external/bsd/less/dist/line.c

   1 /*      $NetBSD: line.c,v 1.4 2013/09/04 19:44:21 tron Exp $    */
   2
   3 /*
   4  * Copyright (C) 1984-2012  Mark Nudelman
   5  *
   6  * You may distribute under the terms of either the GNU General Public
   7  * License or the Less License, as specified in the README file.
   8  *
   9  * For more information, see the README file.
  10  */
  11
  12
  13 /*
  14  * Routines to manipulate the "line buffer".
  15  * The line buffer holds a line of output as it is being built
  16  * in preparation for output to the screen.
  17  */
  18
  19 #include "less.h"
  20 #include "charset.h"
  21
  22 static char *linebuf = NULL;    /* Buffer which holds the current output line */
  23 static char *attr = NULL;       /* Extension of linebuf to hold attributes */
  24 public int size_linebuf = 0;    /* Size of line buffer (and attr buffer) */
  25
  26 static int cshift;              /* Current left-shift of output line buffer */
  27 public int hshift;              /* Desired left-shift of output line buffer */
  28 public int tabstops[TABSTOP_MAX] = { 0 }; /* Custom tabstops */
  29 public int ntabstops = 1;       /* Number of tabstops */
  30 public int tabdefault = 8;      /* Default repeated tabstops */
  31 public POSITION highest_hilite; /* Pos of last hilite in file found so far */
  32
  33 static int curr;                /* Index into linebuf */
  34 static int column;              /* Printable length, accounting for
  35                                    backspaces, etc. */
  36 static int overstrike;          /* Next char should overstrike previous char */
  37 static int last_overstrike = AT_NORMAL;
  38 static int is_null_line;        /* There is no current line */
  39 static int lmargin;             /* Left margin */
  40 static char pendc;
  41 static POSITION pendpos;
  42 static char *end_ansi_chars;
  43 static char *mid_ansi_chars;
  44
  45 static int attr_swidth(int);
  46 static int attr_ewidth(int);
  47 static int do_append(LWCHAR, char *, POSITION);
  48 extern int sigs;
  49 extern int bs_mode;
  50 extern int linenums;
  51 extern int ctldisp;
  52 extern int twiddle;
  53 extern int binattr;
  54 extern int status_col;
  55 extern int auto_wrap, ignaw;
  56 extern int bo_s_width, bo_e_width;
  57 extern int ul_s_width, ul_e_width;
  58 extern int bl_s_width, bl_e_width;
  59 extern int so_s_width, so_e_width;
  60 extern int sc_width, sc_height;
  61 extern int utf_mode;
  62 extern POSITION start_attnpos;
  63 extern POSITION end_attnpos;
  64
  65 static char mbc_buf[MAX_UTF_CHAR_LEN];
  66 static int mbc_buf_len = 0;
  67 static int mbc_buf_index = 0;
  68 static POSITION mbc_pos;
  69
  70 /*
  71  * Initialize from environment variables.
  72  */
  73         public void
  74 init_line()
  75 {
  76         end_ansi_chars = lgetenv("LESSANSIENDCHARS");
  77         if (end_ansi_chars == NULL || *end_ansi_chars == '\0')
  78                 end_ansi_chars = "m";
  79
  80         mid_ansi_chars = lgetenv("LESSANSIMIDCHARS");
  81         if (mid_ansi_chars == NULL || *mid_ansi_chars == '\0')
  82                 mid_ansi_chars = "0123456789;[?!\"'#%()*+ ";
  83
  84         linebuf = (char *) ecalloc(LINEBUF_SIZE, sizeof(char));
  85         attr = (char *) ecalloc(LINEBUF_SIZE, sizeof(char));
  86         size_linebuf = LINEBUF_SIZE;
  87 }
  88
  89 /*
  90  * Expand the line buffer.
  91  */
  92         static int
  93 expand_linebuf()
  94 {
  95         /* Double the size of the line buffer. */
  96         int new_size = size_linebuf * 2;
  97
  98         /* Just realloc to expand the buffer, if we can. */
  99 #if HAVE_REALLOC
 100         char *new_buf = (char *) realloc(linebuf, new_size);
 101         char *new_attr = (char *) realloc(attr, new_size);
 102 #else
 103         char *new_buf = (char *) calloc(new_size, sizeof(char));
 104         char *new_attr = (char *) calloc(new_size, sizeof(char));
 105 #endif
 106         if (new_buf == NULL || new_attr == NULL)
 107         {
 108                 if (new_attr != NULL)
 109                         free(new_attr);
 110                 if (new_buf != NULL)
 111                         free(new_buf);
 112                 return 1;
 113         }
 114 #if HAVE_REALLOC
 115         /*
 116          * We realloc'd the buffers; they already have the old contents.
 117          */
 118         #if 0
 119         memset(new_buf + size_linebuf, 0, new_size - size_linebuf);
 120         memset(new_attr + size_linebuf, 0, new_size - size_linebuf);
 121         #endif
 122 #else
 123         /*
 124          * We just calloc'd the buffers; copy the old contents.
