mdocterm.c

   1 /* $Id: mdocterm.c,v 1.30 2009/03/04 14:04:02 kristaps Exp $ */
   2 /*
   3  * Copyright (c) 2008 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
   7  * above copyright notice and this permission notice appear in all
   8  * copies.
   9  *
  10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL
  11  * WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED
  12  * WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE
  13  * AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL
  14  * DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
  15  * PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
  16  * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
  17  * PERFORMANCE OF THIS SOFTWARE.
  18  */
  19 #include <sys/utsname.h>
  20
  21 #include <assert.h>
  22 #include <ctype.h>
  23 #include <err.h>
  24 #include <getopt.h>
  25 #include <stdio.h>
  26 #include <stdlib.h>
  27 #include <string.h>
  28 #ifndef __OpenBSD__
  29 #include <time.h>
  30 #endif
  31
  32 #include "mmain.h"
  33 #include "term.h"
  34
  35 struct  termenc {
  36         const char       *enc;
  37         int               sym;
  38 };
  39
  40 static  void              body(struct termp *,
  41                                 struct termpair *,
  42                                 const struct mdoc_meta *,
  43                                 const struct mdoc_node *);
  44 static  void              header(struct termp *,
  45                                 const struct mdoc_meta *);
  46 static  void              footer(struct termp *,
  47                                 const struct mdoc_meta *);
  48
  49 static  void              pword(struct termp *, const char *, size_t);
  50 static  void              pescape(struct termp *, const char *,
  51                                 size_t *, size_t);
  52 static  void              nescape(struct termp *,
  53                                 const char *, size_t);
  54 static  void              chara(struct termp *, char);
  55 static  void              stringa(struct termp *,
  56                                 const char *, size_t);
  57 static  void              symbola(struct termp *, enum tsym);
  58 static  void              stylea(struct termp *, enum tstyle);
  59
  60 #ifdef __linux__
  61 extern  size_t            strlcat(char *, const char *, size_t);
  62 extern  size_t            strlcpy(char *, const char *, size_t);
  63 #endif
  64
  65 static  struct termenc    termenc1[] = {
  66         { "\\",           TERMSYM_SLASH },
  67         { "\'",           TERMSYM_RSQUOTE },
  68         { "`",            TERMSYM_LSQUOTE },
  69         { "-",            TERMSYM_HYPHEN },
  70         { " ",            TERMSYM_SPACE },
  71         { ".",            TERMSYM_PERIOD },
  72         { "&",            TERMSYM_BREAK },
  73         { "e",            TERMSYM_SLASH },
  74         { "q",            TERMSYM_DQUOTE },
  75         { NULL,           0 }
  76 };
  77
  78 static  struct termenc    termenc2[] = {
  79         { "rB",           TERMSYM_RBRACK },
  80         { "lB",           TERMSYM_LBRACK },
  81         { "Lq",           TERMSYM_LDQUOTE },
  82         { "lq",           TERMSYM_LDQUOTE },
  83         { "Rq",           TERMSYM_RDQUOTE },
  84         { "rq",           TERMSYM_RDQUOTE },
  85         { "oq",           TERMSYM_LSQUOTE },
  86         { "aq",           TERMSYM_RSQUOTE },
  87
  88         { "<-",           TERMSYM_LARROW },
  89         { "->",           TERMSYM_RARROW },
  90         { "ua",           TERMSYM_UARROW },
  91         { "da",           TERMSYM_DARROW },
