html.c

   1 /*      $Id: html.c,v 1.210 2017/03/15 11:29:53 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008-2011, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2011-2015, 2017 Ingo Schwarze <schwarze@openbsd.org>
   5  *
   6  * Permission to use, copy, modify, and distribute this software for any
   7  * purpose with or without fee is hereby granted, provided that the above
   8  * copyright notice and this permission notice appear in all copies.
   9  *
  10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
  11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
  13  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  14  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  15  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  16  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  17  */
  18 #include "config.h"
  19
  20 #include <sys/types.h>
  21
  22 #include <assert.h>
  23 #include <ctype.h>
  24 #include <stdarg.h>
  25 #include <stdio.h>
  26 #include <stdint.h>
  27 #include <stdlib.h>
  28 #include <string.h>
  29 #include <unistd.h>
  30
  31 #include "mandoc_aux.h"
  32 #include "mandoc.h"
  33 #include "roff.h"
  34 #include "out.h"
  35 #include "html.h"
  36 #include "manconf.h"
  37 #include "main.h"
  38
  39 struct  htmldata {
  40         const char       *name;
  41         int               flags;
  42 #define HTML_NOSTACK     (1 << 0)
  43 #define HTML_AUTOCLOSE   (1 << 1)
  44 #define HTML_NLBEFORE    (1 << 2)
  45 #define HTML_NLBEGIN     (1 << 3)
  46 #define HTML_NLEND       (1 << 4)
  47 #define HTML_NLAFTER     (1 << 5)
  48 #define HTML_NLAROUND    (HTML_NLBEFORE | HTML_NLAFTER)
  49 #define HTML_NLINSIDE    (HTML_NLBEGIN | HTML_NLEND)
  50 #define HTML_NLALL       (HTML_NLAROUND | HTML_NLINSIDE)
  51 #define HTML_INDENT      (1 << 6)
  52 #define HTML_NOINDENT    (1 << 7)
  53 };
  54
  55 static  const struct htmldata htmltags[TAG_MAX] = {
  56         {"html",        HTML_NLALL},
  57         {"head",        HTML_NLALL | HTML_INDENT},
  58         {"body",        HTML_NLALL},
  59         {"meta",        HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL},
  60         {"title",       HTML_NLAROUND},
  61         {"div",         HTML_NLAROUND},
  62         {"h1",          HTML_NLAROUND},
  63         {"h2",          HTML_NLAROUND},
  64         {"span",        0},
  65         {"link",        HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL},
  66         {"br",          HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL},
  67         {"a",           0},
  68         {"table",       HTML_NLALL | HTML_INDENT},
  69         {"colgroup",    HTML_NLALL | HTML_INDENT},
  70         {"col",         HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL},
  71         {"tr",          HTML_NLALL | HTML_INDENT},
  72         {"td",          HTML_NLAROUND},
  73         {"li",          HTML_NLAROUND | HTML_INDENT},
  74         {"ul",          HTML_NLALL | HTML_INDENT},
  75         {"ol",          HTML_NLALL | HTML_INDENT},
  76         {"dl",          HTML_NLALL | HTML_INDENT},
  77         {"dt",          HTML_NLAROUND},
  78         {"dd",          HTML_NLAROUND | HTML_INDENT},
  79         {"pre",         HTML_NLALL | HTML_NOINDENT},
  80         {"var",         0},
  81         {"cite",        0},
  82         {"b",           0},
  83         {"i",           0},
  84         {"code",        0},
  85         {"small",       0},
  86         {"style",       HTML_NLALL | HTML_INDENT},
  87         {"math",        HTML_NLALL | HTML_INDENT},
  88         {"mrow",        0},
  89         {"mi",          0},
  90         {"mo",          0},
  91         {"msup",        0},
  92         {"msub",        0},
  93         {"msubsup",     0},
  94         {"mfrac",       0},
  95         {"msqrt",       0},
  96         {"mfenced",     0},
  97         {"mtable",      0},
  98         {"mtr",         0},
  99         {"mtd",         0},
 100         {"munderover",  0},
 101         {"munder",      0},
 102         {"mover",       0},
 103 };
 104
 105 static  const char      *const roffscales[SCALE_MAX] = {