 125          */
 126         memcpy(new_buf, linebuf, size_linebuf * sizeof(char));
 127         memcpy(new_attr, attr, size_linebuf * sizeof(char));
 128         free(attr);
 129         free(linebuf);
 130 #endif
 131         linebuf = new_buf;
 132         attr = new_attr;
 133         size_linebuf = new_size;
 134         return 0;
 135 }
 136
 137 /*
 138  * Is a character ASCII?
 139  */
 140         public int
 141 is_ascii_char(ch)
 142         LWCHAR ch;
 143 {
 144         return (ch <= 0x7F);
 145 }
 146
 147 /*
 148  * Rewind the line buffer.
 149  */
 150         public void
 151 prewind()
 152 {
 153         curr = 0;
 154         column = 0;
 155         cshift = 0;
 156         overstrike = 0;
 157         last_overstrike = AT_NORMAL;
 158         mbc_buf_len = 0;
 159         is_null_line = 0;
 160         pendc = '\0';
 161         lmargin = 0;
 162         if (status_col)
 163                 lmargin += 1;
 164 }
 165
 166 /*
 167  * Insert the line number (of the given position) into the line buffer.
 168  */
 169         public void
 170 plinenum(pos)
 171         POSITION pos;
 172 {
 173         register LINENUM linenum = 0;
 174         register int i;
 175
 176         if (linenums == OPT_ONPLUS)
 177         {
 178                 /*
 179                  * Get the line number and put it in the current line.
 180                  * {{ Note: since find_linenum calls forw_raw_line,
 181                  *    it may seek in the input file, requiring the caller
 182                  *    of plinenum to re-seek if necessary. }}
 183                  * {{ Since forw_raw_line modifies linebuf, we must
 184                  *    do this first, before storing anything in linebuf. }}
 185                  */
 186                 linenum = find_linenum(pos);
 187         }
 188
 189         /*
 190          * Display a status column if the -J option is set.
 191          */
 192         if (status_col)
 193         {
 194                 linebuf[curr] = ' ';
 195                 if (start_attnpos != NULL_POSITION &&
 196                     pos >= start_attnpos && pos < end_attnpos)
 197                         attr[curr] = AT_NORMAL|AT_HILITE;
 198                 else
 199                         attr[curr] = AT_NORMAL;
 200                 curr++;
 201                 column++;
 202         }
 203         /*
 204          * Display the line number at the start of each line
 205          * if the -N option is set.
 206          */
 207         if (linenums == OPT_ONPLUS)
 208         {
 209                 char buf[INT_STRLEN_BOUND(pos) + 2];
 210                 int n;
 211
 212                 linenumtoa(linenum, buf);
 213                 n = strlen(buf);
 214                 if (n < MIN_LINENUM_WIDTH)
 215                         n = MIN_LINENUM_WIDTH;
 216                 sprintf(linebuf+curr, "%*s ", n, buf);
 217                 n++;  /* One space after the line number. */
 218                 for (i = 0; i < n; i++)
 219                         attr[curr+i] = AT_NORMAL;
 220                 curr += n;
 221                 column += n;
 222                 lmargin += n;
 223         }
 224
 225         /*
 226          * Append enough spaces to bring us to the lmargin.
 227          */
 228         while (column < lmargin)
 229         {
 230                 linebuf[curr] = ' ';
 231                 attr[curr++] = AT_NORMAL;
 232                 column++;
 233         }
 234 }
 235
 236 /*
 237  * Shift the input line left.
 238  * This means discarding N printable chars at the start of the buffer.
 239  */
 240         static void
 241 pshift(shift)
 242         int shift;
 243 {
 244         LWCHAR prev_ch = 0;
 245         unsigned char c;
 246         int shifted = 0;
 247         int to;
 248         int from;
 249         int len;
 250         int width;
 251         int prev_attr;
 252         int next_attr;
 253
 254         if (shift > column - lmargin)
 255                 shift = column - lmargin;
 256         if (shift > curr - lmargin)
 257                 shift = curr - lmargin;
 258
 259         to = from = lmargin;
 260         /*
 261          * We keep on going when shifted == shift
 262          * to get all combining chars.
 263          */
 264         while (shifted <= shift && from < curr)
 265         {
 266                 c = linebuf[from];
 267                 if (ctldisp == OPT_ONPLUS && IS_CSI_START(c))
 268                 {
 269                         /* Keep cumulative effect.  */
 270                         linebuf[to] = c;
 271                         attr[to++] = attr[from++];
 272                         while (from < curr && linebuf[from])
 273                         {
 274                                 linebuf[to] = linebuf[from];
 275                                 attr[to++] = attr[from];
 276                                 if (!is_ansi_middle(linebuf[from++]))
 277                                         break;
 278                         }
 279                         continue;
 280                 }
 281
 282                 width = 0;
 283
 284                 if (!IS_ASCII_OCTET(c) && utf_mode)
 285                 {
 286                         /* Assumes well-formedness validation already done.  */
 287                         LWCHAR ch;
 288
 289                         len = utf_len(c);
 290                         if (from + len > curr)
 291                                 break;
 292                         ch = get_wchar(linebuf + from);
 293                         if (!is_composing_char(ch) && !is_combining_char(prev_ch, ch))
 294                                 width = is_wide_char(ch) ? 2 : 1;
 295                         prev_ch = ch;
 296                 } else
 297                 {
 298                         len = 1;
 299                         if (c == '\b')
 300                                 /* XXX - Incorrect if several '\b' in a row.  */
 301                                 width = (utf_mode && is_wide_char(prev_ch)) ? -2 : -1;
 302                         else if (!control_char(c))
 303                                 width = 1;
 304                         prev_ch = 0;
 305                 }
 306
 307                 if (width == 2 && shift - shifted == 1) {
 308                         /* Should never happen when called by pshift_all().  */
 309                         attr[to] = attr[from];
 310                         /*
 311                          * Assume a wide_char will never be the first half of a
 312                          * combining_char pair, so reset prev_ch in case we're
 313                          * followed by a '\b'.