  92
  93         { "bu",           TERMSYM_BULLET },
  94         { "Ba",           TERMSYM_BAR },
  95         { "ba",           TERMSYM_BAR },
  96         { "co",           TERMSYM_COPY },
  97         { "Am",           TERMSYM_AMP },
  98
  99         { "Le",           TERMSYM_LE },
 100         { "<=",           TERMSYM_LE },
 101         { "Ge",           TERMSYM_GE },
 102         { "=>",           TERMSYM_GE },
 103         { "==",           TERMSYM_EQ },
 104         { "Ne",           TERMSYM_NEQ },
 105         { "!=",           TERMSYM_NEQ },
 106         { "Pm",           TERMSYM_PLUSMINUS },
 107         { "+-",           TERMSYM_PLUSMINUS },
 108         { "If",           TERMSYM_INF2 },
 109         { "if",           TERMSYM_INF },
 110         { "Na",           TERMSYM_NAN },
 111         { "na",           TERMSYM_NAN },
 112         { "**",           TERMSYM_ASTERISK },
 113         { "Gt",           TERMSYM_GT },
 114         { "Lt",           TERMSYM_LT },
 115
 116         { "aa",           TERMSYM_ACUTE },
 117         { "ga",           TERMSYM_GRAVE },
 118
 119         { "en",           TERMSYM_EN },
 120         { "em",           TERMSYM_EM },
 121
 122         { "Pi",           TERMSYM_PI },
 123         { NULL,           0 }
 124 };
 125
 126 static  struct termsym    termsym_ansi[] = {
 127         { "]", 1 },             /* TERMSYM_RBRACK */
 128         { "[", 1 },             /* TERMSYM_LBRACK */
 129         { "<-", 2 },            /* TERMSYM_LARROW */
 130         { "->", 2 },            /* TERMSYM_RARROW */
 131         { "^", 1 },             /* TERMSYM_UARROW */
 132         { "v", 1 },             /* TERMSYM_DARROW */
 133         { "`", 1 },             /* TERMSYM_LSQUOTE */
 134         { "\'", 1 },            /* TERMSYM_RSQUOTE */
 135         { "\'", 1 },            /* TERMSYM_SQUOTE */
 136         { "``", 2 },            /* TERMSYM_LDQUOTE */
 137         { "\'\'", 2 },          /* TERMSYM_RDQUOTE */
 138         { "\"", 1 },            /* TERMSYM_DQUOTE */
 139         { "<", 1 },             /* TERMSYM_LT */
 140         { ">", 1 },             /* TERMSYM_GT */
 141         { "<=", 2 },            /* TERMSYM_LE */
 142         { ">=", 2 },            /* TERMSYM_GE */
 143         { "==", 2 },            /* TERMSYM_EQ */
 144         { "!=", 2 },            /* TERMSYM_NEQ */
 145         { "\'", 1 },            /* TERMSYM_ACUTE */
 146         { "`", 1 },             /* TERMSYM_GRAVE */
 147         { "pi", 2 },            /* TERMSYM_PI */
 148         { "+=", 2 },            /* TERMSYM_PLUSMINUS */
 149         { "oo", 2 },            /* TERMSYM_INF */
 150         { "infinity", 8 },      /* TERMSYM_INF2 */
 151         { "NaN", 3 },           /* TERMSYM_NAN */
 152         { "|", 1 },             /* TERMSYM_BAR */
 153         { "o", 1 },             /* TERMSYM_BULLET */
 154         { "&", 1 },             /* TERMSYM_AMP */
 155         { "--", 2 },            /* TERMSYM_EM */
 156         { "-", 1 },             /* TERMSYM_EN */
 157         { "(C)", 3 },           /* TERMSYM_COPY */
 158         { "*", 1 },             /* TERMSYM_ASTERISK */
 159         { "\\", 1 },            /* TERMSYM_SLASH */
 160         { "-", 1 },             /* TERMSYM_HYPHEN */
 161         { " ", 1 },             /* TERMSYM_SPACE */
 162         { ".", 1 },             /* TERMSYM_PERIOD */
 163         { "", 0 },              /* TERMSYM_BREAK */
 164 };
 165
 166 static  const char        ansi_clear[]  = { 27, '[', '0', 'm' };
 167 static  const char        ansi_bold[]  = { 27, '[', '1', 'm' };
 168 static  const char        ansi_under[]  = { 27, '[', '4', 'm' };
