mdocterm.c

   1 /* $Id: mdocterm.c,v 1.28 2009/03/03 22:17:19 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 = 78; /* XXX */
 193         termp.rmargin = termp.maxrmargin;
 194         termp.maxcols = 1024;
 195         termp.offset = termp.col = 0;
 196         termp.flags = TERMP_NOSPACE;
 197         termp.symtab = termsym_ansi;
 198         termp.styletab = termstyle_ansi;
 199
 200         if (NULL == (termp.buf = malloc(termp.maxcols)))
 201                 err(1, "malloc");
 202
 203         header(&termp, mdoc_meta(mdoc));
 204         body(&termp, NULL, mdoc_meta(mdoc), mdoc_node(mdoc));
 205         footer(&termp, mdoc_meta(mdoc));
 206
 207         free(termp.buf);
 208
 209         mmain_exit(p, 0);
 210         /* NOTREACHED */
 211 }
 212
 213
 214 /*
 215  * Flush a line of text.  A "line" is loosely defined as being something
 216  * that should be followed by a newline, regardless of whether it's
 217  * broken apart by newlines getting there.  A line can also be a
 218  * fragment of a columnar list.
 219  *
 220  * Specifically, a line is whatever's in p->buf of length p->col, which
 221  * is zeroed after this function returns.
 222  *
 223  * The variables TERMP_NOLPAD, TERMP_LITERAL and TERMP_NOBREAK are of
 224  * critical importance here.  Their behaviour follows:
 225  *
 226  *  - TERMP_NOLPAD: when beginning to write the line, don't left-pad the
 227  *    offset value.  This is useful when doing columnar lists where the
 228  *    prior column has right-padded.
 229  *
 230  *  - TERMP_NOBREAK: this is the most important and is used when making
 231  *    columns.  In short: don't print a newline and instead pad to the
 232  *    right margin.  Used in conjunction with TERMP_NOLPAD.
 233  *
 234  *  In-line line breaking:
 235  *
 236  *  If TERMP_NOBREAK is specified and the line overruns the right
 237  *  margin, it will break and pad-right to the right margin after
 238  *  writing.  If maxrmargin is violated, it will break and continue
 239  *  writing from the right-margin, which will lead to the above
 240  *  scenario upon exit.
 241  *
 242  *  Otherwise, the line will break at the right margin.  Extremely long
 243  *  lines will cause the system to emit a warning (TODO: hyphenate, if
 244  *  possible).
 245  */
 246 void
 247 flushln(struct termp *p)
 248 {
 249         size_t           i, j, vsz, vis, maxvis, mmax, bp;
 250
 251         /*
 252          * First, establish the maximum columns of "visible" content.
 253          * This is usually the difference between the right-margin and
 254          * an indentation, but can be, for tagged lists or columns, a
 255          * small set of values.
 256          */
 257
 258         assert(p->offset < p->rmargin);
 259         maxvis = p->rmargin - p->offset;
 260         mmax = p->maxrmargin - p->offset;
 261         bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
 262         vis = 0;
 263
 264         /*
 265          * If in the standard case (left-justified), then begin with our
 266          * indentation, otherwise (columns, etc.) just start spitting
 267          * out text.
 268          */
 269
 270         if ( ! (p->flags & TERMP_NOLPAD))
 271                 /* LINTED */
 272                 for (j = 0; j < p->offset; j++)
 273                         putchar(' ');
 274
 275         for (i = 0; i < p->col; i++) {
 276                 /*
 277                  * Count up visible word characters.  Control sequences
 278                  * (starting with the CSI) aren't counted.  A space
 279                  * generates a non-printing word, which is valid (the
 280                  * space is printed according to regular spacing rules).
 281                  */
 282
 283                 /* FIXME: make non-ANSI friendly. */
 284
 285                 /* LINTED */
 286                 for (j = i, vsz = 0; j < p->col; j++) {
 287                         if (isspace((int)p->buf[j]))
 288                                 break;
 289                         else if (27 == p->buf[j]) {
 290                                 assert(j + 4 <= p->col);
 291                                 j += 3;
 292                         } else
 293                                 vsz++;
 294                 }
 295
 296                 /*
 297                  * Do line-breaking.  If we're greater than our
 298                  * break-point and already in-line, break to the next
 299                  * line and start writing.  If we're at the line start,
 300                  * then write out the word (TODO: hyphenate) and break
 301                  * in a subsequent loop invocation.
