xref: /freebsd/contrib/less/line.c (revision 25ecdc7d52770caf1c9b44b5ec11f468f6b636f3)
1 /*
2  * Copyright (C) 1984-2021  Mark Nudelman
3  *
4  * You may distribute under the terms of either the GNU General Public
5  * License or the Less License, as specified in the README file.
6  *
7  * For more information, see the README file.
8  */
9 
10 /*
11  * Routines to manipulate the "line buffer".
12  * The line buffer holds a line of output as it is being built
13  * in preparation for output to the screen.
14  */
15 
16 #include "less.h"
17 #include "charset.h"
18 #include "position.h"
19 
20 #if MSDOS_COMPILER==WIN32C
21 #define WIN32_LEAN_AND_MEAN
22 #include <windows.h>
23 #endif
24 
25 #define MAX_PFX_WIDTH (MAX_LINENUM_WIDTH + MAX_STATUSCOL_WIDTH + 1)
26 static struct {
27 	char *buf;    /* Buffer which holds the current output line */
28 	int *attr;   /* Parallel to buf, to hold attributes */
29 	int print;    /* Index in buf of first printable char */
30 	int end;      /* Number of chars in buf */
31 	char pfx[MAX_PFX_WIDTH]; /* Holds status column and line number */
32 	int pfx_attr[MAX_PFX_WIDTH];
33 	int pfx_end;  /* Number of chars in pfx */
34 } linebuf;
35 
36 static struct {
37 	char *buf;
38 	int size;
39 	int end;
40 } shifted_ansi;
41 
42 public int size_linebuf = 0; /* Size of line buffer (and attr buffer) */
43 static struct ansi_state *line_ansi = NULL;
44 static int cshift;   /* Current left-shift of output line buffer */
45 public int hshift;   /* Desired left-shift of output line buffer */
46 public int tabstops[TABSTOP_MAX] = { 0 }; /* Custom tabstops */
47 public int ntabstops = 1;        /* Number of tabstops */
48 public int tabdefault = 8;       /* Default repeated tabstops */
49 public POSITION highest_hilite;  /* Pos of last hilite in file found so far */
50 
51 static int end_column;  /* Printable length, accounting for backspaces, etc. */
52 static int right_curr;
53 static int right_column;
54 static int overstrike;  /* Next char should overstrike previous char */
55 static int last_overstrike = AT_NORMAL;
56 static int is_null_line;  /* There is no current line */
57 static LWCHAR pendc;
58 static POSITION pendpos;
59 static char *end_ansi_chars;
60 static char *mid_ansi_chars;
61 
62 static int attr_swidth LESSPARAMS ((int a));
63 static int attr_ewidth LESSPARAMS ((int a));
64 static int do_append LESSPARAMS ((LWCHAR ch, char *rep, POSITION pos));
65 
66 extern int sigs;
67 extern int bs_mode;
68 extern int linenums;
69 extern int ctldisp;
70 extern int twiddle;
71 extern int binattr;
72 extern int status_col;
73 extern int status_col_width;
74 extern int linenum_width;
75 extern int auto_wrap, ignaw;
76 extern int bo_s_width, bo_e_width;
77 extern int ul_s_width, ul_e_width;
78 extern int bl_s_width, bl_e_width;
79 extern int so_s_width, so_e_width;
80 extern int sc_width, sc_height;
81 extern int utf_mode;
82 extern POSITION start_attnpos;
83 extern POSITION end_attnpos;
84 extern char rscroll_char;
85 extern int rscroll_attr;
86 extern int use_color;
87 
88 static char mbc_buf[MAX_UTF_CHAR_LEN];
89 static int mbc_buf_len = 0;
90 static int mbc_buf_index = 0;
91 static POSITION mbc_pos;
92 
93 /* Configurable color map */
94 static char color_map[AT_NUM_COLORS][12] = {
95 	"Wm",  /* AT_COLOR_ATTN */
96 	"kR",  /* AT_COLOR_BIN */
97 	"kR",  /* AT_COLOR_CTRL */
98 	"kY",  /* AT_COLOR_ERROR */
99 	"c",   /* AT_COLOR_LINENUM */
100 	"Wb",  /* AT_COLOR_MARK */
101 	"kC",  /* AT_COLOR_PROMPT */
102 	"kc",  /* AT_COLOR_RSCROLL */
103 	"kG",  /* AT_COLOR_SEARCH */
104 	"",    /* AT_UNDERLINE */
105 	"",    /* AT_BOLD */
106 	"",    /* AT_BLINK */
107 	"",    /* AT_STANDOUT */
108 };
109 
110 /* State while processing an ANSI escape sequence */
111 struct ansi_state {
112 	int hindex;   /* Index into hyperlink prefix */
113 	int hlink;    /* Processing hyperlink address? */
114 	int prev_esc; /* Prev char was ESC (to detect ESC-\ seq) */
115 };
116 
117 /*
118  * Initialize from environment variables.
119  */
120 	public void
121 init_line(VOID_PARAM)
122 {
123 	end_ansi_chars = lgetenv("LESSANSIENDCHARS");
124 	if (isnullenv(end_ansi_chars))
125 		end_ansi_chars = "m";
126 
127 	mid_ansi_chars = lgetenv("LESSANSIMIDCHARS");
128 	if (isnullenv(mid_ansi_chars))
129 		mid_ansi_chars = "0123456789:;[?!\"'#%()*+ ";
130 
131 	linebuf.buf = (char *) ecalloc(LINEBUF_SIZE, sizeof(char));
132 	linebuf.attr = (int *) ecalloc(LINEBUF_SIZE, sizeof(int));
133 	size_linebuf = LINEBUF_SIZE;
134 	shifted_ansi.buf = NULL;
135 	shifted_ansi.size = 0;
136 }
137 
138 /*
139  * Expand the line buffer.
