contrib/less/line.c

   1 /*
   2  * Copyright (C) 1984-2009  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 about less, or for information on how to
   8  * contact the author, see the README file.
   9  */
  10
  11
  12 /*
  13  * Routines to manipulate the "line buffer".
  14  * The line buffer holds a line of output as it is being built
  15  * in preparation for output to the screen.
  16  */
  17
  18 #include "less.h"
  19 #include "charset.h"
  20
  21 static char *linebuf = NULL;    /* Buffer which holds the current output line */
  22 static char *attr = NULL;       /* Extension of linebuf to hold attributes */
  23 public int size_linebuf = 0;    /* Size of line buffer (and attr buffer) */
  24
  25 static int cshift;              /* Current left-shift of output line buffer */
  26 public int hshift;              /* Desired left-shift of output line buffer */
  27 public int tabstops[TABSTOP_MAX] = { 0 }; /* Custom tabstops */
  28 public int ntabstops = 1;       /* Number of tabstops */
  29 public int tabdefault = 8;      /* Default repeated tabstops */
  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 char 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 = 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 = 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                                 a |= AT_HILITE;
 589                 }
 590         }
 591 #endif
 592
 593         if (ctldisp == OPT_ONPLUS && in_ansi_esc_seq())
 594         {
 595                 if (!is_ansi_end(ch) && !is_ansi_middle(ch)) {
 596                         /* Remove whole unrecognized sequence.  */
 597                         char *p = &linebuf[curr];
 598                         LWCHAR bch;
 599                         do {
 600                                 bch = step_char(&p, -1, linebuf);
 601                         } while (p > linebuf && !IS_CSI_START(bch));
 602                         curr = p - linebuf;
 603                         return 0;
 604                 }
 605                 a = AT_ANSI;    /* Will force re-AT_'ing around it.  */
 606                 w = 0;
 607         }
 608         else if (ctldisp == OPT_ONPLUS && IS_CSI_START(ch))
 609         {
 610                 a = AT_ANSI;    /* Will force re-AT_'ing around it.  */
 611                 w = 0;
 612         }
 613         else
 614         {
 615                 char *p = &linebuf[curr];
 616                 LWCHAR prev_ch = step_char(&p, -1, linebuf);
 617                 w = pwidth(ch, a, prev_ch);
 618         }
 619
 620         if (ctldisp != OPT_ON && column + w + attr_ewidth(a) > sc_width)
 621                 /*
 622                  * Won't fit on screen.
 623                  */
 624                 return (1);
 625
 626         if (rep == NULL)
 627         {
 628                 cs = (char) ch;
 629                 rep = &cs;
 630                 replen = 1;
 631         } else
 632         {
 633                 replen = utf_len(rep[0]);
 634         }
 635         if (curr + replen >= size_linebuf-6)
 636         {
 637                 /*
 638                  * Won't fit in line buffer.
 639                  * Try to expand it.
 640                  */
 641                 if (expand_linebuf())
 642                         return (1);
 643         }
 644
 645         while (replen-- > 0)
 646         {
 647                 linebuf[curr] = *rep++;
 648                 attr[curr] = a;
 649                 curr++;
 650         }
 651         column += w;
 652         return (0);
 653 }
 654
 655 /*
 656  * Append a tab to the line buffer.
 657  * Store spaces to represent the tab.
 658  */
 659 #define STORE_TAB(a,pos) \
 660         do { if (store_tab((a),(pos))) return (1); } while (0)
 661
 662         static int
 663 store_tab(attr, pos)
 664         int attr;
 665         POSITION pos;
 666 {
 667         int to_tab = column + cshift - lmargin;
 668         int i;
 669
 670         if (ntabstops < 2 || to_tab >= tabstops[ntabstops-1])
 671                 to_tab = tabdefault -
 672                      ((to_tab - tabstops[ntabstops-1]) % tabdefault);
 673         else
 674         {
 675                 for (i = ntabstops - 2;  i >= 0;  i--)
 676                         if (to_tab >= tabstops[i])
 677                                 break;
 678                 to_tab = tabstops[i+1] - to_tab;
 679         }
 680
 681         if (column + to_tab - 1 + pwidth(' ', attr, 0) + attr_ewidth(attr) > sc_width)
 682                 return 1;
 683
 684         do {
 685                 STORE_CHAR(' ', attr, " ", pos);
 686         } while (--to_tab > 0);
 687         return 0;
 688 }
 689
 690 #define STORE_PRCHAR(c, pos) \
 691         do { if (store_prchar((c), (pos))) return 1; } while (0)
 692
 693         static int
 694 store_prchar(c, pos)
 695         char c;
 696         POSITION pos;
 697 {
 698         char *s;
 699
 700         /*
 701          * Convert to printable representation.