 302                  */
 303
 304                 if ( ! (TERMP_NOBREAK & p->flags)) {
 305                         if (vis && vis + vsz > bp) {
 306                                 putchar('\n');
 307                                 for (j = 0; j < p->offset; j++)
 308                                         putchar(' ');
 309                                 vis = 0;
 310                         } else if (vis + vsz > bp)
 311                                 warnx("word breaks right margin");
 312
 313                         /* TODO: hyphenate. */
 314
 315                 } else {
 316                         if (vis && vis + vsz > bp) {
 317                                 putchar('\n');
 318                                 for (j = 0; j < p->rmargin; j++)
 319                                         putchar(' ');
 320                                 vis = p->rmargin;
 321                         } else if (vis + vsz > bp)
 322                                 warnx("word breaks right margin");
 323
 324                         /* TODO: hyphenate. */
 325                 }
 326
 327                 /*
 328                  * Write out the word and a trailing space.  Omit the
 329                  * space if we're the last word in the line or beyond
 330                  * our breakpoint.
 331                  */
 332
 333                 for ( ; i < p->col; i++) {
 334                         if (isspace((int)p->buf[i]))
 335                                 break;
 336                         putchar(p->buf[i]);
 337                 }
 338                 vis += vsz;
 339                 if (i < p->col && vis <= bp) {
 340                         putchar(' ');
 341                         vis++;
 342                 }
 343         }
 344
 345         /*
 346          * If we've overstepped our maximum visible no-break space, then
 347          * cause a newline and offset at the right margin.
 348          */
 349
 350         if ((TERMP_NOBREAK & p->flags) && vis >= maxvis) {
 351                 putchar('\n');
 352                 for (i = 0; i < p->rmargin; i++)
 353                         putchar(' ');
 354                 p->col = 0;
 355                 return;
 356         }
 357
 358         /*
 359          * If we're not to right-marginalise it (newline), then instead
 360          * pad to the right margin and stay off.
 361          */
 362
 363         if (p->flags & TERMP_NOBREAK)
 364                 for ( ; vis < maxvis; vis++)
 365                         putchar(' ');
 366         else
 367                 putchar('\n');
 368
 369         p->col = 0;
 370 }
 371
 372
 373 /*
 374  * A newline only breaks an existing line; it won't assert vertical
 375  * space.  All data in the output buffer is flushed prior to the newline
 376  * assertion.
 377  */
 378 void
 379 newln(struct termp *p)
 380 {
 381
 382         p->flags |= TERMP_NOSPACE;
 383         if (0 == p->col) {
 384                 p->flags &= ~TERMP_NOLPAD;
 385                 return;
 386         }
 387         flushln(p);
 388         p->flags &= ~TERMP_NOLPAD;
 389 }
 390
 391
 392 /*
 393  * Asserts a vertical space (a full, empty line-break between lines).
 394  * Note that if used twice, this will cause two blank spaces and so on.
 395  * All data in the output buffer is flushed prior to the newline
 396  * assertion.
 397  */
 398 void
 399 vspace(struct termp *p)
 400 {
 401
 402         newln(p);
 403         putchar('\n');
 404 }
 405
 406
 407 /*
 408  * Break apart a word into "pwords" (partial-words, usually from
 409  * breaking up a phrase into individual words) and, eventually, put them
 410  * into the output buffer.  If we're a literal word, then don't break up
 411  * the word and put it verbatim into the output buffer.
 412  */
 413 void
 414 word(struct termp *p, const char *word)
 415 {
 416         size_t           i, j, len;
 417
 418         if (p->flags & TERMP_LITERAL) {
 419                 pword(p, word, strlen(word));
 420                 return;
 421         }
 422
 423         len = strlen(word);
 424         assert(len > 0);
 425
 426         if (mdoc_isdelim(word)) {
 427                 if ( ! (p->flags & TERMP_IGNDELIM))
 428                         p->flags |= TERMP_NOSPACE;
 429                 p->flags &= ~TERMP_IGNDELIM;
 430         }
 431
 432         /* LINTED */
 433         for (j = i = 0; i < len; i++) {
 434                 if ( ! isspace((int)word[i])) {
 435                         j++;
 436                         continue;
 437                 }
 438
 439                 /* Escaped spaces don't delimit... */
 440                 if (i > 0 && isspace((int)word[i]) &&
 441                                 '\\' == word[i - 1]) {
 442                         j++;
 443                         continue;
 444                 }
 445
 446                 if (0 == j)
 447                         continue;
 448                 assert(i >= j);
 449                 pword(p, &word[i - j], j);
 450                 j = 0;
 451         }
 452         if (j > 0) {
 453                 assert(i >= j);
 454                 pword(p, &word[i - j], j);
 455         }
 456 }
 457
 458
 459 /*
 460  * This is the main function for printing out nodes.  It's constituted
 461  * of PRE and POST functions, which correspond to prefix and infix
 462  * processing.  The termpair structure allows data to persist between
 463  * prefix and postfix invocations.