 169
 170 static  struct termsym    termstyle_ansi[] = {
 171         { ansi_clear, 4 },
 172         { ansi_bold, 4 },
 173         { ansi_under, 4 }
 174 };
 175
 176
 177 int
 178 main(int argc, char *argv[])
 179 {
 180         struct mmain    *p;
 181         const struct mdoc *mdoc;
 182         struct termp     termp;
 183
 184         p = mmain_alloc();
 185
 186         if ( ! mmain_getopt(p, argc, argv, NULL, NULL, NULL, NULL))
 187                 mmain_exit(p, 1);
 188
 189         if (NULL == (mdoc = mmain_mdoc(p)))
 190                 mmain_exit(p, 1);
 191
 192         termp.maxrmargin = termp.rmargin = 78; /* XXX */
 193         termp.maxcols = 1024;
 194         termp.offset = termp.col = 0;
 195         termp.flags = TERMP_NOSPACE;
 196         termp.symtab = termsym_ansi;
 197         termp.styletab = termstyle_ansi;
 198
 199         if (NULL == (termp.buf = malloc(termp.maxcols)))
 200                 err(1, "malloc");
 201
 202         header(&termp, mdoc_meta(mdoc));
 203         body(&termp, NULL, mdoc_meta(mdoc), mdoc_node(mdoc));
 204         footer(&termp, mdoc_meta(mdoc));
 205
 206         free(termp.buf);
 207
 208         mmain_exit(p, 0);
 209         /* NOTREACHED */
 210 }
 211
 212
 213 /*
 214  * Flush a line of text.  A "line" is loosely defined as being something
 215  * that should be followed by a newline, regardless of whether it's
 216  * broken apart by newlines getting there.  A line can also be a
 217  * fragment of a columnar list.
 218  *
 219  * Specifically, a line is whatever's in p->buf of length p->col, which
 220  * is zeroed after this function returns.
 221  *
 222  * The variables TERMP_NOLPAD, TERMP_LITERAL and TERMP_NOBREAK are of
 223  * critical importance here.  Their behaviour follows:
 224  *
 225  *  - TERMP_NOLPAD: when beginning to write the line, don't left-pad the
 226  *    offset value.  This is useful when doing columnar lists where the
 227  *    prior column has right-padded.
 228  *
 229  *  - TERMP_NOBREAK: this is the most important and is used when making
 230  *    columns.  In short: don't print a newline and instead pad to the
 231  *    right margin.  Used in conjunction with TERMP_NOLPAD.
 232  *
 233  *  In-line line breaking:
 234  *
 235  *  If TERMP_NOBREAK is specified and the line overruns the right
 236  *  margin, it will break and pad-right to the right margin after
 237  *  writing.  If maxrmargin is violated, it will break and continue
 238  *  writing from the right-margin, which will lead to the above
 239  *  scenario upon exit.
 240  *
 241  *  Otherwise, the line will break at the right margin.  Extremely long
 242  *  lines will cause the system to emit a warning (TODO: hyphenate, if
 243  *  possible).
 244  */
 245 void
 246 flushln(struct termp *p)
 247 {
 248         size_t           i, j, vsz, vis, maxvis, mmax, bp;
 249
 250         /*
 251          * First, establish the maximum columns of "visible" content.
 252          * This is usually the difference between the right-margin and
 253          * an indentation, but can be, for tagged lists or columns, a
 254          * small set of values.
 255          */
 256
 257         assert(p->offset < p->rmargin);
 258         maxvis = p->rmargin - p->offset;
 259         mmax = p->maxrmargin - p->offset;
 260         bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
 261         vis = 0;
 262
 263         /*
 264          * If in the standard case (left-justified), then begin with our
 265          * indentation, otherwise (columns, etc.) just start spitting
 266          * out text.
