mdocterm.c

   1 /* $Id: mdocterm.c,v 1.34 2009/03/06 14:13:47 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 <assert.h>
  20 #include <ctype.h>
  21 #include <err.h>
  22 #include <getopt.h>
  23 #include <stdio.h>
  24 #include <stdlib.h>
  25 #include <string.h>
  26 #ifndef __OpenBSD__
  27 #include <time.h>
  28 #endif
  29
  30 #include "mmain.h"
  31 #include "term.h"
  32
  33 struct  termenc {
  34         const char       *enc;
  35         int               sym;
  36 };
  37
  38 static  void              body(struct termp *,
  39                                 struct termpair *,
  40                                 const struct mdoc_meta *,
  41                                 const struct mdoc_node *);
  42 static  void              header(struct termp *,
  43                                 const struct mdoc_meta *);
  44 static  void              footer(struct termp *,
  45                                 const struct mdoc_meta *);
  46
  47 static  void              pword(struct termp *, const char *, size_t);
  48 static  void              pescape(struct termp *, const char *,
  49                                 size_t *, size_t);
  50 static  void              nescape(struct termp *,
  51                                 const char *, size_t);
  52 static  void              chara(struct termp *, char);
  53 static  void              stringa(struct termp *,
  54                                 const char *, size_t);
  55 static  void              symbola(struct termp *, enum tsym);
  56 static  void              stylea(struct termp *, enum tstyle);
  57
  58 #ifdef __linux__
  59 extern  size_t            strlcat(char *, const char *, size_t);
  60 extern  size_t            strlcpy(char *, const char *, size_t);
  61 #endif
  62
  63 static  struct termenc    termenc1[] = {
  64         { "\\",           TERMSYM_SLASH },
  65         { "\'",           TERMSYM_RSQUOTE },
  66         { "`",            TERMSYM_LSQUOTE },
  67         { "-",            TERMSYM_HYPHEN },
  68         { " ",            TERMSYM_SPACE },
  69         { ".",            TERMSYM_PERIOD },
  70         { "&",            TERMSYM_BREAK },
  71         { "e",            TERMSYM_SLASH },
  72         { "q",            TERMSYM_DQUOTE },
  73         { NULL,           0 }
  74 };
  75
  76 static  struct termenc    termenc2[] = {
  77         { "rB",           TERMSYM_RBRACK },
  78         { "lB",           TERMSYM_LBRACK },
  79         { "ra",           TERMSYM_RANGLE },
  80         { "la",           TERMSYM_LANGLE },
  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         { "<", 1 },             /* TERMSYM_LANGLE */
 165         { ">", 1 },             /* TERMSYM_RANGLE */
 166 };
 167
 168 static  const char        ansi_clear[]  = { 27, '[', '0', 'm' };
 169 static  const char        ansi_bold[]  = { 27, '[', '1', 'm' };
 170 static  const char        ansi_under[]  = { 27, '[', '4', 'm' };
 171
 172 static  struct termsym    termstyle_ansi[] = {
 173         { ansi_clear, 4 },
 174         { ansi_bold, 4 },
 175         { ansi_under, 4 }
 176 };
 177
 178
 179 int
 180 main(int argc, char *argv[])
 181 {
 182         struct mmain      *p;
 183         int                c;
 184         const struct mdoc *mdoc;
 185         struct termp       termp;
 186
 187         p = mmain_alloc();
 188
 189         c = mmain_getopt(p, argc, argv, NULL, NULL, NULL, NULL);
 190         if (1 != c)
 191                 mmain_exit(p, -1 == c ? 1 : 0);
 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((u_char)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((u_char)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                 if ( ! (TERMP_NONOBREAK & p->flags)) {
 355                         putchar('\n');
 356                         for (i = 0; i < p->rmargin; i++)
 357                                 putchar(' ');
 358                 }
 359                 p->col = 0;
 360                 return;
 361         }
 362
 363         /*
 364          * If we're not to right-marginalise it (newline), then instead
 365          * pad to the right margin and stay off.
 366          */
 367
 368         if (p->flags & TERMP_NOBREAK) {
 369                 if ( ! (TERMP_NONOBREAK & p->flags))
 370                         for ( ; vis < maxvis; vis++)
 371                                 putchar(' ');
 372         } else
 373                 putchar('\n');
 374
 375         p->col = 0;
 376 }
 377
 378
 379 /*
 380  * A newline only breaks an existing line; it won't assert vertical
 381  * space.  All data in the output buffer is flushed prior to the newline
 382  * assertion.
