mdocterm.c

   1 /* $Id: mdocterm.c,v 1.40 2009/03/12 06:32:17 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/types.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
  29 #include "mmain.h"
  30 #include "term.h"
  31
  32 struct  termenc {
  33         const char       *enc;
  34         int               sym;
  35 };
  36
  37 static  void              body(struct termp *,
  38                                 struct termpair *,
  39                                 const struct mdoc_meta *,
  40                                 const struct mdoc_node *);
  41 static  void              header(struct termp *,
  42                                 const struct mdoc_meta *);
  43 static  void              footer(struct termp *,
  44                                 const struct mdoc_meta *);
  45
  46 static  void              pword(struct termp *, const char *, size_t);
  47 static  void              pescape(struct termp *, const char *,
  48                                 size_t *, size_t);
  49 static  void              nescape(struct termp *,
  50                                 const char *, size_t);
  51 static  void              chara(struct termp *, char);
  52 static  void              stringa(struct termp *,
  53                                 const char *, size_t);
  54 static  void              symbola(struct termp *, enum tsym);
  55 static  void              sanity(const struct mdoc_node *);
  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         { "rC",           TERMSYM_RBRACE },
  78         { "lC",           TERMSYM_LBRACE },
  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         { "{", 1 },             /* TERMSYM_LBRACE */
 169         { "}", 1 },             /* TERMSYM_RBRACE */
 170 };
 171
 172 static  const char        ansi_clear[]  = { 27, '[', '0', 'm' };
 173 static  const char        ansi_bold[]  = { 27, '[', '1', 'm' };
 174 static  const char        ansi_under[]  = { 27, '[', '4', 'm' };
 175
 176 static  struct termsym    termstyle_ansi[] = {
 177         { ansi_clear, 4 },
 178         { ansi_bold, 4 },
 179         { ansi_under, 4 }
 180 };
 181
 182
 183 int
 184 main(int argc, char *argv[])
 185 {
 186         struct mmain      *p;
 187         int                c;
 188         const struct mdoc *mdoc;
 189         struct termp       termp;
 190
 191         p = mmain_alloc();
 192
 193         c = mmain_getopt(p, argc, argv, NULL, NULL, NULL, NULL);
 194         if (1 != c)
 195                 mmain_exit(p, -1 == c ? 1 : 0);
 196
 197         if (NULL == (mdoc = mmain_mdoc(p)))
 198                 mmain_exit(p, 1);
 199
 200         termp.maxrmargin = termp.rmargin = 78; /* XXX */
 201         termp.maxcols = 1024;
 202         termp.offset = termp.col = 0;
 203         termp.flags = TERMP_NOSPACE;
 204         termp.symtab = termsym_ansi;
 205         termp.styletab = termstyle_ansi;
 206
 207         if (NULL == (termp.buf = malloc(termp.maxcols)))
 208                 err(1, "malloc");
 209
 210         header(&termp, mdoc_meta(mdoc));
 211         body(&termp, NULL, mdoc_meta(mdoc), mdoc_node(mdoc));
 212         footer(&termp, mdoc_meta(mdoc));
 213
 214         free(termp.buf);
 215
 216         mmain_exit(p, 0);
 217         /* NOTREACHED */
 218 }
 219
 220
 221 /*
 222  * Flush a line of text.  A "line" is loosely defined as being something
 223  * that should be followed by a newline, regardless of whether it's
 224  * broken apart by newlines getting there.  A line can also be a
 225  * fragment of a columnar list.
 226  *
 227  * Specifically, a line is whatever's in p->buf of length p->col, which
 228  * is zeroed after this function returns.
 229  *
 230  * The variables TERMP_NOLPAD, TERMP_LITERAL and TERMP_NOBREAK are of
 231  * critical importance here.  Their behaviour follows:
 232  *
 233  *  - TERMP_NOLPAD: when beginning to write the line, don't left-pad the
 234  *    offset value.  This is useful when doing columnar lists where the
 235  *    prior column has right-padded.