 267          */
 268
 269         if ( ! (p->flags & TERMP_NOLPAD))
 270                 /* LINTED */
 271                 for (j = 0; j < p->offset; j++)
 272                         putchar(' ');
 273
 274         for (i = 0; i < p->col; i++) {
 275                 /*
 276                  * Count up visible word characters.  Control sequences
 277                  * (starting with the CSI) aren't counted.  A space
 278                  * generates a non-printing word, which is valid (the
 279                  * space is printed according to regular spacing rules).
 280                  */
 281
 282                 /* FIXME: make non-ANSI friendly. */
 283
 284                 /* LINTED */
 285                 for (j = i, vsz = 0; j < p->col; j++) {
 286                         if (isspace((int)p->buf[j]))
 287                                 break;
 288                         else if (27 == p->buf[j]) {
 289                                 assert(j + 4 <= p->col);
 290                                 j += 3;
 291                         } else
 292                                 vsz++;
 293                 }
 294
 295                 /*
 296                  * Do line-breaking.  If we're greater than our
 297                  * break-point and already in-line, break to the next
 298                  * line and start writing.  If we're at the line start,
 299                  * then write out the word (TODO: hyphenate) and break
 300                  * in a subsequent loop invocation.
 301                  */
 302
 303                 if ( ! (TERMP_NOBREAK & p->flags)) {
 304                         if (vis && vis + vsz > bp) {
 305                                 putchar('\n');
 306                                 for (j = 0; j < p->offset; j++)
 307                                         putchar(' ');
 308                                 vis = 0;
 309                         } else if (vis + vsz > bp)
 310                                 warnx("word breaks right margin");
 311
 312                         /* TODO: hyphenate. */
 313
 314                 } else {
 315                         if (vis && vis + vsz > bp) {
 316                                 putchar('\n');
 317                                 for (j = 0; j < p->rmargin; j++)
 318                                         putchar(' ');
 319                                 vis = p->rmargin - p->offset;
 320                         } else if (vis + vsz > bp)
 321                                 warnx("word breaks right margin");
 322
 323                         /* TODO: hyphenate. */
 324                 }
 325
 326                 /*
 327                  * Write out the word and a trailing space.  Omit the
 328                  * space if we're the last word in the line or beyond
 329                  * our breakpoint.
 330                  */
 331
 332                 for ( ; i < p->col; i++) {
 333                         if (isspace((int)p->buf[i]))
 334                                 break;
 335                         putchar(p->buf[i]);
 336                 }
 337                 vis += vsz;
 338                 if (i < p->col && vis <= bp) {
 339                         putchar(' ');
 340                         vis++;
 341                 }
 342         }
 343
 344         /*
 345          * If we've overstepped our maximum visible no-break space, then
 346          * cause a newline and offset at the right margin.
 347          */
 348
 349         if ((TERMP_NOBREAK & p->flags) && vis >= maxvis) {
 350                 putchar('\n');
 351                 for (i = 0; i < p->rmargin; i++)
 352                         putchar(' ');
 353                 p->col = 0;
 354                 return;
 355         }
 356
 357         /*
 358          * If we're not to right-marginalise it (newline), then instead
 359          * pad to the right margin and stay off.
 360          */
 361
 362         if (p->flags & TERMP_NOBREAK)
 363                 for ( ; vis < maxvis; vis++)
 364                         putchar(' ');
 365         else
 366                 putchar('\n');
 367
 368         p->col = 0;
 369 }
 370
 371
 372 /*
 373  * A newline only breaks an existing line; it won't assert vertical
 374  * space.  All data in the output buffer is flushed prior to the newline
 375  * assertion.
 376  */
 377 void
 378 newln(struct termp *p)
 379 {
 380
 381         p->flags |= TERMP_NOSPACE;
 382         if (0 == p->col) {
 383                 p->flags &= ~TERMP_NOLPAD;
 384                 return;
 385         }
 386         flushln(p);
 387         p->flags &= ~TERMP_NOLPAD;
 388 }
 389
 390
 391 /*
 392  * Asserts a vertical space (a full, empty line-break between lines).
 393  * Note that if used twice, this will cause two blank spaces and so on.
