contrib/less/line.c

   1 /* $FreeBSD$ */
   2 /*
   3  * Copyright (C) 1984-2007  Mark Nudelman
   4  *
   5  * You may distribute under the terms of either the GNU General Public
   6  * License or the Less License, as specified in the README file.
   7  *
   8  * For more information about less, or for information on how to
   9  * contact the author, see the README file.
  10  */
  11
  12
  13 /*
  14  * Routines to manipulate the "line buffer".
  15  * The line buffer holds a line of output as it is being built
  16  * in preparation for output to the screen.
  17  */
  18
  19 #include "less.h"
  20 #include "charset.h"
  21
  22 static char *linebuf = NULL;    /* Buffer which holds the current output line */
  23 static char *attr = NULL;       /* Extension of linebuf to hold attributes */
  24 public int size_linebuf = 0;    /* Size of line buffer (and attr buffer) */
  25
  26 static int cshift;              /* Current left-shift of output line buffer */
  27 public int hshift;              /* Desired left-shift of output line buffer */
  28 public int tabstops[TABSTOP_MAX] = { 0 }; /* Custom tabstops */
  29 public int ntabstops = 1;       /* Number of tabstops */
  30 public int tabdefault = 8;      /* Default repeated tabstops */
  31
  32 static int curr;                /* Index into linebuf */
  33 static int column;              /* Printable length, accounting for
  34                                    backspaces, etc. */
  35 static int overstrike;          /* Next char should overstrike previous char */
  36 static int last_overstrike = AT_NORMAL;
  37 static int is_null_line;        /* There is no current line */
  38 static int lmargin;             /* Left margin */
  39 static int line_matches;        /* Number of search matches in this line */
  40 static char pendc;
  41 static POSITION pendpos;
  42 static char *end_ansi_chars;
  43 static char *mid_ansi_chars;
  44
  45 static int attr_swidth();
  46 static int attr_ewidth();
  47 static int do_append();
  48
  49 extern int sigs;
  50 extern int bs_mode;
  51 extern int linenums;
  52 extern int ctldisp;
  53 extern int twiddle;
  54 extern int binattr;
  55 extern int status_col;
  56 extern int auto_wrap, ignaw;
  57 extern int bo_s_width, bo_e_width;
  58 extern int ul_s_width, ul_e_width;
  59 extern int bl_s_width, bl_e_width;
  60 extern int so_s_width, so_e_width;
  61 extern int sc_width, sc_height;
  62 extern int utf_mode;
  63 extern int oldbot;
  64 extern POSITION start_attnpos;
  65 extern POSITION end_attnpos;
  66
  67 static char mbc_buf[MAX_UTF_CHAR_LEN];
  68 static int mbc_buf_len = 0;
  69 static int mbc_buf_index = 0;
  70 static POSITION mbc_pos;
  71
  72 /*
  73  * Initialize from environment variables.
  74  */
  75         public void
  76 init_line()
  77 {
  78         end_ansi_chars = lgetenv("LESSANSIENDCHARS");
  79         if (end_ansi_chars == NULL || *end_ansi_chars == '\0')
  80                 end_ansi_chars = "m";
  81
  82         mid_ansi_chars = lgetenv("LESSANSIMIDCHARS");
  83         if (mid_ansi_chars == NULL || *mid_ansi_chars == '\0')
  84                 mid_ansi_chars = "0123456789;[?!\"'#%()*+ ";
  85
  86         linebuf = (char *) ecalloc(LINEBUF_SIZE, sizeof(char));
  87         attr = (char *) ecalloc(LINEBUF_SIZE, sizeof(char));
  88         size_linebuf = LINEBUF_SIZE;
  89 }
  90
  91 /*
  92  * Expand the line buffer.
  93  */
  94         static int
  95 expand_linebuf()
  96 {
  97         /* Double the size of the line buffer. */
  98         int new_size = size_linebuf * 2;
  99
 100         /* Just realloc to expand the buffer, if we can. */
 101 #if HAVE_REALLOC
 102         char *new_buf = (char *) realloc(linebuf, new_size);
 103         char *new_attr = (char *) realloc(attr, new_size);
 104 #else
 105         char *new_buf = (char *) calloc(new_size, sizeof(char));
 106         char *new_attr = (char *) calloc(new_size, sizeof(char));
 107 #endif
 108         if (new_buf == NULL || new_attr == NULL)
 109         {
 110                 if (new_attr != NULL)
 111                         free(new_attr);
 112                 if (new_buf != NULL)
 113                         free(new_buf);
 114                 return 1;
 115         }
 116 #if HAVE_REALLOC
 117         /*
 118          * We realloc'd the buffers; they already have the old contents.
 119          */
 120         #if 0
 121         memset(new_buf + size_linebuf, 0, new_size - size_linebuf);
 122         memset(new_attr + size_linebuf, 0, new_size - size_linebuf);
 123         #endif
 124 #else
 125         /*
 126          * We just calloc'd the buffers; copy the old contents.