 236  *
 237  *  - TERMP_NOBREAK: this is the most important and is used when making
 238  *    columns.  In short: don't print a newline and instead pad to the
 239  *    right margin.  Used in conjunction with TERMP_NOLPAD.
 240  *
 241  *  In-line line breaking:
 242  *
 243  *  If TERMP_NOBREAK is specified and the line overruns the right
 244  *  margin, it will break and pad-right to the right margin after
 245  *  writing.  If maxrmargin is violated, it will break and continue
 246  *  writing from the right-margin, which will lead to the above
 247  *  scenario upon exit.
 248  *
 249  *  Otherwise, the line will break at the right margin.  Extremely long
 250  *  lines will cause the system to emit a warning (TODO: hyphenate, if
 251  *  possible).
 252  */
 253 void
 254 flushln(struct termp *p)
 255 {
 256         size_t           i, j, vsz, vis, maxvis, mmax, bp;
 257
 258         /*
 259          * First, establish the maximum columns of "visible" content.
 260          * This is usually the difference between the right-margin and
 261          * an indentation, but can be, for tagged lists or columns, a
 262          * small set of values.
 263          */
 264
 265         assert(p->offset < p->rmargin);
 266         maxvis = p->rmargin - p->offset;
 267         mmax = p->maxrmargin - p->offset;
 268         bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
 269         vis = 0;
 270
 271         /*
 272          * If in the standard case (left-justified), then begin with our
 273          * indentation, otherwise (columns, etc.) just start spitting
 274          * out text.
 275          */
 276
 277         if ( ! (p->flags & TERMP_NOLPAD))
 278                 /* LINTED */
 279                 for (j = 0; j < p->offset; j++)
 280                         putchar(' ');
 281
 282         for (i = 0; i < p->col; i++) {
 283                 /*
 284                  * Count up visible word characters.  Control sequences
 285                  * (starting with the CSI) aren't counted.  A space
 286                  * generates a non-printing word, which is valid (the
 287                  * space is printed according to regular spacing rules).
 288                  */
 289
 290                 /* FIXME: make non-ANSI friendly. */
 291
 292                 /* LINTED */
 293                 for (j = i, vsz = 0; j < p->col; j++) {
 294                         if (isspace((u_char)p->buf[j]))
 295                                 break;
 296                         else if (27 == p->buf[j]) {
 297                                 assert(j + 4 <= p->col);
 298                                 j += 3;
 299                         } else
 300                                 vsz++;
 301                 }
 302
 303                 /*
 304                  * Do line-breaking.  If we're greater than our
 305                  * break-point and already in-line, break to the next
 306                  * line and start writing.  If we're at the line start,
 307                  * then write out the word (TODO: hyphenate) and break
 308                  * in a subsequent loop invocation.
 309                  */
 310
 311                 if ( ! (TERMP_NOBREAK & p->flags)) {
 312                         if (vis && vis + vsz > bp) {
 313                                 putchar('\n');
 314                                 for (j = 0; j < p->offset; j++)
 315                                         putchar(' ');
 316                                 vis = 0;
 317                         } else if (vis + vsz > bp)
 318                                 warnx("word breaks right margin");
 319
 320                         /* TODO: hyphenate. */
 321
 322                 } else {
 323                         if (vis && vis + vsz > bp) {
 324                                 putchar('\n');
 325                                 for (j = 0; j < p->rmargin; j++)
 326                                         putchar(' ');
 327                                 vis = p->rmargin - p->offset;
 328                         } else if (vis + vsz > bp)
 329                                 warnx("word breaks right margin");
 330
 331                         /* TODO: hyphenate. */
 332                 }
 333
 334                 /*
 335                  * Write out the word and a trailing space.  Omit the
 336                  * space if we're the last word in the line or beyond
 337                  * our breakpoint.
 338                  */
 339
 340                 for ( ; i < p->col; i++) {
 341                         if (isspace((u_char)p->buf[i]))
 342                                 break;
 343                         putchar(p->buf[i]);
 344                 }
 345                 vis += vsz;
 346                 if (i < p->col && vis <= bp) {
 347                         putchar(' ');
 348                         vis++;
 349                 }
 350         }
 351
 352         /*
 353          * If we've overstepped our maximum visible no-break space, then
 354          * cause a newline and offset at the right margin.
