contrib/less/line.c

   1 /*
   2  * Copyright (C) 1984-2015  Mark Nudelman
   3  *
   4  * You may distribute under the terms of either the GNU General Public
   5  * License or the Less License, as specified in the README file.
   6  *
   7  * For more information, see the README file.
   8  */
   9
  10
  11 /*
  12  * Routines to manipulate the "line buffer".
  13  * The line buffer holds a line of output as it is being built
  14  * in preparation for output to the screen.
  15  */
  16
  17 #include "less.h"
  18 #include "charset.h"
  19
  20 static char *linebuf = NULL;    /* Buffer which holds the current output line */
  21 static char *attr = NULL;       /* Extension of linebuf to hold attributes */
  22 public int size_linebuf = 0;    /* Size of line buffer (and attr buffer) */
  23
  24 static int cshift;              /* Current left-shift of output line buffer */
  25 public int hshift;              /* Desired left-shift of output line buffer */
  26 public int tabstops[TABSTOP_MAX] = { 0 }; /* Custom tabstops */
  27 public int ntabstops = 1;       /* Number of tabstops */
  28 public int tabdefault = 8;      /* Default repeated tabstops */
  29 public POSITION highest_hilite; /* Pos of last hilite in file found so far */
  30
  31 static int curr;                /* Index into linebuf */
  32 static int column;              /* Printable length, accounting for
  33                                    backspaces, etc. */
  34 static int overstrike;          /* Next char should overstrike previous char */
  35 static int last_overstrike = AT_NORMAL;
  36 static int is_null_line;        /* There is no current line */
  37 static int lmargin;             /* Left margin */
  38 static LWCHAR pendc;
  39 static POSITION pendpos;
  40 static char *end_ansi_chars;
  41 static char *mid_ansi_chars;
  42
  43 static int attr_swidth();
  44 static int attr_ewidth();
  45 static int do_append();
  46
  47 extern int sigs;
  48 extern int bs_mode;
  49 extern int linenums;
  50 extern int ctldisp;
  51 extern int twiddle;
  52 extern int binattr;
  53 extern int status_col;
  54 extern int auto_wrap, ignaw;
  55 extern int bo_s_width, bo_e_width;
  56 extern int ul_s_width, ul_e_width;
  57 extern int bl_s_width, bl_e_width;
  58 extern int so_s_width, so_e_width;
  59 extern int sc_width, sc_height;
  60 extern int utf_mode;
  61 extern POSITION start_attnpos;
  62 extern POSITION end_attnpos;
  63
  64 static char mbc_buf[MAX_UTF_CHAR_LEN];
  65 static int mbc_buf_len = 0;
  66 static int mbc_buf_index = 0;
  67 static POSITION mbc_pos;
  68
  69 /*
  70  * Initialize from environment variables.
  71  */
  72         public void
  73 init_line()
  74 {
  75         end_ansi_chars = lgetenv("LESSANSIENDCHARS");
  76         if (end_ansi_chars == NULL || *end_ansi_chars == '\0')
  77                 end_ansi_chars = "m";
  78
  79         mid_ansi_chars = lgetenv("LESSANSIMIDCHARS");
  80         if (mid_ansi_chars == NULL || *mid_ansi_chars == '\0')
  81                 mid_ansi_chars = "0123456789:;[?!\"'#%()*+ ";
  82
  83         linebuf = (char *) ecalloc(LINEBUF_SIZE, sizeof(char));
  84         attr = (char *) ecalloc(LINEBUF_SIZE, sizeof(char));
  85         size_linebuf = LINEBUF_SIZE;
  86 }
  87
  88 /*
  89  * Expand the line buffer.
  90  */
  91         static int
  92 expand_linebuf()
  93 {
  94         /* Double the size of the line buffer. */
  95         int new_size = size_linebuf * 2;
  96
  97         /* Just realloc to expand the buffer, if we can. */
  98 #if HAVE_REALLOC
  99         char *new_buf = (char *) realloc(linebuf, new_size);
 100         char *new_attr = (char *) realloc(attr, new_size);
 101 #else
 102         char *new_buf = (char *) calloc(new_size, sizeof(char));
 103         char *new_attr = (char *) calloc(new_size, sizeof(char));
 104 #endif
 105         if (new_buf == NULL || new_attr == NULL)
 106         {
 107                 if (new_attr != NULL)
 108                         free(new_attr);
 109                 if (new_buf != NULL)
 110                         free(new_buf);
 111                 return 1;
 112         }
 113 #if HAVE_REALLOC
 114         /*
 115          * We realloc'd the buffers; they already have the old contents.
 116          */
 117         #if 0
 118         memset(new_buf + size_linebuf, 0, new_size - size_linebuf);
 119         memset(new_attr + size_linebuf, 0, new_size - size_linebuf);
 120         #endif
 121 #else
 122         /*
 123          * We just calloc'd the buffers; copy the old contents.
 124          */
 125         memcpy(new_buf, linebuf, size_linebuf * sizeof(char));
 126         memcpy(new_attr, attr, size_linebuf * sizeof(char));
 127         free(attr);
 128         free(linebuf);
 129 #endif
 130         linebuf = new_buf;
 131         attr = new_attr;
 132         size_linebuf = new_size;
 133         return 0;
 134 }
 135
 136 /*
 137  * Is a character ASCII?