 106         "cm", /* SCALE_CM */
 107         "in", /* SCALE_IN */
 108         "pc", /* SCALE_PC */
 109         "pt", /* SCALE_PT */
 110         "em", /* SCALE_EM */
 111         "em", /* SCALE_MM */
 112         "ex", /* SCALE_EN */
 113         "ex", /* SCALE_BU */
 114         "em", /* SCALE_VS */
 115         "ex", /* SCALE_FS */
 116 };
 117
 118 static  void     a2width(const char *, struct roffsu *);
 119 static  void     print_byte(struct html *, char);
 120 static  void     print_endword(struct html *);
 121 static  void     print_indent(struct html *);
 122 static  void     print_word(struct html *, const char *);
 123
 124 static  void     print_ctag(struct html *, struct tag *);
 125 static  int      print_escape(struct html *, char);
 126 static  int      print_encode(struct html *, const char *, const char *, int);
 127 static  void     print_href(struct html *, const char *, const char *, int);
 128 static  void     print_metaf(struct html *, enum mandoc_esc);
 129
 130
 131 void *
 132 html_alloc(const struct manoutput *outopts)
 133 {
 134         struct html     *h;
 135
 136         h = mandoc_calloc(1, sizeof(struct html));
 137
 138         h->tag = NULL;
 139         h->style = outopts->style;
 140         h->base_man = outopts->man;
 141         h->base_includes = outopts->includes;
 142         if (outopts->fragment)
 143                 h->oflags |= HTML_FRAGMENT;
 144
 145         return h;
 146 }
 147
 148 void
 149 html_free(void *p)
 150 {
 151         struct tag      *tag;
 152         struct html     *h;
 153
 154         h = (struct html *)p;
 155
 156         while ((tag = h->tag) != NULL) {
 157                 h->tag = tag->next;
 158                 free(tag);
 159         }
 160
 161         free(h);
 162 }
 163
 164 void
 165 print_gen_head(struct html *h)
 166 {
 167         struct tag      *t;
 168
 169         print_otag(h, TAG_META, "?", "charset", "utf-8");
 170
 171         /*
 172          * Print a default style-sheet.
 173          */
 174
 175         t = print_otag(h, TAG_STYLE, "");
 176         print_text(h, "table.head, table.foot { width: 100%; }");
 177         print_endline(h);
 178         print_text(h, "td.head-rtitle, td.foot-os { text-align: right; }");
 179         print_endline(h);
 180         print_text(h, "td.head-vol { text-align: center; }");
 181         print_endline(h);
 182         print_text(h, "div.Pp { margin: 1ex 0ex; }");
 183         print_tagq(h, t);
 184
 185         if (h->style)
 186                 print_otag(h, TAG_LINK, "?h??", "rel", "stylesheet",
 187                     h->style, "type", "text/css", "media", "all");
 188 }
 189
 190 static void
 191 print_metaf(struct html *h, enum mandoc_esc deco)
 192 {
 193         enum htmlfont    font;
 194
 195         switch (deco) {
 196         case ESCAPE_FONTPREV:
 197                 font = h->metal;
 198                 break;
 199         case ESCAPE_FONTITALIC:
 200                 font = HTMLFONT_ITALIC;
 201                 break;
 202         case ESCAPE_FONTBOLD:
 203                 font = HTMLFONT_BOLD;
 204                 break;
 205         case ESCAPE_FONTBI:
 206                 font = HTMLFONT_BI;
 207                 break;
 208         case ESCAPE_FONT:
 209         case ESCAPE_FONTROMAN:
 210                 font = HTMLFONT_NONE;
 211                 break;
 212         default:
 213                 abort();
 214         }
 215
 216         if (h->metaf) {
 217                 print_tagq(h, h->metaf);
 218                 h->metaf = NULL;
 219         }
 220
 221         h->metal = h->metac;
 222         h->metac = font;
 223
 224         switch (font) {
 225         case HTMLFONT_ITALIC:
 226                 h->metaf = print_otag(h, TAG_I, "");
 227                 break;
 228         case HTMLFONT_BOLD:
 229                 h->metaf = print_otag(h, TAG_B, "");
 230                 break;
 231         case HTMLFONT_BI:
 232                 h->metaf = print_otag(h, TAG_B, "");
 233                 print_otag(h, TAG_I, "");
 234                 break;
 235         default:
