mdocterm.c

   1 /* $Id: mdocterm.c,v 1.31 2009/03/04 14:13:05 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         { "ra",           TERMSYM_RANGLE },
  82         { "la",           TERMSYM_LANGLE },
  83         { "Lq",           TERMSYM_LDQUOTE },
  84         { "lq",           TERMSYM_LDQUOTE },
  85         { "Rq",           TERMSYM_RDQUOTE },
  86         { "rq",           TERMSYM_RDQUOTE },
  87         { "oq",           TERMSYM_LSQUOTE },
  88         { "aq",           TERMSYM_RSQUOTE },
  89
  90         { "<-",           TERMSYM_LARROW },
  91         { "->",           TERMSYM_RARROW },
  92         { "ua",           TERMSYM_UARROW },
  93         { "da",           TERMSYM_DARROW },
  94
  95         { "bu",           TERMSYM_BULLET },
  96         { "Ba",           TERMSYM_BAR },
  97         { "ba",           TERMSYM_BAR },
  98         { "co",           TERMSYM_COPY },
  99         { "Am",           TERMSYM_AMP },
 100
 101         { "Le",           TERMSYM_LE },
 102         { "<=",           TERMSYM_LE },
 103         { "Ge",           TERMSYM_GE },
 104         { ">=",           TERMSYM_GE },
 105         { "==",           TERMSYM_EQ },
 106         { "Ne",           TERMSYM_NEQ },
 107         { "!=",           TERMSYM_NEQ },
 108         { "Pm",           TERMSYM_PLUSMINUS },
 109         { "+-",           TERMSYM_PLUSMINUS },
 110         { "If",           TERMSYM_INF2 },
 111         { "if",           TERMSYM_INF },
 112         { "Na",           TERMSYM_NAN },
 113         { "na",           TERMSYM_NAN },
 114         { "**",           TERMSYM_ASTERISK },
 115         { "Gt",           TERMSYM_GT },
 116         { "Lt",           TERMSYM_LT },
 117
 118         { "aa",           TERMSYM_ACUTE },
 119         { "ga",           TERMSYM_GRAVE },
 120
 121         { "en",           TERMSYM_EN },
 122         { "em",           TERMSYM_EM },
 123
 124         { "Pi",           TERMSYM_PI },
 125         { NULL,           0 }
 126 };
 127
 128 static  struct termsym    termsym_ansi[] = {
 129         { "]", 1 },             /* TERMSYM_RBRACK */
 130         { "[", 1 },             /* TERMSYM_LBRACK */
 131         { "<-", 2 },            /* TERMSYM_LARROW */
 132         { "->", 2 },            /* TERMSYM_RARROW */
 133         { "^", 1 },             /* TERMSYM_UARROW */
 134         { "v", 1 },             /* TERMSYM_DARROW */
 135         { "`", 1 },             /* TERMSYM_LSQUOTE */
 136         { "\'", 1 },            /* TERMSYM_RSQUOTE */
 137         { "\'", 1 },            /* TERMSYM_SQUOTE */
 138         { "``", 2 },            /* TERMSYM_LDQUOTE */
 139         { "\'\'", 2 },          /* TERMSYM_RDQUOTE */
 140         { "\"", 1 },            /* TERMSYM_DQUOTE */
 141         { "<", 1 },             /* TERMSYM_LT */
 142         { ">", 1 },             /* TERMSYM_GT */
 143         { "<=", 2 },            /* TERMSYM_LE */
 144         { ">=", 2 },            /* TERMSYM_GE */
 145         { "==", 2 },            /* TERMSYM_EQ */
 146         { "!=", 2 },            /* TERMSYM_NEQ */
 147         { "\'", 1 },            /* TERMSYM_ACUTE */
 148         { "`", 1 },             /* TERMSYM_GRAVE */
 149         { "pi", 2 },            /* TERMSYM_PI */
 150         { "+=", 2 },            /* TERMSYM_PLUSMINUS */
 151         { "oo", 2 },            /* TERMSYM_INF */
 152         { "infinity", 8 },      /* TERMSYM_INF2 */
 153         { "NaN", 3 },           /* TERMSYM_NAN */
 154         { "|", 1 },             /* TERMSYM_BAR */