 138  */
 139         public int
 140 is_ascii_char(ch)
 141         LWCHAR ch;
 142 {
 143         return (ch <= 0x7F);
 144 }
 145
 146 /*
 147  * Rewind the line buffer.
 148  */
 149         public void
 150 prewind()
 151 {
 152         curr = 0;
 153         column = 0;
 154         cshift = 0;
 155         overstrike = 0;
 156         last_overstrike = AT_NORMAL;
 157         mbc_buf_len = 0;
 158         is_null_line = 0;
 159         pendc = '\0';
 160         lmargin = 0;
 161         if (status_col)
 162                 lmargin += 1;
 163 }
 164
 165 /*
 166  * Insert the line number (of the given position) into the line buffer.
 167  */
 168         public void
 169 plinenum(pos)
 170         POSITION pos;
 171 {
 172         register LINENUM linenum = 0;
 173         register int i;
 174
 175         if (linenums == OPT_ONPLUS)
 176         {
 177                 /*
 178                  * Get the line number and put it in the current line.
 179                  * {{ Note: since find_linenum calls forw_raw_line,
 180                  *    it may seek in the input file, requiring the caller
 181                  *    of plinenum to re-seek if necessary. }}
 182                  * {{ Since forw_raw_line modifies linebuf, we must
 183                  *    do this first, before storing anything in linebuf. }}
 184                  */
 185                 linenum = find_linenum(pos);
 186         }
 187
 188         /*
 189          * Display a status column if the -J option is set.
 190          */
 191         if (status_col)
 192         {
 193                 linebuf[curr] = ' ';
 194                 if (start_attnpos != NULL_POSITION &&
 195                     pos >= start_attnpos && pos < end_attnpos)
 196                         attr[curr] = AT_NORMAL|AT_HILITE;
 197                 else
 198                         attr[curr] = AT_NORMAL;
 199                 curr++;
 200                 column++;
 201         }
 202         /*
 203          * Display the line number at the start of each line
 204          * if the -N option is set.
 205          */
 206         if (linenums == OPT_ONPLUS)
 207         {
 208                 char buf[INT_STRLEN_BOUND(pos) + 2];
 209                 int n;
 210
 211                 linenumtoa(linenum, buf);
 212                 n = (int) strlen(buf);
 213                 if (n < MIN_LINENUM_WIDTH)
 214                         n = MIN_LINENUM_WIDTH;
 215                 sprintf(linebuf+curr, "%*s ", n, buf);
 216                 n++;  /* One space after the line number. */
 217                 for (i = 0; i < n; i++)
 218                         attr[curr+i] = AT_NORMAL;
 219                 curr += n;
 220                 column += n;
 221                 lmargin += n;
 222         }
 223
 224         /*
 225          * Append enough spaces to bring us to the lmargin.
 226          */
 227         while (column < lmargin)
 228         {
 229                 linebuf[curr] = ' ';
 230                 attr[curr++] = AT_NORMAL;
 231                 column++;
 232         }
 233 }
 234
 235 /*
 236  * Shift the input line left.
 237  * This means discarding N printable chars at the start of the buffer.
 238  */
 239         static void
 240 pshift(shift)
 241         int shift;
 242 {
 243         LWCHAR prev_ch = 0;
 244         unsigned char c;
 245         int shifted = 0;
 246         int to;
 247         int from;
 248         int len;
 249         int width;
 250         int prev_attr;
 251         int next_attr;
 252
 253         if (shift > column - lmargin)
 254                 shift = column - lmargin;
 255         if (shift > curr - lmargin)
 256                 shift = curr - lmargin;
 257
 258         to = from = lmargin;
 259         /*
 260          * We keep on going when shifted == shift
 261          * to get all combining chars.
 262          */
 263         while (shifted <= shift && from < curr)
 264         {
 265                 c = linebuf[from];
 266                 if (ctldisp == OPT_ONPLUS && IS_CSI_START(c))
 267                 {
 268                         /* Keep cumulative effect.  */
 269                         linebuf[to] = c;
 270                         attr[to++] = attr[from++];
 271                         while (from < curr && linebuf[from])
 272                         {
 273                                 linebuf[to] = linebuf[from];
 274                                 attr[to++] = attr[from];
 275                                 if (!is_ansi_middle(linebuf[from++]))
 276                                         break;
 277                         }
 278                         continue;
 279                 }
 280
 281                 width = 0;
 282
 283                 if (!IS_ASCII_OCTET(c) && utf_mode)
 284                 {
 285                         /* Assumes well-formedness validation already done.  */
 286                         LWCHAR ch;
 287
 288                         len = utf_len(c);
 289                         if (from + len > curr)
 290                                 break;
 291                         ch = get_wchar(linebuf + from);
 292                         if (!is_composing_char(ch) && !is_combining_char(prev_ch, ch))
 293                                 width = is_wide_char(ch) ? 2 : 1;
 294                         prev_ch = ch;
 295                 } else
 296                 {
 297                         len = 1;
 298                         if (c == '\b')
 299                                 /* XXX - Incorrect if several '\b' in a row.  */
 300                                 width = (utf_mode && is_wide_char(prev_ch)) ? -2 : -1;
 301                         else if (!control_char(c))
 302                                 width = 1;
 303                         prev_ch = 0;
 304                 }
 305
 306                 if (width == 2 && shift - shifted == 1) {
 307                         /* Should never happen when called by pshift_all().  */
 308                         attr[to] = attr[from];
 309                         /*
 310                          * Assume a wide_char will never be the first half of a
 311                          * combining_char pair, so reset prev_ch in case we're
 312                          * followed by a '\b'.