 383  */
 384 void
 385 newln(struct termp *p)
 386 {
 387
 388         p->flags |= TERMP_NOSPACE;
 389         if (0 == p->col) {
 390                 p->flags &= ~TERMP_NOLPAD;
 391                 return;
 392         }
 393         flushln(p);
 394         p->flags &= ~TERMP_NOLPAD;
 395 }
 396
 397
 398 /*
 399  * Asserts a vertical space (a full, empty line-break between lines).
 400  * Note that if used twice, this will cause two blank spaces and so on.
 401  * All data in the output buffer is flushed prior to the newline
 402  * assertion.
 403  */
 404 void
 405 vspace(struct termp *p)
 406 {
 407
 408         newln(p);
 409         putchar('\n');
 410 }
 411
 412
 413 /*
 414  * Break apart a word into "pwords" (partial-words, usually from
 415  * breaking up a phrase into individual words) and, eventually, put them
 416  * into the output buffer.  If we're a literal word, then don't break up
 417  * the word and put it verbatim into the output buffer.
 418  */
 419 void
 420 word(struct termp *p, const char *word)
 421 {
 422         size_t           i, j, len;
 423
 424         if (p->flags & TERMP_LITERAL) {
 425                 pword(p, word, strlen(word));
 426                 return;
 427         }
 428
 429         len = strlen(word);
 430         assert(len > 0);
 431
 432         if (mdoc_isdelim(word)) {
 433                 if ( ! (p->flags & TERMP_IGNDELIM))
 434                         p->flags |= TERMP_NOSPACE;
 435                 p->flags &= ~TERMP_IGNDELIM;
 436         }
 437
 438         /* LINTED */
 439         for (j = i = 0; i < len; i++) {
 440                 if ( ! isspace((u_char)word[i])) {
 441                         j++;
 442                         continue;
 443                 }
 444
 445                 /* Escaped spaces don't delimit... */
 446                 if (i > 0 && isspace((u_char)word[i]) &&
 447                                 '\\' == word[i - 1]) {
 448                         j++;
 449                         continue;
 450                 }
 451
 452                 if (0 == j)
 453                         continue;
 454                 assert(i >= j);
 455                 pword(p, &word[i - j], j);
 456                 j = 0;
 457         }
 458         if (j > 0) {
 459                 assert(i >= j);
 460                 pword(p, &word[i - j], j);
 461         }
 462 }
 463
 464
 465 /*
 466  * This is the main function for printing out nodes.  It's constituted
 467  * of PRE and POST functions, which correspond to prefix and infix
 468  * processing.  The termpair structure allows data to persist between
 469  * prefix and postfix invocations.
 470  */
 471 static void
 472 body(struct termp *p, struct termpair *ppair,
 473                 const struct mdoc_meta *meta,
 474                 const struct mdoc_node *node)
 475 {
 476         int              dochild;
 477         struct termpair  pair;
 478
 479         /* Pre-processing. */
 480
 481         dochild = 1;
 482         pair.ppair = ppair;
 483         pair.type = 0;
 484         pair.offset = pair.rmargin = 0;
 485         pair.flag = 0;
 486         pair.count = 0;
 487
 488         if (MDOC_TEXT != node->type) {
 489                 if (termacts[node->tok].pre)
 490                         if ( ! (*termacts[node->tok].pre)(p, &pair, meta, node))
 491                                 dochild = 0;
 492         } else /* MDOC_TEXT == node->type */
 493                 word(p, node->data.text.string);
 494
 495         /* Children. */
 496
 497         if (TERMPAIR_FLAG & pair.type)
 498                 p->flags |= pair.flag;
 499
 500         if (dochild && node->child)
 501                 body(p, &pair, meta, node->child);
 502
 503         if (TERMPAIR_FLAG & pair.type)
 504                 p->flags &= ~pair.flag;
 505
 506         /* Post-processing. */
 507
 508         if (MDOC_TEXT != node->type)
 509                 if (termacts[node->tok].post)
 510                         (*termacts[node->tok].post)(p, &pair, meta, node);
 511
 512         /* Siblings. */
 513
 514         if (node->next)
 515                 body(p, ppair, meta, node->next);
 516 }
 517
 518
 519 static void
 520 footer(struct termp *p, const struct mdoc_meta *meta)
 521 {
 522         struct tm       *tm;
 523         char            *buf, *os;
 524
 525         if (NULL == (buf = malloc(p->rmargin)))
 526                 err(1, "malloc");
 527         if (NULL == (os = malloc(p->rmargin)))
 528                 err(1, "malloc");
 529
 530         tm = localtime(&meta->date);
 531
 532 #ifdef __OpenBSD__
 533         if (NULL == strftime(buf, p->rmargin, "%B %d, %Y", tm))