140  */
141 	static int
142 expand_linebuf(VOID_PARAM)
143 {
144 	/* Double the size of the line buffer. */
145 	int new_size = size_linebuf * 2;
146 
147 	/* Just realloc to expand the buffer, if we can. */
148 #if HAVE_REALLOC
149 	char *new_buf = (char *) realloc(linebuf.buf, new_size);
150 	int *new_attr = (int *) realloc(linebuf.attr, new_size*sizeof(int));
151 #else
152 	char *new_buf = (char *) calloc(new_size, sizeof(char));
153 	int *new_attr = (int *) calloc(new_size, sizeof(int));
154 #endif
155 	if (new_buf == NULL || new_attr == NULL)
156 	{
157 		if (new_attr != NULL)
158 			free(new_attr);
159 		if (new_buf != NULL)
160 			free(new_buf);
161 		return 1;
162 	}
163 #if !HAVE_REALLOC
164 	/*
165 	 * We just calloc'd the buffers; copy the old contents.
166 	 */
167 	memcpy(new_buf, linebuf.buf, size_linebuf * sizeof(char));
168 	memcpy(new_attr, linebuf.attr, size_linebuf * sizeof(int));
169 	free(linebuf.attr);
170 	free(linebuf.buf);
171 #endif
172 	linebuf.buf = new_buf;
173 	linebuf.attr = new_attr;
174 	size_linebuf = new_size;
175 	return 0;
176 }
177 
178 /*
179  * Is a character ASCII?
180  */
181 	public int
182 is_ascii_char(ch)
183 	LWCHAR ch;
184 {
185 	return (ch <= 0x7F);
186 }
187 
188 /*
189  */
190 	static void
191 inc_end_column(w)
192 	int w;
193 {
194 	if (end_column > right_column && w > 0)
195 	{
196 		right_column = end_column;
197 		right_curr = linebuf.end;
198 	}
199 	end_column += w;
200 }
201 
202 /*
203  * Rewind the line buffer.
204  */
205 	public void
206 prewind(VOID_PARAM)
207 {
208 	linebuf.print = 6; /* big enough for longest UTF-8 sequence */
209 	linebuf.pfx_end = 0;
210 	for (linebuf.end = 0; linebuf.end < linebuf.print; linebuf.end++)
211 	{
212 		linebuf.buf[linebuf.end] = '\0';
213 		linebuf.attr[linebuf.end] = 0;
214 	}
215 
216 	end_column = 0;
217 	right_curr = 0;
218 	right_column = 0;
219 	cshift = 0;
220 	overstrike = 0;
221 	last_overstrike = AT_NORMAL;
222 	mbc_buf_len = 0;
223 	is_null_line = 0;
224 	pendc = '\0';
225 	shifted_ansi.end = 0;
226 }
227 
228 /*
229  * Set a character in the line buffer.
230  */
231 	static void
232 set_linebuf(n, ch, attr)
233 	int n;
234 	char ch;
235 	int attr;
236 {
237 	linebuf.buf[n] = ch;
238 	linebuf.attr[n] = attr;
239 }
240 
241 /*
242  * Append a character to the line buffer.
243  */
244 	static void
245 add_linebuf(ch, attr, w)
246 	char ch;
247 	int attr;
248 	int w;
249 {
250 	set_linebuf(linebuf.end++, ch, attr);
251 	inc_end_column(w);
252 }
253 
254 /*
255  * Set a character in the line prefix buffer.
256  */
257 	static void
258 set_pfx(n, ch, attr)
259 	int n;
260 	char ch;
261 	int attr;
262 {
263 	linebuf.pfx[n] = ch;
264 	linebuf.pfx_attr[n] = attr;
265 }
266 
267 /*
268  * Append a character to the line prefix buffer.
269  */
270 	static void
271 add_pfx(ch, attr)
272 	char ch;
273 	int attr;
274 {
275 	set_pfx(linebuf.pfx_end++, ch, attr);
276 }
277 
278 /*
279  * Insert the status column and line number into the line buffer.
280  */
281 	public void
282 plinestart(pos)
283 	POSITION pos;
284 {
285 	LINENUM linenum = 0;
286 	int i;
287 
288 	if (linenums == OPT_ONPLUS)
289 	{
290 		/*
291 		 * Get the line number and put it in the current line.
292 		 * {{ Note: since find_linenum calls forw_raw_line,
293 		 *    it may seek in the input file, requiring the caller
294 		 *    of plinestart to re-seek if necessary. }}
295 		 * {{ Since forw_raw_line modifies linebuf, we must
296 		 *    do this first, before storing anything in linebuf. }}
297 		 */
298 		linenum = find_linenum(pos);
299 	}
300 
301 	/*
302 	 * Display a status column if the -J option is set.
303 	 */
304 	if (status_col)
305 	{
306 		int a = AT_NORMAL;
307 		char c = posmark(pos);
308 		if (c != 0)
309 			a |= AT_HILITE|AT_COLOR_MARK;
310 		else
311 		{
312 			c = ' ';
313 			if (start_attnpos != NULL_POSITION &&
314 			    pos >= start_attnpos && pos <= end_attnpos)
315 				a |= AT_HILITE|AT_COLOR_ATTN;
316 		}
317 		add_pfx(c, a); /* column 0: status */
318 		while (linebuf.pfx_end < status_col_width)
319 			add_pfx(' ', AT_NORMAL);
320 	}
321 
322 	/*
323 	 * Display the line number at the start of each line
324 	 * if the -N option is set.
325 	 */
326 	if (linenums == OPT_ONPLUS)
327 	{
328 		char buf[INT_STRLEN_BOUND(linenum) + 2];
329 		int len;
330 
331 		linenumtoa(linenum, buf);
332 		len = (int) strlen(buf);
333 		for (i = 0; i < linenum_width - len; i++)
334 			add_pfx(' ', AT_NORMAL);
335 		for (i = 0; i < len; i++)
336 			add_pfx(buf[i], AT_NORMAL|AT_COLOR_LINENUM);
337 		add_pfx(' ', AT_NORMAL);
338 	}
339 	end_column = linebuf.pfx_end;
340 }
341 
342 /*
343  * Return the width of the line prefix (status column and line number).
344  * {{ Actual line number can be wider than linenum_width. }}
345  */
346 	public int
347 line_pfx_width(VOID_PARAM)
348 {
349 	int width = 0;
350 	if (status_col)
351 		width += status_col_width;
352 	if (linenums == OPT_ONPLUS)
353 		width += linenum_width + 1;
354 	return width;
355 }
356 
357 /*
358  * Add char to the shifted_ansi buffer.