 464  */
 465 static void
 466 body(struct termp *p, struct termpair *ppair,
 467                 const struct mdoc_meta *meta,
 468                 const struct mdoc_node *node)
 469 {
 470         int              dochild;
 471         struct termpair  pair;
 472
 473         /* Pre-processing. */
 474
 475         dochild = 1;
 476         pair.ppair = ppair;
 477         pair.type = 0;
 478         pair.offset = pair.rmargin = 0;
 479         pair.flag = 0;
 480         pair.count = 0;
 481
 482         if (MDOC_TEXT != node->type) {
 483                 if (termacts[node->tok].pre)
 484                         if ( ! (*termacts[node->tok].pre)(p, &pair, meta, node))
 485                                 dochild = 0;
 486         } else /* MDOC_TEXT == node->type */
 487                 word(p, node->data.text.string);
 488
 489         /* Children. */
 490
 491         if (TERMPAIR_FLAG & pair.type)
 492                 p->flags |= pair.flag;
 493
 494         if (dochild && node->child)
 495                 body(p, &pair, meta, node->child);
 496
 497         if (TERMPAIR_FLAG & pair.type)
 498                 p->flags &= ~pair.flag;
 499
 500         /* Post-processing. */
 501
 502         if (MDOC_TEXT != node->type)
 503                 if (termacts[node->tok].post)
 504                         (*termacts[node->tok].post)(p, &pair, meta, node);
 505
 506         /* Siblings. */
 507
 508         if (node->next)
 509                 body(p, ppair, meta, node->next);
 510 }
 511
 512
 513 static void
 514 footer(struct termp *p, const struct mdoc_meta *meta)
 515 {
 516         struct tm       *tm;
 517         char            *buf, *os;
 518
 519         if (NULL == (buf = malloc(p->rmargin)))
 520                 err(1, "malloc");
 521         if (NULL == (os = malloc(p->rmargin)))
 522                 err(1, "malloc");
 523
 524         tm = localtime(&meta->date);
 525
 526 #ifdef __OpenBSD__
 527         if (NULL == strftime(buf, p->rmargin, "%B %d, %Y", tm))
 528 #else
 529         if (0 == strftime(buf, p->rmargin, "%B %d, %Y", tm))
 530 #endif
 531                 err(1, "strftime");
 532
 533         (void)strlcpy(os, meta->os, p->rmargin);
 534
 535         /*
 536          * This is /slightly/ different from regular groff output
 537          * because we don't have page numbers.  Print the following:
 538          *
 539          * OS                                            MDOCDATE
 540          */
 541
 542         vspace(p);
 543
 544         p->flags |= TERMP_NOSPACE | TERMP_NOBREAK;
 545         p->rmargin = p->maxrmargin - strlen(buf);
 546         p->offset = 0;
 547
 548         word(p, os);
 549         flushln(p);
 550
 551         p->flags |= TERMP_NOLPAD | TERMP_NOSPACE;
 552         p->offset = p->rmargin;
 553         p->rmargin = p->maxrmargin;
 554         p->flags &= ~TERMP_NOBREAK;
 555
 556         word(p, buf);
 557         flushln(p);
 558
 559         free(buf);
 560         free(os);
 561 }
 562
 563
 564 static void
 565 header(struct termp *p, const struct mdoc_meta *meta)
 566 {
 567         char            *buf, *title, *bufp, *vbuf;
 568         const char      *pp;
 569         struct utsname   uts;
 570
 571         p->rmargin = p->maxrmargin;
 572         p->offset = 0;
 573
 574         if (NULL == (buf = malloc(p->rmargin)))
 575                 err(1, "malloc");
 576         if (NULL == (title = malloc(p->rmargin)))
 577                 err(1, "malloc");
 578         if (NULL == (vbuf = malloc(p->rmargin)))
 579                 err(1, "malloc");
 580
 581         if (NULL == (pp = mdoc_vol2a(meta->vol))) {
 582                 switch (meta->msec) {
 583                 case (MSEC_1):
 584                         /* FALLTHROUGH */
 585                 case (MSEC_6):
 586                         /* FALLTHROUGH */
 587                 case (MSEC_7):
 588                         pp = mdoc_vol2a(VOL_URM);
 589                         break;
 590                 case (MSEC_8):
 591                         pp = mdoc_vol2a(VOL_SMM);
 592                         break;
 593                 case (MSEC_2):
 594                         /* FALLTHROUGH */
 595                 case (MSEC_3):