 236                 break;
 237         }
 238 }
 239
 240 char *
 241 html_make_id(const struct roff_node *n)
 242 {
 243         const struct roff_node  *nch;
 244         char                    *buf, *cp;
 245
 246         for (nch = n->child; nch != NULL; nch = nch->next)
 247                 if (nch->type != ROFFT_TEXT)
 248                         return NULL;
 249
 250         buf = NULL;
 251         deroff(&buf, n);
 252
 253         /* http://www.w3.org/TR/html5/dom.html#the-id-attribute */
 254
 255         for (cp = buf; *cp != '\0'; cp++)
 256                 if (*cp == ' ')
 257                         *cp = '_';
 258
 259         return buf;
 260 }
 261
 262 int
 263 html_strlen(const char *cp)
 264 {
 265         size_t           rsz;
 266         int              skip, sz;
 267
 268         /*
 269          * Account for escaped sequences within string length
 270          * calculations.  This follows the logic in term_strlen() as we
 271          * must calculate the width of produced strings.
 272          * Assume that characters are always width of "1".  This is
 273          * hacky, but it gets the job done for approximation of widths.
 274          */
 275
 276         sz = 0;
 277         skip = 0;
 278         while (1) {
 279                 rsz = strcspn(cp, "\\");
 280                 if (rsz) {
 281                         cp += rsz;
 282                         if (skip) {
 283                                 skip = 0;
 284                                 rsz--;
 285                         }
 286                         sz += rsz;
 287                 }
 288                 if ('\0' == *cp)
 289                         break;
 290                 cp++;
 291                 switch (mandoc_escape(&cp, NULL, NULL)) {
 292                 case ESCAPE_ERROR:
 293                         return sz;
 294                 case ESCAPE_UNICODE:
 295                 case ESCAPE_NUMBERED:
 296                 case ESCAPE_SPECIAL:
 297                 case ESCAPE_OVERSTRIKE:
 298                         if (skip)
 299                                 skip = 0;
 300                         else
 301                                 sz++;
 302                         break;
 303                 case ESCAPE_SKIPCHAR:
 304                         skip = 1;
 305                         break;
 306                 default:
 307                         break;
 308                 }
 309         }
 310         return sz;
 311 }
 312
 313 static int
 314 print_escape(struct html *h, char c)
 315 {
 316
 317         switch (c) {
 318         case '<':
 319                 print_word(h, "&lt;");
 320                 break;
 321         case '>':
 322                 print_word(h, "&gt;");
 323                 break;
 324         case '&':
 325                 print_word(h, "&amp;");
 326                 break;
 327         case '"':
 328                 print_word(h, "&quot;");
 329                 break;
 330         case ASCII_NBRSP:
 331                 print_word(h, "&nbsp;");
 332                 break;
 333         case ASCII_HYPH:
 334                 print_byte(h, '-');
 335                 break;
 336         case ASCII_BREAK:
 337                 break;
 338         default:
 339                 return 0;
 340         }
 341         return 1;
 342 }
 343
 344 static int
 345 print_encode(struct html *h, const char *p, const char *pend, int norecurse)
 346 {
 347         char             numbuf[16];
 348         size_t           sz;
 349         int              c, len, nospace;
 350         const char      *seq;
 351         enum mandoc_esc  esc;
 352         static const char rejs[9] = { '\\', '<', '>', '&', '"',
 353                 ASCII_NBRSP, ASCII_HYPH, ASCII_BREAK, '\0' };
 354
 355         if (pend == NULL)
 356                 pend = strchr(p, '\0');
 357
 358         nospace = 0;
 359
 360         while (p < pend) {
 361                 if (HTML_SKIPCHAR & h->flags && '\\' != *p) {
 362                         h->flags &= ~HTML_SKIPCHAR;
 363                         p++;
 364                         continue;
 365                 }