 394  * All data in the output buffer is flushed prior to the newline
 395  * assertion.
 396  */
 397 void
 398 vspace(struct termp *p)
 399 {
 400
 401         newln(p);
 402         putchar('\n');
 403 }
 404
 405
 406 /*
 407  * Break apart a word into "pwords" (partial-words, usually from
 408  * breaking up a phrase into individual words) and, eventually, put them
 409  * into the output buffer.  If we're a literal word, then don't break up
 410  * the word and put it verbatim into the output buffer.
 411  */
 412 void
 413 word(struct termp *p, const char *word)
 414 {
 415         size_t           i, j, len;
 416
 417         if (p->flags & TERMP_LITERAL) {
 418                 pword(p, word, strlen(word));
 419                 return;
 420         }
 421
 422         len = strlen(word);
 423         assert(len > 0);
 424
 425         if (mdoc_isdelim(word)) {
 426                 if ( ! (p->flags & TERMP_IGNDELIM))
 427                         p->flags |= TERMP_NOSPACE;
 428                 p->flags &= ~TERMP_IGNDELIM;
 429         }
 430
 431         /* LINTED */
 432         for (j = i = 0; i < len; i++) {
 433                 if ( ! isspace((int)word[i])) {
 434                         j++;
 435                         continue;
 436                 }
 437
 438                 /* Escaped spaces don't delimit... */
 439                 if (i > 0 && isspace((int)word[i]) &&
 440                                 '\\' == word[i - 1]) {
 441                         j++;
 442                         continue;
 443                 }
 444
 445                 if (0 == j)
 446                         continue;
 447                 assert(i >= j);
 448                 pword(p, &word[i - j], j);
 449                 j = 0;
 450         }
 451         if (j > 0) {
 452                 assert(i >= j);
 453                 pword(p, &word[i - j], j);
 454         }
 455 }
 456
 457
 458 /*
 459  * This is the main function for printing out nodes.  It's constituted
 460  * of PRE and POST functions, which correspond to prefix and infix
 461  * processing.  The termpair structure allows data to persist between
 462  * prefix and postfix invocations.
 463  */
 464 static void
 465 body(struct termp *p, struct termpair *ppair,
 466                 const struct mdoc_meta *meta,
 467                 const struct mdoc_node *node)
 468 {
 469         int              dochild;
 470         struct termpair  pair;
 471
 472         /* Pre-processing. */
 473
 474         dochild = 1;
 475         pair.ppair = ppair;
 476         pair.type = 0;
 477         pair.offset = pair.rmargin = 0;
 478         pair.flag = 0;
 479         pair.count = 0;
 480
 481         if (MDOC_TEXT != node->type) {
 482                 if (termacts[node->tok].pre)
 483                         if ( ! (*termacts[node->tok].pre)(p, &pair, meta, node))
 484                                 dochild = 0;
 485         } else /* MDOC_TEXT == node->type */
 486                 word(p, node->data.text.string);
 487
 488         /* Children. */
 489
 490         if (TERMPAIR_FLAG & pair.type)
 491                 p->flags |= pair.flag;
 492
 493         if (dochild && node->child)
 494                 body(p, &pair, meta, node->child);
 495
 496         if (TERMPAIR_FLAG & pair.type)
 497                 p->flags &= ~pair.flag;
 498
 499         /* Post-processing. */
 500
 501         if (MDOC_TEXT != node->type)
 502                 if (termacts[node->tok].post)
 503                         (*termacts[node->tok].post)(p, &pair, meta, node);
 504
 505         /* Siblings. */
 506
 507         if (node->next)
 508                 body(p, ppair, meta, node->next);
 509 }
 510
 511
 512 static void
 513 footer(struct termp *p, const struct mdoc_meta *meta)
 514 {
 515         struct tm       *tm;
 516         char            *buf, *os;
 517
 518         if (NULL == (buf = malloc(p->rmargin)))
 519                 err(1, "malloc");
