contrib/less-4/line.c

   1 /*
   2  * Copyright (C) 1984-2007  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 int line_matches;        /* Number of search matches in this line */
  39 static char pendc;
  40 static POSITION pendpos;
  41 static char *end_ansi_chars;
  42 static char *mid_ansi_chars;
  43
  44 static int attr_swidth();
  45 static int attr_ewidth();
  46 static int do_append();
  47
  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 int oldbot;
  63 extern POSITION start_attnpos;
  64 extern POSITION end_attnpos;
  65
  66 static char mbc_buf[MAX_UTF_CHAR_LEN];
  67 static int mbc_buf_len = 0;
  68 static int mbc_buf_index = 0;
  69 static POSITION mbc_pos;
  70
  71 /*
  72  * Initialize from environment variables.
  73  */
  74         public void
  75 init_line()
  76 {
  77         end_ansi_chars = lgetenv("LESSANSIENDCHARS");
  78         if (end_ansi_chars == NULL || *end_ansi_chars == '\0')
  79                 end_ansi_chars = "m";
  80
  81         mid_ansi_chars = lgetenv("LESSANSIMIDCHARS");
  82         if (mid_ansi_chars == NULL || *mid_ansi_chars == '\0')
  83                 mid_ansi_chars = "0123456789;[?!\"'#%()*+ ";
  84
  85         linebuf = (char *) ecalloc(LINEBUF_SIZE, sizeof(char));
  86         attr = (char *) ecalloc(LINEBUF_SIZE, sizeof(char));
  87         size_linebuf = LINEBUF_SIZE;
  88 }
  89
  90 /*
  91  * Expand the line buffer.
  92  */
  93         static int
  94 expand_linebuf()
  95 {
  96         /* Double the size of the line buffer. */
  97         int new_size = size_linebuf * 2;
  98
  99         /* Just realloc to expand the buffer, if we can. */
 100 #if HAVE_REALLOC
 101         char *new_buf = (char *) realloc(linebuf, new_size);
 102         char *new_attr = (char *) realloc(attr, new_size);
 103 #else
 104         char *new_buf = (char *) calloc(new_size, sizeof(char));
 105         char *new_attr = (char *) calloc(new_size, sizeof(char));
 106 #endif
 107         if (new_buf == NULL || new_attr == NULL)
 108         {
 109                 if (new_attr != NULL)
 110                         free(new_attr);
 111                 if (new_buf != NULL)
 112                         free(new_buf);
 113                 return 1;
 114         }
 115 #if HAVE_REALLOC
 116         /*
 117          * We realloc'd the buffers; they already have the old contents.
 118          */
 119         #if 0
 120         memset(new_buf + size_linebuf, 0, new_size - size_linebuf);
 121         memset(new_attr + size_linebuf, 0, new_size - size_linebuf);
 122         #endif
 123 #else
 124         /*
 125          * We just calloc'd the buffers; copy the old contents.
 126          */
 127         memcpy(new_buf, linebuf, size_linebuf * sizeof(char));
 128         memcpy(new_attr, attr, size_linebuf * sizeof(char));
 129         free(attr);
 130         free(linebuf);
 131 #endif
 132         linebuf = new_buf;
 133         attr = new_attr;
 134         size_linebuf = new_size;
 135         return 0;
 136 }
 137
 138 /*
 139  * Is a character ASCII?
 140  */
 141         public int
 142 is_ascii_char(ch)
 143         LWCHAR ch;
 144 {
 145         return (ch <= 0x7F);
 146 }
 147
 148 /*
 149  * Rewind the line buffer.
 150  */
 151         public void
 152 prewind()
 153 {
 154         curr = 0;
 155         column = 0;
 156         cshift = 0;
 157         overstrike = 0;
 158         last_overstrike = AT_NORMAL;
 159         mbc_buf_len = 0;
 160         is_null_line = 0;
 161         pendc = '\0';
 162         lmargin = 0;
 163         if (status_col)
 164                 lmargin += 1;
 165 #if HILITE_SEARCH
 166         line_matches = 0;
 167 #endif
 168 }
 169
 170 /*
 171  * Insert the line number (of the given position) into the line buffer.