 155         { "o", 1 },             /* TERMSYM_BULLET */
 156         { "&", 1 },             /* TERMSYM_AMP */
 157         { "--", 2 },            /* TERMSYM_EM */
 158         { "-", 1 },             /* TERMSYM_EN */
 159         { "(C)", 3 },           /* TERMSYM_COPY */
 160         { "*", 1 },             /* TERMSYM_ASTERISK */
 161         { "\\", 1 },            /* TERMSYM_SLASH */
 162         { "-", 1 },             /* TERMSYM_HYPHEN */
 163         { " ", 1 },             /* TERMSYM_SPACE */
 164         { ".", 1 },             /* TERMSYM_PERIOD */
 165         { "", 0 },              /* TERMSYM_BREAK */
 166         { "<", 1 },             /* TERMSYM_LANGLE */
 167         { ">", 1 },             /* TERMSYM_RANGLE */
 168 };
 169
 170 static  const char        ansi_clear[]  = { 27, '[', '0', 'm' };
 171 static  const char        ansi_bold[]  = { 27, '[', '1', 'm' };
 172 static  const char        ansi_under[]  = { 27, '[', '4', 'm' };
 173
 174 static  struct termsym    termstyle_ansi[] = {
 175         { ansi_clear, 4 },
 176         { ansi_bold, 4 },
 177         { ansi_under, 4 }
 178 };
 179
 180
 181 int
 182 main(int argc, char *argv[])
 183 {
 184         struct mmain    *p;
 185         const struct mdoc *mdoc;
 186         struct termp     termp;
 187
 188         p = mmain_alloc();
 189
 190         if ( ! mmain_getopt(p, argc, argv, NULL, NULL, NULL, NULL))
 191                 mmain_exit(p, 1);
 192
 193         if (NULL == (mdoc = mmain_mdoc(p)))
 194                 mmain_exit(p, 1);
 195
 196         termp.maxrmargin = termp.rmargin = 78; /* XXX */
 197         termp.maxcols = 1024;
 198         termp.offset = termp.col = 0;
 199         termp.flags = TERMP_NOSPACE;
 200         termp.symtab = termsym_ansi;
 201         termp.styletab = termstyle_ansi;
 202
 203         if (NULL == (termp.buf = malloc(termp.maxcols)))
 204                 err(1, "malloc");
 205
 206         header(&termp, mdoc_meta(mdoc));
 207         body(&termp, NULL, mdoc_meta(mdoc), mdoc_node(mdoc));
 208         footer(&termp, mdoc_meta(mdoc));
 209
 210         free(termp.buf);
 211
 212         mmain_exit(p, 0);
 213         /* NOTREACHED */
 214 }
 215
 216
 217 /*
 218  * Flush a line of text.  A "line" is loosely defined as being something
 219  * that should be followed by a newline, regardless of whether it's
 220  * broken apart by newlines getting there.  A line can also be a
 221  * fragment of a columnar list.
 222  *
 223  * Specifically, a line is whatever's in p->buf of length p->col, which
 224  * is zeroed after this function returns.
 225  *
 226  * The variables TERMP_NOLPAD, TERMP_LITERAL and TERMP_NOBREAK are of
 227  * critical importance here.  Their behaviour follows:
 228  *
 229  *  - TERMP_NOLPAD: when beginning to write the line, don't left-pad the
 230  *    offset value.  This is useful when doing columnar lists where the
 231  *    prior column has right-padded.
 232  *
 233  *  - TERMP_NOBREAK: this is the most important and is used when making
 234  *    columns.  In short: don't print a newline and instead pad to the
 235  *    right margin.  Used in conjunction with TERMP_NOLPAD.
 236  *
 237  *  In-line line breaking:
 238  *
 239  *  If TERMP_NOBREAK is specified and the line overruns the right
 240  *  margin, it will break and pad-right to the right margin after
 241  *  writing.  If maxrmargin is violated, it will break and continue
 242  *  writing from the right-margin, which will lead to the above
 243  *  scenario upon exit.