 314                          */
 315                         prev_ch = linebuf[to++] = ' ';
 316                         from += len;
 317                         shifted++;
 318                         continue;
 319                 }
 320
 321                 /* Adjust width for magic cookies. */
 322                 prev_attr = (to > 0) ? attr[to-1] : AT_NORMAL;
 323                 next_attr = (from + len < curr) ? attr[from + len] : prev_attr;
 324                 if (!is_at_equiv(attr[from], prev_attr) &&
 325                         !is_at_equiv(attr[from], next_attr))
 326                 {
 327                         width += attr_swidth(attr[from]);
 328                         if (from + len < curr)
 329                                 width += attr_ewidth(attr[from]);
 330                         if (is_at_equiv(prev_attr, next_attr))
 331                         {
 332                                 width += attr_ewidth(prev_attr);
 333                                 if (from + len < curr)
 334                                         width += attr_swidth(next_attr);
 335                         }
 336                 }
 337
 338                 if (shift - shifted < width)
 339                         break;
 340                 from += len;
 341                 shifted += width;
 342                 if (shifted < 0)
 343                         shifted = 0;
 344         }
 345         while (from < curr)
 346         {
 347                 linebuf[to] = linebuf[from];
 348                 attr[to++] = attr[from++];
 349         }
 350         curr = to;
 351         column -= shifted;
 352         cshift += shifted;
 353 }
 354
 355 /*
 356  *
 357  */
 358         public void
 359 pshift_all()
 360 {
 361         pshift(column);
 362 }
 363
 364 /*
 365  * Return the printing width of the start (enter) sequence
 366  * for a given character attribute.
 367  */
 368         static int
 369 attr_swidth(a)
 370         int a;
 371 {
 372         int w = 0;
 373
 374         a = apply_at_specials(a);
 375
 376         if (a & AT_UNDERLINE)
 377                 w += ul_s_width;
 378         if (a & AT_BOLD)
 379                 w += bo_s_width;
 380         if (a & AT_BLINK)
 381                 w += bl_s_width;
 382         if (a & AT_STANDOUT)
 383                 w += so_s_width;
 384
 385         return w;
 386 }
 387
 388 /*
 389  * Return the printing width of the end (exit) sequence
 390  * for a given character attribute.
 391  */
 392         static int
 393 attr_ewidth(a)
 394         int a;
 395 {
 396         int w = 0;
 397
 398         a = apply_at_specials(a);
 399
 400         if (a & AT_UNDERLINE)
 401                 w += ul_e_width;
 402         if (a & AT_BOLD)
 403                 w += bo_e_width;
 404         if (a & AT_BLINK)
 405                 w += bl_e_width;
 406         if (a & AT_STANDOUT)
 407                 w += so_e_width;
 408
 409         return w;
 410 }
 411
 412 /*
 413  * Return the printing width of a given character and attribute,
 414  * if the character were added to the current position in the line buffer.
 415  * Adding a character with a given attribute may cause an enter or exit
 416  * attribute sequence to be inserted, so this must be taken into account.
 417  */
 418         static int
 419 pwidth(ch, a, prev_ch)
 420         LWCHAR ch;
 421         int a;
 422         LWCHAR prev_ch;
 423 {
 424         int w;
 425
 426         if (ch == '\b')
 427                 /*
 428                  * Backspace moves backwards one or two positions.
 429                  * XXX - Incorrect if several '\b' in a row.
 430                  */
 431                 return (utf_mode && is_wide_char(prev_ch)) ? -2 : -1;
 432
 433         if (!utf_mode || is_ascii_char(ch))
 434         {
 435                 if (control_char((char)ch))
 436                 {
 437                         /*
 438                          * Control characters do unpredictable things,
 439                          * so we don't even try to guess; say it doesn't move.
 440                          * This can only happen if the -r flag is in effect.
 441                          */
 442                         return (0);
 443                 }
 444         } else
 445         {
 446                 if (is_composing_char(ch) || is_combining_char(prev_ch, ch))
 447                 {
 448                         /*
 449                          * Composing and combining chars take up no space.
 450                          *
 451                          * Some terminals, upon failure to compose a
 452                          * composing character with the character(s) that
 453                          * precede(s) it will actually take up one column
 454                          * for the composing character; there isn't much
 455                          * we could do short of testing the (complex)
 456                          * composition process ourselves and printing
 457                          * a binary representation when it fails.