 313                          */
 314                         prev_ch = linebuf[to++] = ' ';
 315                         from += len;
 316                         shifted++;
 317                         continue;
 318                 }
 319
 320                 /* Adjust width for magic cookies. */
 321                 prev_attr = (to > 0) ? attr[to-1] : AT_NORMAL;
 322                 next_attr = (from + len < curr) ? attr[from + len] : prev_attr;
 323                 if (!is_at_equiv(attr[from], prev_attr) &&
 324                         !is_at_equiv(attr[from], next_attr))
 325                 {
 326                         width += attr_swidth(attr[from]);
 327                         if (from + len < curr)
 328                                 width += attr_ewidth(attr[from]);
 329                         if (is_at_equiv(prev_attr, next_attr))
 330                         {
 331                                 width += attr_ewidth(prev_attr);
 332                                 if (from + len < curr)
 333                                         width += attr_swidth(next_attr);
 334                         }
 335                 }
 336
 337                 if (shift - shifted < width)
 338                         break;
 339                 from += len;
 340                 shifted += width;
 341                 if (shifted < 0)
 342                         shifted = 0;
 343         }
 344         while (from < curr)
 345         {
 346                 linebuf[to] = linebuf[from];
 347                 attr[to++] = attr[from++];
 348         }
 349         curr = to;
 350         column -= shifted;
 351         cshift += shifted;
 352 }
 353
 354 /*
 355  *
 356  */
 357         public void
 358 pshift_all()
 359 {
 360         pshift(column);
 361 }
 362
 363 /*
 364  * Return the printing width of the start (enter) sequence
 365  * for a given character attribute.
 366  */
 367         static int
 368 attr_swidth(a)
 369         int a;
 370 {
 371         int w = 0;
 372
 373         a = apply_at_specials(a);
 374
 375         if (a & AT_UNDERLINE)
 376                 w += ul_s_width;
 377         if (a & AT_BOLD)
 378                 w += bo_s_width;
 379         if (a & AT_BLINK)
 380                 w += bl_s_width;
 381         if (a & AT_STANDOUT)
 382                 w += so_s_width;
 383
 384         return w;
 385 }
 386
 387 /*
 388  * Return the printing width of the end (exit) sequence
 389  * for a given character attribute.
 390  */
 391         static int
 392 attr_ewidth(a)
 393         int a;
 394 {
 395         int w = 0;
 396
 397         a = apply_at_specials(a);
 398
 399         if (a & AT_UNDERLINE)
 400                 w += ul_e_width;
 401         if (a & AT_BOLD)
 402                 w += bo_e_width;
 403         if (a & AT_BLINK)
 404                 w += bl_e_width;
 405         if (a & AT_STANDOUT)
 406                 w += so_e_width;
 407
 408         return w;
 409 }
 410
 411 /*
 412  * Return the printing width of a given character and attribute,
 413  * if the character were added to the current position in the line buffer.
 414  * Adding a character with a given attribute may cause an enter or exit
 415  * attribute sequence to be inserted, so this must be taken into account.
 416  */
 417         static int
 418 pwidth(ch, a, prev_ch)
 419         LWCHAR ch;
 420         int a;
 421         LWCHAR prev_ch;
 422 {
 423         int w;
 424
 425         if (ch == '\b')
 426                 /*
 427                  * Backspace moves backwards one or two positions.
 428                  * XXX - Incorrect if several '\b' in a row.
 429                  */
 430                 return (utf_mode && is_wide_char(prev_ch)) ? -2 : -1;
 431
 432         if (!utf_mode || is_ascii_char(ch))
 433         {
 434                 if (control_char((char)ch))
 435                 {
 436                         /*
 437                          * Control characters do unpredictable things,
 438                          * so we don't even try to guess; say it doesn't move.
 439                          * This can only happen if the -r flag is in effect.
 440                          */
 441                         return (0);
 442                 }
 443         } else
 444         {
 445                 if (is_composing_char(ch) || is_combining_char(prev_ch, ch))
 446                 {
 447                         /*
 448                          * Composing and combining chars take up no space.
 449                          *
 450                          * Some terminals, upon failure to compose a
 451                          * composing character with the character(s) that
 452                          * precede(s) it will actually take up one column
 453                          * for the composing character; there isn't much
 454                          * we could do short of testing the (complex)
 455                          * composition process ourselves and printing
 456                          * a binary representation when it fails.
 457                          */
 458                         return (0);
 459                 }
 460         }
 461
 462         /*
 463          * Other characters take one or two columns,
 464          * plus the width of any attribute enter/exit sequence.