 702          */
 703         s = prchar(c);
 704
 705         /*
 706          * Make sure we can get the entire representation
 707          * of the character on this line.
 708          */
 709         if (column + (int) strlen(s) - 1 +
 710             pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
 711                 return 1;
 712
 713         for ( ;  *s != 0;  s++)
 714                 STORE_CHAR(*s, AT_BINARY, NULL, pos);
 715
 716         return 0;
 717 }
 718
 719         static int
 720 flush_mbc_buf(pos)
 721         POSITION pos;
 722 {
 723         int i;
 724
 725         for (i = 0; i < mbc_buf_index; i++)
 726                 if (store_prchar(mbc_buf[i], pos))
 727                         return mbc_buf_index - i;
 728
 729         return 0;
 730 }
 731
 732 /*
 733  * Append a character to the line buffer.
 734  * Expand tabs into spaces, handle underlining, boldfacing, etc.
 735  * Returns 0 if ok, 1 if couldn't fit in buffer.
 736  */
 737         public int
 738 pappend(c, pos)
 739         char c;
 740         POSITION pos;
 741 {
 742         int r;
 743
 744         if (pendc)
 745         {
 746                 if (do_append(pendc, NULL, pendpos))
 747                         /*
 748                          * Oops.  We've probably lost the char which
 749                          * was in pendc, since caller won't back up.
 750                          */
 751                         return (1);
 752                 pendc = '\0';
 753         }
 754
 755         if (c == '\r' && bs_mode == BS_SPECIAL)
 756         {
 757                 if (mbc_buf_len > 0)  /* utf_mode must be on. */
 758                 {
 759                         /* Flush incomplete (truncated) sequence. */
 760                         r = flush_mbc_buf(mbc_pos);
 761                         mbc_buf_index = r + 1;
 762                         mbc_buf_len = 0;
 763                         if (r)
 764                                 return (mbc_buf_index);
 765                 }
 766
 767                 /*
 768                  * Don't put the CR into the buffer until we see
 769                  * the next char.  If the next char is a newline,
 770                  * discard the CR.
 771                  */
 772                 pendc = c;
 773                 pendpos = pos;
 774                 return (0);
 775         }
 776
 777         if (!utf_mode)
 778         {
 779                 r = do_append((LWCHAR) c, NULL, pos);
 780         } else
 781         {
 782                 /* Perform strict validation in all possible cases. */
 783                 if (mbc_buf_len == 0)
 784                 {
 785                 retry:
 786                         mbc_buf_index = 1;
 787                         *mbc_buf = c;
 788                         if (IS_ASCII_OCTET(c))
 789                                 r = do_append((LWCHAR) c, NULL, pos);
 790                         else if (IS_UTF8_LEAD(c))
 791                         {
 792                                 mbc_buf_len = utf_len(c);
 793                                 mbc_pos = pos;
 794                                 return (0);
 795                         } else
 796                                 /* UTF8_INVALID or stray UTF8_TRAIL */
 797                                 r = flush_mbc_buf(pos);
 798                 } else if (IS_UTF8_TRAIL(c))
 799                 {
 800                         mbc_buf[mbc_buf_index++] = c;
 801                         if (mbc_buf_index < mbc_buf_len)
 802                                 return (0);
 803                         if (is_utf8_well_formed(mbc_buf))
 804                                 r = do_append(get_wchar(mbc_buf), mbc_buf, mbc_pos);
 805                         else
 806                                 /* Complete, but not shortest form, sequence. */
 807                                 mbc_buf_index = r = flush_mbc_buf(mbc_pos);
 808                         mbc_buf_len = 0;
 809                 } else
 810                 {
 811                         /* Flush incomplete (truncated) sequence.  */
 812                         r = flush_mbc_buf(mbc_pos);
 813                         mbc_buf_index = r + 1;
 814                         mbc_buf_len = 0;
 815                         /* Handle new char.  */
 816                         if (!r)
 817                                 goto retry;
 818                 }
 819         }
 820
 821         /*
 822          * If we need to shift the line, do it.
 823          * But wait until we get to at least the middle of the screen,
 824          * so shifting it doesn't affect the chars we're currently
 825          * pappending.  (Bold & underline can get messed up otherwise.)