 127          */
 128         memcpy(new_buf, linebuf, size_linebuf * sizeof(char));
 129         memcpy(new_attr, attr, size_linebuf * sizeof(char));
 130         free(attr);
 131         free(linebuf);
 132 #endif
 133         linebuf = new_buf;
 134         attr = new_attr;
 135         size_linebuf = new_size;
 136         return 0;
 137 }
 138
 139 /*
 140  * Is a character ASCII?
 141  */
 142         public int
 143 is_ascii_char(ch)
 144         LWCHAR ch;
 145 {
 146         return (ch <= 0x7F);
 147 }
 148
 149 /*
 150  * Rewind the line buffer.
 151  */
 152         public void
 153 prewind()
 154 {
 155         curr = 0;
 156         column = 0;
 157         cshift = 0;
 158         overstrike = 0;
 159         last_overstrike = AT_NORMAL;
 160         mbc_buf_len = 0;
 161         is_null_line = 0;
 162         pendc = '\0';
 163         lmargin = 0;
 164         if (status_col)
 165                 lmargin += 1;
 166 #if HILITE_SEARCH
 167         line_matches = 0;
 168 #endif
 169 }
 170
 171 /*
 172  * Insert the line number (of the given position) into the line buffer.
 173  */
 174         public void
 175 plinenum(pos)
 176         POSITION pos;
 177 {
 178         register LINENUM linenum = 0;
 179         register int i;
 180
 181         if (linenums == OPT_ONPLUS)
 182         {
 183                 /*
 184                  * Get the line number and put it in the current line.
 185                  * {{ Note: since find_linenum calls forw_raw_line,
 186                  *    it may seek in the input file, requiring the caller
 187                  *    of plinenum to re-seek if necessary. }}
 188                  * {{ Since forw_raw_line modifies linebuf, we must
 189                  *    do this first, before storing anything in linebuf. }}
 190                  */
 191                 linenum = find_linenum(pos);
 192         }
 193
 194         /*
 195          * Display a status column if the -J option is set.
 196          */
 197         if (status_col)
 198         {
 199                 linebuf[curr] = ' ';
 200                 if (start_attnpos != NULL_POSITION &&
 201                     pos >= start_attnpos && pos < end_attnpos)
 202                         attr[curr] = AT_NORMAL|AT_HILITE;
 203                 else
 204                         attr[curr] = AT_NORMAL;
 205                 curr++;
 206                 column++;
 207         }
 208         /*
 209          * Display the line number at the start of each line
 210          * if the -N option is set.
 211          */
 212         if (linenums == OPT_ONPLUS)
 213         {
 214                 char buf[INT_STRLEN_BOUND(pos) + 2];
 215                 int n;
 216
 217                 linenumtoa(linenum, buf);
 218                 n = strlen(buf);
 219                 if (n < MIN_LINENUM_WIDTH)
 220                         n = MIN_LINENUM_WIDTH;
 221                 sprintf(linebuf+curr, "%*s ", n, buf);
 222                 n++;  /* One space after the line number. */
 223                 for (i = 0; i < n; i++)
 224                         attr[curr+i] = AT_NORMAL;
 225                 curr += n;
 226                 column += n;
 227                 lmargin += n;
 228         }
 229
 230         /*
 231          * Append enough spaces to bring us to the lmargin.
 232          */
 233         while (column < lmargin)
 234         {
 235                 linebuf[curr] = ' ';
 236                 attr[curr++] = AT_NORMAL;
 237                 column++;
 238         }
 239 }
 240
 241 /*
 242  * Shift the input line left.
 243  * This means discarding N printable chars at the start of the buffer.
 244  */
 245         static void
 246 pshift(shift)
 247         int shift;
 248 {
 249         LWCHAR prev_ch = 0;
 250         unsigned char c;
 251         int shifted = 0;
 252         int to;
 253         int from;
 254         int len;
 255         int width;
 256         int prev_attr;
 257         int next_attr;
 258
 259         if (shift > column - lmargin)
 260                 shift = column - lmargin;
 261         if (shift > curr - lmargin)
 262                 shift = curr - lmargin;
 263
 264         to = from = lmargin;
 265         /*
 266          * We keep on going when shifted == shift
 267          * to get all combining chars.
 268          */
 269         while (shifted <= shift && from < curr)
 270         {
 271                 c = linebuf[from];
 272                 if (ctldisp == OPT_ONPLUS && IS_CSI_START(c))
 273                 {
 274                         /* Keep cumulative effect.  */
 275                         linebuf[to] = c;
 276                         attr[to++] = attr[from++];
 277                         while (from < curr && linebuf[from])
 278                         {
 279                                 linebuf[to] = linebuf[from];
 280                                 attr[to++] = attr[from];
 281                                 if (!is_ansi_middle(linebuf[from++]))
 282                                         break;
 283                         }
 284                         continue;
 285                 }
 286
 287                 width = 0;
 288
 289                 if (!IS_ASCII_OCTET(c) && utf_mode)
 290                 {
 291                         /* Assumes well-formedness validation already done.  */
 292                         LWCHAR ch;
 293
 294                         len = utf_len(c);
 295                         if (from + len > curr)
 296                                 break;
 297                         ch = get_wchar(linebuf + from);
 298                         if (!is_composing_char(ch) && !is_combining_char(prev_ch, ch))
 299                                 width = is_wide_char(ch) ? 2 : 1;
 300                         prev_ch = ch;
 301                 } else
 302                 {
 303                         len = 1;
 304                         if (c == '\b')
 305                                 /* XXX - Incorrect if several '\b' in a row.  */
 306                                 width = (utf_mode && is_wide_char(prev_ch)) ? -2 : -1;
 307                         else if (!control_char(c))
 308                                 width = 1;
 309                         prev_ch = 0;
 310                 }
 311
 312                 if (width == 2 && shift - shifted == 1) {
 313                         /* Should never happen when called by pshift_all().  */
 314                         attr[to] = attr[from];
 315                         /*
 316                          * Assume a wide_char will never be the first half of a
 317                          * combining_char pair, so reset prev_ch in case we're
 318                          * followed by a '\b'.