359  */
360 	static void
361 add_ansi(ch)
362 	char ch;
363 {
364 	if (shifted_ansi.end == shifted_ansi.size)
365 	{
366 		/* Expand shifted_ansi buffer. */
367 		int size = (shifted_ansi.size == 0) ? 8 : shifted_ansi.size * 2;
368 		char *buf = (char *) ecalloc(size, sizeof(char));
369 		memcpy(buf, shifted_ansi.buf, shifted_ansi.size);
370 		if (shifted_ansi.buf != NULL) free(shifted_ansi.buf);
371 		shifted_ansi.buf = buf;
372 		shifted_ansi.size = size;
373 	}
374 	shifted_ansi.buf[shifted_ansi.end++] = ch;
375 }
376 
377 /*
378  * Shift line left so that the last char is just to the left
379  * of the first visible column.
380  */
381 	public void
382 pshift_all(VOID_PARAM)
383 {
384 	int i;
385 	for (i = linebuf.print;  i < linebuf.end;  i++)
386 		if (linebuf.attr[i] == AT_ANSI)
387 			add_ansi(linebuf.buf[i]);
388 	linebuf.end = linebuf.print;
389 	end_column = linebuf.pfx_end;
390 }
391 
392 /*
393  * Return the printing width of the start (enter) sequence
394  * for a given character attribute.
395  */
396 	static int
397 attr_swidth(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_s_width;
406 	if (a & AT_BOLD)
407 		w += bo_s_width;
408 	if (a & AT_BLINK)
409 		w += bl_s_width;
410 	if (a & AT_STANDOUT)
411 		w += so_s_width;
412 
413 	return w;
414 }
415 
416 /*
417  * Return the printing width of the end (exit) sequence
418  * for a given character attribute.
419  */
420 	static int
421 attr_ewidth(a)
422 	int a;
423 {
424 	int w = 0;
425 
426 	a = apply_at_specials(a);
427 
428 	if (a & AT_UNDERLINE)
429 		w += ul_e_width;
430 	if (a & AT_BOLD)
431 		w += bo_e_width;
432 	if (a & AT_BLINK)
433 		w += bl_e_width;
434 	if (a & AT_STANDOUT)
435 		w += so_e_width;
436 
437 	return w;
438 }
439 
440 /*
441  * Return the printing width of a given character and attribute,
442  * if the character were added after prev_ch.
443  * Adding a character with a given attribute may cause an enter or exit
444  * attribute sequence to be inserted, so this must be taken into account.
445  */
446 	public int
447 pwidth(ch, a, prev_ch, prev_a)
448 	LWCHAR ch;
449 	int a;
450 	LWCHAR prev_ch;
451 	int prev_a;
452 {
453 	int w;
454 
455 	if (ch == '\b')
456 	{
457 		/*
458 		 * Backspace moves backwards one or two positions.
459 		 */
460 		if (prev_a & (AT_ANSI|AT_BINARY))
461 			return strlen(prchar('\b'));
462 		return (utf_mode && is_wide_char(prev_ch)) ? -2 : -1;
463 	}
464 
465 	if (!utf_mode || is_ascii_char(ch))
466 	{
467 		if (control_char((char)ch))
468 		{
469 			/*
470 			 * Control characters do unpredictable things,
471 			 * so we don't even try to guess; say it doesn't move.
472 			 * This can only happen if the -r flag is in effect.
473 			 */
474 			return (0);
475 		}
476 	} else
477 	{
478 		if (is_composing_char(ch) || is_combining_char(prev_ch, ch))
479 		{
480 			/*
481 			 * Composing and combining chars take up no space.
482 			 *
483 			 * Some terminals, upon failure to compose a
484 			 * composing character with the character(s) that
485 			 * precede(s) it will actually take up one end_column
486 			 * for the composing character; there isn't much
487 			 * we could do short of testing the (complex)
488 			 * composition process ourselves and printing
489 			 * a binary representation when it fails.
490 			 */
491 			return (0);
492 		}
493 	}
494 
495 	/*
496 	 * Other characters take one or two columns,
497 	 * plus the width of any attribute enter/exit sequence.
498 	 */
499 	w = 1;
500 	if (is_wide_char(ch))
501 		w++;
502 	if (linebuf.end > 0 && !is_at_equiv(linebuf.attr[linebuf.end-1], a))
503 		w += attr_ewidth(linebuf.attr[linebuf.end-1]);
504 	if (apply_at_specials(a) != AT_NORMAL &&
505 	    (linebuf.end == 0 || !is_at_equiv(linebuf.attr[linebuf.end-1], a)))
506 		w += attr_swidth(a);
507 	return (w);
508 }
509 
510 /*
511  * Delete to the previous base character in the line buffer.
512  */
513 	static int
514 backc(VOID_PARAM)
515 {
516 	LWCHAR ch;
517 	char *p;
518 
519 	if (linebuf.end == 0)
520 		return (0);
521 	p = &linebuf.buf[linebuf.end];
522 	ch = step_char(&p, -1, linebuf.buf);
523 	/* Skip back to the next nonzero-width char. */
524 	while (p > linebuf.buf)
525 	{
526 		LWCHAR prev_ch;
527 		int width;
528 		linebuf.end = (int) (p - linebuf.buf);
529 		prev_ch = step_char(&p, -1, linebuf.buf);
530 		width = pwidth(ch, linebuf.attr[linebuf.end], prev_ch, linebuf.attr[linebuf.end-1]);
531 		end_column -= width;
532 		/* {{ right_column? }} */
533 		if (width > 0)
534 			break;
535 		ch = prev_ch;
536 	}
537 	return (1);
538 }
539 
540 /*
541  * Is a character the end of an ANSI escape sequence?
542  */
543 	public int
544 is_ansi_end(ch)
545 	LWCHAR ch;
546 {
547 	if (!is_ascii_char(ch))
548 		return (0);
549 	return (strchr(end_ansi_chars, (char) ch) != NULL);
550 }
551 
552 /*
553  * Can a char appear in an ANSI escape sequence, before the end char?
554  */
555 	public int
556 is_ansi_middle(ch)
557 	LWCHAR ch;
558 {
559 	if (!is_ascii_char(ch))
560 		return (0);
561 	if (is_ansi_end(ch))
562 		return (0);
563 	return (strchr(mid_ansi_chars, (char) ch) != NULL);
564 }
565 
566 /*
567  * Skip past an ANSI escape sequence.
568  * pp is initially positioned just after the CSI_START char.