 458                          */
 459                         return (0);
 460                 }
 461         }
 462
 463         /*
 464          * Other characters take one or two columns,
 465          * plus the width of any attribute enter/exit sequence.
 466          */
 467         w = 1;
 468         if (is_wide_char(ch))
 469                 w++;
 470         if (curr > 0 && !is_at_equiv(attr[curr-1], a))
 471                 w += attr_ewidth(attr[curr-1]);
 472         if ((apply_at_specials(a) != AT_NORMAL) &&
 473             (curr == 0 || !is_at_equiv(attr[curr-1], a)))
 474                 w += attr_swidth(a);
 475         return (w);
 476 }
 477
 478 /*
 479  * Delete to the previous base character in the line buffer.
 480  * Return 1 if one is found.
 481  */
 482         static int
 483 backc()
 484 {
 485         LWCHAR prev_ch;
 486         char *p = linebuf + curr;
 487         LWCHAR ch = step_char(&p, -1, linebuf + lmargin);
 488         int width;
 489
 490         /* This assumes that there is no '\b' in linebuf.  */
 491         while (   curr > lmargin
 492                && column > lmargin
 493                && (!(attr[curr - 1] & (AT_ANSI|AT_BINARY))))
 494         {
 495                 curr = p - linebuf;
 496                 prev_ch = step_char(&p, -1, linebuf + lmargin);
 497                 width = pwidth(ch, attr[curr], prev_ch);
 498                 column -= width;
 499                 if (width > 0)
 500                         return 1;
 501                 ch = prev_ch;
 502         }
 503
 504         return 0;
 505 }
 506
 507 /*
 508  * Are we currently within a recognized ANSI escape sequence?
 509  */
 510         static int
 511 in_ansi_esc_seq()
 512 {
 513         char *p;
 514
 515         /*
 516          * Search backwards for either an ESC (which means we ARE in a seq);
 517          * or an end char (which means we're NOT in a seq).
 518          */
 519         for (p = &linebuf[curr];  p > linebuf; )
 520         {
 521                 LWCHAR ch = step_char(&p, -1, linebuf);
 522                 if (IS_CSI_START(ch))
 523                         return (1);
 524                 if (!is_ansi_middle(ch))
 525                         return (0);
 526         }
 527         return (0);
 528 }
 529
 530 /*
 531  * Is a character the end of an ANSI escape sequence?
 532  */
 533         public int
 534 is_ansi_end(ch)
 535         LWCHAR ch;
 536 {
 537         if (!is_ascii_char(ch))
 538                 return (0);
 539         return (strchr(end_ansi_chars, (char) ch) != NULL);
 540 }
 541
 542 /*
 543  *
 544  */
 545         public int
 546 is_ansi_middle(ch)
 547         LWCHAR ch;
 548 {
 549         if (!is_ascii_char(ch))
 550                 return (0);
 551         if (is_ansi_end(ch))
 552                 return (0);
 553         return (strchr(mid_ansi_chars, (char) ch) != NULL);
 554 }
 555
 556 /*
 557  * Append a character and attribute to the line buffer.
 558  */
 559 #define STORE_CHAR(ch,a,rep,pos) \
 560         do { \
 561                 if (store_char((ch),(a),(rep),(pos))) return (1); \
 562         } while (0)
 563
 564         static int
 565 store_char(ch, a, rep, pos)
 566         LWCHAR ch;
 567         int a;
 568         char *rep;
 569         POSITION pos;
 570 {
 571         int w;
 572         int replen;
 573         char cs;
 574
 575         w = (a & (AT_UNDERLINE|AT_BOLD));       /* Pre-use w.  */
 576         if (w != AT_NORMAL)
 577                 last_overstrike = w;
 578
 579 #if HILITE_SEARCH
 580         {
 581                 int matches;
 582                 if (is_hilited(pos, pos+1, 0, &matches))
 583                 {
 584                         /*
 585                          * This character should be highlighted.
 586                          * Override the attribute passed in.
 587                          */
 588                         if (a != AT_ANSI)
 589                         {
 590                                 if (highest_hilite != NULL_POSITION &&
 591                                     pos > highest_hilite)
 592                                         highest_hilite = pos;
 593                                 a |= AT_HILITE;
 594                         }
 595                 }
 596         }
 597 #endif
 598
 599         if (ctldisp == OPT_ONPLUS && in_ansi_esc_seq())
 600         {
 601                 if (!is_ansi_end(ch) && !is_ansi_middle(ch)) {
 602                         /* Remove whole unrecognized sequence.  */
 603                         char *p = &linebuf[curr];
 604                         LWCHAR bch;
 605                         do {
 606                                 bch = step_char(&p, -1, linebuf);
 607                         } while (p > linebuf && !IS_CSI_START(bch));
 608                         curr = p - linebuf;
 609                         return 0;
 610                 }
 611                 a = AT_ANSI;    /* Will force re-AT_'ing around it.  */
 612                 w = 0;
 613         }
 614         else if (ctldisp == OPT_ONPLUS && IS_CSI_START(ch))
 615         {
 616                 a = AT_ANSI;    /* Will force re-AT_'ing around it.  */
 617                 w = 0;
 618         }
 619         else
 620         {
 621                 char *p = &linebuf[curr];
 622                 LWCHAR prev_ch = step_char(&p, -1, linebuf);
 623                 w = pwidth(ch, a, prev_ch);
 624         }
 625
 626         if (ctldisp != OPT_ON && column + w + attr_ewidth(a) > sc_width)
 627                 /*
 628                  * Won't fit on screen.