 244  *
 245  *  Otherwise, the line will break at the right margin.  Extremely long
 246  *  lines will cause the system to emit a warning (TODO: hyphenate, if
 247  *  possible).
 248  */
 249 void
 250 flushln(struct termp *p)
 251 {
 252         size_t           i, j, vsz, vis, maxvis, mmax, bp;
 253
 254         /*
 255          * First, establish the maximum columns of "visible" content.
 256          * This is usually the difference between the right-margin and
 257          * an indentation, but can be, for tagged lists or columns, a
 258          * small set of values.
 259          */
 260
 261         assert(p->offset < p->rmargin);
 262         maxvis = p->rmargin - p->offset;
 263         mmax = p->maxrmargin - p->offset;
 264         bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
 265         vis = 0;
 266
 267         /*
 268          * If in the standard case (left-justified), then begin with our
 269          * indentation, otherwise (columns, etc.) just start spitting
 270          * out text.
 271          */
 272
 273         if ( ! (p->flags & TERMP_NOLPAD))
 274                 /* LINTED */
 275                 for (j = 0; j < p->offset; j++)
 276                         putchar(' ');
 277
 278         for (i = 0; i < p->col; i++) {
 279                 /*
 280                  * Count up visible word characters.  Control sequences
 281                  * (starting with the CSI) aren't counted.  A space
 282                  * generates a non-printing word, which is valid (the
 283                  * space is printed according to regular spacing rules).
 284                  */
 285
 286                 /* FIXME: make non-ANSI friendly. */
 287
 288                 /* LINTED */
 289                 for (j = i, vsz = 0; j < p->col; j++) {
 290                         if (isspace((int)p->buf[j]))
 291                                 break;
 292                         else if (27 == p->buf[j]) {
 293                                 assert(j + 4 <= p->col);
 294                                 j += 3;
 295                         } else
 296                                 vsz++;
 297                 }
 298
 299                 /*
 300                  * Do line-breaking.  If we're greater than our
 301                  * break-point and already in-line, break to the next
 302                  * line and start writing.  If we're at the line start,
 303                  * then write out the word (TODO: hyphenate) and break
 304                  * in a subsequent loop invocation.
 305                  */
 306
 307                 if ( ! (TERMP_NOBREAK & p->flags)) {
 308                         if (vis && vis + vsz > bp) {
 309                                 putchar('\n');
 310                                 for (j = 0; j < p->offset; j++)
 311                                         putchar(' ');
 312                                 vis = 0;
 313                         } else if (vis + vsz > bp)
 314                                 warnx("word breaks right margin");
 315
 316                         /* TODO: hyphenate. */
 317
 318                 } else {
 319                         if (vis && vis + vsz > bp) {
 320                                 putchar('\n');
 321                                 for (j = 0; j < p->rmargin; j++)
 322                                         putchar(' ');
 323                                 vis = p->rmargin - p->offset;
 324                         } else if (vis + vsz > bp)
 325                                 warnx("word breaks right margin");
 326
 327                         /* TODO: hyphenate. */
 328                 }
 329
 330                 /*
 331                  * Write out the word and a trailing space.  Omit the
 332                  * space if we're the last word in the line or beyond
 333                  * our breakpoint.
 334                  */
 335
 336                 for ( ; i < p->col; i++) {
 337                         if (isspace((int)p->buf[i]))
 338                                 break;
 339                         putchar(p->buf[i]);
 340                 }
 341                 vis += vsz;
 342                 if (i < p->col && vis <= bp) {
 343                         putchar(' ');
 344                         vis++;
 345                 }
 346         }
 347
 348         /*
 349          * If we've overstepped our maximum visible no-break space, then
 350          * cause a newline and offset at the right margin.
 351          */
 352
 353         if ((TERMP_NOBREAK & p->flags) && vis >= maxvis) {
 354                 putchar('\n');
 355                 for (i = 0; i < p->rmargin; i++)
 356                         putchar(' ');
 357                 p->col = 0;
 358                 return;
 359         }
 360
 361         /*
 362          * If we're not to right-marginalise it (newline), then instead
 363          * pad to the right margin and stay off.