569  */
570 	public void
571 skip_ansi(pansi, pp, limit)
572 	struct ansi_state *pansi;
573 	char **pp;
574 	constant char *limit;
575 {
576 	LWCHAR c;
577 	do {
578 		c = step_char(pp, +1, limit);
579 	} while (*pp < limit && ansi_step(pansi, c) == ANSI_MID);
580 	/* Note that we discard final char, for which is_ansi_end is true. */
581 }
582 
583 /*
584  * Determine if a character starts an ANSI escape sequence.
585  * If so, return an ansi_state struct; otherwise return NULL.
586  */
587 	public struct ansi_state *
588 ansi_start(ch)
589 	LWCHAR ch;
590 {
591 	struct ansi_state *pansi;
592 
593 	if (!IS_CSI_START(ch))
594 		return NULL;
595 	pansi = ecalloc(1, sizeof(struct ansi_state));
596 	pansi->hindex = 0;
597 	pansi->hlink = 0;
598 	pansi->prev_esc = 0;
599 	return pansi;
600 }
601 
602 /*
603  * Determine whether the next char in an ANSI escape sequence
604  * ends the sequence.
605  */
606 	public int
607 ansi_step(pansi, ch)
608 	struct ansi_state *pansi;
609 	LWCHAR ch;
610 {
611 	if (pansi->hlink)
612 	{
613 		/* Hyperlink ends with \7 or ESC-backslash. */
614 		if (ch == '\7')
615 			return ANSI_END;
616 		if (pansi->prev_esc && ch == '\\')
617 			return ANSI_END;
618 		pansi->prev_esc = (ch == ESC);
619 		return ANSI_MID;
620 	}
621 	if (pansi->hindex >= 0)
622 	{
623 		static char hlink_prefix[] = ESCS "]8;";
624 		if (ch == hlink_prefix[pansi->hindex] ||
625 		    (pansi->hindex == 0 && IS_CSI_START(ch)))
626 		{
627 			pansi->hindex++;
628 			if (hlink_prefix[pansi->hindex] == '\0')
629 				pansi->hlink = 1; /* now processing hyperlink addr */
630 			return ANSI_MID;
631 		}
632 		pansi->hindex = -1; /* not a hyperlink */
633 	}
634 	/* Check for SGR sequences */
635 	if (is_ansi_middle(ch))
636 		return ANSI_MID;
637 	if (is_ansi_end(ch))
638 		return ANSI_END;
639 	return ANSI_ERR;
640 }
641 
642 /*
643  * Free an ansi_state structure.
644  */
645 	public void
646 ansi_done(pansi)
647 	struct ansi_state *pansi;
648 {
649 	free(pansi);
650 }
651 
652 /*
653  * Append a character and attribute to the line buffer.
654  */
655 #define STORE_CHAR(ch,a,rep,pos) \
656 	do { \
657 		if (store_char((ch),(a),(rep),(pos))) return (1); \
658 	} while (0)
659 
660 	static int
661 store_char(ch, a, rep, pos)
662 	LWCHAR ch;
663 	int a;
664 	char *rep;
665 	POSITION pos;
666 {
667 	int w;
668 	int i;
669 	int replen;
670 	char cs;
671 
672 	i = (a & (AT_UNDERLINE|AT_BOLD));
673 	if (i != AT_NORMAL)
674 		last_overstrike = i;
675 
676 #if HILITE_SEARCH
677 	{
678 		int matches;
679 		int hl_attr = is_hilited_attr(pos, pos+1, 0, &matches);
680 		if (hl_attr)
681 		{
682 			/*
683 			 * This character should be highlighted.
684 			 * Override the attribute passed in.
685 			 */
686 			if (a != AT_ANSI)
687 			{
688 				if (highest_hilite != NULL_POSITION && pos > highest_hilite)
689 					highest_hilite = pos;
690 				a |= hl_attr;
691 			}
692 		}
693 	}
694 #endif
695 
696 	if (a == AT_ANSI) {
697 		w = 0;
698 	} else {
699 		char *p = &linebuf.buf[linebuf.end];
700 		LWCHAR prev_ch = (linebuf.end > 0) ? step_char(&p, -1, linebuf.buf) : 0;
701 		int prev_a = (linebuf.end > 0) ? linebuf.attr[linebuf.end-1] : 0;
702 		w = pwidth(ch, a, prev_ch, prev_a);
703 	}
704 
705 	if (ctldisp != OPT_ON && end_column - cshift + w + attr_ewidth(a) > sc_width)
706 		/*
707 		 * Won't fit on screen.
708 		 */
709 		return (1);
710 
711 	if (rep == NULL)
712 	{
713 		cs = (char) ch;
714 		rep = &cs;
715 		replen = 1;
716 	} else
717 	{
718 		replen = utf_len(rep[0]);
719 	}
720 	if (linebuf.end + replen >= size_linebuf-6)
721 	{
722 		/*
723 		 * Won't fit in line buffer.
724 		 * Try to expand it.
725 		 */
726 		if (expand_linebuf())
727 			return (1);
728 	}
729 
730 	if (cshift == hshift && shifted_ansi.end > 0)
731 	{
732 		/* Copy shifted ANSI sequences to beginning of line. */
733 		for (i = 0;  i < shifted_ansi.end;  i++)
734 			add_linebuf(shifted_ansi.buf[i], AT_ANSI, 0);
735 		shifted_ansi.end = 0;
736 	}
737 	/* Add the char to the buf, even if we will left-shift it next. */
738 	inc_end_column(w);
739 	for (i = 0;  i < replen;  i++)
740 		add_linebuf(*rep++, a, 0);
741 
742 	if (cshift < hshift)
743 	{
744 		/* We haven't left-shifted enough yet. */
745 		if (a == AT_ANSI)
746 			add_ansi(ch); /* Save ANSI attributes */
747 		if (linebuf.end > linebuf.print)
748 		{
749 			/* Shift left enough to put last byte of this char at print-1. */
750 			memcpy(&linebuf.buf[0], &linebuf.buf[replen], linebuf.print);
751 			memcpy(&linebuf.attr[0], &linebuf.attr[replen], linebuf.print);
752 			linebuf.end -= replen;
753 			cshift += w;
754 			/*
755 			 * If the char we just left-shifted was double width,
756 			 * the 2 spaces we shifted may be too much.