 465          */
 466         w = 1;
 467         if (is_wide_char(ch))
 468                 w++;
 469         if (curr > 0 && !is_at_equiv(attr[curr-1], a))
 470                 w += attr_ewidth(attr[curr-1]);
 471         if ((apply_at_specials(a) != AT_NORMAL) &&
 472             (curr == 0 || !is_at_equiv(attr[curr-1], a)))
 473                 w += attr_swidth(a);
 474         return (w);
 475 }
 476
 477 /*
 478  * Delete to the previous base character in the line buffer.
 479  * Return 1 if one is found.
 480  */
 481         static int
 482 backc()
 483 {
 484         LWCHAR prev_ch;
 485         char *p = linebuf + curr;
 486         LWCHAR ch = step_char(&p, -1, linebuf + lmargin);
 487         int width;
 488
 489         /* This assumes that there is no '\b' in linebuf.  */
 490         while (   curr > lmargin
 491                && column > lmargin
 492                && (!(attr[curr - 1] & (AT_ANSI|AT_BINARY))))
 493         {
 494                 curr = (int) (p - linebuf);
 495                 prev_ch = step_char(&p, -1, linebuf + lmargin);
 496                 width = pwidth(ch, attr[curr], prev_ch);
 497                 column -= width;
 498                 if (width > 0)
 499                         return 1;
 500                 ch = prev_ch;
 501         }
 502
 503         return 0;
 504 }
 505
 506 /*
 507  * Are we currently within a recognized ANSI escape sequence?
 508  */
 509         static int
 510 in_ansi_esc_seq()
 511 {
 512         char *p;
 513
 514         /*
 515          * Search backwards for either an ESC (which means we ARE in a seq);
 516          * or an end char (which means we're NOT in a seq).
 517          */
 518         for (p = &linebuf[curr];  p > linebuf; )
 519         {
 520                 LWCHAR ch = step_char(&p, -1, linebuf);
 521                 if (IS_CSI_START(ch))
 522                         return (1);
 523                 if (!is_ansi_middle(ch))
 524                         return (0);
 525         }
 526         return (0);
 527 }
 528
 529 /*
 530  * Is a character the end of an ANSI escape sequence?
 531  */
 532         public int
 533 is_ansi_end(ch)
 534         LWCHAR ch;
 535 {
 536         if (!is_ascii_char(ch))
 537                 return (0);
 538         return (strchr(end_ansi_chars, (char) ch) != NULL);
 539 }
 540
 541 /*
 542  *
 543  */
 544         public int
 545 is_ansi_middle(ch)
 546         LWCHAR ch;
 547 {
 548         if (!is_ascii_char(ch))
 549                 return (0);
 550         if (is_ansi_end(ch))
 551                 return (0);
 552         return (strchr(mid_ansi_chars, (char) ch) != NULL);
 553 }
 554
 555 /*
 556  * Append a character and attribute to the line buffer.
 557  */
 558 #define STORE_CHAR(ch,a,rep,pos) \
 559         do { \
 560                 if (store_char((ch),(a),(rep),(pos))) return (1); \
 561         } while (0)
 562
 563         static int
 564 store_char(ch, a, rep, pos)
 565         LWCHAR ch;
 566         int a;
 567         char *rep;
 568         POSITION pos;
 569 {
 570         int w;
 571         int replen;
 572         char cs;
 573
 574         w = (a & (AT_UNDERLINE|AT_BOLD));       /* Pre-use w.  */
 575         if (w != AT_NORMAL)
 576                 last_overstrike = w;
 577
 578 #if HILITE_SEARCH
 579         {
 580                 int matches;
 581                 if (is_hilited(pos, pos+1, 0, &matches))
 582                 {
 583                         /*
 584                          * This character should be highlighted.
 585                          * Override the attribute passed in.
 586                          */
 587                         if (a != AT_ANSI)
 588                         {
 589                                 if (highest_hilite != NULL_POSITION &&
 590                                     pos > highest_hilite)
 591                                         highest_hilite = pos;
 592                                 a |= AT_HILITE;
 593                         }
 594                 }
 595         }
 596 #endif
 597
 598         if (ctldisp == OPT_ONPLUS && in_ansi_esc_seq())
 599         {
 600                 if (!is_ansi_end(ch) && !is_ansi_middle(ch)) {
 601                         /* Remove whole unrecognized sequence.  */
 602                         char *p = &linebuf[curr];
 603                         LWCHAR bch;
 604                         do {
 605                                 bch = step_char(&p, -1, linebuf);
 606                         } while (p > linebuf && !IS_CSI_START(bch));
 607                         curr = (int) (p - linebuf);
 608                         return 0;
 609                 }
 610                 a = AT_ANSI;    /* Will force re-AT_'ing around it.  */
 611                 w = 0;
 612         }
 613         else if (ctldisp == OPT_ONPLUS && IS_CSI_START(ch))
 614         {
 615                 a = AT_ANSI;    /* Will force re-AT_'ing around it.  */
 616                 w = 0;
 617         }
 618         else
 619         {
 620                 char *p = &linebuf[curr];
 621                 LWCHAR prev_ch = step_char(&p, -1, linebuf);
 622                 w = pwidth(ch, a, prev_ch);
 623         }
 624
 625         if (ctldisp != OPT_ON && column + w + attr_ewidth(a) > sc_width)
 626                 /*
 627                  * Won't fit on screen.