 172  */
 173         public void
 174 plinenum(pos)
 175         POSITION pos;
 176 {
 177         register LINENUM linenum = 0;
 178         register int i;
 179
 180         if (linenums == OPT_ONPLUS)
 181         {
 182                 /*
 183                  * Get the line number and put it in the current line.
 184                  * {{ Note: since find_linenum calls forw_raw_line,
 185                  *    it may seek in the input file, requiring the caller
 186                  *    of plinenum to re-seek if necessary. }}
 187                  * {{ Since forw_raw_line modifies linebuf, we must
 188                  *    do this first, before storing anything in linebuf. }}
 189                  */
 190                 linenum = find_linenum(pos);
 191         }
 192
 193         /*
 194          * Display a status column if the -J option is set.
 195          */
 196         if (status_col)
 197         {
 198                 linebuf[curr] = ' ';
 199                 if (start_attnpos != NULL_POSITION &&
 200                     pos >= start_attnpos && pos < end_attnpos)
 201                         attr[curr] = AT_NORMAL|AT_HILITE;
 202                 else
 203                         attr[curr] = AT_NORMAL;
 204                 curr++;
 205                 column++;
 206         }
 207         /*
 208          * Display the line number at the start of each line
 209          * if the -N option is set.
 210          */
 211         if (linenums == OPT_ONPLUS)
 212         {
 213                 char buf[INT_STRLEN_BOUND(pos) + 2];
 214                 int n;
 215
 216                 linenumtoa(linenum, buf);
 217                 n = strlen(buf);
 218                 if (n < MIN_LINENUM_WIDTH)
 219                         n = MIN_LINENUM_WIDTH;
 220                 sprintf(linebuf+curr, "%*s ", n, buf);
 221                 n++;  /* One space after the line number. */
 222                 for (i = 0; i < n; i++)
 223                         attr[curr+i] = AT_NORMAL;
 224                 curr += n;
 225                 column += n;
 226                 lmargin += n;
 227         }
 228
 229         /*
 230          * Append enough spaces to bring us to the lmargin.
 231          */
 232         while (column < lmargin)
 233         {
 234                 linebuf[curr] = ' ';
 235                 attr[curr++] = AT_NORMAL;
 236                 column++;
 237         }
 238 }
 239
 240 /*
 241  * Shift the input line left.
 242  * This means discarding N printable chars at the start of the buffer.
 243  */
 244         static void
 245 pshift(shift)
 246         int shift;
 247 {
 248         LWCHAR prev_ch = 0;
 249         unsigned char c;
 250         int shifted = 0;
 251         int to;
 252         int from;
 253         int len;
 254         int width;
 255         int prev_attr;
 256         int next_attr;
 257
 258         if (shift > column - lmargin)
 259                 shift = column - lmargin;
 260         if (shift > curr - lmargin)
 261                 shift = curr - lmargin;
 262
 263         to = from = lmargin;
 264         /*
 265          * We keep on going when shifted == shift
 266          * to get all combining chars.
 267          */
 268         while (shifted <= shift && from < curr)
 269         {
 270                 c = linebuf[from];
 271                 if (ctldisp == OPT_ONPLUS && IS_CSI_START(c))
 272                 {
 273                         /* Keep cumulative effect.  */
 274                         linebuf[to] = c;
 275                         attr[to++] = attr[from++];
 276                         while (from < curr && linebuf[from])
 277                         {
 278                                 linebuf[to] = linebuf[from];
 279                                 attr[to++] = attr[from];
 280                                 if (!is_ansi_middle(linebuf[from++]))
 281                                         break;
 282                         }
 283                         continue;
 284                 }
 285
 286                 width = 0;
 287
 288                 if (!IS_ASCII_OCTET(c) && utf_mode)
 289                 {
 290                         /* Assumes well-formedness validation already done.  */
 291                         LWCHAR ch;
 292
 293                         len = utf_len(c);
 294                         if (from + len > curr)
 295                                 break;
 296                         ch = get_wchar(linebuf + from);
 297                         if (!is_composing_char(ch) && !is_combining_char(prev_ch, ch))
 298                                 width = is_wide_char(ch) ? 2 : 1;
 299                         prev_ch = ch;
 300                 } else
 301                 {
 302                         len = 1;
 303                         if (c == '\b')
 304                                 /* XXX - Incorrect if several '\b' in a row.  */
 305                                 width = (utf_mode && is_wide_char(prev_ch)) ? -2 : -1;
 306                         else if (!control_char(c))
 307                                 width = 1;
 308                         prev_ch = 0;
 309                 }
 310
 311                 if (width == 2 && shift - shifted == 1) {
 312                         /* Should never happen when called by pshift_all().  */
 313                         attr[to] = attr[from];
 314                         /*
 315                          * Assume a wide_char will never be the first half of a
 316                          * combining_char pair, so reset prev_ch in case we're
 317                          * followed by a '\b'.