 366
 367                 for (sz = strcspn(p, rejs); sz-- && p < pend; p++)
 368                         if (*p == ' ')
 369                                 print_endword(h);
 370                         else
 371                                 print_byte(h, *p);
 372
 373                 if (p >= pend)
 374                         break;
 375
 376                 if (print_escape(h, *p++))
 377                         continue;
 378
 379                 esc = mandoc_escape(&p, &seq, &len);
 380                 if (ESCAPE_ERROR == esc)
 381                         break;
 382
 383                 switch (esc) {
 384                 case ESCAPE_FONT:
 385                 case ESCAPE_FONTPREV:
 386                 case ESCAPE_FONTBOLD:
 387                 case ESCAPE_FONTITALIC:
 388                 case ESCAPE_FONTBI:
 389                 case ESCAPE_FONTROMAN:
 390                         if (0 == norecurse)
 391                                 print_metaf(h, esc);
 392                         continue;
 393                 case ESCAPE_SKIPCHAR:
 394                         h->flags |= HTML_SKIPCHAR;
 395                         continue;
 396                 default:
 397                         break;
 398                 }
 399
 400                 if (h->flags & HTML_SKIPCHAR) {
 401                         h->flags &= ~HTML_SKIPCHAR;
 402                         continue;
 403                 }
 404
 405                 switch (esc) {
 406                 case ESCAPE_UNICODE:
 407                         /* Skip past "u" header. */
 408                         c = mchars_num2uc(seq + 1, len - 1);
 409                         break;
 410                 case ESCAPE_NUMBERED:
 411                         c = mchars_num2char(seq, len);
 412                         if (c < 0)
 413                                 continue;
 414                         break;
 415                 case ESCAPE_SPECIAL:
 416                         c = mchars_spec2cp(seq, len);
 417                         if (c <= 0)
 418                                 continue;
 419                         break;
 420                 case ESCAPE_NOSPACE:
 421                         if ('\0' == *p)
 422                                 nospace = 1;
 423                         continue;
 424                 case ESCAPE_OVERSTRIKE:
 425                         if (len == 0)
 426                                 continue;
 427                         c = seq[len - 1];
 428                         break;
 429                 default:
 430                         continue;
 431                 }
 432                 if ((c < 0x20 && c != 0x09) ||
 433                     (c > 0x7E && c < 0xA0))
 434                         c = 0xFFFD;
 435                 if (c > 0x7E) {
 436                         (void)snprintf(numbuf, sizeof(numbuf), "&#%d;", c);
 437                         print_word(h, numbuf);
 438                 } else if (print_escape(h, c) == 0)
 439                         print_byte(h, c);
 440         }
 441
 442         return nospace;
 443 }
 444
 445 static void
 446 print_href(struct html *h, const char *name, const char *sec, int man)
 447 {
 448         const char      *p, *pp;
 449
 450         pp = man ? h->base_man : h->base_includes;
 451         while ((p = strchr(pp, '%')) != NULL) {
 452                 print_encode(h, pp, p, 1);
 453                 if (man && p[1] == 'S') {
 454                         if (sec == NULL)
 455                                 print_byte(h, '1');
 456                         else
 457                                 print_encode(h, sec, NULL, 1);
 458                 } else if ((man && p[1] == 'N') ||
 459                     (man == 0 && p[1] == 'I'))
 460                         print_encode(h, name, NULL, 1);
 461                 else
 462                         print_encode(h, p, p + 2, 1);
 463                 pp = p + 2;
 464         }
 465         if (*pp != '\0')
 466                 print_encode(h, pp, NULL, 1);
 467 }
 468
 469 struct tag *
 470 print_otag(struct html *h, enum htmltag tag, const char *fmt, ...)
