xref: /openbsd-src/usr.bin/less/line.c (revision 4c1e55dc91edd6e69ccc60ce855900fbc12cf34f)
1 /*
2  * Copyright (C) 1984-2011  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 #include <err.h>
22 
23 static char *linebuf = NULL;	/* Buffer which holds the current output line */
24 static char *attr = NULL;	/* Extension of linebuf to hold attributes */
25 public int size_linebuf = 0;	/* Size of line buffer (and attr buffer) */
26 
27 static int cshift;		/* Current left-shift of output line buffer */
28 public int hshift;		/* Desired left-shift of output line buffer */
29 public int tabstops[TABSTOP_MAX] = { 0 }; /* Custom tabstops */
30 public int ntabstops = 1;	/* Number of tabstops */
31 public int tabdefault = 8;	/* Default repeated tabstops */
32 
33 static int curr;		/* Index into linebuf */
34 static int column;		/* Printable length, accounting for
35 				   backspaces, etc. */
36 static int overstrike;		/* Next char should overstrike previous char */
37 static int last_overstrike = AT_NORMAL;
38 static int is_null_line;	/* There is no current line */
39 static int lmargin;		/* Left margin */
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 volatile sig_atomic_t 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 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 	char *new_buf;
101 	char *new_attr;
102 
103 	new_buf = realloc(linebuf, new_size);
104 	if (new_buf == NULL)
105 		return 1;
106 	new_attr = realloc(attr, new_size);
107 	if (new_attr == NULL) {
108 		/* realloc linebuf back to original size */
109 		linebuf = realloc(new_buf, size_linebuf);
110 		if (linebuf == NULL)
111 			err(1, NULL);
112 		return 1;
113 	}
114 	/*
115 	 * We realloc'd the buffers; they already have the old contents.
116 	 */
117 	memset(new_buf + size_linebuf, 0, new_size - size_linebuf);
118 	memset(new_attr + size_linebuf, 0, new_size - size_linebuf);
119 	linebuf = new_buf;
120 	attr = new_attr;
121 	size_linebuf = new_size;
122 	return 0;
123 }
124 
125 /*
126  * Is a character ASCII?
127  */
128 	public int
129 is_ascii_char(ch)
130 	LWCHAR ch;
131 {
132 	return (ch <= 0x7F);
133 }
134 
135 /*
136  * Rewind the line buffer.
137  */
138 	public void
139 prewind()
140 {
141 	curr = 0;
142 	column = 0;
143 	cshift = 0;
144 	overstrike = 0;
145 	last_overstrike = AT_NORMAL;
146 	mbc_buf_len = 0;
147 	is_null_line = 0;
148 	pendc = '\0';
149 	lmargin = 0;
150 	if (status_col)
151 		lmargin += 1;
152 }
153 
154 /*
155  * Insert the line number (of the given position) into the line buffer.
156  */
157 	public void
158 plinenum(pos)
159 	POSITION pos;
160 {
161 	register LINENUM linenum = 0;
162 	register int i;
163 
164 	if (linenums == OPT_ONPLUS)
165 	{
166 		/*
167 		 * Get the line number and put it in the current line.
168 		 * {{ Note: since find_linenum calls forw_raw_line,
169 		 *    it may seek in the input file, requiring the caller
170 		 *    of plinenum to re-seek if necessary. }}
171 		 * {{ Since forw_raw_line modifies linebuf, we must
172 		 *    do this first, before storing anything in linebuf. }}
173 		 */
174 		linenum = find_linenum(pos);
175 	}
176 
177 	/*
178 	 * Display a status column if the -J option is set.
179 	 */
180 	if (status_col)
181 	{
182 		linebuf[curr] = ' ';
183 		if (start_attnpos != NULL_POSITION &&
184 		    pos >= start_attnpos && pos < end_attnpos)
185 			attr[curr] = AT_NORMAL|AT_HILITE;
186 		else
187 			attr[curr] = AT_NORMAL;
188 		curr++;
189 		column++;
190 	}
191 	/*
192 	 * Display the line number at the start of each line
193 	 * if the -N option is set.
194 	 */
195 	if (linenums == OPT_ONPLUS)
196 	{
197 		char buf[INT_STRLEN_BOUND(pos) + 2];
198 		int n;
199 
200 		linenumtoa(linenum, buf, sizeof(buf));
201 		n = strlen(buf);
202 		if (n < MIN_LINENUM_WIDTH)
203 			n = MIN_LINENUM_WIDTH;
204 		snprintf(linebuf+curr, size_linebuf-curr, "%*s ", n, buf);
205 		n++;  /* One space after the line number. */
206 		for (i = 0; i < n; i++)
207 			attr[curr+i] = AT_NORMAL;
208 		curr += n;
209 		column += n;
210 		lmargin += n;
211 	}
212 
213 	/*
214 	 * Append enough spaces to bring us to the lmargin.