 520         if (NULL == (os = malloc(p->rmargin)))
 521                 err(1, "malloc");
 522
 523         tm = localtime(&meta->date);
 524
 525 #ifdef __OpenBSD__
 526         if (NULL == strftime(buf, p->rmargin, "%B %d, %Y", tm))
 527 #else
 528         if (0 == strftime(buf, p->rmargin, "%B %d, %Y", tm))
 529 #endif
 530                 err(1, "strftime");
 531
 532         (void)strlcpy(os, meta->os, p->rmargin);
 533
 534         /*
 535          * This is /slightly/ different from regular groff output
 536          * because we don't have page numbers.  Print the following:
 537          *
 538          * OS                                            MDOCDATE
 539          */
 540
 541         vspace(p);
 542
 543         p->flags |= TERMP_NOSPACE | TERMP_NOBREAK;
 544         p->rmargin = p->maxrmargin - strlen(buf);
 545         p->offset = 0;
 546
 547         word(p, os);
 548         flushln(p);
 549
 550         p->flags |= TERMP_NOLPAD | TERMP_NOSPACE;
 551         p->offset = p->rmargin;
 552         p->rmargin = p->maxrmargin;
 553         p->flags &= ~TERMP_NOBREAK;
 554
 555         word(p, buf);
 556         flushln(p);
 557
 558         free(buf);
 559         free(os);
 560 }
 561
 562
 563 static void
 564 header(struct termp *p, const struct mdoc_meta *meta)
 565 {
 566         char            *buf, *title, *bufp, *vbuf;
 567         const char      *pp;
 568         struct utsname   uts;
 569
 570         p->rmargin = p->maxrmargin;
 571         p->offset = 0;
 572
 573         if (NULL == (buf = malloc(p->rmargin)))
 574                 err(1, "malloc");
 575         if (NULL == (title = malloc(p->rmargin)))
 576                 err(1, "malloc");
 577         if (NULL == (vbuf = malloc(p->rmargin)))
 578                 err(1, "malloc");
 579
 580         if (NULL == (pp = mdoc_vol2a(meta->vol))) {
 581                 switch (meta->msec) {
 582                 case (MSEC_1):
 583                         /* FALLTHROUGH */
 584                 case (MSEC_6):
 585                         /* FALLTHROUGH */
 586                 case (MSEC_7):
 587                         pp = mdoc_vol2a(VOL_URM);
 588                         break;
 589                 case (MSEC_8):
 590                         pp = mdoc_vol2a(VOL_SMM);
 591                         break;
 592                 case (MSEC_2):
 593                         /* FALLTHROUGH */
 594                 case (MSEC_3):
 595                         /* FALLTHROUGH */
 596                 case (MSEC_4):
 597                         /* FALLTHROUGH */
 598                 case (MSEC_5):
 599                         pp = mdoc_vol2a(VOL_PRM);
 600                         break;
 601                 case (MSEC_9):
 602                         pp = mdoc_vol2a(VOL_KM);
 603                         break;
 604                 default:
 605                         break;
 606                 }
 607         }
 608         vbuf[0] = 0;
 609
 610         if (pp) {
 611                 if (-1 == uname(&uts))
 612                         err(1, "uname");
 613                 (void)strlcat(vbuf, uts.sysname, p->rmargin);
 614                 (void)strlcat(vbuf, " ", p->rmargin);
 615         } else if (NULL == (pp = mdoc_msec2a(meta->msec)))
 616                 pp = mdoc_msec2a(MSEC_local);
 617
 618         (void)strlcat(vbuf, pp, p->rmargin);
 619
 620         /*
 621          * The header is strange.  It has three components, which are
 622          * really two with the first duplicated.  It goes like this:
 623          *
 624          * IDENTIFIER              TITLE                   IDENTIFIER
 625          *
 626          * The IDENTIFIER is NAME(SECTION), which is the command-name
 627          * (if given, or "unknown" if not) followed by the manual page
 628          * section.  These are given in `Dt'.  The TITLE is a free-form
 629          * string depending on the manual volume.  If not specified, it
 630          * switches on the manual section.