 596                         /* FALLTHROUGH */
 597                 case (MSEC_4):
 598                         /* FALLTHROUGH */
 599                 case (MSEC_5):
 600                         pp = mdoc_vol2a(VOL_PRM);
 601                         break;
 602                 case (MSEC_9):
 603                         pp = mdoc_vol2a(VOL_KM);
 604                         break;
 605                 default:
 606                         break;
 607                 }
 608         }
 609         vbuf[0] = 0;
 610
 611         if (pp) {
 612                 if (-1 == uname(&uts))
 613                         err(1, "uname");
 614                 (void)strlcat(vbuf, uts.sysname, p->rmargin);
 615                 (void)strlcat(vbuf, " ", p->rmargin);
 616         } else if (NULL == (pp = mdoc_msec2a(meta->msec)))
 617                 pp = mdoc_msec2a(MSEC_local);
 618
 619         (void)strlcat(vbuf, pp, p->rmargin);
 620
 621         /*
 622          * The header is strange.  It has three components, which are
 623          * really two with the first duplicated.  It goes like this:
 624          *
 625          * IDENTIFIER              TITLE                   IDENTIFIER
 626          *
 627          * The IDENTIFIER is NAME(SECTION), which is the command-name
 628          * (if given, or "unknown" if not) followed by the manual page
 629          * section.  These are given in `Dt'.  The TITLE is a free-form
 630          * string depending on the manual volume.  If not specified, it
 631          * switches on the manual section.
 632          */
 633
 634         if (mdoc_arch2a(meta->arch))
 635                 (void)snprintf(buf, p->rmargin, "%s (%s)",
 636                                 vbuf, mdoc_arch2a(meta->arch));
 637         else
 638                 (void)strlcpy(buf, vbuf, p->rmargin);
 639
 640         pp = mdoc_msec2a(meta->msec);
 641
 642         (void)snprintf(title, p->rmargin, "%s(%s)",
 643                         meta->title, pp ? pp : "");
 644
 645         for (bufp = title; *bufp; bufp++)
 646                 *bufp = toupper(*bufp);
 647
 648         p->offset = 0;
 649         p->rmargin = (p->maxrmargin - strlen(buf)) / 2;
 650         p->flags |= TERMP_NOBREAK | TERMP_NOSPACE;
 651
 652         word(p, title);
 653         flushln(p);
 654
 655         p->flags |= TERMP_NOLPAD | TERMP_NOSPACE;
 656         p->offset = p->rmargin;
 657         p->rmargin = p->maxrmargin - strlen(title);
 658
 659         word(p, buf);
 660         flushln(p);
 661
 662         p->offset = p->rmargin;
 663         p->rmargin = p->maxrmargin;
 664         p->flags &= ~TERMP_NOBREAK;
 665         p->flags |= TERMP_NOLPAD | TERMP_NOSPACE;
 666
 667         word(p, title);
 668         flushln(p);
 669
 670         p->rmargin = p->maxrmargin;
 671         p->offset = 0;
 672         p->flags &= ~TERMP_NOSPACE;
 673
 674         free(title);
 675         free(vbuf);
 676         free(buf);
 677 }
 678
 679
 680 /*
 681  * Determine the symbol indicated by an escape sequences, that is, one
 682  * starting with a backslash.  Once done, we pass this value into the
 683  * output buffer by way of the symbol table.
 684  */
 685 static void
 686 nescape(struct termp *p, const char *word, size_t len)
 687 {
 688         struct termenc  *enc;
 689
 690         switch (len) {
 691         case (1):
 692                 enc = termenc1;
 693                 break;
 694         case (2):
 695                 enc = termenc2;
 696                 break;
 697         default:
 698                 warnx("unsupported %zu-byte escape sequence", len);
 699                 return;
 700         }
 701
 702         for ( ; enc->enc; enc++)
 703                 if (0 == memcmp(enc->enc, word, len)) {
 704                         symbola(p, enc->sym);
 705                         return;
 706                 }
 707
 708         warnx("unsupported %zu-byte escape sequence", len);
 709 }
 710
 711
 712 /*
 713  * Handle an escape sequence: determine its length and pass it to the
 714  * escape-symbol look table.  Note that we assume mdoc(3) has validated
 715  * the escape sequence (we assert upon badly-formed escape sequences).