 319                          */
 320                         prev_ch = linebuf[to++] = ' ';
 321                         from += len;
 322                         shifted++;
 323                         continue;
 324                 }
 325
 326                 /* Adjust width for magic cookies. */
 327                 prev_attr = (to > 0) ? attr[to-1] : AT_NORMAL;
 328                 next_attr = (from + len < curr) ? attr[from + len] : prev_attr;
 329                 if (!is_at_equiv(attr[from], prev_attr) &&
 330                         !is_at_equiv(attr[from], next_attr))
 331                 {
 332                         width += attr_swidth(attr[from]);
 333                         if (from + len < curr)
 334                                 width += attr_ewidth(attr[from]);
 335                         if (is_at_equiv(prev_attr, next_attr))
 336                         {
 337                                 width += attr_ewidth(prev_attr);
 338                                 if (from + len < curr)
 339                                         width += attr_swidth(next_attr);
 340                         }
 341                 }
 342
 343                 if (shift - shifted < width)
 344                         break;
 345                 from += len;
 346                 shifted += width;
 347                 if (shifted < 0)
 348                         shifted = 0;
 349         }
 350         while (from < curr)
 351         {
 352                 linebuf[to] = linebuf[from];
 353                 attr[to++] = attr[from++];
 354         }
 355         curr = to;
 356         column -= shifted;
 357         cshift += shifted;
 358 }
 359
 360 /*
 361  *
 362  */
 363         public void
 364 pshift_all()
 365 {
 366         pshift(column);
 367 }
 368
 369 /*
 370  * Return the printing width of the start (enter) sequence
 371  * for a given character attribute.
 372  */
 373         static int
 374 attr_swidth(a)
 375         int a;
 376 {
 377         int w = 0;
 378
 379         a = apply_at_specials(a);
 380
 381         if (a & AT_UNDERLINE)
 382                 w += ul_s_width;
 383         if (a & AT_BOLD)
 384                 w += bo_s_width;
 385         if (a & AT_BLINK)
 386                 w += bl_s_width;
 387         if (a & AT_STANDOUT)
 388                 w += so_s_width;
 389
 390         return w;
 391 }
 392
 393 /*
 394  * Return the printing width of the end (exit) sequence
 395  * for a given character attribute.
 396  */
 397         static int
 398 attr_ewidth(a)
 399         int a;
 400 {
 401         int w = 0;
 402
 403         a = apply_at_specials(a);
 404
 405         if (a & AT_UNDERLINE)
 406                 w += ul_e_width;
 407         if (a & AT_BOLD)
 408                 w += bo_e_width;
 409         if (a & AT_BLINK)
 410                 w += bl_e_width;
 411         if (a & AT_STANDOUT)
 412                 w += so_e_width;
 413
 414         return w;
 415 }
 416
 417 /*
 418  * Return the printing width of a given character and attribute,
 419  * if the character were added to the current position in the line buffer.
 420  * Adding a character with a given attribute may cause an enter or exit
 421  * attribute sequence to be inserted, so this must be taken into account.
 422  */
 423         static int
 424 pwidth(ch, a, prev_ch)
 425         LWCHAR ch;
 426         int a;
 427         LWCHAR prev_ch;
 428 {
 429         int w;
 430
 431         if (ch == '\b')
 432                 /*
 433                  * Backspace moves backwards one or two positions.
 434                  * XXX - Incorrect if several '\b' in a row.
 435                  */
 436                 return (utf_mode && is_wide_char(prev_ch)) ? -2 : -1;
 437
 438         if (!utf_mode || is_ascii_char(ch))
 439         {
 440                 if (control_char((char)ch))
 441                 {
 442                         /*
 443                          * Control characters do unpredictable things,
 444                          * so we don't even try to guess; say it doesn't move.
 445                          * This can only happen if the -r flag is in effect.
 446                          */
 447                         return (0);
 448                 }
 449         } else
 450         {
 451                 if (is_composing_char(ch) || is_combining_char(prev_ch, ch))
 452                 {
 453                         /*
 454                          * Composing and combining chars take up no space.
 455                          *
 456                          * Some terminals, upon failure to compose a
 457                          * composing character with the character(s) that
 458                          * precede(s) it will actually take up one column
 459                          * for the composing character; there isn't much
 460                          * we could do short of testing the (complex)
 461                          * composition process ourselves and printing
 462                          * a binary representation when it fails.
