term.c

   1 /*      $Id: term.c,v 1.150 2010/06/26 15:36:37 kristaps Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@bsd.lv>
   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 above
   7  * copyright notice and this permission notice appear in all copies.
   8  *
   9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  16  */
  17 #ifdef HAVE_CONFIG_H
  18 #include "config.h"
  19 #endif
  20
  21 #include <sys/types.h>
  22
  23 #include <assert.h>
  24 #include <ctype.h>
  25 #include <stdint.h>
  26 #include <stdio.h>
  27 #include <stdlib.h>
  28 #include <string.h>
  29
  30 #include "mandoc.h"
  31 #include "chars.h"
  32 #include "out.h"
  33 #include "term.h"
  34 #include "main.h"
  35
  36 static  void              spec(struct termp *, const char *, size_t);
  37 static  void              res(struct termp *, const char *, size_t);
  38 static  void              buffera(struct termp *, const char *, size_t);
  39 static  void              bufferc(struct termp *, char);
  40 static  void              adjbuf(struct termp *p, size_t);
  41 static  void              encode(struct termp *, const char *, size_t);
  42
  43
  44 void
  45 term_free(struct termp *p)
  46 {
  47
  48         if (p->buf)
  49                 free(p->buf);
  50         if (p->symtab)
  51                 chars_free(p->symtab);
  52
  53         free(p);
  54 }
  55
  56
  57 void
  58 term_begin(struct termp *p, term_margin head,
  59                 term_margin foot, const void *arg)
  60 {
  61
  62         p->headf = head;
  63         p->footf = foot;
  64         p->argf = arg;
  65         (*p->begin)(p);
  66 }
  67
  68
  69 void
  70 term_end(struct termp *p)
  71 {
  72
  73         (*p->end)(p);
  74 }
  75
  76
  77 struct termp *
  78 term_alloc(enum termenc enc)
  79 {
  80         struct termp    *p;
  81
  82         p = calloc(1, sizeof(struct termp));
  83         if (NULL == p) {
  84                 perror(NULL);
  85                 exit(EXIT_FAILURE);
  86         }
  87
  88         p->enc = enc;
  89         return(p);
  90 }
  91
  92
  93 /*
  94  * Flush a line of text.  A "line" is loosely defined as being something
  95  * that should be followed by a newline, regardless of whether it's
  96  * broken apart by newlines getting there.  A line can also be a
  97  * fragment of a columnar list (`Bl -tag' or `Bl -column'), which does
  98  * not have a trailing newline.
  99  *
 100  * The following flags may be specified:
 101  *
 102  *  - TERMP_NOLPAD: when beginning to write the line, don't left-pad the
 103  *    offset value.  This is useful when doing columnar lists where the
 104  *    prior column has right-padded.
 105  *
 106  *  - TERMP_NOBREAK: this is the most important and is used when making
 107  *    columns.  In short: don't print a newline and instead pad to the
 108  *    right margin.  Used in conjunction with TERMP_NOLPAD.
 109  *
 110  *  - TERMP_TWOSPACE: when padding, make sure there are at least two
 111  *    space characters of padding.  Otherwise, rather break the line.
 112  *
 113  *  - TERMP_DANGLE: don't newline when TERMP_NOBREAK is specified and
 114  *    the line is overrun, and don't pad-right if it's underrun.
 115  *
 116  *  - TERMP_HANG: like TERMP_DANGLE, but doesn't newline when
 117  *    overruning, instead save the position and continue at that point
 118  *    when the next invocation.
 119  *
 120  *  In-line line breaking:
 121  *
 122  *  If TERMP_NOBREAK is specified and the line overruns the right
 123  *  margin, it will break and pad-right to the right margin after
 124  *  writing.  If maxrmargin is violated, it will break and continue
 125  *  writing from the right-margin, which will lead to the above scenario
 126  *  upon exit.  Otherwise, the line will break at the right margin.
 127  */
 128 void
 129 term_flushln(struct termp *p)
 130 {
 131         int              i;     /* current input position in p->buf */
 132         size_t           vis;   /* current visual position on output */
 133         size_t           vbl;   /* number of blanks to prepend to output */
 134         size_t           vend;  /* end of word visual position on output */
 135         size_t           bp;    /* visual right border position */
 136         int              j;     /* temporary loop index */
 137         int              jhy;   /* last hyphen before line overflow */
 138         size_t           maxvis, mmax;
 139
 140         /*
 141          * First, establish the maximum columns of "visible" content.