757 			 * Represent the "half char" at start of line with a highlighted space.
758 			 */
759 			while (cshift > hshift)
760 			{
761 				add_linebuf(' ', rscroll_attr, 0);
762 				cshift--;
763 			}
764 		}
765 	}
766 	return (0);
767 }
768 
769 /*
770  * Append a tab to the line buffer.
771  * Store spaces to represent the tab.
772  */
773 #define STORE_TAB(a,pos) \
774 	do { if (store_tab((a),(pos))) return (1); } while (0)
775 
776 	static int
777 store_tab(attr, pos)
778 	int attr;
779 	POSITION pos;
780 {
781 	int to_tab = end_column - linebuf.pfx_end;
782 
783 	if (ntabstops < 2 || to_tab >= tabstops[ntabstops-1])
784 		to_tab = tabdefault -
785 		     ((to_tab - tabstops[ntabstops-1]) % tabdefault);
786 	else
787 	{
788 		int i;
789 		for (i = ntabstops - 2;  i >= 0;  i--)
790 			if (to_tab >= tabstops[i])
791 				break;
792 		to_tab = tabstops[i+1] - to_tab;
793 	}
794 
795 	do {
796 		STORE_CHAR(' ', attr, " ", pos);
797 	} while (--to_tab > 0);
798 	return 0;
799 }
800 
801 #define STORE_PRCHAR(c, pos) \
802 	do { if (store_prchar((c), (pos))) return 1; } while (0)
803 
804 	static int
805 store_prchar(c, pos)
806 	LWCHAR c;
807 	POSITION pos;
808 {
809 	char *s;
810 
811 	/*
812 	 * Convert to printable representation.
813 	 */
814 	s = prchar(c);
815 	for ( ;  *s != 0;  s++)
816 		STORE_CHAR(*s, AT_BINARY|AT_COLOR_CTRL, NULL, pos);
817 	return 0;
818 }
819 
820 	static int
821 flush_mbc_buf(pos)
822 	POSITION pos;
823 {
824 	int i;
825 
826 	for (i = 0; i < mbc_buf_index; i++)
827 		if (store_prchar(mbc_buf[i], pos))
828 			return mbc_buf_index - i;
829 	return 0;
830 }
831 
832 /*
833  * Append a character to the line buffer.
834  * Expand tabs into spaces, handle underlining, boldfacing, etc.
835  * Returns 0 if ok, 1 if couldn't fit in buffer.
836  */
837 	public int
838 pappend(c, pos)
839 	int c;
840 	POSITION pos;
841 {
842 	int r;
843 
844 	if (pendc)
845 	{
846 		if (c == '\r' && pendc == '\r')
847 			return (0);
848 		if (do_append(pendc, NULL, pendpos))
849 			/*
850 			 * Oops.  We've probably lost the char which
851 			 * was in pendc, since caller won't back up.
852 			 */
853 			return (1);
854 		pendc = '\0';
855 	}
856 
857 	if (c == '\r' && bs_mode == BS_SPECIAL)
858 	{
859 		if (mbc_buf_len > 0)  /* utf_mode must be on. */
860 		{
861 			/* Flush incomplete (truncated) sequence. */
862 			r = flush_mbc_buf(mbc_pos);
863 			mbc_buf_index = r + 1;
864 			mbc_buf_len = 0;
865 			if (r)
866 				return (mbc_buf_index);
867 		}
868 
869 		/*
870 		 * Don't put the CR into the buffer until we see
871 		 * the next char.  If the next char is a newline,
872 		 * discard the CR.
873 		 */
874 		pendc = c;
875 		pendpos = pos;
876 		return (0);
877 	}
878 
879 	if (!utf_mode)
880 	{
881 		r = do_append(c, NULL, pos);
882 	} else
883 	{
884 		/* Perform strict validation in all possible cases. */
885 		if (mbc_buf_len == 0)
886 		{
887 		retry:
888 			mbc_buf_index = 1;
889 			*mbc_buf = c;
890 			if (IS_ASCII_OCTET(c))
891 				r = do_append(c, NULL, pos);
892 			else if (IS_UTF8_LEAD(c))
893 			{
894 				mbc_buf_len = utf_len(c);
895 				mbc_pos = pos;
896 				return (0);
897 			} else
898 				/* UTF8_INVALID or stray UTF8_TRAIL */
899 				r = flush_mbc_buf(pos);
900 		} else if (IS_UTF8_TRAIL(c))
901 		{
902 			mbc_buf[mbc_buf_index++] = c;
903 			if (mbc_buf_index < mbc_buf_len)
904 				return (0);
905 			if (is_utf8_well_formed(mbc_buf, mbc_buf_index))
906 				r = do_append(get_wchar(mbc_buf), mbc_buf, mbc_pos);
907 			else
908 				/* Complete, but not shortest form, sequence. */
909 				mbc_buf_index = r = flush_mbc_buf(mbc_pos);
910 			mbc_buf_len = 0;
911 		} else
912 		{
913 			/* Flush incomplete (truncated) sequence.  */
914 			r = flush_mbc_buf(mbc_pos);
915 			mbc_buf_index = r + 1;
916 			mbc_buf_len = 0;
917 			/* Handle new char.  */
918 			if (!r)
919 				goto retry;
920 		}
921 	}
922 	if (r)
923 	{
924 		/* How many chars should caller back up? */
925 		r = (!utf_mode) ? 1 : mbc_buf_index;
926 	}
927 	return (r);
928 }
929 
930 	static int
931 store_control_char(ch, rep, pos)
932 	LWCHAR ch;
933 	char *rep;
934 	POSITION pos;
935 {
936 	if (ctldisp == OPT_ON)
937 	{
938 		/* Output the character itself. */
939 		STORE_CHAR(ch, AT_NORMAL, rep, pos);
940 	} else
941 	{
942 		/* Output a printable representation of the character. */
943 		STORE_PRCHAR((char) ch, pos);
944 	}
945 	return (0);
946 }
947 
948 	static int
949 store_ansi(ch, rep, pos)
950 	LWCHAR ch;
951 	char *rep;
952 	POSITION pos;
953 {
954 	switch (ansi_step(line_ansi, ch))