 463                          */
 464                         return (0);
 465                 }
 466         }
 467
 468         /*
 469          * Other characters take one or two columns,
 470          * plus the width of any attribute enter/exit sequence.
 471          */
 472         w = 1;
 473         if (is_wide_char(ch))
 474                 w++;
 475         if (curr > 0 && !is_at_equiv(attr[curr-1], a))
 476                 w += attr_ewidth(attr[curr-1]);
 477         if ((apply_at_specials(a) != AT_NORMAL) &&
 478             (curr == 0 || !is_at_equiv(attr[curr-1], a)))
 479                 w += attr_swidth(a);
 480         return (w);
 481 }
 482
 483 /*
 484  * Delete to the previous base character in the line buffer.
 485  * Return 1 if one is found.
 486  */
 487         static int
 488 backc()
 489 {
 490         LWCHAR prev_ch;
 491         char *p = linebuf + curr;
 492         LWCHAR ch = step_char(&p, -1, linebuf + lmargin);
 493         int width;
 494
 495         /* This assumes that there is no '\b' in linebuf.  */
 496         while (   curr > lmargin
 497                && column > lmargin
 498                && (!(attr[curr - 1] & (AT_ANSI|AT_BINARY))))
 499         {
 500                 curr = p - linebuf;
 501                 prev_ch = step_char(&p, -1, linebuf + lmargin);
 502                 width = pwidth(ch, attr[curr], prev_ch);
 503                 column -= width;
 504                 if (width > 0)
 505                         return 1;
 506                 ch = prev_ch;
 507         }
 508
 509         return 0;
 510 }
 511
 512 /*
 513  * Are we currently within a recognized ANSI escape sequence?
 514  */
 515         static int
 516 in_ansi_esc_seq()
 517 {
 518         char *p;
 519
 520         /*
 521          * Search backwards for either an ESC (which means we ARE in a seq);
 522          * or an end char (which means we're NOT in a seq).
 523          */
 524         for (p = &linebuf[curr];  p > linebuf; )
 525         {
 526                 LWCHAR ch = step_char(&p, -1, linebuf);
 527                 if (IS_CSI_START(ch))
 528                         return (1);
 529                 if (!is_ansi_middle(ch))
 530                         return (0);
 531         }
 532         return (0);
 533 }
 534
 535 /*
 536  * Is a character the end of an ANSI escape sequence?
 537  */
 538         public int
 539 is_ansi_end(ch)
 540         LWCHAR ch;
 541 {
 542         if (!is_ascii_char(ch))
 543                 return (0);
 544         return (strchr(end_ansi_chars, (char) ch) != NULL);
 545 }
 546
 547 /*
 548  *
 549  */
 550         public int
 551 is_ansi_middle(ch)
 552         LWCHAR ch;
 553 {
 554         if (!is_ascii_char(ch))
 555                 return (0);
 556         if (is_ansi_end(ch))
 557                 return (0);
 558         return (strchr(mid_ansi_chars, (char) ch) != NULL);
 559 }
 560
 561 /*
 562  * Append a character and attribute to the line buffer.
 563  */
 564 #define STORE_CHAR(ch,a,rep,pos) \
 565         do { \
 566                 if (store_char((ch),(a),(rep),(pos))) return (1); \
 567         } while (0)
 568
 569         static int
 570 store_char(ch, a, rep, pos)
 571         LWCHAR ch;
 572         int a;
 573         char *rep;
 574         POSITION pos;
 575 {
 576         int w;
 577         int replen;
 578         char cs;
 579
 580         w = (a & (AT_UNDERLINE|AT_BOLD));       /* Pre-use w.  */
 581         if (w != AT_NORMAL)
 582                 last_overstrike = w;
 583
 584 #if HILITE_SEARCH
 585         {
 586                 int matches;
 587                 if (is_hilited(pos, pos+1, 0, &matches))
 588                 {
 589                         /*
 590                          * This character should be highlighted.
 591                          * Override the attribute passed in.
 592                          */
 593                         if (a != AT_ANSI)
 594                                 a |= AT_HILITE;
 595                 }
 596                 line_matches += matches;
 597         }
 598 #endif
 599
 600         if (ctldisp == OPT_ONPLUS && in_ansi_esc_seq())
 601         {
 602                 if (!is_ansi_end(ch) && !is_ansi_middle(ch)) {
 603                         /* Remove whole unrecognized sequence.  */
 604                         do {
 605                                 --curr;
 606                         } while (!IS_CSI_START(linebuf[curr]));
 607                         return 0;
 608                 }
 609                 a = AT_ANSI;    /* Will force re-AT_'ing around it.  */
 610                 w = 0;
 611         }
 612         else if (ctldisp == OPT_ONPLUS && IS_CSI_START(ch))
 613         {
 614                 a = AT_ANSI;    /* Will force re-AT_'ing around it.  */
 615                 w = 0;
 616         }
 617         else
 618         {
 619                 char *p = &linebuf[curr];
 620                 LWCHAR prev_ch = step_char(&p, -1, linebuf);
 621                 w = pwidth(ch, a, prev_ch);
 622         }
 623
 624         if (ctldisp != OPT_ON && column + w + attr_ewidth(a) > sc_width)
 625                 /*
 626                  * Won't fit on screen.