 142          * This is usually the difference between the right-margin and
 143          * an indentation, but can be, for tagged lists or columns, a
 144          * small set of values.
 145          */
 146
 147         assert(p->offset < p->rmargin);
 148
 149         maxvis = (int)(p->rmargin - p->offset) - p->overstep < 0 ?
 150                 /* LINTED */
 151                 0 : p->rmargin - p->offset - p->overstep;
 152         mmax = (int)(p->maxrmargin - p->offset) - p->overstep < 0 ?
 153                 /* LINTED */
 154                 0 : p->maxrmargin - p->offset - p->overstep;
 155
 156         bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
 157
 158         /*
 159          * Indent the first line of a paragraph.
 160          */
 161         vbl = p->flags & TERMP_NOLPAD ? 0 : p->offset;
 162
 163         /*
 164          * FIXME: if bp is zero, we still output the first word before
 165          * breaking the line.
 166          */
 167
 168         vis = vend = i = 0;
 169         while (i < (int)p->col) {
 170
 171                 /*
 172                  * Handle literal tab characters.
 173                  */
 174                 for (j = i; j < (int)p->col; j++) {
 175                         if ('\t' != p->buf[j])
 176                                 break;
 177                         vend = (vis/p->tabwidth+1)*p->tabwidth;
 178                         vbl += vend - vis;
 179                         vis = vend;
 180                 }
 181
 182                 /*
 183                  * Count up visible word characters.  Control sequences
 184                  * (starting with the CSI) aren't counted.  A space
 185                  * generates a non-printing word, which is valid (the
 186                  * space is printed according to regular spacing rules).
 187                  */
 188
 189                 /* LINTED */
 190                 for (jhy = 0; j < (int)p->col; j++) {
 191                         if ((j && ' ' == p->buf[j]) || '\t' == p->buf[j])
 192                                 break;
 193                         if (8 != p->buf[j]) {
 194                                 if (vend > vis && vend < bp &&
 195                                     ASCII_HYPH == p->buf[j])
 196                                         jhy = j;
 197                                 vend++;
 198                         } else
 199                                 vend--;
 200                 }
 201
 202                 /*
 203                  * Find out whether we would exceed the right margin.
 204                  * If so, break to the next line.
 205                  */
 206                 if (vend > bp && 0 == jhy && vis > 0) {
 207                         vend -= vis;
 208                         (*p->endline)(p);
 209                         if (TERMP_NOBREAK & p->flags) {
 210                                 p->viscol = p->rmargin;
 211                                 (*p->advance)(p, p->rmargin);
 212                                 vend += p->rmargin - p->offset;
 213                         } else {
 214                                 p->viscol = 0;
 215                                 vbl = p->offset;
 216                         }
 217
 218                         /* Remove the p->overstep width. */
 219
 220                         bp += (int)/* LINTED */
 221                                 p->overstep;
 222                         p->overstep = 0;
 223                 }
 224
 225                 /*
 226                  * Skip leading tabs, they were handled above.
 227                  */
 228                 while (i < (int)p->col && '\t' == p->buf[i])
 229                         i++;
 230
 231                 /* Write out the [remaining] word. */
 232                 for ( ; i < (int)p->col; i++) {
 233                         if (vend > bp && jhy > 0 && i > jhy)
 234                                 break;
 235                         if ('\t' == p->buf[i])
 236                                 break;
 237                         if (' ' == p->buf[i]) {
 238                                 while (' ' == p->buf[i]) {
 239                                         vbl++;
 240                                         i++;
 241                                 }
 242                                 break;
 243                         }
 244                         if (ASCII_NBRSP == p->buf[i]) {
 245                                 vbl++;
 246                                 continue;
 247                         }
 248
 249                         /*
 250                          * Now we definitely know there will be
 251                          * printable characters to output,
 252                          * so write preceding white space now.