 318                          */
 319                         prev_ch = linebuf[to++] = ' ';
 320                         from += len;
 321                         shifted++;
 322                         continue;
 323                 }
 324
 325                 /* Adjust width for magic cookies. */
 326                 prev_attr = (to > 0) ? attr[to-1] : AT_NORMAL;
 327                 next_attr = (from + len < curr) ? attr[from + len] : prev_attr;
 328                 if (!is_at_equiv(attr[from], prev_attr) &&
 329                         !is_at_equiv(attr[from], next_attr))
 330                 {
 331                         width += attr_swidth(attr[from]);
 332                         if (from + len < curr)
 333                                 width += attr_ewidth(attr[from]);
 334                         if (is_at_equiv(prev_attr, next_attr))
 335                         {
 336                                 width += attr_ewidth(prev_attr);
 337                                 if (from + len < curr)
 338                                         width += attr_swidth(next_attr);
 339                         }
 340                 }
 341
 342                 if (shift - shifted < width)
 343                         break;
 344                 from += len;
 345                 shifted += width;
 346                 if (shifted < 0)
 347                         shifted = 0;
 348         }
 349         while (from < curr)
 350         {
 351                 linebuf[to] = linebuf[from];
 352                 attr[to++] = attr[from++];
 353         }
 354         curr = to;
 355         column -= shifted;
 356         cshift += shifted;
 357 }
 358
 359 /*
 360  *
 361  */
 362         public void
 363 pshift_all()
 364 {
 365         pshift(column);
 366 }
 367
 368 /*
 369  * Return the printing width of the start (enter) sequence
 370  * for a given character attribute.
 371  */
 372         static int
 373 attr_swidth(a)
 374         int a;
 375 {
 376         int w = 0;
 377
 378         a = apply_at_specials(a);
 379
 380         if (a & AT_UNDERLINE)
 381                 w += ul_s_width;
 382         if (a & AT_BOLD)
 383                 w += bo_s_width;
 384         if (a & AT_BLINK)
 385                 w += bl_s_width;
 386         if (a & AT_STANDOUT)
 387                 w += so_s_width;
 388
 389         return w;
 390 }
 391
 392 /*
 393  * Return the printing width of the end (exit) sequence
 394  * for a given character attribute.
 395  */
 396         static int
 397 attr_ewidth(a)
 398         int a;
 399 {
 400         int w = 0;
 401
 402         a = apply_at_specials(a);
 403
 404         if (a & AT_UNDERLINE)
 405                 w += ul_e_width;
 406         if (a & AT_BOLD)
 407                 w += bo_e_width;
 408         if (a & AT_BLINK)
 409                 w += bl_e_width;
 410         if (a & AT_STANDOUT)
 411                 w += so_e_width;
 412
 413         return w;
 414 }
 415
 416 /*
 417  * Return the printing width of a given character and attribute,
 418  * if the character were added to the current position in the line buffer.
 419  * Adding a character with a given attribute may cause an enter or exit
 420  * attribute sequence to be inserted, so this must be taken into account.
 421  */
 422         static int
 423 pwidth(ch, a, prev_ch)
 424         LWCHAR ch;
 425         int a;
 426         LWCHAR prev_ch;
 427 {
 428         int w;
 429
 430         if (ch == '\b')
 431                 /*
 432                  * Backspace moves backwards one or two positions.