 364          */
 365
 366         if (p->flags & TERMP_NOBREAK)
 367                 for ( ; vis < maxvis; vis++)
 368                         putchar(' ');
 369         else
 370                 putchar('\n');
 371
 372         p->col = 0;
 373 }
 374
 375
 376 /*
 377  * A newline only breaks an existing line; it won't assert vertical
 378  * space.  All data in the output buffer is flushed prior to the newline
 379  * assertion.
 380  */
 381 void
 382 newln(struct termp *p)
 383 {
 384
 385         p->flags |= TERMP_NOSPACE;
 386         if (0 == p->col) {
 387                 p->flags &= ~TERMP_NOLPAD;
 388                 return;
 389         }
 390         flushln(p);
 391         p->flags &= ~TERMP_NOLPAD;
 392 }
 393
 394
 395 /*
 396  * Asserts a vertical space (a full, empty line-break between lines).
 397  * Note that if used twice, this will cause two blank spaces and so on.
 398  * All data in the output buffer is flushed prior to the newline
 399  * assertion.
 400  */
 401 void
 402 vspace(struct termp *p)
 403 {
 404
 405         newln(p);
 406         putchar('\n');
 407 }
 408
 409
 410 /*
 411  * Break apart a word into "pwords" (partial-words, usually from
 412  * breaking up a phrase into individual words) and, eventually, put them
 413  * into the output buffer.  If we're a literal word, then don't break up
 414  * the word and put it verbatim into the output buffer.
 415  */
 416 void
 417 word(struct termp *p, const char *word)
 418 {
 419         size_t           i, j, len;
 420
 421         if (p->flags & TERMP_LITERAL) {
 422                 pword(p, word, strlen(word));
 423                 return;
 424         }
 425
 426         len = strlen(word);
 427         assert(len > 0);
 428
 429         if (mdoc_isdelim(word)) {
 430                 if ( ! (p->flags & TERMP_IGNDELIM))
 431                         p->flags |= TERMP_NOSPACE;
 432                 p->flags &= ~TERMP_IGNDELIM;
 433         }
 434
 435         /* LINTED */
 436         for (j = i = 0; i < len; i++) {
 437                 if ( ! isspace((int)word[i])) {
 438                         j++;
 439                         continue;
 440                 }
 441
 442                 /* Escaped spaces don't delimit... */
 443                 if (i > 0 && isspace((int)word[i]) &&
 444                                 '\\' == word[i - 1]) {
 445                         j++;
 446                         continue;
 447                 }
 448
 449                 if (0 == j)
 450                         continue;
 451                 assert(i >= j);
 452                 pword(p, &word[i - j], j);
 453                 j = 0;
 454         }
 455         if (j > 0) {
 456                 assert(i >= j);
 457                 pword(p, &word[i - j], j);
 458         }
 459 }
 460
 461
 462 /*
 463  * This is the main function for printing out nodes.  It's constituted
 464  * of PRE and POST functions, which correspond to prefix and infix
 465  * processing.  The termpair structure allows data to persist between
 466  * prefix and postfix invocations.