 253                          */
 254                         if (vbl) {
 255                                 (*p->advance)(p, vbl);
 256                                 p->viscol += vbl;
 257                                 vbl = 0;
 258                         }
 259
 260                         if (ASCII_HYPH == p->buf[i])
 261                                 (*p->letter)(p, '-');
 262                         else
 263                                 (*p->letter)(p, p->buf[i]);
 264
 265                         p->viscol += 1;
 266                 }
 267                 vend += vbl;
 268                 vis = vend;
 269         }
 270
 271         p->col = 0;
 272         p->overstep = 0;
 273
 274         if ( ! (TERMP_NOBREAK & p->flags)) {
 275                 p->viscol = 0;
 276                 (*p->endline)(p);
 277                 return;
 278         }
 279
 280         if (TERMP_HANG & p->flags) {
 281                 /* We need one blank after the tag. */
 282                 p->overstep = /* LINTED */
 283                         vis - maxvis + 1;
 284
 285                 /*
 286                  * Behave exactly the same way as groff:
 287                  * If we have overstepped the margin, temporarily move
 288                  * it to the right and flag the rest of the line to be
 289                  * shorter.
 290                  * If we landed right at the margin, be happy.
 291                  * If we are one step before the margin, temporarily
 292                  * move it one step LEFT and flag the rest of the line
 293                  * to be longer.
 294                  */
 295                 if (p->overstep >= -1) {
 296                         assert((int)maxvis + p->overstep >= 0);
 297                         /* LINTED */
 298                         maxvis += p->overstep;
 299                 } else
 300                         p->overstep = 0;
 301
 302         } else if (TERMP_DANGLE & p->flags)
 303                 return;
 304
 305         /* Right-pad. */
 306         if (maxvis > vis + /* LINTED */
 307                         ((TERMP_TWOSPACE & p->flags) ? 1 : 0)) {
 308                 p->viscol += maxvis - vis;
 309                 (*p->advance)(p, maxvis - vis);
 310                 vis += (maxvis - vis);
 311         } else {        /* ...or newline break. */
 312                 (*p->endline)(p);
 313                 p->viscol = p->rmargin;
 314                 (*p->advance)(p, p->rmargin);
 315         }
 316 }
 317
 318
 319 /*
 320  * A newline only breaks an existing line; it won't assert vertical
 321  * space.  All data in the output buffer is flushed prior to the newline
 322  * assertion.
 323  */
 324 void
 325 term_newln(struct termp *p)
 326 {
 327
 328         p->flags |= TERMP_NOSPACE;
 329         if (0 == p->col && 0 == p->viscol) {
 330                 p->flags &= ~TERMP_NOLPAD;
 331                 return;
 332         }
 333         term_flushln(p);
 334         p->flags &= ~TERMP_NOLPAD;
 335 }
 336
 337
 338 /*
 339  * Asserts a vertical space (a full, empty line-break between lines).
 340  * Note that if used twice, this will cause two blank spaces and so on.
 341  * All data in the output buffer is flushed prior to the newline
 342  * assertion.
 343  */
 344 void
 345 term_vspace(struct termp *p)
 346 {
 347
 348         term_newln(p);
 349         p->viscol = 0;
 350         (*p->endline)(p);
 351 }
 352
 353
 354 static void
 355 spec(struct termp *p, const char *word, size_t len)
 356 {
 357         const char      *rhs;
 358         size_t           sz;
 359
 360         rhs = chars_a2ascii(p->symtab, word, len, &sz);
 361         if (rhs)
 362                 encode(p, rhs, sz);
 363 }
 364
 365
 366 static void
 367 res(struct termp *p, const char *word, size_t len)
 368 {
 369         const char      *rhs;
 370         size_t           sz;
 371
 372         rhs = chars_a2res(p->symtab, word, len, &sz);
 373         if (rhs)
 374                 encode(p, rhs, sz);
 375 }
 376
 377
 378 void
 379 term_fontlast(struct termp *p)
 380 {
 381         enum termfont    f;
 382
 383         f = p->fontl;
 384         p->fontl = p->fontq[p->fonti];
 385         p->fontq[p->fonti] = f;
 386 }
 387
 388
 389 void
 390 term_fontrepl(struct termp *p, enum termfont f)
 391 {
 392
 393         p->fontl = p->fontq[p->fonti];
 394         p->fontq[p->fonti] = f;
 395 }
 396
 397
 398 void
 399 term_fontpush(struct termp *p, enum termfont f)
 400 {
 401
 402         assert(p->fonti + 1 < 10);
 403         p->fontl = p->fontq[p->fonti];
 404         p->fontq[++p->fonti] = f;
 405 }
 406
 407
 408 const void *
 409 term_fontq(struct termp *p)
 410 {
 411
 412         return(&p->fontq[p->fonti]);
 413 }
 414
 415
 416 enum termfont
 417 term_fonttop(struct termp *p)
 418 {
 419
 420         return(p->fontq[p->fonti]);
 421 }
 422
 423
 424 void
 425 term_fontpopq(struct termp *p, const void *key)
 426 {
 427
 428         while (p->fonti >= 0 && key != &p->fontq[p->fonti])
 429                 p->fonti--;
 430         assert(p->fonti >= 0);
 431 }
 432
 433
 434 void
 435 term_fontpop(struct termp *p)
 436 {
 437
 438         assert(p->fonti);
 439         p->fonti--;
 440 }
 441
 442
 443 /*
 444  * Handle pwords, partial words, which may be either a single word or a
 445  * phrase that cannot be broken down (such as a literal string).  This
 446  * handles word styling.