 627                  */
 628                 return (1);
 629
 630         if (rep == NULL)
 631         {
 632                 cs = (char) ch;
 633                 rep = &cs;
 634                 replen = 1;
 635         } else
 636         {
 637                 replen = utf_len(rep[0]);
 638         }
 639         if (curr + replen >= size_linebuf-6)
 640         {
 641                 /*
 642                  * Won't fit in line buffer.
 643                  * Try to expand it.
 644                  */
 645                 if (expand_linebuf())
 646                         return (1);
 647         }
 648
 649         while (replen-- > 0)
 650         {
 651                 linebuf[curr] = *rep++;
 652                 attr[curr] = a;
 653                 curr++;
 654         }
 655         column += w;
 656         return (0);
 657 }
 658
 659 /*
 660  * Append a tab to the line buffer.
 661  * Store spaces to represent the tab.
 662  */
 663 #define STORE_TAB(a,pos) \
 664         do { if (store_tab((a),(pos))) return (1); } while (0)
 665
 666         static int
 667 store_tab(attr, pos)
 668         int attr;
 669         POSITION pos;
 670 {
 671         int to_tab = column + cshift - lmargin;
 672         int i;
 673
 674         if (ntabstops < 2 || to_tab >= tabstops[ntabstops-1])
 675                 to_tab = tabdefault -
 676                      ((to_tab - tabstops[ntabstops-1]) % tabdefault);
 677         else
 678         {
 679                 for (i = ntabstops - 2;  i >= 0;  i--)
 680                         if (to_tab >= tabstops[i])
 681                                 break;
 682                 to_tab = tabstops[i+1] - to_tab;
 683         }
 684
 685         if (column + to_tab - 1 + pwidth(' ', attr, 0) + attr_ewidth(attr) > sc_width)
 686                 return 1;
 687
 688         do {
 689                 STORE_CHAR(' ', attr, " ", pos);
 690         } while (--to_tab > 0);
 691         return 0;
 692 }
 693
 694 #define STORE_PRCHAR(c, pos) \
 695         do { if (store_prchar((c), (pos))) return 1; } while (0)
 696
 697         static int
 698 store_prchar(c, pos)
 699         char c;
 700         POSITION pos;
 701 {
 702         char *s;
 703
 704         /*
 705          * Convert to printable representation.
 706          */
 707         s = prchar(c);
 708
 709         /*
 710          * Make sure we can get the entire representation
 711          * of the character on this line.
 712          */
 713         if (column + (int) strlen(s) - 1 +
 714             pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
 715                 return 1;
 716
 717         for ( ;  *s != 0;  s++)
 718                 STORE_CHAR(*s, AT_BINARY, NULL, pos);
 719
 720         return 0;
 721 }
 722
 723         static int
 724 flush_mbc_buf(pos)
 725         POSITION pos;
 726 {
 727         int i;
 728
 729         for (i = 0; i < mbc_buf_index; i++)
 730                 if (store_prchar(mbc_buf[i], pos))
 731                         return mbc_buf_index - i;
 732
 733         return 0;
 734 }
 735
 736 /*
 737  * Append a character to the line buffer.
 738  * Expand tabs into spaces, handle underlining, boldfacing, etc.
 739  * Returns 0 if ok, 1 if couldn't fit in buffer.
 740  */
 741         public int
 742 pappend(c, pos)
 743         char c;
 744         POSITION pos;
 745 {
 746         int r;
 747
 748         if (pendc)
 749         {
 750                 if (do_append(pendc, NULL, pendpos))
 751                         /*
 752                          * Oops.  We've probably lost the char which
 753                          * was in pendc, since caller won't back up.
 754                          */
 755                         return (1);
 756                 pendc = '\0';
 757         }
 758
 759         if (c == '\r' && bs_mode == BS_SPECIAL)
 760         {
 761                 if (mbc_buf_len > 0)  /* utf_mode must be on. */
 762                 {
 763                         /* Flush incomplete (truncated) sequence. */
 764                         r = flush_mbc_buf(mbc_pos);
 765                         mbc_buf_index = r + 1;
 766                         mbc_buf_len = 0;
 767                         if (r)
 768                                 return (mbc_buf_index);
 769                 }
 770
 771                 /*
 772                  * Don't put the CR into the buffer until we see
 773                  * the next char.  If the next char is a newline,
 774                  * discard the CR.