 467  */
 468 static void
 469 body(struct termp *p, struct termpair *ppair,
 470                 const struct mdoc_meta *meta,
 471                 const struct mdoc_node *node)
 472 {
 473         int              dochild;
 474         struct termpair  pair;
 475
 476         /* Pre-processing. */
 477
 478         dochild = 1;
 479         pair.ppair = ppair;
 480         pair.type = 0;
 481         pair.offset = pair.rmargin = 0;
 482         pair.flag = 0;
 483         pair.count = 0;
 484
 485         if (MDOC_TEXT != node->type) {
 486                 if (termacts[node->tok].pre)
 487                         if ( ! (*termacts[node->tok].pre)(p, &pair, meta, node))
 488                                 dochild = 0;
 489         } else /* MDOC_TEXT == node->type */
 490                 word(p, node->data.text.string);
 491
 492         /* Children. */
 493
 494         if (TERMPAIR_FLAG & pair.type)
 495                 p->flags |= pair.flag;
 496
 497         if (dochild && node->child)
 498                 body(p, &pair, meta, node->child);
 499
 500         if (TERMPAIR_FLAG & pair.type)
 501                 p->flags &= ~pair.flag;
 502
 503         /* Post-processing. */
 504
 505         if (MDOC_TEXT != node->type)
 506                 if (termacts[node->tok].post)
 507                         (*termacts[node->tok].post)(p, &pair, meta, node);
 508
 509         /* Siblings. */
 510
 511         if (node->next)
 512                 body(p, ppair, meta, node->next);
 513 }
 514
 515
 516 static void
 517 footer(struct termp *p, const struct mdoc_meta *meta)
 518 {
 519         struct tm       *tm;
 520         char            *buf, *os;
 521
 522         if (NULL == (buf = malloc(p->rmargin)))
 523                 err(1, "malloc");
 524         if (NULL == (os = malloc(p->rmargin)))
 525                 err(1, "malloc");
 526
 527         tm = localtime(&meta->date);
 528
 529 #ifdef __OpenBSD__
 530         if (NULL == strftime(buf, p->rmargin, "%B %d, %Y", tm))
 531 #else
 532         if (0 == strftime(buf, p->rmargin, "%B %d, %Y", tm))
 533 #endif
 534                 err(1, "strftime");
 535
 536         (void)strlcpy(os, meta->os, p->rmargin);
 537
 538         /*
 539          * This is /slightly/ different from regular groff output
 540          * because we don't have page numbers.  Print the following:
 541          *
 542          * OS                                            MDOCDATE
 543          */
 544
 545         vspace(p);
 546
 547         p->flags |= TERMP_NOSPACE | TERMP_NOBREAK;
 548         p->rmargin = p->maxrmargin - strlen(buf);
 549         p->offset = 0;
 550
 551         word(p, os);
 552         flushln(p);
 553
 554         p->flags |= TERMP_NOLPAD | TERMP_NOSPACE;
 555         p->offset = p->rmargin;
 556         p->rmargin = p->maxrmargin;
 557         p->flags &= ~TERMP_NOBREAK;
 558
 559         word(p, buf);
 560         flushln(p);
 561
 562         free(buf);
 563         free(os);
 564 }
 565
 566
 567 static void
 568 header(struct termp *p, const struct mdoc_meta *meta)
 569 {
 570         char            *buf, *title, *bufp, *vbuf;
 571         const char      *pp;
 572         struct utsname   uts;
 573
 574         p->rmargin = p->maxrmargin;
 575         p->offset = 0;
 576
 577         if (NULL == (buf = malloc(p->rmargin)))
 578                 err(1, "malloc");
 579         if (NULL == (title = malloc(p->rmargin)))
 580                 err(1, "malloc");
 581         if (NULL == (vbuf = malloc(p->rmargin)))
 582                 err(1, "malloc");
 583
 584         if (NULL == (pp = mdoc_vol2a(meta->vol))) {
 585                 switch (meta->msec) {
 586                 case (MSEC_1):
 587                         /* FALLTHROUGH */
 588                 case (MSEC_6):
 589                         /* FALLTHROUGH */
 590                 case (MSEC_7):
 591                         pp = mdoc_vol2a(VOL_URM);
 592                         break;
 593                 case (MSEC_8):
 594                         pp = mdoc_vol2a(VOL_SMM);
 595                         break;
 596                 case (MSEC_2):
 597                         /* FALLTHROUGH */
 598                 case (MSEC_3):
 599                         /* FALLTHROUGH */
 600                 case (MSEC_4):
 601                         /* FALLTHROUGH */
 602                 case (MSEC_5):
 603                         pp = mdoc_vol2a(VOL_PRM);
 604                         break;
 605                 case (MSEC_9):
 606                         pp = mdoc_vol2a(VOL_KM);
 607                         break;
 608                 default:
 609                         break;
 610                 }
 611         }
 612         vbuf[0] = 0;
 613
 614         if (pp) {
 615                 if (-1 == uname(&uts))
 616                         err(1, "uname");
 617                 (void)strlcat(vbuf, uts.sysname, p->rmargin);
 618                 (void)strlcat(vbuf, " ", p->rmargin);
 619         } else if (NULL == (pp = mdoc_msec2a(meta->msec)))
 620                 pp = mdoc_msec2a(MSEC_local);
 621
 622         (void)strlcat(vbuf, pp, p->rmargin);
 623
 624         /*
 625          * The header is strange.  It has three components, which are
 626          * really two with the first duplicated.  It goes like this:
 627          *
 628          * IDENTIFIER              TITLE                   IDENTIFIER
 629          *
 630          * The IDENTIFIER is NAME(SECTION), which is the command-name
 631          * (if given, or "unknown" if not) followed by the manual page
 632          * section.  These are given in `Dt'.  The TITLE is a free-form
 633          * string depending on the manual volume.  If not specified, it
 634          * switches on the manual section.