 471 {
 472         va_list          ap;
 473         struct roffsu    mysu, *su;
 474         char             numbuf[16];
 475         struct tag      *t;
 476         const char      *attr;
 477         char            *arg1, *arg2;
 478         double           v;
 479         int              i, have_style, tflags;
 480
 481         tflags = htmltags[tag].flags;
 482
 483         /* Push this tag onto the stack of open scopes. */
 484
 485         if ((tflags & HTML_NOSTACK) == 0) {
 486                 t = mandoc_malloc(sizeof(struct tag));
 487                 t->tag = tag;
 488                 t->next = h->tag;
 489                 h->tag = t;
 490         } else
 491                 t = NULL;
 492
 493         if (tflags & HTML_NLBEFORE)
 494                 print_endline(h);
 495         if (h->col == 0)
 496                 print_indent(h);
 497         else if ((h->flags & HTML_NOSPACE) == 0) {
 498                 if (h->flags & HTML_KEEP)
 499                         print_word(h, "&#160;");
 500                 else {
 501                         if (h->flags & HTML_PREKEEP)
 502                                 h->flags |= HTML_KEEP;
 503                         print_endword(h);
 504                 }
 505         }
 506
 507         if ( ! (h->flags & HTML_NONOSPACE))
 508                 h->flags &= ~HTML_NOSPACE;
 509         else
 510                 h->flags |= HTML_NOSPACE;
 511
 512         /* Print out the tag name and attributes. */
 513
 514         print_byte(h, '<');
 515         print_word(h, htmltags[tag].name);
 516
 517         va_start(ap, fmt);
 518
 519         have_style = 0;
 520         while (*fmt != '\0') {
 521                 if (*fmt == 's') {
 522                         have_style = 1;
 523                         fmt++;
 524                         break;
 525                 }
 526
 527                 /* Parse a non-style attribute and its arguments. */
 528
 529                 arg1 = va_arg(ap, char *);
 530                 switch (*fmt++) {
 531                 case 'c':
 532                         attr = "class";
 533                         break;
 534                 case 'h':
 535                         attr = "href";
 536                         break;
 537                 case 'i':
 538                         attr = "id";
 539                         break;
 540                 case '?':
 541                         attr = arg1;
 542                         arg1 = va_arg(ap, char *);
 543                         break;
 544                 default:
 545                         abort();
 546                 }
 547                 arg2 = NULL;
 548                 if (*fmt == 'M')
 549                         arg2 = va_arg(ap, char *);
 550                 if (arg1 == NULL)
 551                         continue;
 552
 553                 /* Print the non-style attributes. */
 554
 555                 print_byte(h, ' ');
 556                 print_word(h, attr);
 557                 print_byte(h, '=');
 558                 print_byte(h, '"');
 559                 switch (*fmt) {
 560                 case 'I':
 561                         print_href(h, arg1, NULL, 0);
 562                         fmt++;
 563                         break;
 564                 case 'M':
 565                         print_href(h, arg1, arg2, 1);
 566                         fmt++;
 567                         break;
 568                 case 'R':
 569                         print_byte(h, '#');
 570                         print_encode(h, arg1, NULL, 1);
 571                         fmt++;
 572                         break;
 573                 case 'T':
 574                         print_encode(h, arg1, NULL, 1);
 575                         print_word(h, "\" title=\"");
 576                         print_encode(h, arg1, NULL, 1);
 577                         fmt++;
 578                         break;
 579                 default:
 580                         print_encode(h, arg1, NULL, 1);
 581                         break;
 582                 }
 583                 print_byte(h, '"');
 584         }
 585
 586         /* Print out styles. */
 587
 588         while (*fmt != '\0') {
 589                 arg1 = NULL;
 590                 su = NULL;
 591
 592                 /* First letter: input argument type. */
 593
 594                 switch (*fmt++) {
 595                 case 'h':