 629                  */
 630                 return (1);
 631
 632         if (rep == NULL)
 633         {
 634                 cs = (char) ch;
 635                 rep = &cs;
 636                 replen = 1;
 637         } else
 638         {
 639                 replen = utf_len(rep[0]);
 640         }
 641         if (curr + replen >= size_linebuf-6)
 642         {
 643                 /*
 644                  * Won't fit in line buffer.
 645                  * Try to expand it.
 646                  */
 647                 if (expand_linebuf())
 648                         return (1);
 649         }
 650
 651         while (replen-- > 0)
 652         {
 653                 linebuf[curr] = *rep++;
 654                 attr[curr] = a;
 655                 curr++;
 656         }
 657         column += w;
 658         return (0);
 659 }
 660
 661 /*
 662  * Append a tab to the line buffer.
 663  * Store spaces to represent the tab.
 664  */
 665 #define STORE_TAB(a,pos) \
 666         do { if (store_tab((a),(pos))) return (1); } while (0)
 667
 668         static int
 669 store_tab(attr, pos)
 670         int attr;
 671         POSITION pos;
 672 {
 673         int to_tab = column + cshift - lmargin;
 674         int i;
 675
 676         if (ntabstops < 2 || to_tab >= tabstops[ntabstops-1])
 677                 to_tab = tabdefault -
 678                      ((to_tab - tabstops[ntabstops-1]) % tabdefault);
 679         else
 680         {
 681                 for (i = ntabstops - 2;  i >= 0;  i--)
 682                         if (to_tab >= tabstops[i])
 683                                 break;
 684                 to_tab = tabstops[i+1] - to_tab;
 685         }
 686
 687         if (column + to_tab - 1 + pwidth(' ', attr, 0) + attr_ewidth(attr) > sc_width)
 688                 return 1;
 689
 690         do {
 691                 STORE_CHAR(' ', attr, " ", pos);
 692         } while (--to_tab > 0);
 693         return 0;
 694 }
 695
 696 #define STORE_PRCHAR(c, pos) \
 697         do { if (store_prchar((c), (pos))) return 1; } while (0)
 698
 699         static int
 700 store_prchar(c, pos)
 701         char c;
 702         POSITION pos;
 703 {
 704         char *s;
 705
 706         /*
 707          * Convert to printable representation.
 708          */
 709         s = prchar(c);
 710
 711         /*
 712          * Make sure we can get the entire representation
 713          * of the character on this line.
 714          */
 715         if (column + (int) strlen(s) - 1 +
 716             pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
 717                 return 1;
 718
 719         for ( ;  *s != 0;  s++)
 720                 STORE_CHAR(*s, AT_BINARY, NULL, pos);
 721
 722         return 0;
 723 }
 724
 725         static int
 726 flush_mbc_buf(pos)
 727         POSITION pos;
 728 {
 729         int i;
 730
 731         for (i = 0; i < mbc_buf_index; i++)
 732                 if (store_prchar(mbc_buf[i], pos))
 733                         return mbc_buf_index - i;
 734
 735         return 0;
 736 }
 737
 738 /*
 739  * Append a character to the line buffer.
 740  * Expand tabs into spaces, handle underlining, boldfacing, etc.
 741  * Returns 0 if ok, 1 if couldn't fit in buffer.
 742  */
 743         public int
 744 pappend(c, pos)
 745         char c;
 746         POSITION pos;
 747 {
 748         int r;
 749
 750         if (pendc)
 751         {
 752                 if (do_append(pendc, NULL, pendpos))
 753                         /*
 754                          * Oops.  We've probably lost the char which
 755                          * was in pendc, since caller won't back up.
 756                          */
 757                         return (1);
 758                 pendc = '\0';
 759         }
 760
 761         if (c == '\r' && bs_mode == BS_SPECIAL)
 762         {
 763                 if (mbc_buf_len > 0)  /* utf_mode must be on. */
 764                 {
 765                         /* Flush incomplete (truncated) sequence. */
 766                         r = flush_mbc_buf(mbc_pos);
 767                         mbc_buf_index = r + 1;
 768                         mbc_buf_len = 0;
 769                         if (r)
 770                                 return (mbc_buf_index);
 771                 }
 772
 773                 /*
 774                  * Don't put the CR into the buffer until we see
 775                  * the next char.  If the next char is a newline,
 776                  * discard the CR.