955 	{
956 	case ANSI_MID:
957 		STORE_CHAR(ch, AT_ANSI, rep, pos);
958 		break;
959 	case ANSI_END:
960 		STORE_CHAR(ch, AT_ANSI, rep, pos);
961 		ansi_done(line_ansi);
962 		line_ansi = NULL;
963 		break;
964 	case ANSI_ERR: {
965 		/* Remove whole unrecognized sequence.  */
966 		char *start = (cshift < hshift) ? shifted_ansi.buf : linebuf.buf;
967 		int *end = (cshift < hshift) ? &shifted_ansi.end : &linebuf.end;
968 		char *p = start + *end;
969 		LWCHAR bch;
970 		do {
971 			bch = step_char(&p, -1, start);
972 		} while (p > start && !IS_CSI_START(bch));
973 		*end = (int) (p - start);
974 		ansi_done(line_ansi);
975 		line_ansi = NULL;
976 		break; }
977 	}
978 	return (0);
979 }
980 
981 	static int
982 store_bs(ch, rep, pos)
983 	LWCHAR ch;
984 	char *rep;
985 	POSITION pos;
986 {
987 	if (bs_mode == BS_CONTROL)
988 		return store_control_char(ch, rep, pos);
989 	if (linebuf.end > 0 &&
990 		((linebuf.end <= linebuf.print && linebuf.buf[linebuf.end-1] == '\0') ||
991 	     (linebuf.end > 0 && linebuf.attr[linebuf.end - 1] & (AT_ANSI|AT_BINARY))))
992 		STORE_PRCHAR('\b', pos);
993 	else if (bs_mode == BS_NORMAL)
994 		STORE_CHAR(ch, AT_NORMAL, NULL, pos);
995 	else if (bs_mode == BS_SPECIAL)
996 		overstrike = backc();
997 	return 0;
998 }
999 
1000 	static int
1001 do_append(ch, rep, pos)
1002 	LWCHAR ch;
1003 	char *rep;
1004 	POSITION pos;
1005 {
1006 	int a = AT_NORMAL;
1007 
1008 	if (ctldisp == OPT_ONPLUS && line_ansi == NULL)
1009 		line_ansi = ansi_start(ch);
1010 
1011 	if (line_ansi != NULL)
1012 		return store_ansi(ch, rep, pos);
1013 
1014 	if (ch == '\b')
1015 		return store_bs(ch, rep, pos);
1016 
1017 	if (overstrike > 0)
1018 	{
1019 		/*
1020 		 * Overstrike the character at the current position
1021 		 * in the line buffer.  This will cause either
1022 		 * underline (if a "_" is overstruck),
1023 		 * bold (if an identical character is overstruck),
1024 		 * or just replacing the character in the buffer.
1025 		 */
1026 		LWCHAR prev_ch;
1027 		overstrike = utf_mode ? -1 : 0;
1028 		if (utf_mode)
1029 		{
1030 			/* To be correct, this must be a base character.  */
1031 			prev_ch = get_wchar(&linebuf.buf[linebuf.end]);
1032 		} else
1033 		{
1034 			prev_ch = (unsigned char) linebuf.buf[linebuf.end];
1035 		}
1036 		a = linebuf.attr[linebuf.end];
1037 		if (ch == prev_ch)
1038 		{
1039 			/*
1040 			 * Overstriking a char with itself means make it bold.
1041 			 * But overstriking an underscore with itself is
1042 			 * ambiguous.  It could mean make it bold, or
1043 			 * it could mean make it underlined.
1044 			 * Use the previous overstrike to resolve it.
1045 			 */
1046 			if (ch == '_')
1047 			{
1048 				if ((a & (AT_BOLD|AT_UNDERLINE)) != AT_NORMAL)
1049 					a |= (AT_BOLD|AT_UNDERLINE);
1050 				else if (last_overstrike != AT_NORMAL)
1051 					a |= last_overstrike;
1052 				else
1053 					a |= AT_BOLD;
1054 			} else
1055 				a |= AT_BOLD;
1056 		} else if (ch == '_')
1057 		{
1058 			a |= AT_UNDERLINE;
1059 			ch = prev_ch;
1060 			rep = &linebuf.buf[linebuf.end];
1061 		} else if (prev_ch == '_')
1062 		{
1063 			a |= AT_UNDERLINE;
1064 		}
1065 		/* Else we replace prev_ch, but we keep its attributes.  */
1066 	} else if (overstrike < 0)
1067 	{
1068 		if (   is_composing_char(ch)
1069 		    || is_combining_char(get_wchar(&linebuf.buf[linebuf.end]), ch))
1070 			/* Continuation of the same overstrike.  */
1071 			a = last_overstrike;
1072 		else
1073 			overstrike = 0;
1074 	}
1075 
1076 	if (ch == '\t')
1077 	{
1078 		/*
1079 		 * Expand a tab into spaces.
1080 		 */
1081 		switch (bs_mode)
1082 		{
1083 		case BS_CONTROL:
1084 			return store_control_char(ch, rep, pos);
1085 		case BS_NORMAL:
1086 		case BS_SPECIAL:
1087 			STORE_TAB(a, pos);
1088 			break;
1089 		}
1090 		return (0);
1091 	}
1092 	if ((!utf_mode || is_ascii_char(ch)) && control_char((char)ch))
1093 	{
1094 		return store_control_char(ch, rep, pos);
1095 	} else if (utf_mode && ctldisp != OPT_ON && is_ubin_char(ch))
1096 	{
1097 		char *s = prutfchar(ch);
1098 		for ( ;  *s != 0;  s++)
1099 			STORE_CHAR(*s, AT_BINARY, NULL, pos);
1100 	} else
1101 	{
1102 		STORE_CHAR(ch, a, rep, pos);
1103 	}
1104 	return (0);
1105 }
1106 
1107 /*
1108  *
1109  */
1110 	public int
1111 pflushmbc(VOID_PARAM)
1112 {
1113 	int r = 0;
1114 
1115 	if (mbc_buf_len > 0)
1116 	{
1117 		/* Flush incomplete (truncated) sequence.  */
1118 		r = flush_mbc_buf(mbc_pos);
1119 		mbc_buf_len = 0;
1120 	}
1121 	return r;
1122 }
1123 
1124 /*
1125  * Switch to normal attribute at end of line.