 534 #else
 535         if (0 == strftime(buf, p->rmargin, "%B %d, %Y", tm))
 536 #endif
 537                 err(1, "strftime");
 538
 539         (void)strlcpy(os, meta->os, p->rmargin);
 540
 541         /*
 542          * This is /slightly/ different from regular groff output
 543          * because we don't have page numbers.  Print the following:
 544          *
 545          * OS                                            MDOCDATE
 546          */
 547
 548         vspace(p);
 549
 550         p->flags |= TERMP_NOSPACE | TERMP_NOBREAK;
 551         p->rmargin = p->maxrmargin - strlen(buf);
 552         p->offset = 0;
 553
 554         word(p, os);
 555         flushln(p);
 556
 557         p->flags |= TERMP_NOLPAD | TERMP_NOSPACE;
 558         p->offset = p->rmargin;
 559         p->rmargin = p->maxrmargin;
 560         p->flags &= ~TERMP_NOBREAK;
 561
 562         word(p, buf);
 563         flushln(p);
 564
 565         free(buf);
 566         free(os);
 567 }
 568
 569
 570 static void
 571 header(struct termp *p, const struct mdoc_meta *meta)
 572 {
 573         char            *buf, *title, *bufp;
 574
 575         p->rmargin = p->maxrmargin;
 576         p->offset = 0;
 577
 578         if (NULL == (buf = malloc(p->rmargin)))
 579                 err(1, "malloc");
 580         if (NULL == (title = malloc(p->rmargin)))
 581                 err(1, "malloc");
 582
 583         /*
 584          * The header is strange.  It has three components, which are
 585          * really two with the first duplicated.  It goes like this:
 586          *
 587          * IDENTIFIER              TITLE                   IDENTIFIER
 588          *
 589          * The IDENTIFIER is NAME(SECTION), which is the command-name
 590          * (if given, or "unknown" if not) followed by the manual page
 591          * section.  These are given in `Dt'.  The TITLE is a free-form
 592          * string depending on the manual volume.  If not specified, it
 593          * switches on the manual section.
 594          */
 595
 596         assert(meta->vol);
 597         (void)strlcpy(buf, meta->vol, p->rmargin);
 598
 599         if (meta->arch) {
 600                 (void)strlcat(buf, " (", p->rmargin);
 601                 (void)strlcat(buf, meta->arch, p->rmargin);
 602                 (void)strlcat(buf, ")", p->rmargin);
 603         }
 604
 605         (void)snprintf(title, p->rmargin, "%s(%d)",
 606                         meta->title, meta->msec);
 607
 608         for (bufp = title; *bufp; bufp++)
 609                 *bufp = toupper((u_char)*bufp);
 610
 611         p->offset = 0;
 612         p->rmargin = (p->maxrmargin - strlen(buf)) / 2;
 613         p->flags |= TERMP_NOBREAK | TERMP_NOSPACE;
 614
 615         word(p, title);
 616         flushln(p);
 617
 618         p->flags |= TERMP_NOLPAD | TERMP_NOSPACE;
 619         p->offset = p->rmargin;
 620         p->rmargin = p->maxrmargin - strlen(title);
 621
 622         word(p, buf);
 623         flushln(p);
 624
 625         p->offset = p->rmargin;
 626         p->rmargin = p->maxrmargin;
 627         p->flags &= ~TERMP_NOBREAK;
 628         p->flags |= TERMP_NOLPAD | TERMP_NOSPACE;
 629
 630         word(p, title);
 631         flushln(p);
 632
 633         p->rmargin = p->maxrmargin;
 634         p->offset = 0;
 635         p->flags &= ~TERMP_NOSPACE;
 636
 637         free(title);
 638         free(buf);
 639 }
 640
 641
 642 /*
 643  * Determine the symbol indicated by an escape sequences, that is, one
 644  * starting with a backslash.  Once done, we pass this value into the
 645  * output buffer by way of the symbol table.
 646  */
 647 static void
 648 nescape(struct termp *p, const char *word, size_t len)
 649 {
 650         struct termenc  *enc;
 651
 652         switch (len) {
 653         case (1):
 654                 enc = termenc1;
 655                 break;
 656         case (2):
 657                 enc = termenc2;
 658                 break;
 659         default:
 660                 warnx("unsupported %zu-byte escape sequence", len);
 661                 return;
 662         }
 663
 664         for ( ; enc->enc; enc++)
 665                 if (0 == memcmp(enc->enc, word, len)) {
 666                         symbola(p, enc->sym);
 667                         return;
 668                 }
 669
 670         warnx("unsupported %zu-byte escape sequence", len);
 671 }
 672
 673
 674 /*
 675  * Handle an escape sequence: determine its length and pass it to the
 676  * escape-symbol look table.  Note that we assume mdoc(3) has validated
 677  * the escape sequence (we assert upon badly-formed escape sequences).