 826          */
 827         if (cshift < hshift && column > sc_width / 2)
 828         {
 829                 linebuf[curr] = '\0';
 830                 pshift(hshift - cshift);
 831         }
 832         if (r)
 833         {
 834                 /* How many chars should caller back up? */
 835                 r = (!utf_mode) ? 1 : mbc_buf_index;
 836         }
 837         return (r);
 838 }
 839
 840         static int
 841 do_append(ch, rep, pos)
 842         LWCHAR ch;
 843         char *rep;
 844         POSITION pos;
 845 {
 846         register int a;
 847         LWCHAR prev_ch;
 848
 849         a = AT_NORMAL;
 850
 851         if (ch == '\b')
 852         {
 853                 if (bs_mode == BS_CONTROL)
 854                         goto do_control_char;
 855
 856                 /*
 857                  * A better test is needed here so we don't
 858                  * backspace over part of the printed
 859                  * representation of a binary character.
 860                  */
 861                 if (   curr <= lmargin
 862                     || column <= lmargin
 863                     || (attr[curr - 1] & (AT_ANSI|AT_BINARY)))
 864                         STORE_PRCHAR('\b', pos);
 865                 else if (bs_mode == BS_NORMAL)
 866                         STORE_CHAR(ch, AT_NORMAL, NULL, pos);
 867                 else if (bs_mode == BS_SPECIAL)
 868                         overstrike = backc();
 869
 870                 return 0;
 871         }
 872
 873         if (overstrike > 0)
 874         {
 875                 /*
 876                  * Overstrike the character at the current position
 877                  * in the line buffer.  This will cause either
 878                  * underline (if a "_" is overstruck),
 879                  * bold (if an identical character is overstruck),
 880                  * or just deletion of the character in the buffer.
 881                  */
 882                 overstrike = utf_mode ? -1 : 0;
 883                 /* To be correct, this must be a base character.  */
 884                 prev_ch = get_wchar(linebuf + curr);
 885                 a = attr[curr];
 886                 if (ch == prev_ch)
 887                 {
 888                         /*
 889                          * Overstriking a char with itself means make it bold.
 890                          * But overstriking an underscore with itself is
 891                          * ambiguous.  It could mean make it bold, or
 892                          * it could mean make it underlined.
 893                          * Use the previous overstrike to resolve it.
 894                          */
 895                         if (ch == '_')
 896                         {
 897                                 if ((a & (AT_BOLD|AT_UNDERLINE)) != AT_NORMAL)
 898                                         a |= (AT_BOLD|AT_UNDERLINE);
 899                                 else if (last_overstrike != AT_NORMAL)
 900                                         a |= last_overstrike;
 901                                 else
 902                                         a |= AT_BOLD;
 903                         } else
 904                                 a |= AT_BOLD;
 905                 } else if (ch == '_')
 906                 {
 907                         a |= AT_UNDERLINE;
 908                         ch = prev_ch;
 909                         rep = linebuf + curr;
 910                 } else if (prev_ch == '_')
 911                 {
 912                         a |= AT_UNDERLINE;
 913                 }
 914                 /* Else we replace prev_ch, but we keep its attributes.  */
 915         } else if (overstrike < 0)
 916         {
 917                 if (   is_composing_char(ch)
 918                     || is_combining_char(get_wchar(linebuf + curr), ch))
 919                         /* Continuation of the same overstrike.  */
 920                         a = last_overstrike;
 921                 else
 922                         overstrike = 0;
 923         }
 924
 925         if (ch == '\t')
 926         {
 927                 /*
 928                  * Expand a tab into spaces.
 929                  */
 930                 switch (bs_mode)
 931                 {
 932                 case BS_CONTROL:
 933                         goto do_control_char;
 934                 case BS_NORMAL:
 935                 case BS_SPECIAL:
 936                         STORE_TAB(a, pos);
 937                         break;
 938                 }
 939         } else if ((!utf_mode || is_ascii_char(ch)) && control_char((char)ch))
 940         {
 941         do_control_char:
 942                 if (ctldisp == OPT_ON || (ctldisp == OPT_ONPLUS && IS_CSI_START(ch)))
 943                 {
 944                         /*
 945                          * Output as a normal character.