215 	 */
216 	while (column < lmargin)
217 	{
218 		linebuf[curr] = ' ';
219 		attr[curr++] = AT_NORMAL;
220 		column++;
221 	}
222 }
223 
224 /*
225  * Shift the input line left.
226  * This means discarding N printable chars at the start of the buffer.
227  */
228 	static void
229 pshift(shift)
230 	int shift;
231 {
232 	LWCHAR prev_ch = 0;
233 	unsigned char c;
234 	int shifted = 0;
235 	int to;
236 	int from;
237 	int len;
238 	int width;
239 	int prev_attr;
240 	int next_attr;
241 
242 	if (shift > column - lmargin)
243 		shift = column - lmargin;
244 	if (shift > curr - lmargin)
245 		shift = curr - lmargin;
246 
247 	to = from = lmargin;
248 	/*
249 	 * We keep on going when shifted == shift
250 	 * to get all combining chars.
251 	 */
252 	while (shifted <= shift && from < curr)
253 	{
254 		c = linebuf[from];
255 		if (ctldisp == OPT_ONPLUS && IS_CSI_START(c))
256 		{
257 			/* Keep cumulative effect.  */
258 			linebuf[to] = c;
259 			attr[to++] = attr[from++];
260 			while (from < curr && linebuf[from])
261 			{
262 				linebuf[to] = linebuf[from];
263 				attr[to++] = attr[from];
264 				if (!is_ansi_middle(linebuf[from++]))
265 					break;
266 			}
267 			continue;
268 		}
269 
270 		width = 0;
271 
272 #if !SMALL
273 		if (!IS_ASCII_OCTET(c) && utf_mode)
274 		{
275 			/* Assumes well-formedness validation already done.  */
276 			LWCHAR ch;
277 
278 			len = utf_len(c);
279 			if (from + len > curr)
280 				break;
281 			ch = get_wchar(linebuf + from);
282 			if (!is_composing_char(ch) && !is_combining_char(prev_ch, ch))
283 				width = is_wide_char(ch) ? 2 : 1;
284 			prev_ch = ch;
285 		} else
286 #endif /* !SMALL */
287 		{
288 			len = 1;
289 			if (c == '\b')
290 				/* XXX - Incorrect if several '\b' in a row.  */
291 #if !SMALL
292 				width = (utf_mode && is_wide_char(prev_ch)) ? -2 : -1;
293 #else
294 				width = -1;
295 #endif /* !SMALL */
296 			else if (!control_char(c))
297 				width = 1;
298 			prev_ch = 0;
299 		}
300 
301 		if (width == 2 && shift - shifted == 1) {
302 			/* Should never happen when called by pshift_all().  */
303 			attr[to] = attr[from];
304 			/*
305 			 * Assume a wide_char will never be the first half of a
306 			 * combining_char pair, so reset prev_ch in case we're
307 			 * followed by a '\b'.
308 			 */
309 			prev_ch = linebuf[to++] = ' ';
310 			from += len;
311 			shifted++;
312 			continue;
313 		}
314 
315 		/* Adjust width for magic cookies. */
316 		prev_attr = (to > 0) ? attr[to-1] : AT_NORMAL;
317 		next_attr = (from + len < curr) ? attr[from + len] : prev_attr;
318 		if (!is_at_equiv(attr[from], prev_attr) &&
319 			!is_at_equiv(attr[from], next_attr))
320 		{
321 			width += attr_swidth(attr[from]);
322 			if (from + len < curr)
323 				width += attr_ewidth(attr[from]);
324 			if (is_at_equiv(prev_attr, next_attr))
325 			{
326 				width += attr_ewidth(prev_attr);
327 				if (from + len < curr)
328 					width += attr_swidth(next_attr);
329 			}
330 		}
331 
332 		if (shift - shifted < width)
333 			break;
334 		from += len;
335 		shifted += width;
336 		if (shifted < 0)
337 			shifted = 0;
338 	}
339 	while (from < curr)
340 	{
341 		linebuf[to] = linebuf[from];
342 		attr[to++] = attr[from++];
343 	}
344 	curr = to;
345 	column -= shifted;
346 	cshift += shifted;
347 }
348 
349 /*
350  *
351  */
352 	public void
353 pshift_all()
354 {
355 	pshift(column);
356 }
357 
358 /*
359  * Return the printing width of the start (enter) sequence
360  * for a given character attribute.