 678  */
 679 static void
 680 pescape(struct termp *p, const char *word, size_t *i, size_t len)
 681 {
 682         size_t           j;
 683
 684         (*i)++;
 685         assert(*i < len);
 686
 687         if ('(' == word[*i]) {
 688                 (*i)++;
 689                 assert(*i + 1 < len);
 690                 nescape(p, &word[*i], 2);
 691                 (*i)++;
 692                 return;
 693
 694         } else if ('*' == word[*i]) {
 695                 /* XXX - deprecated! */
 696                 (*i)++;
 697                 assert(*i < len);
 698                 switch (word[*i]) {
 699                 case ('('):
 700                         (*i)++;
 701                         assert(*i + 1 < len);
 702                         nescape(p, &word[*i], 2);
 703                         (*i)++;
 704                         return;
 705                 case ('['):
 706                         break;
 707                 default:
 708                         nescape(p, &word[*i], 1);
 709                         return;
 710                 }
 711
 712         } else if ('[' != word[*i]) {
 713                 nescape(p, &word[*i], 1);
 714                 return;
 715         }
 716
 717         (*i)++;
 718         for (j = 0; word[*i] && ']' != word[*i]; (*i)++, j++)
 719                 /* Loop... */ ;
 720
 721         assert(word[*i]);
 722         nescape(p, &word[*i - j], j);
 723 }
 724
 725
 726 /*
 727  * Handle pwords, partial words, which may be either a single word or a
 728  * phrase that cannot be broken down (such as a literal string).  This
 729  * handles word styling.
 730  */
 731 static void
 732 pword(struct termp *p, const char *word, size_t len)
 733 {
 734         size_t           i;
 735
 736         if ( ! (TERMP_NOSPACE & p->flags) &&
 737                         ! (TERMP_LITERAL & p->flags))
 738                 chara(p, ' ');
 739
 740         if ( ! (p->flags & TERMP_NONOSPACE))
 741                 p->flags &= ~TERMP_NOSPACE;
 742
 743         /*
 744          * XXX - if literal and underlining, this will underline the
 745          * spaces between literal words.
 746          */
 747
 748         if (p->flags & TERMP_BOLD)
 749                 stylea(p, TERMSTYLE_BOLD);
 750         if (p->flags & TERMP_UNDERLINE)
 751                 stylea(p, TERMSTYLE_UNDER);
 752
 753         for (i = 0; i < len; i++) {
 754                 if ('\\' == word[i]) {
 755                         pescape(p, word, &i, len);
 756                         continue;
 757                 }
 758                 chara(p, word[i]);
 759         }
 760
 761         if (p->flags & TERMP_BOLD ||
 762                         p->flags & TERMP_UNDERLINE)
 763                 stylea(p, TERMSTYLE_CLEAR);
 764 }
 765
 766
 767 /*
 768  * Add a symbol to the output line buffer.
 769  */
 770 static void
 771 symbola(struct termp *p, enum tsym sym)
 772 {
 773
 774         assert(p->symtab[sym].sym);
 775         stringa(p, p->symtab[sym].sym, p->symtab[sym].sz);
 776 }
 777
 778
 779 /*
 780  * Add a style to the output line buffer.
 781  */
 782 static void
 783 stylea(struct termp *p, enum tstyle style)
 784 {
 785
 786         assert(p->styletab[style].sym);
 787         stringa(p, p->styletab[style].sym, p->styletab[style].sz);
 788 }
 789
 790
 791 /*
 792  * Like chara() but for arbitrary-length buffers.  Resize the buffer by
 793  * a factor of two (if the buffer is less than that) or the buffer's
 794  * size.
 795  */
 796 static void
 797 stringa(struct termp *p, const char *c, size_t sz)
 798 {
 799         size_t           s;
 800
 801         if (0 == sz)
 802                 return;
 803
 804         s = sz > p->maxcols * 2 ? sz : p->maxcols * 2;
 805
 806         assert(c);
 807         if (p->col + sz >= p->maxcols) {
 808                 p->buf = realloc(p->buf, s);
 809                 if (NULL == p->buf)
 810                         err(1, "realloc");
 811                 p->maxcols = s;
 812         }
 813
 814         (void)memcpy(&p->buf[p->col], c, sz);
 815         p->col += sz;
 816 }
 817
 818
 819 /*
 820  * Insert a single character into the line-buffer.  If the buffer's
 821  * space is exceeded, then allocate more space by doubling the buffer
 822  * size.
 823  */
 824 static void
 825 chara(struct termp *p, char c)
 826 {
 827
 828         if (p->col + 1 >= p->maxcols) {
 829                 p->buf = realloc(p->buf, p->maxcols * 2);
 830                 if (NULL == p->buf)
 831                         err(1, "malloc");
 832                 p->maxcols *= 2;
 833         }
 834         p->buf[(p->col)++] = c;
 835 }