 433                  * XXX - Incorrect if several '\b' in a row.
 434                  */
 435                 return (utf_mode && is_wide_char(prev_ch)) ? -2 : -1;
 436
 437         if (!utf_mode || is_ascii_char(ch))
 438         {
 439                 if (control_char((char)ch))
 440                 {
 441                         /*
 442                          * Control characters do unpredictable things,
 443                          * so we don't even try to guess; say it doesn't move.
 444                          * This can only happen if the -r flag is in effect.
 445                          */
 446                         return (0);
 447                 }
 448         } else
 449         {
 450                 if (is_composing_char(ch) || is_combining_char(prev_ch, ch))
 451                 {
 452                         /*
 453                          * Composing and combining chars take up no space.
 454                          *
 455                          * Some terminals, upon failure to compose a
 456                          * composing character with the character(s) that
 457                          * precede(s) it will actually take up one column
 458                          * for the composing character; there isn't much
 459                          * we could do short of testing the (complex)
 460                          * composition process ourselves and printing
 461                          * a binary representation when it fails.
 462                          */
 463                         return (0);
 464                 }
 465         }
 466
 467         /*
 468          * Other characters take one or two columns,
 469          * plus the width of any attribute enter/exit sequence.
 470          */
 471         w = 1;
 472         if (is_wide_char(ch))
 473                 w++;
 474         if (curr > 0 && !is_at_equiv(attr[curr-1], a))
 475                 w += attr_ewidth(attr[curr-1]);
 476         if ((apply_at_specials(a) != AT_NORMAL) &&
 477             (curr == 0 || !is_at_equiv(attr[curr-1], a)))
 478                 w += attr_swidth(a);
 479         return (w);
 480 }
 481
 482 /*
 483  * Delete to the previous base character in the line buffer.
 484  * Return 1 if one is found.
 485  */
 486         static int
 487 backc()
 488 {
 489         LWCHAR prev_ch;
 490         char *p = linebuf + curr;
 491         LWCHAR ch = step_char(&p, -1, linebuf + lmargin);
 492         int width;
 493
 494         /* This assumes that there is no '\b' in linebuf.  */
 495         while (   curr > lmargin
 496                && column > lmargin
 497                && (!(attr[curr - 1] & (AT_ANSI|AT_BINARY))))
 498         {
 499                 curr = p - linebuf;
 500                 prev_ch = step_char(&p, -1, linebuf + lmargin);
 501                 width = pwidth(ch, attr[curr], prev_ch);
 502                 column -= width;
 503                 if (width > 0)
 504                         return 1;
 505                 ch = prev_ch;
 506         }
 507
 508         return 0;
 509 }
 510
 511 /*
 512  * Are we currently within a recognized ANSI escape sequence?
 513  */
 514         static int
 515 in_ansi_esc_seq()
 516 {
 517         char *p;
 518
 519         /*
 520          * Search backwards for either an ESC (which means we ARE in a seq);
 521          * or an end char (which means we're NOT in a seq).
 522          */
 523         for (p = &linebuf[curr];  p > linebuf; )
 524         {
 525                 LWCHAR ch = step_char(&p, -1, linebuf);
 526                 if (IS_CSI_START(ch))
 527                         return (1);
 528                 if (!is_ansi_middle(ch))
 529                         return (0);
 530         }
 531         return (0);
 532 }
 533
 534 /*
 535  * Is a character the end of an ANSI escape sequence?
 536  */
 537         public int
 538 is_ansi_end(ch)
 539         LWCHAR ch;
 540 {
 541         if (!is_ascii_char(ch))
 542                 return (0);
 543         return (strchr(end_ansi_chars, (char) ch) != NULL);
 544 }
 545
 546 /*
 547  *
 548  */
 549         public int
 550 is_ansi_middle(ch)
 551         LWCHAR ch;
 552 {
 553         if (!is_ascii_char(ch))
 554                 return (0);
 555         if (is_ansi_end(ch))
 556                 return (0);
 557         return (strchr(mid_ansi_chars, (char) ch) != NULL);
 558 }
 559
 560 /*
 561  * Append a character and attribute to the line buffer.