 631          */
 632
 633         if (mdoc_arch2a(meta->arch))
 634                 (void)snprintf(buf, p->rmargin, "%s (%s)",
 635                                 vbuf, mdoc_arch2a(meta->arch));
 636         else
 637                 (void)strlcpy(buf, vbuf, p->rmargin);
 638
 639         pp = mdoc_msec2a(meta->msec);
 640
 641         (void)snprintf(title, p->rmargin, "%s(%s)",
 642                         meta->title, pp ? pp : "");
 643
 644         for (bufp = title; *bufp; bufp++)
 645                 *bufp = toupper(*bufp);
 646
 647         p->offset = 0;
 648         p->rmargin = (p->maxrmargin - strlen(buf)) / 2;
 649         p->flags |= TERMP_NOBREAK | TERMP_NOSPACE;
 650
 651         word(p, title);
 652         flushln(p);
 653
 654         p->flags |= TERMP_NOLPAD | TERMP_NOSPACE;
 655         p->offset = p->rmargin;
 656         p->rmargin = p->maxrmargin - strlen(title);
 657
 658         word(p, buf);
 659         flushln(p);
 660
 661         p->offset = p->rmargin;
 662         p->rmargin = p->maxrmargin;
 663         p->flags &= ~TERMP_NOBREAK;
 664         p->flags |= TERMP_NOLPAD | TERMP_NOSPACE;
 665
 666         word(p, title);
 667         flushln(p);
 668
 669         p->rmargin = p->maxrmargin;
 670         p->offset = 0;
 671         p->flags &= ~TERMP_NOSPACE;
 672
 673         free(title);
 674         free(vbuf);
 675         free(buf);
 676 }
 677
 678
 679 /*
 680  * Determine the symbol indicated by an escape sequences, that is, one
 681  * starting with a backslash.  Once done, we pass this value into the
 682  * output buffer by way of the symbol table.
 683  */
 684 static void
 685 nescape(struct termp *p, const char *word, size_t len)
 686 {
 687         struct termenc  *enc;
 688
 689         switch (len) {
 690         case (1):
 691                 enc = termenc1;
 692                 break;
 693         case (2):
 694                 enc = termenc2;
 695                 break;
 696         default:
 697                 warnx("unsupported %zu-byte escape sequence", len);
 698                 return;
 699         }
 700
 701         for ( ; enc->enc; enc++)
 702                 if (0 == memcmp(enc->enc, word, len)) {
 703                         symbola(p, enc->sym);
 704                         return;
 705                 }
 706
 707         warnx("unsupported %zu-byte escape sequence", len);
 708 }
 709
 710
 711 /*
 712  * Handle an escape sequence: determine its length and pass it to the
 713  * escape-symbol look table.  Note that we assume mdoc(3) has validated
 714  * the escape sequence (we assert upon badly-formed escape sequences).
 715  */
 716 static void
 717 pescape(struct termp *p, const char *word, size_t *i, size_t len)
 718 {
 719         size_t           j;
 720
 721         (*i)++;
 722         assert(*i < len);
 723
 724         if ('(' == word[*i]) {
 725                 (*i)++;
 726                 assert(*i + 1 < len);
 727                 nescape(p, &word[*i], 2);
 728                 (*i)++;
 729                 return;
 730
 731         } else if ('*' == word[*i]) {
 732                 /* XXX - deprecated! */
 733                 (*i)++;
 734                 assert(*i < len);
 735                 switch (word[*i]) {
 736                 case ('('):
 737                         (*i)++;
 738                         assert(*i + 1 < len);
 739                         nescape(p, &word[*i], 2);
 740                         (*i)++;
 741                         return;
 742                 case ('['):
 743                         break;
 744                 default:
 745                         nescape(p, &word[*i], 1);
 746                         return;
 747                 }
 748
 749         } else if ('[' != word[*i]) {
 750                 nescape(p, &word[*i], 1);
 751                 return;
 752         }
 753
 754         (*i)++;
 755         for (j = 0; word[*i] && ']' != word[*i]; (*i)++, j++)
 756                 /* Loop... */ ;
 757
 758         assert(word[*i]);
 759         nescape(p, &word[*i - j], j);
 760 }
 761
 762
 763 /*
 764  * Handle pwords, partial words, which may be either a single word or a
 765  * phrase that cannot be broken down (such as a literal string).  This
 766  * handles word styling.