 596                         i = va_arg(ap, int);
 597                         su = &mysu;
 598                         SCALE_HS_INIT(su, i);
 599                         break;
 600                 case 's':
 601                         arg1 = va_arg(ap, char *);
 602                         break;
 603                 case 'u':
 604                         su = va_arg(ap, struct roffsu *);
 605                         break;
 606                 case 'v':
 607                         i = va_arg(ap, int);
 608                         su = &mysu;
 609                         SCALE_VS_INIT(su, i);
 610                         break;
 611                 case 'w':
 612                 case 'W':
 613                         if ((arg2 = va_arg(ap, char *)) == NULL)
 614                                 break;
 615                         su = &mysu;
 616                         a2width(arg2, su);
 617                         /* Increase width to make even bold text fit. */
 618                         su->scale *= 1.1;
 619                         if (fmt[-1] == 'W')
 620                                 su->scale *= -1.0;
 621                         break;
 622                 default:
 623                         abort();
 624                 }
 625
 626                 /* Second letter: style name. */
 627
 628                 switch (*fmt++) {
 629                 case 'b':
 630                         attr = "margin-bottom";
 631                         break;
 632                 case 'h':
 633                         attr = "height";
 634                         break;
 635                 case 'i':
 636                         attr = "text-indent";
 637                         break;
 638                 case 'l':
 639                         attr = "margin-left";
 640                         break;
 641                 case 't':
 642                         attr = "margin-top";
 643                         break;
 644                 case 'w':
 645                         attr = "width";
 646                         break;
 647                 case 'W':
 648                         attr = "min-width";
 649                         break;
 650                 case '?':
 651                         attr = arg1;
 652                         arg1 = va_arg(ap, char *);
 653                         break;
 654                 default:
 655                         abort();
 656                 }
 657                 if (su == NULL && arg1 == NULL)
 658                         continue;
 659
 660                 if (have_style == 1)
 661                         print_word(h, " style=\"");
 662                 else
 663                         print_byte(h, ' ');
 664                 print_word(h, attr);
 665                 print_byte(h, ':');
 666                 print_byte(h, ' ');
 667                 if (su != NULL) {
 668                         v = su->scale;
 669                         if (su->unit == SCALE_MM && (v /= 100.0) == 0.0)
 670                                 v = 1.0;
 671                         else if (su->unit == SCALE_BU)
 672                                 v /= 24.0;
 673                         (void)snprintf(numbuf, sizeof(numbuf), "%.2f", v);
 674                         print_word(h, numbuf);
 675                         print_word(h, roffscales[su->unit]);
 676                 } else
 677                         print_word(h, arg1);
 678                 print_byte(h, ';');
 679                 have_style = 2;
 680         }
 681         if (have_style == 2)
 682                 print_byte(h, '"');
 683
 684         va_end(ap);
 685
 686         /* Accommodate for "well-formed" singleton escaping. */
 687
 688         if (HTML_AUTOCLOSE & htmltags[tag].flags)
 689                 print_byte(h, '/');
 690
 691         print_byte(h, '>');
 692
 693         if (tflags & HTML_NLBEGIN)
 694                 print_endline(h);
 695         else
 696                 h->flags |= HTML_NOSPACE;
 697
 698         if (tflags & HTML_INDENT)
 699                 h->indent++;
 700         if (tflags & HTML_NOINDENT)
 701                 h->noindent++;
 702
 703         return t;
 704 }
 705
 706 static void
 707 print_ctag(struct html *h, struct tag *tag)
 708 {
 709         int      tflags;
 710
 711         /*
 712          * Remember to close out and nullify the current
 713          * meta-font and table, if applicable.