 355          */
 356
 357         if ((TERMP_NOBREAK & p->flags) && vis >= maxvis) {
 358                 if ( ! (TERMP_NONOBREAK & p->flags)) {
 359                         putchar('\n');
 360                         for (i = 0; i < p->rmargin; i++)
 361                                 putchar(' ');
 362                 }
 363                 p->col = 0;
 364                 return;
 365         }
 366
 367         /*
 368          * If we're not to right-marginalise it (newline), then instead
 369          * pad to the right margin and stay off.
 370          */
 371
 372         if (p->flags & TERMP_NOBREAK) {
 373                 if ( ! (TERMP_NONOBREAK & p->flags))
 374                         for ( ; vis < maxvis; vis++)
 375                                 putchar(' ');
 376         } else
 377                 putchar('\n');
 378
 379         p->col = 0;
 380 }
 381
 382
 383 /*
 384  * A newline only breaks an existing line; it won't assert vertical
 385  * space.  All data in the output buffer is flushed prior to the newline
 386  * assertion.
 387  */
 388 void
 389 newln(struct termp *p)
 390 {
 391
 392         p->flags |= TERMP_NOSPACE;
 393         if (0 == p->col) {
 394                 p->flags &= ~TERMP_NOLPAD;
 395                 return;
 396         }
 397         flushln(p);
 398         p->flags &= ~TERMP_NOLPAD;
 399 }
 400
 401
 402 /*
 403  * Asserts a vertical space (a full, empty line-break between lines).
 404  * Note that if used twice, this will cause two blank spaces and so on.
 405  * All data in the output buffer is flushed prior to the newline
 406  * assertion.
 407  */
 408 void
 409 vspace(struct termp *p)
 410 {
 411
 412         newln(p);
 413         putchar('\n');
 414 }
 415
 416
 417 /*
 418  * Break apart a word into "pwords" (partial-words, usually from
 419  * breaking up a phrase into individual words) and, eventually, put them
 420  * into the output buffer.  If we're a literal word, then don't break up
 421  * the word and put it verbatim into the output buffer.
 422  */
 423 void
 424 word(struct termp *p, const char *word)
 425 {
 426         size_t           i, j, len;
 427
 428         if (p->flags & TERMP_LITERAL) {
 429                 pword(p, word, strlen(word));
 430                 return;
 431         }
 432
 433         if (0 == (len = strlen(word)))
 434                 errx(1, "blank line not in literal context");
 435
 436         if (mdoc_isdelim(word)) {
 437                 if ( ! (p->flags & TERMP_IGNDELIM))
 438                         p->flags |= TERMP_NOSPACE;
 439                 p->flags &= ~TERMP_IGNDELIM;
 440         }
 441
 442         /* LINTED */
 443         for (j = i = 0; i < len; i++) {
 444                 if ( ! isspace((u_char)word[i])) {
 445                         j++;
 446                         continue;
 447                 }
 448
 449                 /* Escaped spaces don't delimit... */
 450                 if (i > 0 && isspace((u_char)word[i]) &&
 451                                 '\\' == word[i - 1]) {
 452                         j++;
 453                         continue;
 454                 }
 455
 456                 if (0 == j)
 457                         continue;
 458                 assert(i >= j);
 459                 pword(p, &word[i - j], j);
 460                 j = 0;
 461         }
 462         if (j > 0) {
 463                 assert(i >= j);
 464                 pword(p, &word[i - j], j);
 465         }
 466 }
 467
 468
 469 /*
 470  * This is the main function for printing out nodes.  It's constituted
 471  * of PRE and POST functions, which correspond to prefix and infix
 472  * processing.  The termpair structure allows data to persist between
 473  * prefix and postfix invocations.