 767  */
 768 static void
 769 pword(struct termp *p, const char *word, size_t len)
 770 {
 771         size_t           i;
 772
 773         if ( ! (TERMP_NOSPACE & p->flags) &&
 774                         ! (TERMP_LITERAL & p->flags))
 775                 chara(p, ' ');
 776
 777         if ( ! (p->flags & TERMP_NONOSPACE))
 778                 p->flags &= ~TERMP_NOSPACE;
 779
 780         /*
 781          * XXX - if literal and underlining, this will underline the
 782          * spaces between literal words.
 783          */
 784
 785         if (p->flags & TERMP_BOLD)
 786                 stylea(p, TERMSTYLE_BOLD);
 787         if (p->flags & TERMP_UNDERLINE)
 788                 stylea(p, TERMSTYLE_UNDER);
 789
 790         for (i = 0; i < len; i++) {
 791                 if ('\\' == word[i]) {
 792                         pescape(p, word, &i, len);
 793                         continue;
 794                 }
 795                 chara(p, word[i]);
 796         }
 797
 798         if (p->flags & TERMP_BOLD ||
 799                         p->flags & TERMP_UNDERLINE)
 800                 stylea(p, TERMSTYLE_CLEAR);
 801 }
 802
 803
 804 /*
 805  * Add a symbol to the output line buffer.
 806  */
 807 static void
 808 symbola(struct termp *p, enum tsym sym)
 809 {
 810
 811         assert(p->symtab[sym].sym);
 812         stringa(p, p->symtab[sym].sym, p->symtab[sym].sz);
 813 }
 814
 815
 816 /*
 817  * Add a style to the output line buffer.
 818  */
 819 static void
 820 stylea(struct termp *p, enum tstyle style)
 821 {
 822
 823         assert(p->styletab[style].sym);
 824         stringa(p, p->styletab[style].sym, p->styletab[style].sz);
 825 }
 826
 827
 828 /*
 829  * Like chara() but for arbitrary-length buffers.  Resize the buffer by
 830  * a factor of two (if the buffer is less than that) or the buffer's
 831  * size.
 832  */
 833 static void
 834 stringa(struct termp *p, const char *c, size_t sz)
 835 {
 836         size_t           s;
 837
 838         if (0 == sz)
 839                 return;
 840
 841         s = sz > p->maxcols * 2 ? sz : p->maxcols * 2;
 842
 843         assert(c);
 844         if (p->col + sz >= p->maxcols) {
 845                 p->buf = realloc(p->buf, s);
 846                 if (NULL == p->buf)
 847                         err(1, "realloc");
 848                 p->maxcols = s;
 849         }
 850
 851         (void)memcpy(&p->buf[p->col], c, sz);
 852         p->col += sz;
 853 }
 854
 855
 856 /*
 857  * Insert a single character into the line-buffer.  If the buffer's
 858  * space is exceeded, then allocate more space by doubling the buffer
 859  * size.
 860  */
 861 static void
 862 chara(struct termp *p, char c)
 863 {
 864
 865         if (p->col + 1 >= p->maxcols) {
 866                 p->buf = realloc(p->buf, p->maxcols * 2);
 867                 if (NULL == p->buf)
 868                         err(1, "malloc");
 869                 p->maxcols *= 2;
 870         }
 871         p->buf[(p->col)++] = c;
 872 }