 777                  */
 778                 pendc = c;
 779                 pendpos = pos;
 780                 return (0);
 781         }
 782
 783         if (!utf_mode)
 784         {
 785                 r = do_append((LWCHAR) c, NULL, pos);
 786         } else
 787         {
 788                 /* Perform strict validation in all possible cases. */
 789                 if (mbc_buf_len == 0)
 790                 {
 791                 retry:
 792                         mbc_buf_index = 1;
 793                         *mbc_buf = c;
 794                         if (IS_ASCII_OCTET(c))
 795                                 r = do_append((LWCHAR) c, NULL, pos);
 796                         else if (IS_UTF8_LEAD(c))
 797                         {
 798                                 mbc_buf_len = utf_len(c);
 799                                 mbc_pos = pos;
 800                                 return (0);
 801                         } else
 802                                 /* UTF8_INVALID or stray UTF8_TRAIL */
 803                                 r = flush_mbc_buf(pos);
 804                 } else if (IS_UTF8_TRAIL(c))
 805                 {
 806                         mbc_buf[mbc_buf_index++] = c;
 807                         if (mbc_buf_index < mbc_buf_len)
 808                                 return (0);
 809                         if (is_utf8_well_formed(mbc_buf))
 810                                 r = do_append(get_wchar(mbc_buf), mbc_buf, mbc_pos);
 811                         else
 812                                 /* Complete, but not shortest form, sequence. */
 813                                 mbc_buf_index = r = flush_mbc_buf(mbc_pos);
 814                         mbc_buf_len = 0;
 815                 } else
 816                 {
 817                         /* Flush incomplete (truncated) sequence.  */
 818                         r = flush_mbc_buf(mbc_pos);
 819                         mbc_buf_index = r + 1;
 820                         mbc_buf_len = 0;
 821                         /* Handle new char.  */
 822                         if (!r)
 823                                 goto retry;
 824                 }
 825         }
 826
 827         /*
 828          * If we need to shift the line, do it.
 829          * But wait until we get to at least the middle of the screen,
 830          * so shifting it doesn't affect the chars we're currently
 831          * pappending.  (Bold & underline can get messed up otherwise.)
 832          */
 833         if (cshift < hshift && column > sc_width / 2)
 834         {
 835                 linebuf[curr] = '\0';
 836                 pshift(hshift - cshift);
 837         }
 838         if (r)
 839         {
 840                 /* How many chars should caller back up? */
 841                 r = (!utf_mode) ? 1 : mbc_buf_index;
 842         }
 843         return (r);
 844 }
 845
 846         static int
 847 do_append(ch, rep, pos)
 848         LWCHAR ch;
 849         char *rep;
 850         POSITION pos;
 851 {
 852         register int a;
 853         LWCHAR prev_ch;
 854
 855         a = AT_NORMAL;
 856
 857         if (ch == '\b')
 858         {
 859                 if (bs_mode == BS_CONTROL)
 860                         goto do_control_char;
 861
 862                 /*
 863                  * A better test is needed here so we don't
 864                  * backspace over part of the printed
 865                  * representation of a binary character.
 866                  */
 867                 if (   curr <= lmargin
 868                     || column <= lmargin
 869                     || (attr[curr - 1] & (AT_ANSI|AT_BINARY)))
 870                         STORE_PRCHAR('\b', pos);
 871                 else if (bs_mode == BS_NORMAL)
 872                         STORE_CHAR(ch, AT_NORMAL, NULL, pos);
 873                 else if (bs_mode == BS_SPECIAL)
 874                         overstrike = backc();
 875
 876                 return 0;
 877         }
 878
 879         if (overstrike > 0)
 880         {
 881                 /*
 882                  * Overstrike the character at the current position
 883                  * in the line buffer.  This will cause either
 884                  * underline (if a "_" is overstruck),
 885                  * bold (if an identical character is overstruck),
 886                  * or just deletion of the character in the buffer.
 887                  */
 888                 overstrike = utf_mode ? -1 : 0;
 889                 /* To be correct, this must be a base character.  */
 890                 prev_ch = get_wchar(linebuf + curr);
 891                 a = attr[curr];
 892                 if (ch == prev_ch)
 893                 {
 894                         /*
 895                          * Overstriking a char with itself means make it bold.
 896                          * But overstriking an underscore with itself is
 897                          * ambiguous.  It could mean make it bold, or
 898                          * it could mean make it underlined.
 899                          * Use the previous overstrike to resolve it.
 900                          */
 901                         if (ch == '_')
 902                         {
 903                                 if ((a & (AT_BOLD|AT_UNDERLINE)) != AT_NORMAL)
 904                                         a |= (AT_BOLD|AT_UNDERLINE);
 905                                 else if (last_overstrike != AT_NORMAL)
 906                                         a |= last_overstrike;
 907                                 else
 908                                         a |= AT_BOLD;
 909                         } else
 910                                 a |= AT_BOLD;
 911                 } else if (ch == '_')
 912                 {
 913                         a |= AT_UNDERLINE;
 914                         ch = prev_ch;
 915                         rep = linebuf + curr;
 916                 } else if (prev_ch == '_')
 917                 {
 918                         a |= AT_UNDERLINE;
 919                 }
 920                 /* Else we replace prev_ch, but we keep its attributes.  */
 921         } else if (overstrike < 0)
 922         {
 923                 if (   is_composing_char(ch)
 924                     || is_combining_char(get_wchar(linebuf + curr), ch))
 925                         /* Continuation of the same overstrike.  */
 926                         a = last_overstrike;
 927                 else
 928                         overstrike = 0;
 929         }
 930
 931         if (ch == '\t')
 932         {
 933                 /*
 934                  * Expand a tab into spaces.