 474  */
 475 static void
 476 body(struct termp *p, struct termpair *ppair,
 477                 const struct mdoc_meta *meta,
 478                 const struct mdoc_node *node)
 479 {
 480         int              dochild;
 481         struct termpair  pair;
 482
 483         /* Some quick sanity-checking. */
 484
 485         sanity(node);
 486
 487         /* Pre-processing. */
 488
 489         dochild = 1;
 490         pair.ppair = ppair;
 491         pair.type = 0;
 492         pair.offset = pair.rmargin = 0;
 493         pair.flag = 0;
 494         pair.count = 0;
 495
 496         if (MDOC_TEXT != node->type) {
 497                 if (termacts[node->tok].pre)
 498                         if ( ! (*termacts[node->tok].pre)(p, &pair, meta, node))
 499                                 dochild = 0;
 500         } else /* MDOC_TEXT == node->type */
 501                 word(p, node->string);
 502
 503         /* Children. */
 504
 505         if (TERMPAIR_FLAG & pair.type)
 506                 p->flags |= pair.flag;
 507
 508         if (dochild && node->child)
 509                 body(p, &pair, meta, node->child);
 510
 511         if (TERMPAIR_FLAG & pair.type)
 512                 p->flags &= ~pair.flag;
 513
 514         /* Post-processing. */
 515
 516         if (MDOC_TEXT != node->type)
 517                 if (termacts[node->tok].post)
 518                         (*termacts[node->tok].post)(p, &pair, meta, node);
 519
 520         /* Siblings. */
 521
 522         if (node->next)
 523                 body(p, ppair, meta, node->next);
 524 }
 525
 526
 527 static void
 528 footer(struct termp *p, const struct mdoc_meta *meta)
 529 {
 530         struct tm       *tm;
 531         char            *buf, *os;
 532
 533         if (NULL == (buf = malloc(p->rmargin)))
 534                 err(1, "malloc");
 535         if (NULL == (os = malloc(p->rmargin)))
 536                 err(1, "malloc");
 537
 538         tm = localtime(&meta->date);
 539
 540 #ifdef __OpenBSD__
 541         if (NULL == strftime(buf, p->rmargin, "%B %d, %Y", tm))
 542 #else
 543         if (0 == strftime(buf, p->rmargin, "%B %d, %Y", tm))
 544 #endif
 545                 err(1, "strftime");
 546
 547         (void)strlcpy(os, meta->os, p->rmargin);
 548
 549         /*
 550          * This is /slightly/ different from regular groff output
 551          * because we don't have page numbers.  Print the following:
 552          *
 553          * OS                                            MDOCDATE
 554          */
 555
 556         vspace(p);
 557
 558         p->flags |= TERMP_NOSPACE | TERMP_NOBREAK;
 559         p->rmargin = p->maxrmargin - strlen(buf);
 560         p->offset = 0;
 561
 562         word(p, os);
 563         flushln(p);
 564
 565         p->flags |= TERMP_NOLPAD | TERMP_NOSPACE;
 566         p->offset = p->rmargin;
 567         p->rmargin = p->maxrmargin;
 568         p->flags &= ~TERMP_NOBREAK;
 569
 570         word(p, buf);
 571         flushln(p);
 572
 573         free(buf);
 574         free(os);
 575 }
 576
 577
 578 static void
 579 header(struct termp *p, const struct mdoc_meta *meta)
 580 {
 581         char            *buf, *title, *bufp;
 582
 583         p->rmargin = p->maxrmargin;
 584         p->offset = 0;
 585
 586         if (NULL == (buf = malloc(p->rmargin)))
 587                 err(1, "malloc");
 588         if (NULL == (title = malloc(p->rmargin)))
 589                 err(1, "malloc");
 590
 591         /*
 592          * The header is strange.  It has three components, which are
 593          * really two with the first duplicated.  It goes like this:
 594          *
 595          * IDENTIFIER              TITLE                   IDENTIFIER
 596          *
 597          * The IDENTIFIER is NAME(SECTION), which is the command-name
 598          * (if given, or "unknown" if not) followed by the manual page
 599          * section.  These are given in `Dt'.  The TITLE is a free-form
 600          * string depending on the manual volume.  If not specified, it
 601          * switches on the manual section.