 562  */
 563 #define STORE_CHAR(ch,a,rep,pos) \
 564         do { \
 565                 if (store_char((ch),(a),(rep),(pos))) return (1); \
 566         } while (0)
 567
 568         static int
 569 store_char(ch, a, rep, pos)
 570         LWCHAR ch;
 571         int a;
 572         char *rep;
 573         POSITION pos;
 574 {
 575         int w;
 576         int replen;
 577         char cs;
 578
 579         w = (a & (AT_UNDERLINE|AT_BOLD));       /* Pre-use w.  */
 580         if (w != AT_NORMAL)
 581                 last_overstrike = w;
 582
 583 #if HILITE_SEARCH
 584         {
 585                 int matches;
 586                 if (is_hilited(pos, pos+1, 0, &matches))
 587                 {
 588                         /*
 589                          * This character should be highlighted.
 590                          * Override the attribute passed in.
 591                          */
 592                         if (a != AT_ANSI)
 593                                 a |= AT_HILITE;
 594                 }
 595                 line_matches += matches;
 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                         do {
 604                                 if (curr == 0)
 605                                         break;
 606                                 --curr;
 607                         } while (!IS_CSI_START(linebuf[curr]));
 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         char 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         char c;
 745         POSITION pos;
 746 {
 747         int r;
 748
 749         if (pendc)
 750         {
 751                 if (do_append(pendc, NULL, pendpos))
 752                         /*
 753                          * Oops.  We've probably lost the char which
 754                          * was in pendc, since caller won't back up.
 755                          */
 756                         return (1);
 757                 pendc = '\0';
 758         }
 759
 760         if (c == '\r' && bs_mode == BS_SPECIAL)
 761         {
 762                 if (mbc_buf_len > 0)  /* utf_mode must be on. */
 763                 {
 764                         /* Flush incomplete (truncated) sequence. */
 765                         r = flush_mbc_buf(mbc_pos);
 766                         mbc_buf_index = r + 1;
 767                         mbc_buf_len = 0;
 768                         if (r)
 769                                 return (mbc_buf_index);
 770                 }
 771
 772                 /*
 773                  * Don't put the CR into the buffer until we see
 774                  * the next char.  If the next char is a newline,
 775                  * discard the CR.
 776                  */
 777                 pendc = c;
 778                 pendpos = pos;
 779                 return (0);
 780         }
 781
 782         if (!utf_mode)
 783         {
 784                 r = do_append((LWCHAR) c, NULL, pos);
 785         } else
 786         {
 787                 /* Perform strict validation in all possible cases. */
 788                 if (mbc_buf_len == 0)
 789                 {
 790                 retry:
 791                         mbc_buf_index = 1;
 792                         *mbc_buf = c;
 793                         if (IS_ASCII_OCTET(c))
 794                                 r = do_append((LWCHAR) c, NULL, pos);
 795                         else if (IS_UTF8_LEAD(c))
 796                         {
 797                                 mbc_buf_len = utf_len(c);
 798                                 mbc_pos = pos;
 799                                 return (0);
 800                         } else
 801                                 /* UTF8_INVALID or stray UTF8_TRAIL */
 802                                 r = flush_mbc_buf(pos);
 803                 } else if (IS_UTF8_TRAIL(c))
 804                 {
 805                         mbc_buf[mbc_buf_index++] = c;
 806                         if (mbc_buf_index < mbc_buf_len)
 807                                 return (0);
 808                         if (is_utf8_well_formed(mbc_buf))
 809                                 r = do_append(get_wchar(mbc_buf), mbc_buf, mbc_pos);
 810                         else
 811                                 /* Complete, but not shortest form, sequence. */
 812                                 mbc_buf_index = r = flush_mbc_buf(mbc_pos);
 813                         mbc_buf_len = 0;
 814                 } else
 815                 {
 816                         /* Flush incomplete (truncated) sequence.  */
 817                         r = flush_mbc_buf(mbc_pos);
 818                         mbc_buf_index = r + 1;
 819                         mbc_buf_len = 0;
 820                         /* Handle new char.  */
 821                         if (!r)
 822                                 goto retry;
 823                 }
 824         }
 825
 826         /*
 827          * If we need to shift the line, do it.