 628                  */
 629                 return (1);
 630
 631         if (rep == NULL)
 632         {
 633                 cs = (char) ch;
 634                 rep = &cs;
 635                 replen = 1;
 636         } else
 637         {
 638                 replen = utf_len(rep[0]);
 639         }
 640         if (curr + replen >= size_linebuf-6)
 641         {
 642                 /*
 643                  * Won't fit in line buffer.
 644                  * Try to expand it.
 645                  */
 646                 if (expand_linebuf())
 647                         return (1);
 648         }
 649
 650         while (replen-- > 0)
 651         {
 652                 linebuf[curr] = *rep++;
 653                 attr[curr] = a;
 654                 curr++;
 655         }
 656         column += w;
 657         return (0);
 658 }
 659
 660 /*
 661  * Append a tab to the line buffer.
 662  * Store spaces to represent the tab.
 663  */
 664 #define STORE_TAB(a,pos) \
 665         do { if (store_tab((a),(pos))) return (1); } while (0)
 666
 667         static int
 668 store_tab(attr, pos)
 669         int attr;
 670         POSITION pos;
 671 {
 672         int to_tab = column + cshift - lmargin;
 673         int i;
 674
 675         if (ntabstops < 2 || to_tab >= tabstops[ntabstops-1])
 676                 to_tab = tabdefault -
 677                      ((to_tab - tabstops[ntabstops-1]) % tabdefault);
 678         else
 679         {
 680                 for (i = ntabstops - 2;  i >= 0;  i--)
 681                         if (to_tab >= tabstops[i])
 682                                 break;
 683                 to_tab = tabstops[i+1] - to_tab;
 684         }
 685
 686         if (column + to_tab - 1 + pwidth(' ', attr, 0) + attr_ewidth(attr) > sc_width)
 687                 return 1;
 688
 689         do {
 690                 STORE_CHAR(' ', attr, " ", pos);
 691         } while (--to_tab > 0);
 692         return 0;
 693 }
 694
 695 #define STORE_PRCHAR(c, pos) \
 696         do { if (store_prchar((c), (pos))) return 1; } while (0)
 697
 698         static int
 699 store_prchar(c, pos)
 700         LWCHAR c;
 701         POSITION pos;
 702 {
 703         char *s;
 704
 705         /*
 706          * Convert to printable representation.
 707          */
 708         s = prchar(c);
 709
 710         /*
 711          * Make sure we can get the entire representation
 712          * of the character on this line.
 713          */
 714         if (column + (int) strlen(s) - 1 +
 715             pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
 716                 return 1;
 717
 718         for ( ;  *s != 0;  s++)
 719                 STORE_CHAR(*s, AT_BINARY, NULL, pos);
 720
 721         return 0;
 722 }
 723
 724         static int
 725 flush_mbc_buf(pos)
 726         POSITION pos;
 727 {
 728         int i;
 729
 730         for (i = 0; i < mbc_buf_index; i++)
 731                 if (store_prchar(mbc_buf[i], pos))
 732                         return mbc_buf_index - i;
 733
 734         return 0;
 735 }
 736
 737 /*
 738  * Append a character to the line buffer.
 739  * Expand tabs into spaces, handle underlining, boldfacing, etc.
 740  * Returns 0 if ok, 1 if couldn't fit in buffer.
 741  */
 742         public int
 743 pappend(c, pos)
 744         unsigned char c;
 745         POSITION pos;
 746 {
 747         int r;
 748
 749         if (pendc)
 750         {
 751                 if (c == '\r' && pendc == '\r')
 752                         return (0);
 753                 if (do_append(pendc, NULL, pendpos))
 754                         /*
 755                          * Oops.  We've probably lost the char which
 756                          * was in pendc, since caller won't back up.
 757                          */
 758                         return (1);
 759                 pendc = '\0';
 760         }
 761
 762         if (c == '\r' && bs_mode == BS_SPECIAL)
 763         {
 764                 if (mbc_buf_len > 0)  /* utf_mode must be on. */
 765                 {
 766                         /* Flush incomplete (truncated) sequence. */
 767                         r = flush_mbc_buf(mbc_pos);
 768                         mbc_buf_index = r + 1;
 769                         mbc_buf_len = 0;
 770                         if (r)
 771                                 return (mbc_buf_index);
 772                 }
 773
 774                 /*
 775                  * Don't put the CR into the buffer until we see
 776                  * the next char.  If the next char is a newline,
 777                  * discard the CR.