 447  */
 448 void
 449 term_word(struct termp *p, const char *word)
 450 {
 451         const char      *sv, *seq;
 452         int              sz;
 453         size_t           ssz;
 454         enum roffdeco    deco;
 455
 456         sv = word;
 457
 458         if (word[0] && '\0' == word[1])
 459                 switch (word[0]) {
 460                 case('.'):
 461                         /* FALLTHROUGH */
 462                 case(','):
 463                         /* FALLTHROUGH */
 464                 case(';'):
 465                         /* FALLTHROUGH */
 466                 case(':'):
 467                         /* FALLTHROUGH */
 468                 case('?'):
 469                         /* FALLTHROUGH */
 470                 case('!'):
 471                         /* FALLTHROUGH */
 472                 case(')'):
 473                         /* FALLTHROUGH */
 474                 case(']'):
 475                         if ( ! (TERMP_IGNDELIM & p->flags))
 476                                 p->flags |= TERMP_NOSPACE;
 477                         break;
 478                 default:
 479                         break;
 480                 }
 481
 482         if ( ! (TERMP_NOSPACE & p->flags)) {
 483                 bufferc(p, ' ');
 484                 if (TERMP_SENTENCE & p->flags)
 485                         bufferc(p, ' ');
 486         }
 487
 488         if ( ! (p->flags & TERMP_NONOSPACE))
 489                 p->flags &= ~TERMP_NOSPACE;
 490
 491         p->flags &= ~TERMP_SENTENCE;
 492
 493         /* FIXME: use strcspn. */
 494
 495         while (*word) {
 496                 if ('\\' != *word) {
 497                         encode(p, word, 1);
 498                         word++;
 499                         continue;
 500                 }
 501
 502                 seq = ++word;
 503                 sz = a2roffdeco(&deco, &seq, &ssz);
 504
 505                 switch (deco) {
 506                 case (DECO_RESERVED):
 507                         res(p, seq, ssz);
 508                         break;
 509                 case (DECO_SPECIAL):
 510                         spec(p, seq, ssz);
 511                         break;
 512                 case (DECO_BOLD):
 513                         term_fontrepl(p, TERMFONT_BOLD);
 514                         break;
 515                 case (DECO_ITALIC):
 516                         term_fontrepl(p, TERMFONT_UNDER);
 517                         break;
 518                 case (DECO_ROMAN):
 519                         term_fontrepl(p, TERMFONT_NONE);
 520                         break;
 521                 case (DECO_PREVIOUS):
 522                         term_fontlast(p);
 523                         break;
 524                 default:
 525                         break;
 526                 }
 527
 528                 word += sz;
 529                 if (DECO_NOSPACE == deco && '\0' == *word)
 530                         p->flags |= TERMP_NOSPACE;
 531         }
 532
 533         /*
 534          * Note that we don't process the pipe: the parser sees it as
 535          * punctuation, but we don't in terms of typography.