 828          * But wait until we get to at least the middle of the screen,
 829          * so shifting it doesn't affect the chars we're currently
 830          * pappending.  (Bold & underline can get messed up otherwise.)
 831          */
 832         if (cshift < hshift && column > sc_width / 2)
 833         {
 834                 linebuf[curr] = '\0';
 835                 pshift(hshift - cshift);
 836         }
 837         if (r)
 838         {
 839                 /* How many chars should caller back up? */
 840                 r = (!utf_mode) ? 1 : mbc_buf_index;
 841         }
 842         return (r);
 843 }
 844
 845         static int
 846 do_append(ch, rep, pos)
 847         LWCHAR ch;
 848         char *rep;
 849         POSITION pos;
 850 {
 851         register int a;
 852         LWCHAR prev_ch;
 853
 854         a = AT_NORMAL;
 855
 856         if (ch == '\b')
 857         {
 858                 if (bs_mode == BS_CONTROL)
 859                         goto do_control_char;
 860
 861                 /*
 862                  * A better test is needed here so we don't
 863                  * backspace over part of the printed
 864                  * representation of a binary character.
 865                  */
 866                 if (   curr <= lmargin
 867                     || column <= lmargin
 868                     || (attr[curr - 1] & (AT_ANSI|AT_BINARY)))
 869                         STORE_PRCHAR('\b', pos);
 870                 else if (bs_mode == BS_NORMAL)
 871                         STORE_CHAR(ch, AT_NORMAL, NULL, pos);
 872                 else if (bs_mode == BS_SPECIAL)
 873                         overstrike = backc();
 874
 875                 return 0;
 876         }
 877
 878         if (overstrike > 0)
 879         {
 880                 /*
 881                  * Overstrike the character at the current position
 882                  * in the line buffer.  This will cause either
 883                  * underline (if a "_" is overstruck),
 884                  * bold (if an identical character is overstruck),
 885                  * or just deletion of the character in the buffer.
 886                  */
 887                 overstrike = utf_mode ? -1 : 0;
 888                 /* To be correct, this must be a base character.  */
 889                 prev_ch = get_wchar(linebuf + curr);
 890                 a = attr[curr];
 891                 if (ch == prev_ch)
 892                 {
 893                         /*
 894                          * Overstriking a char with itself means make it bold.
 895                          * But overstriking an underscore with itself is
 896                          * ambiguous.  It could mean make it bold, or
 897                          * it could mean make it underlined.
 898                          * Use the previous overstrike to resolve it.
 899                          */
 900                         if (ch == '_')
 901                         {
 902                                 if ((a & (AT_BOLD|AT_UNDERLINE)) != AT_NORMAL)
 903                                         a |= (AT_BOLD|AT_UNDERLINE);
 904                                 else if (last_overstrike != AT_NORMAL)
 905                                         a |= last_overstrike;
 906                                 else
 907                                         a |= AT_BOLD;
 908                         } else
 909                                 a |= AT_BOLD;
 910                 } else if (ch == '_')
 911                 {
 912                         a |= AT_UNDERLINE;
 913                         ch = prev_ch;
 914                         rep = linebuf + curr;
 915                 } else if (prev_ch == '_')
 916                 {
 917                         a |= AT_UNDERLINE;
 918                 }
 919                 /* Else we replace prev_ch, but we keep its attributes.  */
 920         } else if (overstrike < 0)
 921         {
 922                 if (   is_composing_char(ch)
 923                     || is_combining_char(get_wchar(linebuf + curr), ch))
 924                         /* Continuation of the same overstrike.  */
 925                         a = last_overstrike;
 926                 else
 927                         overstrike = 0;
 928         }
 929
 930         if (ch == '\t')
 931         {
 932                 /*
 933                  * Expand a tab into spaces.