 775                  */
 776                 pendc = c;
 777                 pendpos = pos;
 778                 return (0);
 779         }
 780
 781         if (!utf_mode)
 782         {
 783                 r = do_append((LWCHAR) c, NULL, pos);
 784         } else
 785         {
 786                 /* Perform strict validation in all possible cases. */
 787                 if (mbc_buf_len == 0)
 788                 {
 789                 retry:
 790                         mbc_buf_index = 1;
 791                         *mbc_buf = c;
 792                         if (IS_ASCII_OCTET(c))
 793                                 r = do_append((LWCHAR) c, NULL, pos);
 794                         else if (IS_UTF8_LEAD(c))
 795                         {
 796                                 mbc_buf_len = utf_len(c);
 797                                 mbc_pos = pos;
 798                                 return (0);
 799                         } else
 800                                 /* UTF8_INVALID or stray UTF8_TRAIL */
 801                                 r = flush_mbc_buf(pos);
 802                 } else if (IS_UTF8_TRAIL(c))
 803                 {
 804                         mbc_buf[mbc_buf_index++] = c;
 805                         if (mbc_buf_index < mbc_buf_len)
 806                                 return (0);
 807                         if (is_utf8_well_formed(mbc_buf))
 808                                 r = do_append(get_wchar(mbc_buf), mbc_buf, mbc_pos);
 809                         else
 810                                 /* Complete, but not shortest form, sequence. */
 811                                 mbc_buf_index = r = flush_mbc_buf(mbc_pos);
 812                         mbc_buf_len = 0;
 813                 } else
 814                 {
 815                         /* Flush incomplete (truncated) sequence.  */
 816                         r = flush_mbc_buf(mbc_pos);
 817                         mbc_buf_index = r + 1;
 818                         mbc_buf_len = 0;
 819                         /* Handle new char.  */
 820                         if (!r)
 821                                 goto retry;
 822                 }
 823         }
 824
 825         /*
 826          * If we need to shift the line, do it.
 827          * But wait until we get to at least the middle of the screen,
 828          * so shifting it doesn't affect the chars we're currently
 829          * pappending.  (Bold & underline can get messed up otherwise.)
 830          */
 831         if (cshift < hshift && column > sc_width / 2)
 832         {
 833                 linebuf[curr] = '\0';
 834                 pshift(hshift - cshift);
 835         }
 836         if (r)
 837         {
 838                 /* How many chars should caller back up? */
 839                 r = (!utf_mode) ? 1 : mbc_buf_index;
 840         }
 841         return (r);
 842 }
 843
 844         static int
 845 do_append(ch, rep, pos)
 846         LWCHAR ch;
 847         char *rep;
 848         POSITION pos;
 849 {
 850         register int a;
 851         LWCHAR prev_ch;
 852
 853         a = AT_NORMAL;
 854
 855         if (ch == '\b')
 856         {
 857                 if (bs_mode == BS_CONTROL)
 858                         goto do_control_char;
 859
 860                 /*
 861                  * A better test is needed here so we don't
 862                  * backspace over part of the printed
 863                  * representation of a binary character.
 864                  */
 865                 if (   curr <= lmargin
 866                     || column <= lmargin
 867                     || (attr[curr - 1] & (AT_ANSI|AT_BINARY)))
 868                         STORE_PRCHAR('\b', pos);
 869                 else if (bs_mode == BS_NORMAL)
 870                         STORE_CHAR(ch, AT_NORMAL, NULL, pos);
 871                 else if (bs_mode == BS_SPECIAL)
 872                         overstrike = backc();
 873
 874                 return 0;
 875         }
 876
 877         if (overstrike > 0)
 878         {
 879                 /*
 880                  * Overstrike the character at the current position
 881                  * in the line buffer.  This will cause either
 882                  * underline (if a "_" is overstruck),
 883                  * bold (if an identical character is overstruck),
 884                  * or just deletion of the character in the buffer.
 885                  */
 886                 overstrike = utf_mode ? -1 : 0;
 887                 /* To be correct, this must be a base character.  */
 888                 prev_ch = get_wchar(linebuf + curr);
 889                 a = attr[curr];
 890                 if (ch == prev_ch)
 891                 {
 892                         /*
 893                          * Overstriking a char with itself means make it bold.
 894                          * But overstriking an underscore with itself is
 895                          * ambiguous.  It could mean make it bold, or
 896                          * it could mean make it underlined.
 897                          * Use the previous overstrike to resolve it.
 898                          */
 899                         if (ch == '_')
 900                         {
 901                                 if ((a & (AT_BOLD|AT_UNDERLINE)) != AT_NORMAL)
 902                                         a |= (AT_BOLD|AT_UNDERLINE);
 903                                 else if (last_overstrike != AT_NORMAL)
 904                                         a |= last_overstrike;
 905                                 else
 906                                         a |= AT_BOLD;
 907                         } else
 908                                 a |= AT_BOLD;
 909                 } else if (ch == '_')
 910                 {
 911                         a |= AT_UNDERLINE;
 912                         ch = prev_ch;
 913                         rep = linebuf + curr;
 914                 } else if (prev_ch == '_')
 915                 {
 916                         a |= AT_UNDERLINE;
 917                 }
 918                 /* Else we replace prev_ch, but we keep its attributes.  */
 919         } else if (overstrike < 0)
 920         {
 921                 if (   is_composing_char(ch)
 922                     || is_combining_char(get_wchar(linebuf + curr), ch))
 923                         /* Continuation of the same overstrike.  */
 924                         a = last_overstrike;
 925                 else
 926                         overstrike = 0;
 927         }
 928
 929         if (ch == '\t')
 930         {
 931                 /*
 932                  * Expand a tab into spaces.