 778                  */
 779                 pendc = c;
 780                 pendpos = pos;
 781                 return (0);
 782         }
 783
 784         if (!utf_mode)
 785         {
 786                 r = do_append(c, NULL, pos);
 787         } else
 788         {
 789                 /* Perform strict validation in all possible cases. */
 790                 if (mbc_buf_len == 0)
 791                 {
 792                 retry:
 793                         mbc_buf_index = 1;
 794                         *mbc_buf = c;
 795                         if (IS_ASCII_OCTET(c))
 796                                 r = do_append(c, NULL, pos);
 797                         else if (IS_UTF8_LEAD(c))
 798                         {
 799                                 mbc_buf_len = utf_len(c);
 800                                 mbc_pos = pos;
 801                                 return (0);
 802                         } else
 803                                 /* UTF8_INVALID or stray UTF8_TRAIL */
 804                                 r = flush_mbc_buf(pos);
 805                 } else if (IS_UTF8_TRAIL(c))
 806                 {
 807                         mbc_buf[mbc_buf_index++] = c;
 808                         if (mbc_buf_index < mbc_buf_len)
 809                                 return (0);
 810                         if (is_utf8_well_formed(mbc_buf, mbc_buf_index))
 811                                 r = do_append(get_wchar(mbc_buf), mbc_buf, mbc_pos);
 812                         else
 813                                 /* Complete, but not shortest form, sequence. */
 814                                 mbc_buf_index = r = flush_mbc_buf(mbc_pos);
 815                         mbc_buf_len = 0;
 816                 } else
 817                 {
 818                         /* Flush incomplete (truncated) sequence.  */
 819                         r = flush_mbc_buf(mbc_pos);
 820                         mbc_buf_index = r + 1;
 821                         mbc_buf_len = 0;
 822                         /* Handle new char.  */
 823                         if (!r)
 824                                 goto retry;
 825                 }
 826         }
 827
 828         /*
 829          * If we need to shift the line, do it.
 830          * But wait until we get to at least the middle of the screen,
 831          * so shifting it doesn't affect the chars we're currently
 832          * pappending.  (Bold & underline can get messed up otherwise.)
 833          */
 834         if (cshift < hshift && column > sc_width / 2)
 835         {
 836                 linebuf[curr] = '\0';
 837                 pshift(hshift - cshift);
 838         }
 839         if (r)
 840         {
 841                 /* How many chars should caller back up? */
 842                 r = (!utf_mode) ? 1 : mbc_buf_index;
 843         }
 844         return (r);
 845 }
 846
 847         static int
 848 do_append(ch, rep, pos)
 849         LWCHAR ch;
 850         char *rep;
 851         POSITION pos;
 852 {
 853         register int a;
 854         LWCHAR prev_ch;
 855
 856         a = AT_NORMAL;
 857
 858         if (ch == '\b')
 859         {
 860                 if (bs_mode == BS_CONTROL)
 861                         goto do_control_char;
 862
 863                 /*
 864                  * A better test is needed here so we don't
 865                  * backspace over part of the printed
 866                  * representation of a binary character.
 867                  */
 868                 if (   curr <= lmargin
 869                     || column <= lmargin
 870                     || (attr[curr - 1] & (AT_ANSI|AT_BINARY)))
 871                         STORE_PRCHAR('\b', pos);
 872                 else if (bs_mode == BS_NORMAL)
 873                         STORE_CHAR(ch, AT_NORMAL, NULL, pos);
 874                 else if (bs_mode == BS_SPECIAL)
 875                         overstrike = backc();
 876
 877                 return 0;
 878         }
 879
 880         if (overstrike > 0)
 881         {
 882                 /*
 883                  * Overstrike the character at the current position
 884                  * in the line buffer.  This will cause either
 885                  * underline (if a "_" is overstruck),
 886                  * bold (if an identical character is overstruck),
 887                  * or just deletion of the character in the buffer.
 888                  */
 889                 overstrike = utf_mode ? -1 : 0;
 890                 if (utf_mode)
 891                 {
 892                         /* To be correct, this must be a base character.  */
 893                         prev_ch = get_wchar(linebuf + curr);
 894                 } else
 895                 {
 896                         prev_ch = (unsigned char) linebuf[curr];
 897                 }
 898                 a = attr[curr];
 899                 if (ch == prev_ch)
 900                 {
 901                         /*
 902                          * Overstriking a char with itself means make it bold.
 903                          * But overstriking an underscore with itself is
 904                          * ambiguous.  It could mean make it bold, or
 905                          * it could mean make it underlined.
 906                          * Use the previous overstrike to resolve it.
 907                          */
 908                         if (ch == '_')
 909                         {
 910                                 if ((a & (AT_BOLD|AT_UNDERLINE)) != AT_NORMAL)
 911                                         a |= (AT_BOLD|AT_UNDERLINE);
 912                                 else if (last_overstrike != AT_NORMAL)
 913                                         a |= last_overstrike;
 914                                 else
 915                                         a |= AT_BOLD;
 916                         } else
 917                                 a |= AT_BOLD;
 918                 } else if (ch == '_')
 919                 {
 920                         a |= AT_UNDERLINE;
 921                         ch = prev_ch;
 922                         rep = linebuf + curr;
 923                 } else if (prev_ch == '_')
 924                 {
 925                         a |= AT_UNDERLINE;
 926                 }
 927                 /* Else we replace prev_ch, but we keep its attributes.  */
 928         } else if (overstrike < 0)
 929         {
 930                 if (   is_composing_char(ch)
 931                     || is_combining_char(get_wchar(linebuf + curr), ch))
 932                         /* Continuation of the same overstrike.  */
 933                         a = last_overstrike;
 934                 else
 935                         overstrike = 0;
 936         }
 937
 938         if (ch == '\t')
 939         {
 940                 /*
 941                  * Expand a tab into spaces.