 934                  */
 935                 switch (bs_mode)
 936                 {
 937                 case BS_CONTROL:
 938                         goto do_control_char;
 939                 case BS_NORMAL:
 940                 case BS_SPECIAL:
 941                         STORE_TAB(a, pos);
 942                         break;
 943                 }
 944         } else if ((!utf_mode || is_ascii_char(ch)) && control_char((char)ch))
 945         {
 946         do_control_char:
 947                 if (ctldisp == OPT_ON || (ctldisp == OPT_ONPLUS && IS_CSI_START(ch)))
 948                 {
 949                         /*
 950                          * Output as a normal character.
 951                          */
 952                         STORE_CHAR(ch, AT_NORMAL, rep, pos);
 953                 } else
 954                 {
 955                         STORE_PRCHAR((char) ch, pos);
 956                 }
 957         } else if (utf_mode && ctldisp != OPT_ON && is_ubin_char(ch))
 958         {
 959                 char *s;
 960
 961                 s = prutfchar(ch);
 962
 963                 if (column + (int) strlen(s) - 1 +
 964                     pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
 965                         return (1);
 966
 967                 for ( ;  *s != 0;  s++)
 968                         STORE_CHAR(*s, AT_BINARY, NULL, pos);
 969         } else
 970         {
 971                 STORE_CHAR(ch, a, rep, pos);
 972         }
 973         return (0);
 974 }
 975
 976 /*
 977  *
 978  */
 979         public int
 980 pflushmbc()
 981 {
 982         int r = 0;
 983
 984         if (mbc_buf_len > 0)
 985         {
 986                 /* Flush incomplete (truncated) sequence.  */
 987                 r = flush_mbc_buf(mbc_pos);
 988                 mbc_buf_len = 0;
 989         }
 990         return r;
 991 }
 992
 993 /*
 994  * Terminate the line in the line buffer.
 995  */
 996         public void
 997 pdone(endline)
 998         int endline;
 999 {
1000         int nl;
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 (!oldbot)
1043                 nl = (column < sc_width || !auto_wrap || (endline && ignaw) || ctldisp == OPT_ON);
1044         else
1045                 nl = (column < sc_width || !auto_wrap || ignaw || ctldisp == OPT_ON);
1046         if (nl)
1047         {
1048                 linebuf[curr] = '\n';
1049                 attr[curr] = AT_NORMAL;
1050                 curr++;
1051         }
1052         else if (ignaw && !auto_wrap && column >= sc_width)
1053         {
1054                 /*
1055                  * Big horrible kludge.
1056                  * No-wrap terminals are too hard to deal with when they get in
1057                  * the state where a full screen width of characters have been
1058                  * output but the cursor is sitting on the right edge instead
1059                  * of at the start of the next line.
1060                  * So after we output a full line, we output an extra
1061                  * space and backspace to force the cursor to the
1062                  * beginning of the next line, like a sane terminal.
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 #if HILITE_SEARCH
1073         if (status_col && line_matches > 0)
1074         {
1075                 linebuf[0] = '*';
1076                 attr[0] = AT_NORMAL|AT_HILITE;
1077         }
1078 #endif
1079 }
1080
1081 /*
1082  * Get a character from the current line.
1083  * Return the character as the function return value,
1084  * and the character attribute in *ap.
1085  */
1086         public int
1087 gline(i, ap)
1088         register int i;
1089         register int *ap;
1090 {
1091         if (is_null_line)
1092         {
1093                 /*
1094                  * If there is no current line, we pretend the line is
1095                  * either "~" or "", depending on the "twiddle" flag.
1096                  */
1097                 if (twiddle)
1098                 {
1099                         if (i == 0)
1100                         {
1101                                 *ap = AT_BOLD;
1102                                 return '~';
1103                         }
1104                         --i;
1105                 }
1106                 /* Make sure we're back to AT_NORMAL before the '\n'.  */
1107                 *ap = AT_NORMAL;
1108                 return i ? '\0' : '\n';
1109         }
1110
1111         *ap = attr[i];
1112         return (linebuf[i] & 0xFF);
1113 }
1114
1115 /*
1116  * Indicate that there is no current line.