 935                  */
 936                 switch (bs_mode)
 937                 {
 938                 case BS_CONTROL:
 939                         goto do_control_char;
 940                 case BS_NORMAL:
 941                 case BS_SPECIAL:
 942                         STORE_TAB(a, pos);
 943                         break;
 944                 }
 945         } else if ((!utf_mode || is_ascii_char(ch)) && control_char((char)ch))
 946         {
 947         do_control_char:
 948                 if (ctldisp == OPT_ON || (ctldisp == OPT_ONPLUS && IS_CSI_START(ch)))
 949                 {
 950                         /*
 951                          * Output as a normal character.
 952                          */
 953                         STORE_CHAR(ch, AT_NORMAL, rep, pos);
 954                 } else
 955                 {
 956                         STORE_PRCHAR((char) ch, pos);
 957                 }
 958         } else if (utf_mode && ctldisp != OPT_ON && is_ubin_char(ch))
 959         {
 960                 char *s;
 961
 962                 s = prutfchar(ch);
 963
 964                 if (column + (int) strlen(s) - 1 +
 965                     pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
 966                         return (1);
 967
 968                 for ( ;  *s != 0;  s++)
 969                         STORE_CHAR(*s, AT_BINARY, NULL, pos);
 970         } else
 971         {
 972                 STORE_CHAR(ch, a, rep, pos);
 973         }
 974         return (0);
 975 }
 976
 977 /*
 978  *
 979  */
 980         public int
 981 pflushmbc()
 982 {
 983         int r = 0;
 984
 985         if (mbc_buf_len > 0)
 986         {
 987                 /* Flush incomplete (truncated) sequence.  */
 988                 r = flush_mbc_buf(mbc_pos);
 989                 mbc_buf_len = 0;
 990         }
 991         return r;
 992 }
 993
 994 /*
 995  * Terminate the line in the line buffer.
 996  */
 997         public void
 998 pdone(endline, forw)
 999         int endline;
1000         int forw;
1001 {
1002         (void) pflushmbc();
1003
1004         if (pendc && (pendc != '\r' || !endline))
1005                 /*
1006                  * If we had a pending character, put it in the buffer.
1007                  * But discard a pending CR if we are at end of line
1008                  * (that is, discard the CR in a CR/LF sequence).
1009                  */
1010                 (void) do_append(pendc, NULL, pendpos);
1011
1012         /*
1013          * Make sure we've shifted the line, if we need to.
1014          */
1015         if (cshift < hshift)
1016                 pshift(hshift - cshift);
1017
1018         if (ctldisp == OPT_ONPLUS && is_ansi_end('m'))
1019         {
1020                 /* Switch to normal attribute at end of line. */
1021                 char *p = "\033[m";
1022                 for ( ;  *p != '\0';  p++)
1023                 {
1024                         linebuf[curr] = *p;
1025                         attr[curr++] = AT_ANSI;
1026                 }
1027         }
1028
1029         /*
1030          * Add a newline if necessary,
1031          * and append a '\0' to the end of the line.
1032          * We output a newline if we're not at the right edge of the screen,
1033          * or if the terminal doesn't auto wrap,
1034          * or if this is really the end of the line AND the terminal ignores
1035          * a newline at the right edge.
1036          * (In the last case we don't want to output a newline if the terminal
1037          * doesn't ignore it since that would produce an extra blank line.
1038          * But we do want to output a newline if the terminal ignores it in case
1039          * the next line is blank.  In that case the single newline output for
1040          * that blank line would be ignored!)
1041          */
1042         if (column < sc_width || !auto_wrap || (endline && ignaw) || ctldisp == OPT_ON)
1043         {
1044                 linebuf[curr] = '\n';
1045                 attr[curr] = AT_NORMAL;
1046                 curr++;
1047         }
1048         else if (ignaw && column >= sc_width && forw)
1049         {
1050                 /*
1051                  * Terminals with "ignaw" don't wrap until they *really* need
1052                  * to, i.e. when the character *after* the last one to fit on a
1053                  * line is output. But they are too hard to deal with when they
1054                  * get in the state where a full screen width of characters
1055                  * have been output but the cursor is sitting on the right edge
1056                  * instead of at the start of the next line.
1057                  * So we nudge them into wrapping by outputting a space
1058                  * character plus a backspace.  But do this only if moving
1059                  * forward; if we're moving backward and drawing this line at
1060                  * the top of the screen, the space would overwrite the first
1061                  * char on the next line.  We don't need to do this "nudge"
1062                  * at the top of the screen anyway.
1063                  */
1064                 linebuf[curr] = ' ';
1065                 attr[curr++] = AT_NORMAL;
1066                 linebuf[curr] = '\b';
1067                 attr[curr++] = AT_NORMAL;
1068         }
1069         linebuf[curr] = '\0';
1070         attr[curr] = AT_NORMAL;
1071 }
1072
1073 /*
1074  *
1075  */
1076         public void
1077 set_status_col(c)
1078         char c;
1079 {
1080         linebuf[0] = c;
1081         attr[0] = AT_NORMAL|AT_HILITE;
1082 }
1083
1084 /*
1085  * Get a character from the current line.
