 {
 156         struct htmlpair  tag[4];
 157
 158         tag[0].key = ATTR_HTTPEQUIV;
 159         tag[0].val = "Content-Type";
 160         tag[1].key = ATTR_CONTENT;
 161         tag[1].val = "text/html; charset=utf-8";
 162         print_otag(h, TAG_META, 2, tag);
 163
 164         tag[0].key = ATTR_NAME;
 165         tag[0].val = "resource-type";
 166         tag[1].key = ATTR_CONTENT;
 167         tag[1].val = "document";
 168         print_otag(h, TAG_META, 2, tag);
 169
 170         if (h->style) {
 171                 tag[0].key = ATTR_REL;
 172                 tag[0].val = "stylesheet";
 173                 tag[1].key = ATTR_HREF;
 174                 tag[1].val = h->style;
 175                 tag[2].key = ATTR_TYPE;
 176                 tag[2].val = "text/css";
 177                 tag[3].key = ATTR_MEDIA;
 178                 tag[3].val = "all";
 179                 print_otag(h, TAG_LINK, 4, tag);
 180         }
 181 }
 182
 183
 184 static void
 185 print_spec(struct html *h, const char *p, int len)
 186 {
 187         const char      *rhs;
 188         int              i;
 189         size_t           sz;
 190
 191         rhs = chars_a2ascii(h->symtab, p, (size_t)len, &sz);
 192
 193         if (NULL == rhs)
 194                 return;
 195         for (i = 0; i < (int)sz; i++)
 196                 putchar(rhs[i]);
 197 }
 198
 199
 200 static void
 201 print_res(struct html *h, const char *p, int len)
 202 {
 203         const char      *rhs;
 204         int              i;
 205         size_t           sz;
 206
 207         rhs = chars_a2res(h->symtab, p, (size_t)len, &sz);
 208
 209         if (NULL == rhs)
 210                 return;
 211         for (i = 0; i < (int)sz; i++)
 212                 putchar(rhs[i]);
 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
 320         for (; *p; p++) {
 321                 if ('\\' == *p) {
 322                         print_escape(h, &p);
 323                         continue;
 324                 }
 325                 switch (*p) {
 326                 case ('<'):
 327                         printf("&lt;");
 328                         break;
 329                 case ('>'):
 330                         printf("&gt;");
 331                         break;
 332                 case ('&'):
 333                         printf("&amp;");
 334                         break;
 335                 default:
 336                         putchar(*p);
 337                         break;
 338                 }
 339         }
 340 }
 341
 342
 343 struct tag *
 344 print_otag(struct html *h, enum htmltag tag,
 345                 int sz, const struct htmlpair *p)
 346 {
 347         int              i;
 348         struct tag      *t;
 349
 350         if ( ! (HTML_NOSTACK & htmltags[tag].flags)) {
 351                 if (NULL == (t = malloc(sizeof(struct tag))))
 352                         err(EXIT_FAILURE, "malloc");
 353                 t->tag = tag;
 354                 t->next = h->tags.head;
 355                 h->tags.head = t;
 356         } else
 357                 t = NULL;
 358
 359         if ( ! (HTML_NOSPACE & h->flags))
 360                 if ( ! (HTML_CLRLINE & htmltags[tag].flags))
 361                         printf(" ");
 362
 363         printf("<%s", htmltags[tag].name);
 364         for (i = 0; i < sz; i++) {
 365                 printf(" %s=\"", htmlattrs[p[i].key]);
 366                 assert(p->val);
 367                 print_encode(h, p[i].val);
 368                 printf("\"");
 369         }
 370         printf(">");
 371
 372         h->flags |= HTML_NOSPACE;
 373         if (HTML_CLRLINE & htmltags[tag].flags)
 374                 h->flags |= HTML_NEWLINE;
 375         else
 376                 h->flags &= ~HTML_NEWLINE;
 377
 378         return(t);
 379 }
 380
 381
 382 /* ARGSUSED */
 383 static void
 384 print_ctag(struct html *h, enum htmltag tag)
 385 {
 386
 387         printf("</%s>", htmltags[tag].name);
 388         if (HTML_CLRLINE & htmltags[tag].flags)
 389                 h->flags |= HTML_NOSPACE;
 390         if (HTML_CLRLINE & htmltags[tag].flags)
 391                 h->flags |= HTML_NEWLINE;
 392         else
 393                 h->flags &= ~HTML_NEWLINE;
 394 }
 395
 396
 397 /* ARGSUSED */
 398 void
 399 print_gen_doctype(struct html *h)
 400 {
 401
 402         printf("<!DOCTYPE HTML PUBLIC \"%s\" \"%s\">", DOCTYPE, DTD);
 403 }
 404
 405
 406 void
 407 print_text(struct html *h, const char *p)
 408 {
 409
 410         if (*p && 0 == *(p + 1))
 411                 switch (*p) {
 412                 case('.'):
 413                         /* FALLTHROUGH */
 414                 case(','):
 415                         /* FALLTHROUGH */
 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                         if ( ! (HTML_IGNDELIM & h->flags))
 430                                 h->flags |= HTML_NOSPACE;
 431                         break;
 432                 default:
 433                         break;
 434                 }
 435
 436         if ( ! (h->flags & HTML_NOSPACE))
 437                 printf(" ");
 438
 439         h->flags &= ~HTML_NOSPACE;
 440         h->flags &= ~HTML_NEWLINE;
 441
 442         if (p)
 443                 print_encode(h, p);
 444