 933                  */
 934                 switch (bs_mode)
 935                 {
 936                 case BS_CONTROL:
 937                         goto do_control_char;
 938                 case BS_NORMAL:
 939                 case BS_SPECIAL:
 940                         STORE_TAB(a, pos);
 941                         break;
 942                 }
 943         } else if ((!utf_mode || is_ascii_char(ch)) && control_char((char)ch))
 944         {
 945         do_control_char:
 946                 if (ctldisp == OPT_ON || (ctldisp == OPT_ONPLUS && IS_CSI_START(ch)))
 947                 {
 948                         /*
 949                          * Output as a normal character.
 950                          */
 951                         STORE_CHAR(ch, AT_NORMAL, rep, pos);
 952                 } else
 953                 {
 954                         STORE_PRCHAR((char) ch, pos);
 955                 }
 956         } else if (utf_mode && ctldisp != OPT_ON && is_ubin_char(ch))
 957         {
 958                 char *s;
 959
 960                 s = prutfchar(ch);
 961
 962                 if (column + (int) strlen(s) - 1 +
 963                     pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
 964                         return (1);
 965
 966                 for ( ;  *s != 0;  s++)
 967                         STORE_CHAR(*s, AT_BINARY, NULL, pos);
 968         } else
 969         {
 970                 STORE_CHAR(ch, a, rep, pos);
 971         }
 972         return (0);
 973 }
 974
 975 /*
 976  *
 977  */
 978         public int
 979 pflushmbc()
 980 {
 981         int r = 0;
 982
 983         if (mbc_buf_len > 0)
 984         {
 985                 /* Flush incomplete (truncated) sequence.  */
 986                 r = flush_mbc_buf(mbc_pos);
 987                 mbc_buf_len = 0;
 988         }
 989         return r;
 990 }
 991
 992 /*
 993  * Terminate the line in the line buffer.
 994  */
 995         public void
 996 pdone(endline)
 997         int endline;
 998 {
 999         int nl;
1000
1001         (void) pflushmbc();
1002
1003         if (pendc && (pendc != '\r' || !endline))
1004                 /*
1005                  * If we had a pending character, put it in the buffer.
1006                  * But discard a pending CR if we are at end of line
1007                  * (that is, discard the CR in a CR/LF sequence).
1008                  */
1009                 (void) do_append(pendc, NULL, pendpos);
1010
1011         /*
1012          * Make sure we've shifted the line, if we need to.
1013          */
1014         if (cshift < hshift)
1015                 pshift(hshift - cshift);
1016
1017         if (ctldisp == OPT_ONPLUS && is_ansi_end('m'))
1018         {
1019                 /* Switch to normal attribute at end of line. */
1020                 char *p = "\033[m";
1021                 for ( ;  *p != '\0';  p++)
1022                 {
1023                         linebuf[curr] = *p;
1024                         attr[curr++] = AT_ANSI;
1025                 }
1026         }
1027
1028         /*
1029          * Add a newline if necessary,
1030          * and append a '\0' to the end of the line.
1031          * We output a newline if we're not at the right edge of the screen,
1032          * or if the terminal doesn't auto wrap,
1033          * or if this is really the end of the line AND the terminal ignores
1034          * a newline at the right edge.
1035          * (In the last case we don't want to output a newline if the terminal
1036          * doesn't ignore it since that would produce an extra blank line.
1037          * But we do want to output a newline if the terminal ignores it in case
1038          * the next line is blank.  In that case the single newline output for
1039          * that blank line would be ignored!)
1040          */
1041         if (!oldbot)
1042                 nl = (column < sc_width || !auto_wrap || (endline && ignaw) || ctldisp == OPT_ON);
1043         else
1044                 nl = (column < sc_width || !auto_wrap || ignaw || ctldisp == OPT_ON);
1045         if (nl)
1046         {
1047                 linebuf[curr] = '\n';
1048                 attr[curr] = AT_NORMAL;
1049                 curr++;
1050         }
1051         else if (ignaw && !auto_wrap && column >= sc_width)
1052         {
1053                 /*
1054                  * Big horrible kludge.
1055                  * No-wrap terminals are too hard to deal with when they get in
1056                  * the state where a full screen width of characters have been
1057                  * output but the cursor is sitting on the right edge instead
1058                  * of at the start of the next line.
1059                  * So after we output a full line, we output an extra
1060                  * space and backspace to force the cursor to the
1061                  * beginning of the next line, like a sane terminal.
1062                  */
1063                 linebuf[curr] = ' ';
1064                 attr[curr++] = AT_NORMAL;
1065                 linebuf[curr] = '\b';
1066                 attr[curr++] = AT_NORMAL;
1067         }
1068         linebuf[curr] = '\0';
1069         attr[curr] = AT_NORMAL;
1070
1071 #if HILITE_SEARCH
1072         if (status_col && line_matches > 0)
1073         {
1074                 linebuf[0] = '*';
1075                 attr[0] = AT_NORMAL|AT_HILITE;
1076         }
1077 #endif
1078 }
1079
1080 /*
1081  * Get a character from the current line.