1086  * Return the character as the function return value,
1087  * and the character attribute in *ap.
1088  */
1089         public int
1090 gline(i, ap)
1091         register int i;
1092         register int *ap;
1093 {
1094         if (is_null_line)
1095         {
1096                 /*
1097                  * If there is no current line, we pretend the line is
1098                  * either "~" or "", depending on the "twiddle" flag.
1099                  */
1100                 if (twiddle)
1101                 {
1102                         if (i == 0)
1103                         {
1104                                 *ap = AT_BOLD;
1105                                 return '~';
1106                         }
1107                         --i;
1108                 }
1109                 /* Make sure we're back to AT_NORMAL before the '\n'.  */
1110                 *ap = AT_NORMAL;
1111                 return i ? '\0' : '\n';
1112         }
1113
1114         *ap = attr[i];
1115         return (linebuf[i] & 0xFF);
1116 }
1117
1118 /*
1119  * Indicate that there is no current line.
1120  */
1121         public void
1122 null_line()
1123 {
1124         is_null_line = 1;
1125         cshift = 0;
1126 }
1127
1128 /*
1129  * Analogous to forw_line(), but deals with "raw lines":
1130  * lines which are not split for screen width.
1131  * {{ This is supposed to be more efficient than forw_line(). }}
1132  */
1133         public POSITION
1134 forw_raw_line(curr_pos, linep, line_lenp)
1135         POSITION curr_pos;
1136         char **linep;
1137         int *line_lenp;
1138 {
1139         register int n;
1140         register int c;
1141         POSITION new_pos;
1142
1143         if (curr_pos == NULL_POSITION || ch_seek(curr_pos) ||
1144                 (c = ch_forw_get()) == EOI)
1145                 return (NULL_POSITION);
1146
1147         n = 0;
1148         for (;;)
1149         {
1150                 if (c == '\n' || c == EOI || ABORT_SIGS())
1151                 {
1152                         new_pos = ch_tell();
1153                         break;
1154                 }
1155                 if (n >= size_linebuf-1)
1156                 {
1157                         if (expand_linebuf())
1158                         {
1159                                 /*
1160                                  * Overflowed the input buffer.
1161                                  * Pretend the line ended here.
1162                                  */
1163                                 new_pos = ch_tell() - 1;
1164                                 break;
1165                         }
1166                 }
1167                 linebuf[n++] = c;
1168                 c = ch_forw_get();
1169         }
1170         linebuf[n] = '\0';
1171         if (linep != NULL)
1172                 *linep = linebuf;
1173         if (line_lenp != NULL)
1174                 *line_lenp = n;
1175         return (new_pos);
1176 }
1177
1178 /*
1179  * Analogous to back_line(), but deals with "raw lines".
1180  * {{ This is supposed to be more efficient than back_line(). }}
1181  */
1182         public POSITION
1183 back_raw_line(curr_pos, linep, line_lenp)
1184         POSITION curr_pos;
1185         char **linep;
1186         int *line_lenp;
1187 {
1188         register int n;
1189         register int c;
1190         POSITION new_pos;
1191
1192         if (curr_pos == NULL_POSITION || curr_pos <= ch_zero() ||
1193                 ch_seek(curr_pos-1))
1194                 return (NULL_POSITION);
1195
1196         n = size_linebuf;
1197         linebuf[--n] = '\0';
1198         for (;;)
1199         {
1200                 c = ch_back_get();
1201                 if (c == '\n' || ABORT_SIGS())
1202                 {
1203                         /*
1204                          * This is the newline ending the previous line.
1205                          * We have hit the beginning of the line.
1206                          */
1207                         new_pos = ch_tell() + 1;
1208                         break;
1209                 }
1210                 if (c == EOI)
1211                 {
1212                         /*
1213                          * We have hit the beginning of the file.
1214                          * This must be the first line in the file.
1215                          * This must, of course, be the beginning of the line.
1216                          */
1217                         new_pos = ch_zero();
1218                         break;
1219                 }
1220                 if (n <= 0)
1221                 {
1222                         int old_size_linebuf = size_linebuf;
1223                         char *fm;
1224                         char *to;
1225                         if (expand_linebuf())
1226                         {
1227                                 /*
1228                                  * Overflowed the input buffer.
1229                                  * Pretend the line ended here.
1230                                  */
1231                                 new_pos = ch_tell() + 1;
1232                                 break;
1233                         }
1234                         /*
1235                          * Shift the data to the end of the new linebuf.
1236                          */
1237                         for (fm = linebuf + old_size_linebuf - 1,
1238                               to = linebuf + size_linebuf - 1;
1239                              fm >= linebuf;  fm--, to--)
1240                                 *to = *fm;
1241                         n = size_linebuf - old_size_linebuf;
1242                 }
1243                 linebuf[--n] = c;
1244         }
1245         if (linep != NULL)
1246                 *linep = &linebuf[n];
1247         if (line_lenp != NULL)
1248                 *line_lenp = size_linebuf - 1 - n;
1249         return (new_pos);
1250 }