 602          */
 603
 604         assert(meta->vol);
 605         (void)strlcpy(buf, meta->vol, p->rmargin);
 606
 607         if (meta->arch) {
 608                 (void)strlcat(buf, " (", p->rmargin);
 609                 (void)strlcat(buf, meta->arch, p->rmargin);
 610                 (void)strlcat(buf, ")", p->rmargin);
 611         }
 612
 613         (void)snprintf(title, p->rmargin, "%s(%d)",
 614                         meta->title, meta->msec);
 615
 616         for (bufp = title; *bufp; bufp++)
 617                 *bufp = toupper((u_char)*bufp);
 618
 619         p->offset = 0;
 620         p->rmargin = (p->maxrmargin - strlen(buf)) / 2;
 621         p->flags |= TERMP_NOBREAK | TERMP_NOSPACE;
 622
 623         word(p, title);
 624         flushln(p);
 625
 626         p->flags |= TERMP_NOLPAD | TERMP_NOSPACE;
 627         p->offset = p->rmargin;
 628         p->rmargin = p->maxrmargin - strlen(title);
 629
 630         word(p, buf);
 631         flushln(p);
 632
 633         p->offset = p->rmargin;
 634         p->rmargin = p->maxrmargin;
 635         p->flags &= ~TERMP_NOBREAK;
 636         p->flags |= TERMP_NOLPAD | TERMP_NOSPACE;
 637
 638         word(p, title);
 639         flushln(p);
 640
 641         p->rmargin = p->maxrmargin;
 642         p->offset = 0;
 643         p->flags &= ~TERMP_NOSPACE;
 644
 645         free(title);
 646         free(buf);
 647 }
 648
 649
 650 /*
 651  * Determine the symbol indicated by an escape sequences, that is, one
 652  * starting with a backslash.  Once done, we pass this value into the
 653  * output buffer by way of the symbol table.
 654  */
 655 static void
 656 nescape(struct termp *p, const char *word, size_t len)
 657 {
 658         struct termenc  *enc;
 659
 660         switch (len) {
 661         case (1):
 662                 enc = termenc1;
 663                 break;
 664         case (2):
 665                 enc = termenc2;
 666                 break;
 667         default:
 668                 warnx("unsupported %zu-byte escape sequence", len);
 669                 return;
 670         }
 671
 672         for ( ; enc->enc; enc++)
 673                 if (0 == memcmp(enc->enc, word, len)) {
 674                         symbola(p, enc->sym);
 675                         return;
 676                 }
 677
 678         warnx("unsupported %zu-byte escape sequence", len);
 679 }
 680
 681
 682 /*
 683  * Handle an escape sequence: determine its length and pass it to the
 684  * escape-symbol look table.  Note that we assume mdoc(3) has validated
 685  * the escape sequence (we assert upon badly-formed escape sequences).
 686  */
 687 static void
 688 pescape(struct termp *p, const char *word, size_t *i, size_t len)
 689 {
 690         size_t           j;
 691
 692         if (++(*i) >= len) {
 693                 warnx("ignoring bad escape sequence");
 694                 return;
 695         }
 696
 697         if ('(' == word[*i]) {
 698                 (*i)++;
 699                 if (*i + 1 >= len) {
 700                         warnx("ignoring bad escape sequence");
 701                         return;
 702                 }
 703                 nescape(p, &word[*i], 2);
 704                 (*i)++;
 705                 return;
 706
 707         } else if ('*' == word[*i]) {
 708                 (*i)++;
 709                 if (*i >= len) {
 710                         warnx("ignoring bad escape sequence");
 711                         return;
 712                 }
 713                 switch (word[*i]) {
 714                 case ('('):
 715                         (*i)++;
 716                         if (*i + 1 >= len) {
 717                                 warnx("ignoring bad escape sequence");
 718                                 return;
 719                         }
 720                         nescape(p, &word[*i], 2);
 721                         (*i)++;
 722                         return;
 723                 case ('['):
 724                         break;
 725                 default:
 726                         nescape(p, &word[*i], 1);
 727                         return;
 728                 }
 729
 730         } else if ('[' != word[*i]) {
 731                 nescape(p, &word[*i], 1);
 732                 return;
 733         }
 734
 735         (*i)++;
 736         for (j = 0; word[*i] && ']' != word[*i]; (*i)++, j++)
 737                 /* Loop... */ ;
 738
 739         if (0 == word[*i]) {
 740                 warnx("ignoring bad escape sequence");
 741                 return;
 742         }
 743         nescape(p, &word[*i - j], j);
 744 }
 745
 746
 747 /*
 748  * Handle pwords, partial words, which may be either a single word or a
 749  * phrase that cannot be broken down (such as a literal string).  This
 750  * handles word styling.