 714          */
 715         if (tag == h->metaf)
 716                 h->metaf = NULL;
 717         if (tag == h->tblt)
 718                 h->tblt = NULL;
 719
 720         tflags = htmltags[tag->tag].flags;
 721
 722         if (tflags & HTML_INDENT)
 723                 h->indent--;
 724         if (tflags & HTML_NOINDENT)
 725                 h->noindent--;
 726         if (tflags & HTML_NLEND)
 727                 print_endline(h);
 728         print_indent(h);
 729         print_byte(h, '<');
 730         print_byte(h, '/');
 731         print_word(h, htmltags[tag->tag].name);
 732         print_byte(h, '>');
 733         if (tflags & HTML_NLAFTER)
 734                 print_endline(h);
 735
 736         h->tag = tag->next;
 737         free(tag);
 738 }
 739
 740 void
 741 print_gen_decls(struct html *h)
 742 {
 743         print_word(h, "<!DOCTYPE html>");
 744         print_endline(h);
 745 }
 746
 747 void
 748 print_text(struct html *h, const char *word)
 749 {
 750         if (h->col && (h->flags & HTML_NOSPACE) == 0) {
 751                 if ( ! (HTML_KEEP & h->flags)) {
 752                         if (HTML_PREKEEP & h->flags)
 753                                 h->flags |= HTML_KEEP;
 754                         print_endword(h);
 755                 } else
 756                         print_word(h, "&#160;");
 757         }
 758
 759         assert(NULL == h->metaf);
 760         switch (h->metac) {
 761         case HTMLFONT_ITALIC:
 762                 h->metaf = print_otag(h, TAG_I, "");
 763                 break;
 764         case HTMLFONT_BOLD:
 765                 h->metaf = print_otag(h, TAG_B, "");
 766                 break;
 767         case HTMLFONT_BI:
 768                 h->metaf = print_otag(h, TAG_B, "");
 769                 print_otag(h, TAG_I, "");
 770                 break;
 771         default:
 772                 print_indent(h);
 773                 break;
 774         }
 775
 776         assert(word);
 777         if ( ! print_encode(h, word, NULL, 0)) {
 778                 if ( ! (h->flags & HTML_NONOSPACE))
 779                         h->flags &= ~HTML_NOSPACE;
 780                 h->flags &= ~HTML_NONEWLINE;
 781         } else
 782                 h->flags |= HTML_NOSPACE | HTML_NONEWLINE;
 783
 784         if (h->metaf) {
 785                 print_tagq(h, h->metaf);
 786                 h->metaf = NULL;
 787         }
 788
 789         h->flags &= ~HTML_IGNDELIM;
 790 }
 791
 792 void
 793 print_tagq(struct html *h, const struct tag *until)
 794 {
 795         struct tag      *tag;
 796
 797         while ((tag = h->tag) != NULL) {
 798                 print_ctag(h, tag);
 799                 if (until && tag == until)
 800                         return;
 801         }
 802 }
 803
 804 void
 805 print_stagq(struct html *h, const struct tag *suntil)
 806 {
 807         struct tag      *tag;
 808
 809         while ((tag = h->tag) != NULL) {
 810                 if (suntil && tag == suntil)
 811                         return;
 812                 print_ctag(h, tag);
 813         }
 814 }
 815
 816 void
 817 print_paragraph(struct html *h)
 818 {
 819         struct tag      *t;
 820
 821         t = print_otag(h, TAG_DIV, "c", "Pp");
 822         print_tagq(h, t);
 823 }
 824
 825
 826 /***********************************************************************
 827  * Low level output functions.