1082  * Return the character as the function return value,
1083  * and the character attribute in *ap.
1084  */
1085         public int
1086 gline(i, ap)
1087         register int i;
1088         register int *ap;
1089 {
1090         if (is_null_line)
1091         {
1092                 /*
1093                  * If there is no current line, we pretend the line is
1094                  * either "~" or "", depending on the "twiddle" flag.
1095                  */
1096                 if (twiddle)
1097                 {
1098                         if (i == 0)
1099                         {
1100                                 *ap = AT_BOLD;
1101                                 return '~';
1102                         }
1103                         --i;
1104                 }
1105                 /* Make sure we're back to AT_NORMAL before the '\n'.  */
1106                 *ap = AT_NORMAL;
1107                 return i ? '\0' : '\n';
1108         }
1109
1110         *ap = attr[i];
1111         return (linebuf[i] & 0xFF);
1112 }
1113
1114 /*
1115  * Indicate that there is no current line.
1116  */
1117         public void
1118 null_line()
1119 {
1120         is_null_line = 1;
1121         cshift = 0;
1122 }
1123
1124 /*
1125  * Analogous to forw_line(), but deals with "raw lines":
1126  * lines which are not split for screen width.
1127  * {{ This is supposed to be more efficient than forw_line(). }}
1128  */
1129         public POSITION
1130 forw_raw_line(curr_pos, linep, line_lenp)
1131         POSITION curr_pos;
1132         char **linep;
1133         int *line_lenp;
1134 {
1135         register int n;
1136         register int c;
1137         POSITION new_pos;
1138
1139         if (curr_pos == NULL_POSITION || ch_seek(curr_pos) ||
1140                 (c = ch_forw_get()) == EOI)
1141                 return (NULL_POSITION);
1142
1143         n = 0;
1144         for (;;)
1145         {
1146                 if (c == '\n' || c == EOI || ABORT_SIGS())
1147                 {
1148                         new_pos = ch_tell();
1149                         break;
1150                 }
1151                 if (n >= size_linebuf-1)
1152                 {
1153                         if (expand_linebuf())
1154                         {
1155                                 /*
1156                                  * Overflowed the input buffer.
1157                                  * Pretend the line ended here.
1158                                  */
1159                                 new_pos = ch_tell() - 1;
1160                                 break;
1161                         }
1162                 }
1163                 linebuf[n++] = c;
1164                 c = ch_forw_get();
1165         }
1166         linebuf[n] = '\0';
1167         if (linep != NULL)
1168                 *linep = linebuf;
1169         if (line_lenp != NULL)
1170                 *line_lenp = n;
1171         return (new_pos);
1172 }
1173
1174 /*
1175  * Analogous to back_line(), but deals with "raw lines".
1176  * {{ This is supposed to be more efficient than back_line(). }}
1177  */
1178         public POSITION
1179 back_raw_line(curr_pos, linep, line_lenp)
1180         POSITION curr_pos;
1181         char **linep;
1182         int *line_lenp;
1183 {
1184         register int n;
1185         register int c;
1186         POSITION new_pos;
1187
1188         if (curr_pos == NULL_POSITION || curr_pos <= ch_zero() ||
1189                 ch_seek(curr_pos-1))
1190                 return (NULL_POSITION);
1191
1192         n = size_linebuf;
1193         linebuf[--n] = '\0';
1194         for (;;)
1195         {
1196                 c = ch_back_get();
1197                 if (c == '\n' || ABORT_SIGS())
1198                 {
1199                         /*
1200                          * This is the newline ending the previous line.
1201                          * We have hit the beginning of the line.
1202                          */
1203                         new_pos = ch_tell() + 1;
1204                         break;
1205                 }
1206                 if (c == EOI)
1207                 {
1208                         /*
1209                          * We have hit the beginning of the file.
1210                          * This must be the first line in the file.
1211                          * This must, of course, be the beginning of the line.
1212                          */
1213                         new_pos = ch_zero();
1214                         break;
1215                 }
1216                 if (n <= 0)
1217                 {
1218                         int old_size_linebuf = size_linebuf;
1219                         char *fm;
1220                         char *to;
1221                         if (expand_linebuf())
1222                         {
1223                                 /*
1224                                  * Overflowed the input buffer.
1225                                  * Pretend the line ended here.
1226                                  */
1227                                 new_pos = ch_tell() + 1;
1228                                 break;
1229                         }
1230                         /*
1231                          * Shift the data to the end of the new linebuf.
1232                          */
1233                         for (fm = linebuf + old_size_linebuf - 1,
1234                               to = linebuf + size_linebuf - 1;
1235                              fm >= linebuf;  fm--, to--)
1236                                 *to = *fm;
1237                         n = size_linebuf - old_size_linebuf;
1238                 }
1239                 linebuf[--n] = c;
1240         }
1241         if (linep != NULL)
1242                 *linep = &linebuf[n];
1243         if (line_lenp != NULL)
1244                 *line_lenp = size_linebuf - 1 - n;
1245         return (new_pos);
1246 }