 946                          */
 947                         STORE_CHAR(ch, AT_NORMAL, rep, pos);
 948                 } else
 949                 {
 950                         STORE_PRCHAR((char) ch, pos);
 951                 }
 952         } else if (utf_mode && ctldisp != OPT_ON && is_ubin_char(ch))
 953         {
 954                 char *s;
 955
 956                 s = prutfchar(ch);
 957
 958                 if (column + (int) strlen(s) - 1 +
 959                     pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
 960                         return (1);
 961
 962                 for ( ;  *s != 0;  s++)
 963                         STORE_CHAR(*s, AT_BINARY, NULL, pos);
 964         } else
 965         {
 966                 STORE_CHAR(ch, a, rep, pos);
 967         }
 968         return (0);
 969 }
 970
 971 /*
 972  *
 973  */
 974         public int
 975 pflushmbc()
 976 {
 977         int r = 0;
 978
 979         if (mbc_buf_len > 0)
 980         {
 981                 /* Flush incomplete (truncated) sequence.  */
 982                 r = flush_mbc_buf(mbc_pos);
 983                 mbc_buf_len = 0;
 984         }
 985         return r;
 986 }
 987
 988 /*
 989  * Terminate the line in the line buffer.
 990  */
 991         public void
 992 pdone(endline, forw)
 993         int endline;
 994         int forw;
 995 {
 996         (void) pflushmbc();
 997
 998         if (pendc && (pendc != '\r' || !endline))
 999                 /*
1000                  * If we had a pending character, put it in the buffer.
1001                  * But discard a pending CR if we are at end of line
1002                  * (that is, discard the CR in a CR/LF sequence).
1003                  */
1004                 (void) do_append(pendc, NULL, pendpos);
1005
1006         /*
1007          * Make sure we've shifted the line, if we need to.
1008          */
1009         if (cshift < hshift)
1010                 pshift(hshift - cshift);
1011
1012         if (ctldisp == OPT_ONPLUS && is_ansi_end('m'))
1013         {
1014                 /* Switch to normal attribute at end of line. */
1015                 char *p = "\033[m";
1016                 for ( ;  *p != '\0';  p++)
1017                 {
1018                         linebuf[curr] = *p;
1019                         attr[curr++] = AT_ANSI;
1020                 }
1021         }
1022
1023         /*
1024          * Add a newline if necessary,
1025          * and append a '\0' to the end of the line.
1026          * We output a newline if we're not at the right edge of the screen,
1027          * or if the terminal doesn't auto wrap,
1028          * or if this is really the end of the line AND the terminal ignores
1029          * a newline at the right edge.
1030          * (In the last case we don't want to output a newline if the terminal
1031          * doesn't ignore it since that would produce an extra blank line.
1032          * But we do want to output a newline if the terminal ignores it in case
1033          * the next line is blank.  In that case the single newline output for
1034          * that blank line would be ignored!)
1035          */
1036         if (column < sc_width || !auto_wrap || (endline && ignaw) || ctldisp == OPT_ON)
1037         {
1038                 linebuf[curr] = '\n';
1039                 attr[curr] = AT_NORMAL;
1040                 curr++;
1041         }
1042         else if (ignaw && column >= sc_width && forw)
1043         {
1044                 /*
1045                  * Terminals with "ignaw" don't wrap until they *really* need
1046                  * to, i.e. when the character *after* the last one to fit on a
1047                  * line is output. But they are too hard to deal with when they
1048                  * get in the state where a full screen width of characters
1049                  * have been output but the cursor is sitting on the right edge
1050                  * instead of at the start of the next line.
1051                  * So we nudge them into wrapping by outputting a space
1052                  * character plus a backspace.  But do this only if moving
1053                  * forward; if we're moving backward and drawing this line at
1054                  * the top of the screen, the space would overwrite the first
1055                  * char on the next line.  We don't need to do this "nudge"
1056                  * at the top of the screen anyway.
1057                  */
1058                 linebuf[curr] = ' ';
1059                 attr[curr++] = AT_NORMAL;
1060                 linebuf[curr] = '\b';
1061                 attr[curr++] = AT_NORMAL;
1062         }
1063         linebuf[curr] = '\0';
1064         attr[curr] = AT_NORMAL;
1065 }
1066
1067 /*
1068  *
1069  */
1070         public void
1071 set_status_col(c)
1072         char c;
1073 {
1074         linebuf[0] = c;
1075         attr[0] = AT_NORMAL|AT_HILITE;
1076 }
1077
1078 /*
1079  * Get a character from the current line.
1080  * Return the character as the function return value,
1081  * and the character attribute in *ap.
1082  */
1083         public int
1084 gline(i, ap)
1085         register int i;
1086         register int *ap;
1087 {
1088         if (is_null_line)
1089         {
1090                 /*
1091                  * If there is no current line, we pretend the line is
1092                  * either "~" or "", depending on the "twiddle" flag.