 635          */
 636
 637         if (mdoc_arch2a(meta->arch))
 638                 (void)snprintf(buf, p->rmargin, "%s (%s)",
 639                                 vbuf, mdoc_arch2a(meta->arch));
 640         else
 641                 (void)strlcpy(buf, vbuf, p->rmargin);
 642
 643         pp = mdoc_msec2a(meta->msec);
 644
 645         (void)snprintf(title, p->rmargin, "%s(%s)",
 646                         meta->title, pp ? pp : "");
 647
 648         for (bufp = title; *bufp; bufp++)
 649                 *bufp = toupper(*bufp);
 650
 651         p->offset = 0;
 652         p->rmargin = (p->maxrmargin - strlen(buf)) / 2;
 653         p->flags |= TERMP_NOBREAK | TERMP_NOSPACE;
 654
 655         word(p, title);
 656         flushln(p);
 657
 658         p->flags |= TERMP_NOLPAD | TERMP_NOSPACE;
 659         p->offset = p->rmargin;
 660         p->rmargin = p->maxrmargin - strlen(title);
 661
 662         word(p, buf);
 663         flushln(p);
 664
 665         p->offset = p->rmargin;
 666         p->rmargin = p->maxrmargin;
 667         p->flags &= ~TERMP_NOBREAK;
 668         p->flags |= TERMP_NOLPAD | TERMP_NOSPACE;
 669
 670         word(p, title);
 671         flushln(p);
 672
 673         p->rmargin = p->maxrmargin;
 674         p->offset = 0;
 675         p->flags &= ~TERMP_NOSPACE;
 676
 677         free(title);
 678         free(vbuf);
 679         free(buf);
 680 }
 681
 682
 683 /*
 684  * Determine the symbol indicated by an escape sequences, that is, one
 685  * starting with a backslash.  Once done, we pass this value into the
 686  * output buffer by way of the symbol table.
 687  */
 688 static void
 689 nescape(struct termp *p, const char *word, size_t len)
 690 {
 691         struct termenc  *enc;
 692
 693         switch (len) {
 694         case (1):
 695                 enc = termenc1;
 696                 break;
 697         case (2):
 698                 enc = termenc2;
 699                 break;
 700         default:
 701                 warnx("unsupported %zu-byte escape sequence", len);
 702                 return;
 703         }
 704
 705         for ( ; enc->enc; enc++)
 706                 if (0 == memcmp(enc->enc, word, len)) {
 707                         symbola(p, enc->sym);
 708                         return;
 709                 }
 710
 711         warnx("unsupported %zu-byte escape sequence", len);
 712 }
 713
 714
 715 /*
 716  * Handle an escape sequence: determine its length and pass it to the
 717  * escape-symbol look table.  Note that we assume mdoc(3) has validated
 718  * the escape sequence (we assert upon badly-formed escape sequences).