 751  */
 752 static void
 753 pword(struct termp *p, const char *word, size_t len)
 754 {
 755         size_t           i;
 756
 757         if ( ! (TERMP_NOSPACE & p->flags) &&
 758                         ! (TERMP_LITERAL & p->flags))
 759                 chara(p, ' ');
 760
 761         if ( ! (p->flags & TERMP_NONOSPACE))
 762                 p->flags &= ~TERMP_NOSPACE;
 763
 764         /*
 765          * XXX - if literal and underlining, this will underline the
 766          * spaces between literal words.
 767          */
 768
 769         if (p->flags & TERMP_BOLD)
 770                 stylea(p, TERMSTYLE_BOLD);
 771         if (p->flags & TERMP_UNDERLINE)
 772                 stylea(p, TERMSTYLE_UNDER);
 773
 774         for (i = 0; i < len; i++) {
 775                 if ('\\' == word[i]) {
 776                         pescape(p, word, &i, len);
 777                         continue;
 778                 }
 779                 chara(p, word[i]);
 780         }
 781
 782         if (p->flags & TERMP_BOLD ||
 783                         p->flags & TERMP_UNDERLINE)
 784                 stylea(p, TERMSTYLE_CLEAR);
 785 }
 786
 787
 788 /*
 789  * Add a symbol to the output line buffer.
 790  */
 791 static void
 792 symbola(struct termp *p, enum tsym sym)
 793 {
 794
 795         assert(p->symtab[sym].sym);
 796         stringa(p, p->symtab[sym].sym, p->symtab[sym].sz);
 797 }
 798
 799
 800 /*
 801  * Add a style to the output line buffer.
 802  */
 803 static void
 804 stylea(struct termp *p, enum tstyle style)
 805 {
 806
 807         assert(p->styletab[style].sym);
 808         stringa(p, p->styletab[style].sym, p->styletab[style].sz);
 809 }
 810
 811
 812 /*
 813  * Like chara() but for arbitrary-length buffers.  Resize the buffer by
 814  * a factor of two (if the buffer is less than that) or the buffer's
 815  * size.
 816  */
 817 static void
 818 stringa(struct termp *p, const char *c, size_t sz)
 819 {
 820         size_t           s;
 821
 822         if (0 == sz)
 823                 return;
 824
 825         s = sz > p->maxcols * 2 ? sz : p->maxcols * 2;
 826
 827         assert(c);
 828         if (p->col + sz >= p->maxcols) {
 829                 p->buf = realloc(p->buf, s);
 830                 if (NULL == p->buf)
 831                         err(1, "realloc");
 832                 p->maxcols = s;
 833         }
 834
 835         (void)memcpy(&p->buf[p->col], c, sz);
 836         p->col += sz;
 837 }
 838
 839
 840 /*
 841  * Insert a single character into the line-buffer.  If the buffer's
 842  * space is exceeded, then allocate more space by doubling the buffer
 843  * size.