1126  */
1127 	static void
1128 add_attr_normal(VOID_PARAM)
1129 {
1130 	char *p = "\033[m";
1131 
1132 	if (ctldisp != OPT_ONPLUS || !is_ansi_end('m'))
1133 		return;
1134 	for ( ;  *p != '\0';  p++)
1135 		add_linebuf(*p, AT_ANSI, 0);
1136 }
1137 
1138 /*
1139  * Terminate the line in the line buffer.
1140  */
1141 	public void
1142 pdone(endline, chopped, forw)
1143 	int endline;
1144 	int chopped;
1145 	int forw;
1146 {
1147 	(void) pflushmbc();
1148 
1149 	if (pendc && (pendc != '\r' || !endline))
1150 		/*
1151 		 * If we had a pending character, put it in the buffer.
1152 		 * But discard a pending CR if we are at end of line
1153 		 * (that is, discard the CR in a CR/LF sequence).
1154 		 */
1155 		(void) do_append(pendc, NULL, pendpos);
1156 
1157 	if (chopped && rscroll_char)
1158 	{
1159 		/*
1160 		 * Display the right scrolling char.
1161 		 * If we've already filled the rightmost screen char
1162 		 * (in the buffer), overwrite it.
1163 		 */
1164 		if (end_column >= sc_width + cshift)
1165 		{
1166 			/* We've already written in the rightmost char. */
1167 			end_column = right_column;
1168 			linebuf.end = right_curr;
1169 		}
1170 		add_attr_normal();
1171 		while (end_column < sc_width-1 + cshift)
1172 		{
1173 			/*
1174 			 * Space to last (rightmost) char on screen.
1175 			 * This may be necessary if the char we overwrote
1176 			 * was double-width.
1177 			 */
1178 			add_linebuf(' ', rscroll_attr, 1);
1179 		}
1180 		/* Print rscroll char. It must be single-width. */
1181 		add_linebuf(rscroll_char, rscroll_attr, 1);
1182 	} else
1183 	{
1184 		add_attr_normal();
1185 	}
1186 
1187 	/*
1188 	 * Add a newline if necessary,
1189 	 * and append a '\0' to the end of the line.
1190 	 * We output a newline if we're not at the right edge of the screen,
1191 	 * or if the terminal doesn't auto wrap,
1192 	 * or if this is really the end of the line AND the terminal ignores
1193 	 * a newline at the right edge.
1194 	 * (In the last case we don't want to output a newline if the terminal
1195 	 * doesn't ignore it since that would produce an extra blank line.
1196 	 * But we do want to output a newline if the terminal ignores it in case
1197 	 * the next line is blank.  In that case the single newline output for
1198 	 * that blank line would be ignored!)
1199 	 */
1200 	if (end_column < sc_width + cshift || !auto_wrap || (endline && ignaw) || ctldisp == OPT_ON)
1201 	{
1202 		add_linebuf('\n', AT_NORMAL, 0);
1203 	}
1204 	else if (ignaw && end_column >= sc_width + cshift && forw)
1205 	{
1206 		/*
1207 		 * Terminals with "ignaw" don't wrap until they *really* need
1208 		 * to, i.e. when the character *after* the last one to fit on a
1209 		 * line is output. But they are too hard to deal with when they
1210 		 * get in the state where a full screen width of characters
1211 		 * have been output but the cursor is sitting on the right edge
1212 		 * instead of at the start of the next line.
1213 		 * So we nudge them into wrapping by outputting a space
1214 		 * character plus a backspace.  But do this only if moving
1215 		 * forward; if we're moving backward and drawing this line at
1216 		 * the top of the screen, the space would overwrite the first
1217 		 * char on the next line.  We don't need to do this "nudge"
1218 		 * at the top of the screen anyway.
1219 		 */
1220 		add_linebuf(' ', AT_NORMAL, 1);
1221 		add_linebuf('\b', AT_NORMAL, -1);
1222 	}
1223 	set_linebuf(linebuf.end, '\0', AT_NORMAL);
1224 }
1225 
1226 /*
1227  *
1228  */
1229 	public void
1230 set_status_col(c, attr)
1231 	int c;
1232 	int attr;
1233 {
1234 	set_pfx(0, c, attr);
1235 }
1236 
1237 /*
1238  * Get a character from the current line.
1239  * Return the character as the function return value,
1240  * and the character attribute in *ap.
1241  */
1242 	public int
1243 gline(i, ap)
1244 	int i;
1245 	int *ap;
1246 {
1247 	if (is_null_line)
1248 	{
1249 		/*
1250 		 * If there is no current line, we pretend the line is
1251 		 * either "~" or "", depending on the "twiddle" flag.
1252 		 */
1253 		if (twiddle)
1254 		{
1255 			if (i == 0)
1256 			{
1257 				*ap = AT_BOLD;
1258 				return '~';
1259 			}
1260 			--i;
1261 		}
1262 		/* Make sure we're back to AT_NORMAL before the '\n'.  */
1263 		*ap = AT_NORMAL;
1264 		return i ? '\0' : '\n';
1265 	}
1266 
1267 	if (i < linebuf.pfx_end)
1268 	{
1269 		*ap = linebuf.pfx_attr[i];
1270 		return linebuf.pfx[i];
1271 	}
1272 	i += linebuf.print - linebuf.pfx_end;
1273 	*ap = linebuf.attr[i];
1274 	return (linebuf.buf[i] & 0xFF);
1275 }
1276 
1277 /*
1278  * Indicate that there is no current line.
1279  */
1280 	public void
1281 null_line(VOID_PARAM)
1282 {
1283 	is_null_line = 1;
1284 	cshift = 0;
1285 }
1286 
1287 /*
1288  * Analogous to forw_line(), but deals with "raw lines":
1289  * lines which are not split for screen width.
1290  * {{ This is supposed to be more efficient than forw_line(). }}
1291  */
1292 	public POSITION
1293 forw_raw_line(curr_pos, linep, line_lenp)
1294 	POSITION curr_pos;
1295 	char **linep;
1296 	int *line_lenp;
1297 {
1298 	int n;
1299 	int c;
1300 	POSITION new_pos;
1301 
1302 	if (curr_pos == NULL_POSITION || ch_seek(curr_pos) ||
1303 		(c = ch_forw_get()) == EOI)
1304 		return (NULL_POSITION);
1305 
1306 	n = 0;
1307 	for (;;)
1308 	{
1309 		if (c == '\n' || c == EOI || ABORT_SIGS())
1310 		{
1311 			new_pos = ch_tell();
1312 			break;
1313 		}
1314 		if (n >= size_linebuf-1)
1315 		{
1316 			if (expand_linebuf())
1317 			{
1318 				/*
1319 				 * Overflowed the input buffer.