 445         if (*p && 0 == *(p + 1))
 446                 switch (*p) {
 447                 case('('):
 448                         /* FALLTHROUGH */
 449                 case('['):
 450                         /* FALLTHROUGH */
 451                 case('{'):
 452                         h->flags |= HTML_NOSPACE;
 453                         break;
 454                 default:
 455                         break;
 456                 }
 457 }
 458
 459
 460 void
 461 print_tagq(struct html *h, const struct tag *until)
 462 {
 463         struct tag      *tag;
 464
 465         while ((tag = h->tags.head) != NULL) {
 466                 print_ctag(h, tag->tag);
 467                 h->tags.head = tag->next;
 468                 free(tag);
 469                 if (until && tag == until)
 470                         return;
 471         }
 472 }
 473
 474
 475 void
 476 print_stagq(struct html *h, const struct tag *suntil)
 477 {
 478         struct tag      *tag;
 479
 480         while ((tag = h->tags.head) != NULL) {
 481                 if (suntil && tag == suntil)
 482                         return;
 483                 print_ctag(h, tag->tag);
 484                 h->tags.head = tag->next;
 485                 free(tag);
 486         }
 487 }
 488
 489
 490 void
 491 bufinit(struct html *h)
 492 {
 493
 494         h->buf[0] = '\0';
 495         h->buflen = 0;
 496 }
 497
 498
 499 void
 500 bufcat_style(struct html *h, const char *key, const char *val)
 501 {
 502
 503         bufcat(h, key);
 504         bufncat(h, ":", 1);
 505         bufcat(h, val);
 506         bufncat(h, ";", 1);
 507 }
 508
 509
 510 void
 511 bufcat(struct html *h, const char *p)
 512 {
 513
 514         bufncat(h, p, strlen(p));
 515 }
 516
 517
 518 void
 519 buffmt(struct html *h, const char *fmt, ...)
 520 {
 521         va_list          ap;
 522
 523         va_start(ap, fmt);
 524         (void)vsnprintf(h->buf + (int)h->buflen,
 525                         BUFSIZ - h->buflen - 1, fmt, ap);
 526         va_end(ap);
 527         h->buflen = strlen(h->buf);
 528 }
 529
 530
 531 void
 532 bufncat(struct html *h, const char *p, size_t sz)
 533 {
 534
 535         if (h->buflen + sz > BUFSIZ - 1)
 536                 sz = BUFSIZ - 1 - h->buflen;
 537
 538         (void)strncat(h->buf, p, sz);
 539         h->buflen += sz;
 540 }
 541
 542
 543 void
 544 buffmt_includes(struct html *h, const char *name)
 545 {
 546         const char      *p, *pp;
 547
 548         pp = h->base_includes;
 549
 550         while (NULL != (p = strchr(pp, '%'))) {
 551                 bufncat(h, pp, (size_t)(p - pp));
 552                 switch (*(p + 1)) {
 553                 case('I'):
 554                         bufcat(h, name);
 555                         break;
 556                 default:
 557                         bufncat(h, p, 2);
 558                         break;
 559                 }
 560                 pp = p + 2;
 561         }
 562         if (pp)
 563                 bufcat(h, pp);
 564 }
 565
 566
 567 void
 568 buffmt_man(struct html *h,
 569                 const char *name, const char *sec)
 570 {
 571         const char      *p, *pp;
 572
 573         pp = h->base_man;
 574
 575         /* LINTED */
 576         while (NULL != (p = strchr(pp, '%'))) {
 577                 bufncat(h, pp, (size_t)(p - pp));
 578                 switch (*(p + 1)) {
 579                 case('S'):
 580                         bufcat(h, sec ? sec : "1");
 581                         break;
 582                 case('N'):
 583                         buffmt(h, name);
 584                         break;
 585                 default:
 586                         bufncat(h, p, 2);
 587                         break;
 588                 }
 589                 pp = p + 2;
 590         }
 591         if (pp)
 592                 bufcat(h, pp);
 593 }
 594
 595
 596 void
 597 bufcat_su(struct html *h, const char *p, const struct roffsu *su)
 598 {
 599         double           v;
 600         const char      *u;
 601
 602         v = su->scale;
 603
 604         switch (su->unit) {
 605         case (SCALE_CM):
 606                 u = "cm";
 607                 break;
 608         case (SCALE_IN):
 609                 u = "in";
 610                 break;
 611         case (SCALE_PC):
 612                 u = "pc";
 613                 break;
 614         case (SCALE_PT):
 615                 u = "pt";
 616                 break;
 617         case (SCALE_EM):
 618                 u = "em";
 619                 break;
 620         case (SCALE_MM):
 621                 if (0 == (v /= 100))
 622                         v = 1;
 623                 u = "em";
 624                 break;
 625         case (SCALE_EN):
 626                 u = "ex";
 627                 break;
 628         case (SCALE_BU):
 629                 u = "ex";
 630                 break;
 631         case (SCALE_VS):
 632                 u = "em";
 633                 break;
 634         default:
 635                 u = "ex";
 636                 break;
 637         }
 638
 639         if (su->pt)
 640                 buffmt(h, "%s: %f%s;", p, v, u);
 641         else
 642                 /* LINTED */
 643                 buffmt(h, "%s: %d%s;", p, (int)v, u);
 644 }