 716  */
 717 static void
 718 pescape(struct termp *p, const char *word, size_t *i, size_t len)
 719 {
 720         size_t           j;
 721
 722         (*i)++;
 723         assert(*i < len);
 724
 725         if ('(' == word[*i]) {
 726                 (*i)++;
 727                 assert(*i + 1 < len);
 728                 nescape(p, &word[*i], 2);
 729                 (*i)++;
 730                 return;
 731
 732         } else if ('*' == word[*i]) {
 733                 /* XXX - deprecated! */
 734                 (*i)++;
 735                 assert(*i < len);
 736                 switch (word[*i]) {
 737                 case ('('):
 738                         (*i)++;
 739                         assert(*i + 1 < len);
 740                         nescape(p, &word[*i], 2);
 741                         (*i)++;
 742                         return;
 743                 case ('['):
 744                         break;
 745                 default:
 746                         nescape(p, &word[*i], 1);
 747                         return;
 748                 }
 749
 750         } else if ('[' != word[*i]) {
 751                 nescape(p, &word[*i], 1);
 752                 return;
 753         }
 754
 755         (*i)++;
 756         for (j = 0; word[*i] && ']' != word[*i]; (*i)++, j++)
 757                 /* Loop... */ ;
 758
 759         assert(word[*i]);
 760         nescape(p, &word[*i - j], j);
 761 }
 762
 763
 764 /*
 765  * Handle pwords, partial words, which may be either a single word or a
 766  * phrase that cannot be broken down (such as a literal string).  This
 767  * handles word styling.
 768  */
 769 static void
 770 pword(struct termp *p, const char *word, size_t len)
 771 {
 772         size_t           i;
 773
 774         if ( ! (TERMP_NOSPACE & p->flags) &&
 775                         ! (TERMP_LITERAL & p->flags))
 776                 chara(p, ' ');
 777
 778         if ( ! (p->flags & TERMP_NONOSPACE))
 779                 p->flags &= ~TERMP_NOSPACE;
 780
 781         /*
 782          * XXX - if literal and underlining, this will underline the
 783          * spaces between literal words.
 784          */
 785
 786         if (p->flags & TERMP_BOLD)
 787                 stylea(p, TERMSTYLE_BOLD);
 788         if (p->flags & TERMP_UNDERLINE)
 789                 stylea(p, TERMSTYLE_UNDER);
 790
 791         for (i = 0; i < len; i++) {
 792                 if ('\\' == word[i]) {
 793                         pescape(p, word, &i, len);
 794                         continue;
 795                 }
 796                 chara(p, word[i]);
 797         }
 798
 799         if (p->flags & TERMP_BOLD ||
 800                         p->flags & TERMP_UNDERLINE)
 801                 stylea(p, TERMSTYLE_CLEAR);
 802 }
 803
 804
 805 /*
 806  * Add a symbol to the output line buffer.
 807  */
 808 static void
 809 symbola(struct termp *p, enum tsym sym)
 810 {
 811
 812         assert(p->symtab[sym].sym);
 813         stringa(p, p->symtab[sym].sym, p->symtab[sym].sz);
 814 }
 815
 816
 817 /*
 818  * Add a style to the output line buffer.
 819  */
 820 static void
 821 stylea(struct termp *p, enum tstyle style)
 822 {
 823
 824         assert(p->styletab[style].sym);
 825         stringa(p, p->styletab[style].sym, p->styletab[style].sz);
 826 }
 827
 828
 829 /*
 830  * Like chara() but for arbitrary-length buffers.  Resize the buffer by
 831  * a factor of two (if the buffer is less than that) or the buffer's
 832  * size.
 833  */
 834 static void
 835 stringa(struct termp *p, const char *c, size_t sz)
 836 {
 837         size_t           s;
 838
 839         if (0 == sz)
 840                 return;
 841
 842         s = sz > p->maxcols * 2 ? sz : p->maxcols * 2;
 843
 844         assert(c);
 845         if (p->col + sz >= p->maxcols) {
 846                 p->buf = realloc(p->buf, s);
 847                 if (NULL == p->buf)
 848                         err(1, "realloc");
 849                 p->maxcols = s;
 850         }
 851
 852         (void)memcpy(&p->buf[p->col], c, sz);
 853         p->col += sz;
 854 }
 855
 856
 857 /*
 858  * Insert a single character into the line-buffer.  If the buffer's
 859  * space is exceeded, then allocate more space by doubling the buffer
 860  * size.
 861  */
 862 static void
 863 chara(struct termp *p, char c)
 864 {
 865
 866         if (p->col + 1 >= p->maxcols) {
 867                 p->buf = realloc(p->buf, p->maxcols * 2);
 868                 if (NULL == p->buf)
 869                         err(1, "malloc");
 870                 p->maxcols *= 2;
 871         }
 872         p->buf[(p->col)++] = c;
 873 }