 536          */
 537         if (sv[0] && 0 == sv[1])
 538                 switch (sv[0]) {
 539                 case('('):
 540                         /* FALLTHROUGH */
 541                 case('['):
 542                         p->flags |= TERMP_NOSPACE;
 543                         break;
 544                 default:
 545                         break;
 546                 }
 547 }
 548
 549
 550 static void
 551 adjbuf(struct termp *p, size_t sz)
 552 {
 553
 554         if (0 == p->maxcols)
 555                 p->maxcols = 1024;
 556         while (sz >= p->maxcols)
 557                 p->maxcols <<= 2;
 558
 559         p->buf = realloc(p->buf, p->maxcols);
 560         if (NULL == p->buf) {
 561                 perror(NULL);
 562                 exit(EXIT_FAILURE);
 563         }
 564 }
 565
 566
 567 static void
 568 buffera(struct termp *p, const char *word, size_t sz)
 569 {
 570
 571         if (p->col + sz >= p->maxcols)
 572                 adjbuf(p, p->col + sz);
 573
 574         memcpy(&p->buf[(int)p->col], word, sz);
 575         p->col += sz;
 576 }
 577
 578
 579 static void
 580 bufferc(struct termp *p, char c)
 581 {
 582
 583         if (p->col + 1 >= p->maxcols)
 584                 adjbuf(p, p->col + 1);
 585
 586         p->buf[(int)p->col++] = c;
 587 }
 588
 589
 590 static void
 591 encode(struct termp *p, const char *word, size_t sz)
 592 {
 593         enum termfont     f;
 594         int               i;
 595
 596         /*
 597          * Encode and buffer a string of characters.  If the current
 598          * font mode is unset, buffer directly, else encode then buffer
 599          * character by character.
 600          */
 601
 602         if (TERMFONT_NONE == (f = term_fonttop(p))) {
 603                 buffera(p, word, sz);
 604                 return;
 605         }
 606
 607         for (i = 0; i < (int)sz; i++) {
 608                 if ( ! isgraph((u_char)word[i])) {
 609                         bufferc(p, word[i]);
 610                         continue;
 611                 }
 612
 613                 if (TERMFONT_UNDER == f)
 614                         bufferc(p, '_');
 615                 else
 616                         bufferc(p, word[i]);
 617
 618                 bufferc(p, 8);
 619                 bufferc(p, word[i]);
 620         }
 621 }
 622
 623
 624 size_t
 625 term_len(const struct termp *p, size_t sz)
 626 {
 627
 628         return((*p->width)(p, ' ') * sz);
 629 }
 630
 631
 632 size_t
 633 term_strlen(const struct termp *p, const char *cp)
 634 {
 635         size_t           sz;
 636
 637         for (sz = 0; *cp; cp++)
 638                 sz += (*p->width)(p, *cp);
 639
 640         return(sz);
 641 }
 642
 643
 644 size_t
 645 term_vspan(const struct termp *p, const struct roffsu *su)
 646 {
 647         double           r;
 648
 649         switch (su->unit) {
 650         case (SCALE_CM):
 651                 r = su->scale * 2;
 652                 break;
 653         case (SCALE_IN):
 654                 r = su->scale * 6;
 655                 break;
 656         case (SCALE_PC):
 657                 r = su->scale;
 658                 break;
 659         case (SCALE_PT):
 660                 r = su->scale / 8;
 661                 break;
 662         case (SCALE_MM):
 663                 r = su->scale / 1000;
 664                 break;
 665         case (SCALE_VS):
 666                 r = su->scale;
 667                 break;
 668         default:
 669                 r = su->scale - 1;
 670                 break;
 671         }
 672
 673         if (r < 0.0)
 674                 r = 0.0;
 675         return(/* LINTED */(size_t)
 676                         r);
 677 }
 678
 679
 680 size_t
 681 term_hspan(const struct termp *p, const struct roffsu *su)
 682 {
 683         double           r;
 684
 685         /* XXX: CM, IN, and PT are approximations. */
 686
 687         switch (su->unit) {
 688         case (SCALE_CM):
 689                 r = 4 * su->scale;
 690                 break;
 691         case (SCALE_IN):
 692                 /* XXX: this is an approximation. */
 693                 r = 10 * su->scale;
 694                 break;
 695         case (SCALE_PC):
 696                 r = (10 * su->scale) / 6;
 697                 break;
 698         case (SCALE_PT):
 699                 r = (10 * su->scale) / 72;
 700                 break;
 701         case (SCALE_MM):
 702                 r = su->scale / 1000; /* FIXME: double-check. */
 703                 break;
 704         case (SCALE_VS):
 705                 r = su->scale * 2 - 1; /* FIXME: double-check. */
 706                 break;
 707         default:
 708                 r = su->scale;
 709                 break;
 710         }
 711
 712         if (r < 0.0)
 713                 r = 0.0;
 714         return((size_t)/* LINTED */
 715                         r);
 716 }
 717
 718