 942                  */
 943                 switch (bs_mode)
 944                 {
 945                 case BS_CONTROL:
 946                         goto do_control_char;
 947                 case BS_NORMAL:
 948                 case BS_SPECIAL:
 949                         STORE_TAB(a, pos);
 950                         break;
 951                 }
 952         } else if ((!utf_mode || is_ascii_char(ch)) && control_char((char)ch))
 953         {
 954         do_control_char:
 955                 if (ctldisp == OPT_ON || (ctldisp == OPT_ONPLUS && IS_CSI_START(ch)))
 956                 {
 957                         /*
 958                          * Output as a normal character.
 959                          */
 960                         STORE_CHAR(ch, AT_NORMAL, rep, pos);
 961                 } else
 962                 {
 963                         STORE_PRCHAR((char) ch, pos);
 964                 }
 965         } else if (utf_mode && ctldisp != OPT_ON && is_ubin_char(ch))
 966         {
 967                 char *s;
 968
 969                 s = prutfchar(ch);
 970
 971                 if (column + (int) strlen(s) - 1 +
 972                     pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
 973                         return (1);
 974
 975                 for ( ;  *s != 0;  s++)
 976                         STORE_CHAR(*s, AT_BINARY, NULL, pos);
 977         } else
 978         {
 979                 STORE_CHAR(ch, a, rep, pos);
 980         }
 981         return (0);
 982 }
 983
 984 /*
 985  *
 986  */
 987         public int
 988 pflushmbc()
 989 {
 990         int r = 0;
 991
 992         if (mbc_buf_len > 0)
 993         {
 994                 /* Flush incomplete (truncated) sequence.  */
 995                 r = flush_mbc_buf(mbc_pos);
 996                 mbc_buf_len = 0;
 997         }
 998         return r;
 999 }
1000
1001 /*
1002  * Terminate the line in the line buffer.
1003  */
1004         public void
1005 pdone(endline, forw)
1006         int endline;
1007         int forw;
1008 {
1009         (void) pflushmbc();
1010
1011         if (pendc && (pendc != '\r' || !endline))
1012                 /*
1013                  * If we had a pending character, put it in the buffer.
1014                  * But discard a pending CR if we are at end of line
1015                  * (that is, discard the CR in a CR/LF sequence).
1016                  */
1017                 (void) do_append(pendc, NULL, pendpos);
1018
1019         /*
1020          * Make sure we've shifted the line, if we need to.
1021          */
1022         if (cshift < hshift)
1023                 pshift(hshift - cshift);
1024
1025         if (ctldisp == OPT_ONPLUS && is_ansi_end('m'))
1026         {
1027                 /* Switch to normal attribute at end of line. */
1028                 char *p = "\033[m";
1029                 for ( ;  *p != '\0';  p++)
1030                 {
1031                         linebuf[curr] = *p;
1032                         attr[curr++] = AT_ANSI;
1033                 }
1034         }
1035
1036         /*
1037          * Add a newline if necessary,
1038          * and append a '\0' to the end of the line.
1039          * We output a newline if we're not at the right edge of the screen,
1040          * or if the terminal doesn't auto wrap,
1041          * or if this is really the end of the line AND the terminal ignores
1042          * a newline at the right edge.
1043          * (In the last case we don't want to output a newline if the terminal
1044          * doesn't ignore it since that would produce an extra blank line.
1045          * But we do want to output a newline if the terminal ignores it in case
1046          * the next line is blank.  In that case the single newline output for
1047          * that blank line would be ignored!)
1048          */
1049         if (column < sc_width || !auto_wrap || (endline && ignaw) || ctldisp == OPT_ON)
1050         {
1051                 linebuf[curr] = '\n';
1052                 attr[curr] = AT_NORMAL;
1053                 curr++;
1054         }
1055         else if (ignaw && column >= sc_width && forw)
1056         {
1057                 /*
1058                  * Terminals with "ignaw" don't wrap until they *really* need
1059                  * to, i.e. when the character *after* the last one to fit on a
1060                  * line is output. But they are too hard to deal with when they
1061                  * get in the state where a full screen width of characters
1062                  * have been output but the cursor is sitting on the right edge
1063                  * instead of at the start of the next line.
1064                  * So we nudge them into wrapping by outputting a space
1065                  * character plus a backspace.  But do this only if moving
1066                  * forward; if we're moving backward and drawing this line at
1067                  * the top of the screen, the space would overwrite the first
1068                  * char on the next line.  We don't need to do this "nudge"
1069                  * at the top of the screen anyway.
1070                  */
1071                 linebuf[curr] = ' ';
1072                 attr[curr++] = AT_NORMAL;
1073                 linebuf[curr] = '\b';
1074                 attr[curr++] = AT_NORMAL;
1075         }
1076         linebuf[curr] = '\0';
1077         attr[curr] = AT_NORMAL;
1078 }
1079
1080 /*
1081  *
1082  */
1083         public void
1084 set_status_col(c)
1085         char c;
1086 {
1087         linebuf[0] = c;
1088         attr[0] = AT_NORMAL|AT_HILITE;
1089 }
1090
1091 /*
1092  * Get a character from the current line.