 828  * They implement line breaking using a short static buffer.
 829  ***********************************************************************/
 830
 831 /*
 832  * Buffer one HTML output byte.
 833  * If the buffer is full, flush and deactivate it and start a new line.
 834  * If the buffer is inactive, print directly.
 835  */
 836 static void
 837 print_byte(struct html *h, char c)
 838 {
 839         if ((h->flags & HTML_BUFFER) == 0) {
 840                 putchar(c);
 841                 h->col++;
 842                 return;
 843         }
 844
 845         if (h->col + h->bufcol < sizeof(h->buf)) {
 846                 h->buf[h->bufcol++] = c;
 847                 return;
 848         }
 849
 850         putchar('\n');
 851         h->col = 0;
 852         print_indent(h);
 853         putchar(' ');
 854         putchar(' ');
 855         fwrite(h->buf, h->bufcol, 1, stdout);
 856         putchar(c);
 857         h->col = (h->indent + 1) * 2 + h->bufcol + 1;
 858         h->bufcol = 0;
 859         h->flags &= ~HTML_BUFFER;
 860 }
 861
 862 /*
 863  * If something was printed on the current output line, end it.
 864  * Not to be called right after print_indent().
 865  */
 866 void
 867 print_endline(struct html *h)
 868 {
 869         if (h->col == 0)
 870                 return;
 871
 872         if (h->bufcol) {
 873                 putchar(' ');
 874                 fwrite(h->buf, h->bufcol, 1, stdout);
 875                 h->bufcol = 0;
 876         }
 877         putchar('\n');
 878         h->col = 0;
 879         h->flags |= HTML_NOSPACE;
 880         h->flags &= ~HTML_BUFFER;
 881 }
 882
 883 /*
 884  * Flush the HTML output buffer.
 885  * If it is inactive, activate it.
 886  */
 887 static void
 888 print_endword(struct html *h)
 889 {
 890         if (h->noindent) {
 891                 print_byte(h, ' ');
 892                 return;
 893         }
 894
 895         if ((h->flags & HTML_BUFFER) == 0) {
 896                 h->col++;
 897                 h->flags |= HTML_BUFFER;
 898         } else if (h->bufcol) {
 899                 putchar(' ');
 900                 fwrite(h->buf, h->bufcol, 1, stdout);
 901                 h->col += h->bufcol + 1;
 902         }
 903         h->bufcol = 0;
 904 }
 905
 906 /*
 907  * If at the beginning of a new output line,
 908  * perform indentation and mark the line as containing output.
 909  * Make sure to really produce some output right afterwards,
 910  * but do not use print_otag() for producing it.
 911  */
 912 static void
 913 print_indent(struct html *h)
 914 {
 915         size_t   i;
 916
 917         if (h->col)
 918                 return;
 919
 920         if (h->noindent == 0) {
 921                 h->col = h->indent * 2;
 922                 for (i = 0; i < h->col; i++)
 923                         putchar(' ');
 924         }
 925         h->flags &= ~HTML_NOSPACE;
 926 }
 927
 928 /*
 929  * Print or buffer some characters
 930  * depending on the current HTML output buffer state.
 931  */
 932 static void
 933 print_word(struct html *h, const char *cp)
 934 {
 935         while (*cp != '\0')
 936                 print_byte(h, *cp++);
 937 }
 938
 939 /*
 940  * Calculate the scaling unit passed in a `-width' argument.  This uses
 941  * either a native scaling unit (e.g., 1i, 2m) or the string length of
 942  * the value.
 943  */
 944 static void
 945 a2width(const char *p, struct roffsu *su)
 946 {
 947         if (a2roffsu(p, su, SCALE_MAX) < 2) {
 948                 su->unit = SCALE_EN;
 949                 su->scale = html_strlen(p);
 950         } else if (su->scale < 0.0)
 951                 su->scale = 0.0;
 952 }