 844  */
 845 static void
 846 chara(struct termp *p, char c)
 847 {
 848
 849         if (p->col + 1 >= p->maxcols) {
 850                 p->buf = realloc(p->buf, p->maxcols * 2);
 851                 if (NULL == p->buf)
 852                         err(1, "malloc");
 853                 p->maxcols *= 2;
 854         }
 855         p->buf[(p->col)++] = c;
 856 }
 857
 858
 859 static void
 860 sanity(const struct mdoc_node *n)
 861 {
 862
 863         switch (n->type) {
 864         case (MDOC_TEXT):
 865                 if (n->child)
 866                         errx(1, "regular form violated (1)");
 867                 if (NULL == n->parent)
 868                         errx(1, "regular form violated (2)");
 869                 if (NULL == n->string)
 870                         errx(1, "regular form violated (3)");
 871                 switch (n->parent->type) {
 872                 case (MDOC_TEXT):
 873                         /* FALLTHROUGH */
 874                 case (MDOC_ROOT):
 875                         errx(1, "regular form violated (4)");
 876                         /* NOTREACHED */
 877                 default:
 878                         break;
 879                 }
 880                 break;
 881         case (MDOC_ELEM):
 882                 if (NULL == n->parent)
 883                         errx(1, "regular form violated (5)");
 884                 switch (n->parent->type) {
 885                 case (MDOC_TAIL):
 886                         /* FALLTHROUGH */
 887                 case (MDOC_BODY):
 888                         /* FALLTHROUGH */
 889                 case (MDOC_HEAD):
 890                         break;
 891                 default:
 892                         errx(1, "regular form violated (6)");
 893                         /* NOTREACHED */
 894                 }
 895                 if (n->child) switch (n->child->type) {
 896                 case (MDOC_TEXT):
 897                         break;
 898                 default:
 899                         errx(1, "regular form violated (7(");
 900                         /* NOTREACHED */
 901                 }
 902                 break;
 903         case (MDOC_HEAD):
 904                 /* FALLTHROUGH */
 905         case (MDOC_BODY):
 906                 /* FALLTHROUGH */
 907         case (MDOC_TAIL):
 908                 if (NULL == n->parent)
 909                         errx(1, "regular form violated (8)");
 910                 if (MDOC_BLOCK != n->parent->type)
 911                         errx(1, "regular form violated (9)");
 912                 if (n->child) switch (n->child->type) {
 913                 case (MDOC_BLOCK):
 914                         /* FALLTHROUGH */
 915                 case (MDOC_ELEM):
 916                         /* FALLTHROUGH */
 917                 case (MDOC_TEXT):
 918                         break;
 919                 default:
 920                         errx(1, "regular form violated (a)");
 921                         /* NOTREACHED */
 922                 }
 923                 break;
 924         case (MDOC_BLOCK):
 925                 if (NULL == n->parent)
 926                         errx(1, "regular form violated (b)");
 927                 if (NULL == n->child)
 928                         errx(1, "regular form violated (c)");
 929                 switch (n->parent->type) {
 930                 case (MDOC_ROOT):
 931                         /* FALLTHROUGH */
 932                 case (MDOC_HEAD):
 933                         /* FALLTHROUGH */
 934                 case (MDOC_BODY):
 935                         /* FALLTHROUGH */
 936                 case (MDOC_TAIL):
 937                         break;
 938                 default:
 939                         errx(1, "regular form violated (d)");
 940                         /* NOTREACHED */
 941                 }
 942                 switch (n->child->type) {
 943                 case (MDOC_ROOT):
 944                         /* FALLTHROUGH */
 945                 case (MDOC_ELEM):
 946                         errx(1, "regular form violated (e)");
 947                         /* NOTREACHED */
 948                 default:
 949                         break;
 950                 }
 951                 break;
 952         case (MDOC_ROOT):
 953                 if (n->parent)
 954                         errx(1, "regular form violated (f)");
 955                 if (NULL == n->child)
 956                         errx(1, "regular form violated (10)");
 957                 switch (n->child->type) {
 958                 case (MDOC_BLOCK):
 959                         break;
 960                 default:
 961                         errx(1, "regular form violated (11)");
 962                         /* NOTREACHED */
 963                 }
 964                 break;
 965         }
 966 }
 967