1093                  */
1094                 if (twiddle)
1095                 {
1096                         if (i == 0)
1097                         {
1098                                 *ap = AT_BOLD;
1099                                 return '~';
1100                         }
1101                         --i;
1102                 }
1103                 /* Make sure we're back to AT_NORMAL before the '\n'.  */
1104                 *ap = AT_NORMAL;
1105                 return i ? '\0' : '\n';
1106         }
1107
1108         *ap = attr[i];
1109         return (linebuf[i] & 0xFF);
1110 }
1111
1112 /*
1113  * Indicate that there is no current line.
1114  */
1115         public void
1116 null_line()
1117 {
1118         is_null_line = 1;
1119         cshift = 0;
1120 }
1121
1122 /*
1123  * Analogous to forw_line(), but deals with "raw lines":
1124  * lines which are not split for screen width.
1125  * {{ This is supposed to be more efficient than forw_line(). }}
1126  */
1127         public POSITION
1128 forw_raw_line(curr_pos, linep, line_lenp)
1129         POSITION curr_pos;
1130         char **linep;
1131         int *line_lenp;
1132 {
1133         register int n;
1134         register int c;
1135         POSITION new_pos;
1136
1137         if (curr_pos == NULL_POSITION || ch_seek(curr_pos) ||
1138                 (c = ch_forw_get()) == EOI)
1139                 return (NULL_POSITION);
1140
1141         n = 0;
1142         for (;;)
1143         {
1144                 if (c == '\n' || c == EOI || ABORT_SIGS())
1145                 {
1146                         new_pos = ch_tell();
1147                         break;
1148                 }
1149                 if (n >= size_linebuf-1)
1150                 {
1151                         if (expand_linebuf())
1152                         {
1153                                 /*
1154                                  * Overflowed the input buffer.
1155                                  * Pretend the line ended here.
1156                                  */
1157                                 new_pos = ch_tell() - 1;
1158                                 break;
1159                         }
1160                 }
1161                 linebuf[n++] = c;
1162                 c = ch_forw_get();
1163         }
1164         linebuf[n] = '\0';
1165         if (linep != NULL)
1166                 *linep = linebuf;
1167         if (line_lenp != NULL)
1168                 *line_lenp = n;
1169         return (new_pos);
1170 }
1171
1172 /*
1173  * Analogous to back_line(), but deals with "raw lines".
1174  * {{ This is supposed to be more efficient than back_line(). }}
1175  */
1176         public POSITION
1177 back_raw_line(curr_pos, linep, line_lenp)
1178         POSITION curr_pos;
1179         char **linep;
1180         int *line_lenp;
1181 {
1182         register int n;
1183         register int c;
1184         POSITION new_pos;
1185
1186         if (curr_pos == NULL_POSITION || curr_pos <= ch_zero() ||
1187                 ch_seek(curr_pos-1))
1188                 return (NULL_POSITION);
1189
1190         n = size_linebuf;
1191         linebuf[--n] = '\0';
1192         for (;;)
1193         {
1194                 c = ch_back_get();
1195                 if (c == '\n' || ABORT_SIGS())
1196                 {
1197                         /*
1198                          * This is the newline ending the previous line.
1199                          * We have hit the beginning of the line.
1200                          */
1201                         new_pos = ch_tell() + 1;
1202                         break;
1203                 }
1204                 if (c == EOI)
1205                 {
1206                         /*
1207                          * We have hit the beginning of the file.
1208                          * This must be the first line in the file.
1209                          * This must, of course, be the beginning of the line.
1210                          */
1211                         new_pos = ch_zero();
1212                         break;
1213                 }
1214                 if (n <= 0)
1215                 {
1216                         int old_size_linebuf = size_linebuf;
1217                         char *fm;
1218                         char *to;
1219                         if (expand_linebuf())
1220                         {
1221                                 /*
1222                                  * Overflowed the input buffer.
1223                                  * Pretend the line ended here.
1224                                  */
1225                                 new_pos = ch_tell() + 1;
1226                                 break;
1227                         }
1228                         /*
1229                          * Shift the data to the end of the new linebuf.
1230                          */
1231                         for (fm = linebuf + old_size_linebuf - 1,
1232                               to = linebuf + size_linebuf - 1;
1233                              fm >= linebuf;  fm--, to--)
1234                                 *to = *fm;
1235                         n = size_linebuf - old_size_linebuf;
1236                 }
1237                 linebuf[--n] = c;
1238         }
1239         if (linep != NULL)
1240                 *linep = &linebuf[n];
1241         if (line_lenp != NULL)
1242                 *line_lenp = size_linebuf - 1 - n;
1243         return (new_pos);
1244 }