361  */
362 	static int
363 attr_swidth(a)
364 	int a;
365 {
366 	int w = 0;
367 
368 	a = apply_at_specials(a);
369 
370 	if (a & AT_UNDERLINE)
371 		w += ul_s_width;
372 	if (a & AT_BOLD)
373 		w += bo_s_width;
374 	if (a & AT_BLINK)
375 		w += bl_s_width;
376 	if (a & AT_STANDOUT)
377 		w += so_s_width;
378 
379 	return w;
380 }
381 
382 /*
383  * Return the printing width of the end (exit) sequence
384  * for a given character attribute.
385  */
386 	static int
387 attr_ewidth(a)
388 	int a;
389 {
390 	int w = 0;
391 
392 	a = apply_at_specials(a);
393 
394 	if (a & AT_UNDERLINE)
395 		w += ul_e_width;
396 	if (a & AT_BOLD)
397 		w += bo_e_width;
398 	if (a & AT_BLINK)
399 		w += bl_e_width;
400 	if (a & AT_STANDOUT)
401 		w += so_e_width;
402 
403 	return w;
404 }
405 
406 /*
407  * Return the printing width of a given character and attribute,
408  * if the character were added to the current position in the line buffer.
409  * Adding a character with a given attribute may cause an enter or exit
410  * attribute sequence to be inserted, so this must be taken into account.
411  */
412 	static int
413 pwidth(ch, a, prev_ch)
414 	LWCHAR ch;
415 	int a;
416 	LWCHAR prev_ch;
417 {
418 	int w;
419 
420 	if (ch == '\b')
421 		/*
422 		 * Backspace moves backwards one or two positions.
423 		 * XXX - Incorrect if several '\b' in a row.
424 		 */
425 #if !SMALL
426 		return (utf_mode && is_wide_char(prev_ch)) ? -2 : -1;
427 #else
428 		return -1;
429 #endif /* !SMALL */
430 
431 	if (!utf_mode || is_ascii_char(ch))
432 	{
433 		if (control_char((char)ch))
434 		{
435 			/*
436 			 * Control characters do unpredictable things,
437 			 * so we don't even try to guess; say it doesn't move.
438 			 * This can only happen if the -r flag is in effect.
439 			 */
440 			return (0);
441 		}
442 	}
443 #if !SMALL
444 	else
445 	{
446 		if (is_composing_char(ch) || is_combining_char(prev_ch, ch))
447 		{
448 			/*
449 			 * Composing and combining chars take up no space.
450 			 *
451 			 * Some terminals, upon failure to compose a
452 			 * composing character with the character(s) that
453 			 * precede(s) it will actually take up one column
454 			 * for the composing character; there isn't much
455 			 * we could do short of testing the (complex)
456 			 * composition process ourselves and printing
457 			 * a binary representation when it fails.
458 			 */
459 			return (0);
460 		}
461 	}
462 #endif /* !SMALL */
463 
464 	/*
465 	 * Other characters take one or two columns,
466 	 * plus the width of any attribute enter/exit sequence.
467 	 */
468 	w = 1;
469 #if !SMALL
470 	if (is_wide_char(ch))
471 		w++;
472 #endif /* !SMALL */
473 	if (curr > 0 && !is_at_equiv(attr[curr-1], a))
474 		w += attr_ewidth(attr[curr-1]);
475 	if ((apply_at_specials(a) != AT_NORMAL) &&
476 	    (curr == 0 || !is_at_equiv(attr[curr-1], a)))
477 		w += attr_swidth(a);
478 	return (w);
479 }
480 
481 /*
482  * Delete to the previous base character in the line buffer.
483  * Return 1 if one is found.
484  */
485 	static int
486 backc()
487 {
488 	LWCHAR prev_ch;
489 	char *p = linebuf + curr;
490 	LWCHAR ch = step_char(&p, -1, linebuf + lmargin);
491 	int width;
492 
493 	/* This assumes that there is no '\b' in linebuf.  */
494 	while (   curr > lmargin
495 	       && column > lmargin
496 	       && (!(attr[curr - 1] & (AT_ANSI|AT_BINARY))))
497 	{
498 		curr = p - linebuf;
499 		prev_ch = step_char(&p, -1, linebuf + lmargin);
500 		width = pwidth(ch, attr[curr], prev_ch);
501 		column -= width;
502 		if (width > 0)
503 			return 1;
504 		ch = prev_ch;
505 	}
506 
507 	return 0;
508 }
509 
510 /*
511  * Are we currently within a recognized ANSI escape sequence?