1320 				 * Pretend the line ended here.
1321 				 */
1322 				new_pos = ch_tell() - 1;
1323 				break;
1324 			}
1325 		}
1326 		linebuf.buf[n++] = c;
1327 		c = ch_forw_get();
1328 	}
1329 	linebuf.buf[n] = '\0';
1330 	if (linep != NULL)
1331 		*linep = linebuf.buf;
1332 	if (line_lenp != NULL)
1333 		*line_lenp = n;
1334 	return (new_pos);
1335 }
1336 
1337 /*
1338  * Analogous to back_line(), but deals with "raw lines".
1339  * {{ This is supposed to be more efficient than back_line(). }}
1340  */
1341 	public POSITION
1342 back_raw_line(curr_pos, linep, line_lenp)
1343 	POSITION curr_pos;
1344 	char **linep;
1345 	int *line_lenp;
1346 {
1347 	int n;
1348 	int c;
1349 	POSITION new_pos;
1350 
1351 	if (curr_pos == NULL_POSITION || curr_pos <= ch_zero() ||
1352 		ch_seek(curr_pos-1))
1353 		return (NULL_POSITION);
1354 
1355 	n = size_linebuf;
1356 	linebuf.buf[--n] = '\0';
1357 	for (;;)
1358 	{
1359 		c = ch_back_get();
1360 		if (c == '\n' || ABORT_SIGS())
1361 		{
1362 			/*
1363 			 * This is the newline ending the previous line.
1364 			 * We have hit the beginning of the line.
1365 			 */
1366 			new_pos = ch_tell() + 1;
1367 			break;
1368 		}
1369 		if (c == EOI)
1370 		{
1371 			/*
1372 			 * We have hit the beginning of the file.
1373 			 * This must be the first line in the file.
1374 			 * This must, of course, be the beginning of the line.
1375 			 */
1376 			new_pos = ch_zero();
1377 			break;
1378 		}
1379 		if (n <= 0)
1380 		{
1381 			int old_size_linebuf = size_linebuf;
1382 			char *fm;
1383 			char *to;
1384 			if (expand_linebuf())
1385 			{
1386 				/*
1387 				 * Overflowed the input buffer.
1388 				 * Pretend the line ended here.
1389 				 */
1390 				new_pos = ch_tell() + 1;
1391 				break;
1392 			}
1393 			/*
1394 			 * Shift the data to the end of the new linebuf.
1395 			 */
1396 			for (fm = linebuf.buf + old_size_linebuf - 1,
1397 			      to = linebuf.buf + size_linebuf - 1;
1398 			     fm >= linebuf.buf;  fm--, to--)
1399 				*to = *fm;
1400 			n = size_linebuf - old_size_linebuf;
1401 		}
1402 		linebuf.buf[--n] = c;
1403 	}
1404 	if (linep != NULL)
1405 		*linep = &linebuf.buf[n];
1406 	if (line_lenp != NULL)
1407 		*line_lenp = size_linebuf - 1 - n;
1408 	return (new_pos);
1409 }
1410 
1411 /*
1412  * Find the shift necessary to show the end of the longest displayed line.
1413  */
1414 	public int
1415 rrshift(VOID_PARAM)
1416 {
1417 	POSITION pos;
1418 	int save_width;
1419 	int line;
1420 	int longest = 0;
1421 
1422 	save_width = sc_width;
1423 	sc_width = INT_MAX;
1424 	pos = position(TOP);
1425 	for (line = 0; line < sc_height && pos != NULL_POSITION; line++)
1426 	{
1427 		pos = forw_line(pos);
1428 		if (end_column > longest)
1429 			longest = end_column;
1430 	}
1431 	sc_width = save_width;
1432 	if (longest < sc_width)
1433 		return 0;
1434 	return longest - sc_width;
1435 }
1436 
1437 /*
1438  * Get the color_map index associated with a given attribute.
1439  */
1440 	static int
1441 color_index(attr)
1442 	int attr;
1443 {
1444 	if (use_color)
1445 	{
1446 		switch (attr & AT_COLOR)
1447 		{
1448 		case AT_COLOR_ATTN:    return 0;
1449 		case AT_COLOR_BIN:     return 1;
1450 		case AT_COLOR_CTRL:    return 2;
1451 		case AT_COLOR_ERROR:   return 3;
1452 		case AT_COLOR_LINENUM: return 4;
1453 		case AT_COLOR_MARK:    return 5;
1454 		case AT_COLOR_PROMPT:  return 6;
1455 		case AT_COLOR_RSCROLL: return 7;
1456 		case AT_COLOR_SEARCH:  return 8;
1457 		}
1458 	}
1459 	if (attr & AT_UNDERLINE)
1460 		return 9;
1461 	if (attr & AT_BOLD)
1462 		return 10;
1463 	if (attr & AT_BLINK)
1464 		return 11;
1465 	if (attr & AT_STANDOUT)
1466 		return 12;
1467 	return -1;
1468 }
1469 
1470 /*
1471  * Set the color string to use for a given attribute.
1472  */
1473 	public int
1474 set_color_map(attr, colorstr)
1475 	int attr;
1476 	char *colorstr;
1477 {
1478 	int cx = color_index(attr);
1479 	if (cx < 0)
1480 		return -1;
1481 	if (strlen(colorstr)+1 > sizeof(color_map[cx]))
1482 		return -1;
1483 	if (*colorstr != '\0' && parse_color(colorstr, NULL, NULL) == CT_NULL)
1484 		return -1;
1485 	strcpy(color_map[cx], colorstr);
1486 	return 0;
1487 }
1488 
1489 /*
1490  * Get the color string to use for a given attribute.
1491  */
1492 	public char *
1493 get_color_map(attr)
1494 	int attr;
1495 {
1496 	int cx = color_index(attr);
1497 	if (cx < 0)
1498 		return NULL;
1499 	return color_map[cx];
1500 }
1501