1093  * Return the character as the function return value,
1094  * and the character attribute in *ap.
1095  */
1096         public int
1097 gline(i, ap)
1098         register int i;
1099         register int *ap;
1100 {
1101         if (is_null_line)
1102         {
1103                 /*
1104                  * If there is no current line, we pretend the line is
1105                  * either "~" or "", depending on the "twiddle" flag.
1106                  */
1107                 if (twiddle)
1108                 {
1109                         if (i == 0)
1110                         {
1111                                 *ap = AT_BOLD;
1112                                 return '~';
1113                         }
1114                         --i;
1115                 }
1116                 /* Make sure we're back to AT_NORMAL before the '\n'.  */
1117                 *ap = AT_NORMAL;
1118                 return i ? '\0' : '\n';
1119         }
1120
1121         *ap = attr[i];
1122         return (linebuf[i] & 0xFF);
1123 }
1124
1125 /*
1126  * Indicate that there is no current line.
1127  */
1128         public void
1129 null_line()
1130 {
1131         is_null_line = 1;
1132         cshift = 0;
1133 }
1134
1135 /*
1136  * Analogous to forw_line(), but deals with "raw lines":
1137  * lines which are not split for screen width.
1138  * {{ This is supposed to be more efficient than forw_line(). }}
1139  */
1140         public POSITION
1141 forw_raw_line(curr_pos, linep, line_lenp)
1142         POSITION curr_pos;
1143         char **linep;
1144         int *line_lenp;
1145 {
1146         register int n;
1147         register int c;
1148         POSITION new_pos;
1149
1150         if (curr_pos == NULL_POSITION || ch_seek(curr_pos) ||
1151                 (c = ch_forw_get()) == EOI)
1152                 return (NULL_POSITION);
1153
1154         n = 0;
1155         for (;;)
1156         {
1157                 if (c == '\n' || c == EOI || ABORT_SIGS())
1158                 {
1159                         new_pos = ch_tell();
1160                         break;
1161                 }
1162                 if (n >= size_linebuf-1)
1163                 {
1164                         if (expand_linebuf())
1165                         {
1166                                 /*
1167                                  * Overflowed the input buffer.
1168                                  * Pretend the line ended here.
1169                                  */
1170                                 new_pos = ch_tell() - 1;
1171                                 break;
1172                         }
1173                 }
1174                 linebuf[n++] = c;
1175                 c = ch_forw_get();
1176         }
1177         linebuf[n] = '\0';
1178         if (linep != NULL)
1179                 *linep = linebuf;
1180         if (line_lenp != NULL)
1181                 *line_lenp = n;
1182         return (new_pos);
1183 }
1184
1185 /*
1186  * Analogous to back_line(), but deals with "raw lines".
1187  * {{ This is supposed to be more efficient than back_line(). }}
1188  */
1189         public POSITION
1190 back_raw_line(curr_pos, linep, line_lenp)
1191         POSITION curr_pos;
1192         char **linep;
1193         int *line_lenp;
1194 {
1195         register int n;
1196         register int c;
1197         POSITION new_pos;
1198
1199         if (curr_pos == NULL_POSITION || curr_pos <= ch_zero() ||
1200                 ch_seek(curr_pos-1))
1201                 return (NULL_POSITION);
1202
1203         n = size_linebuf;
1204         linebuf[--n] = '\0';
1205         for (;;)
1206         {
1207                 c = ch_back_get();
1208                 if (c == '\n' || ABORT_SIGS())
1209                 {
1210                         /*
1211                          * This is the newline ending the previous line.
1212                          * We have hit the beginning of the line.
1213                          */
1214                         new_pos = ch_tell() + 1;
1215                         break;
1216                 }
1217                 if (c == EOI)
1218                 {
1219                         /*
1220                          * We have hit the beginning of the file.
1221                          * This must be the first line in the file.
1222                          * This must, of course, be the beginning of the line.
1223                          */
1224                         new_pos = ch_zero();
1225                         break;
1226                 }
1227                 if (n <= 0)
1228                 {
1229                         int old_size_linebuf = size_linebuf;
1230                         char *fm;
1231                         char *to;
1232                         if (expand_linebuf())
1233                         {
1234                                 /*
1235                                  * Overflowed the input buffer.
1236                                  * Pretend the line ended here.
1237                                  */
1238                                 new_pos = ch_tell() + 1;
1239                                 break;
1240                         }
1241                         /*
1242                          * Shift the data to the end of the new linebuf.
1243                          */
1244                         for (fm = linebuf + old_size_linebuf - 1,
1245                               to = linebuf + size_linebuf - 1;
1246                              fm >= linebuf;  fm--, to--)
1247                                 *to = *fm;
1248                         n = size_linebuf - old_size_linebuf;
1249                 }
1250                 linebuf[--n] = c;
1251         }
1252         if (linep != NULL)
1253                 *linep = &linebuf[n];
1254         if (line_lenp != NULL)
1255                 *line_lenp = size_linebuf - 1 - n;
1256         return (new_pos);
1257 }