1117  */
1118         public void
1119 null_line()
1120 {
1121         is_null_line = 1;
1122         cshift = 0;
1123 }
1124
1125 /*
1126  * Analogous to forw_line(), but deals with "raw lines":
1127  * lines which are not split for screen width.
1128  * {{ This is supposed to be more efficient than forw_line(). }}
1129  */
1130         public POSITION
1131 forw_raw_line(curr_pos, linep, line_lenp)
1132         POSITION curr_pos;
1133         char **linep;
1134         int *line_lenp;
1135 {
1136         register int n;
1137         register int c;
1138         POSITION new_pos;
1139
1140         if (curr_pos == NULL_POSITION || ch_seek(curr_pos) ||
1141                 (c = ch_forw_get()) == EOI)
1142                 return (NULL_POSITION);
1143
1144         n = 0;
1145         for (;;)
1146         {
1147                 if (c == '\n' || c == EOI || ABORT_SIGS())
1148                 {
1149                         new_pos = ch_tell();
1150                         break;
1151                 }
1152                 if (n >= size_linebuf-1)
1153                 {
1154                         if (expand_linebuf())
1155                         {
1156                                 /*
1157                                  * Overflowed the input buffer.
1158                                  * Pretend the line ended here.
1159                                  */
1160                                 new_pos = ch_tell() - 1;
1161                                 break;
1162                         }
1163                 }
1164                 linebuf[n++] = c;
1165                 c = ch_forw_get();
1166         }
1167         linebuf[n] = '\0';
1168         if (linep != NULL)
1169                 *linep = linebuf;
1170         if (line_lenp != NULL)
1171                 *line_lenp = n;
1172         return (new_pos);
1173 }
1174
1175 /*
1176  * Analogous to back_line(), but deals with "raw lines".
1177  * {{ This is supposed to be more efficient than back_line(). }}
1178  */
1179         public POSITION
1180 back_raw_line(curr_pos, linep, line_lenp)
1181         POSITION curr_pos;
1182         char **linep;
1183         int *line_lenp;
1184 {
1185         register int n;
1186         register int c;
1187         POSITION new_pos;
1188
1189         if (curr_pos == NULL_POSITION || curr_pos <= ch_zero() ||
1190                 ch_seek(curr_pos-1))
1191                 return (NULL_POSITION);
1192
1193         n = size_linebuf;
1194         linebuf[--n] = '\0';
1195         for (;;)
1196         {
1197                 c = ch_back_get();
1198                 if (c == '\n' || ABORT_SIGS())
1199                 {
1200                         /*
1201                          * This is the newline ending the previous line.
1202                          * We have hit the beginning of the line.
1203                          */
1204                         new_pos = ch_tell() + 1;
1205                         break;
1206                 }
1207                 if (c == EOI)
1208                 {
1209                         /*
1210                          * We have hit the beginning of the file.
1211                          * This must be the first line in the file.
1212                          * This must, of course, be the beginning of the line.
1213                          */
1214                         new_pos = ch_zero();
1215                         break;
1216                 }
1217                 if (n <= 0)
1218                 {
1219                         int old_size_linebuf = size_linebuf;
1220                         char *fm;
1221                         char *to;
1222                         if (expand_linebuf())
1223                         {
1224                                 /*
1225                                  * Overflowed the input buffer.
1226                                  * Pretend the line ended here.
1227                                  */
1228                                 new_pos = ch_tell() + 1;
1229                                 break;
1230                         }
1231                         /*
1232                          * Shift the data to the end of the new linebuf.
1233                          */
1234                         for (fm = linebuf + old_size_linebuf - 1,
1235                               to = linebuf + size_linebuf - 1;
1236                              fm >= linebuf;  fm--, to--)
1237                                 *to = *fm;
1238                         n = size_linebuf - old_size_linebuf;
1239                 }
1240                 linebuf[--n] = c;
1241         }
1242         if (linep != NULL)
1243                 *linep = &linebuf[n];
1244         if (line_lenp != NULL)
1245                 *line_lenp = size_linebuf - 1 - n;
1246         return (new_pos);
1247 }