 719  */
 720 static void
 721 pescape(struct termp *p, const char *word, size_t *i, size_t len)
 722 {
 723         size_t           j;
 724
 725         (*i)++;
 726         assert(*i < len);
 727
 728         if ('(' == word[*i]) {
 729                 (*i)++;
 730                 assert(*i + 1 < len);
 731                 nescape(p, &word[*i], 2);
 732                 (*i)++;
 733                 return;
 734
 735         } else if ('*' == word[*i]) {
 736                 /* XXX - deprecated! */
 737                 (*i)++;
 738                 assert(*i < len);
 739                 switch (word[*i]) {
 740                 case ('('):
 741                         (*i)++;
 742                         assert(*i + 1 < len);
 743                         nescape(p, &word[*i], 2);
 744                         (*i)++;
 745                         return;
 746                 case ('['):
 747                         break;
 748                 default:
 749                         nescape(p, &word[*i], 1);
 750                         return;
 751                 }
 752
 753         } else if ('[' != word[*i]) {
 754                 nescape(p, &word[*i], 1);
 755                 return;
 756         }
 757
 758         (*i)++;
 759         for (j = 0; word[*i] && ']' != word[*i]; (*i)++, j++)
 760                 /* Loop... */ ;
 761
 762         assert(word[*i]);
 763         nescape(p, &word[*i - j], j);
 764 }
 765
 766
 767 /*
 768  * Handle pwords, partial words, which may be either a single word or a
 769  * phrase that cannot be broken down (such as a literal string).  This
 770  * handles word styling.
 771  */
 772 static void
 773 pword(struct termp *p, const char *word, size_t len)
 774 {
 775         size_t           i;
 776
 777         if ( ! (TERMP_NOSPACE & p->flags) &&
 778                         ! (TERMP_LITERAL & p->flags))
 779                 chara(p, ' ');
 780
 781         if ( ! (p->flags & TERMP_NONOSPACE))
 782                 p->flags &= ~TERMP_NOSPACE;
 783
 784         /*
 785          * XXX - if literal and underlining, this will underline the
 786          * spaces between literal words.
 787          */
 788
 789         if (p->flags & TERMP_BOLD)
 790                 stylea(p, TERMSTYLE_BOLD);
 791         if (p->flags & TERMP_UNDERLINE)
 792                 stylea(p, TERMSTYLE_UNDER);
 793
 794         for (i = 0; i < len; i++) {
 795                 if ('\\' == word[i]) {
 796                         pescape(p, word, &i, len);
 797                         continue;
 798                 }
 799                 chara(p, word[i]);
 800         }
 801
 802         if (p->flags & TERMP_BOLD ||
 803                         p->flags & TERMP_UNDERLINE)
 804                 stylea(p, TERMSTYLE_CLEAR);
 805 }
 806
 807
 808 /*
 809  * Add a symbol to the output line buffer.
 810  */
 811 static void
 812 symbola(struct termp *p, enum tsym sym)
 813 {
 814
 815         assert(p->symtab[sym].sym);
 816         stringa(p, p->symtab[sym].sym, p->symtab[sym].sz);
 817 }
 818
 819
 820 /*
 821  * Add a style to the output line buffer.
 822  */
 823 static void
 824 stylea(struct termp *p, enum tstyle style)
 825 {
 826
 827         assert(p->styletab[style].sym);
 828         stringa(p, p->styletab[style].sym, p->styletab[style].sz);
 829 }
 830
 831
 832 /*
 833  * Like chara() but for arbitrary-length buffers.  Resize the buffer by
 834  * a factor of two (if the buffer is less than that) or the buffer's
 835  * size.
 836  */
 837 static void
 838 stringa(struct termp *p, const char *c, size_t sz)
 839 {
 840         size_t           s;
 841
 842         if (0 == sz)
 843                 return;
 844
 845         s = sz > p->maxcols * 2 ? sz : p->maxcols * 2;
 846
 847         assert(c);
 848         if (p->col + sz >= p->maxcols) {
 849                 p->buf = realloc(p->buf, s);
 850                 if (NULL == p->buf)
 851                         err(1, "realloc");
 852                 p->maxcols = s;
 853         }
 854
 855         (void)memcpy(&p->buf[p->col], c, sz);
 856         p->col += sz;
 857 }
 858
 859
 860 /*
 861  * Insert a single character into the line-buffer.  If the buffer's
 862  * space is exceeded, then allocate more space by doubling the buffer
 863  * size.
 864  */
 865 static void
 866 chara(struct termp *p, char c)
 867 {
 868
 869         if (p->col + 1 >= p->maxcols) {
 870                 p->buf = realloc(p->buf, p->maxcols * 2);
 871                 if (NULL == p->buf)
 872                         err(1, "malloc");
 873                 p->maxcols *= 2;
 874         }
 875         p->buf[(p->col)++] = c;
 876 }