512  */
513 	static int
514 in_ansi_esc_seq()
515 {
516 	char *p;
517 
518 	/*
519 	 * Search backwards for either an ESC (which means we ARE in a seq);
520 	 * or an end char (which means we're NOT in a seq).
521 	 */
522 	for (p = &linebuf[curr];  p > linebuf; )
523 	{
524 		LWCHAR ch = step_char(&p, -1, linebuf);
525 		if (IS_CSI_START(ch))
526 			return (1);
527 		if (!is_ansi_middle(ch))
528 			return (0);
529 	}
530 	return (0);
531 }
532 
533 /*
534  * Is a character the end of an ANSI escape sequence?
535  */
536 	public int
537 is_ansi_end(ch)
538 	LWCHAR ch;
539 {
540 	if (!is_ascii_char(ch))
541 		return (0);
542 	return (strchr(end_ansi_chars, (char) ch) != NULL);
543 }
544 
545 /*
546  *
547  */
548 	public int
549 is_ansi_middle(ch)
550 	LWCHAR ch;
551 {
552 	if (!is_ascii_char(ch))
553 		return (0);
554 	if (is_ansi_end(ch))
555 		return (0);
556 	return (strchr(mid_ansi_chars, (char) ch) != NULL);
557 }
558 
559 /*
560  * Append a character and attribute to the line buffer.
561  */
562 #define	STORE_CHAR(ch,a,rep,pos) \
563 	do { \
564 		if (store_char((ch),(a),(rep),(pos))) return (1); \
565 	} while (0)
566 
567 	static int
568 store_char(ch, a, rep, pos)
569 	LWCHAR ch;
570 	int a;
571 	char *rep;
572 	POSITION pos;
573 {
574 	int w;
575 	int replen;
576 	char cs;
577 
578 	w = (a & (AT_UNDERLINE|AT_BOLD));	/* Pre-use w.  */
579 	if (w != AT_NORMAL)
580 		last_overstrike = w;
581 
582 #if HILITE_SEARCH
583 	{
584 		int matches;
585 		if (is_hilited(pos, pos+1, 0, &matches))
586 		{
587 			/*
588 			 * This character should be highlighted.
589 			 * Override the attribute passed in.
590 			 */
591 			if (a != AT_ANSI)
592 				a |= AT_HILITE;
593 		}
594 	}
595 #endif
596 
597 	if (ctldisp == OPT_ONPLUS && in_ansi_esc_seq())
598 	{
599 		if (!is_ansi_end(ch) && !is_ansi_middle(ch)) {
600 			/* Remove whole unrecognized sequence.  */
601 			char *p = &linebuf[curr];
602 			LWCHAR bch;
603 			do {
604 				bch = step_char(&p, -1, linebuf);
605 			} while (p > linebuf && !IS_CSI_START(bch));
606 			curr = p - linebuf;
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 #if !SMALL
638 		replen = utf_len(rep[0]);
639 #else
640 		replen = 1;
641 #endif
642 	}
643 	if (curr + replen >= size_linebuf-6)
644 	{
645 		/*
646 		 * Won't fit in line buffer.
647 		 * Try to expand it.
648 		 */
649 		if (expand_linebuf())
650 			return (1);
651 	}
652 
653 	while (replen-- > 0)
654 	{
655 		linebuf[curr] = *rep++;
656 		attr[curr] = a;
657 		curr++;
658 	}
659 	column += w;
660 	return (0);
661 }
662 
663 /*
664  * Append a tab to the line buffer.
665  * Store spaces to represent the tab.
666  */
667 #define	STORE_TAB(a,pos) \
668 	do { if (store_tab((a),(pos))) return (1); } while (0)
669 
670 	static int
671 store_tab(attr, pos)
672 	int attr;
673 	POSITION pos;
674 {
675 	int to_tab = column + cshift - lmargin;
676 	int i;
677 
678 	if (ntabstops < 2 || to_tab >= tabstops[ntabstops-1])
679 		to_tab = tabdefault -
680 		     ((to_tab - tabstops[ntabstops-1]) % tabdefault);
681 	else
682 	{
683 		for (i = ntabstops - 2;  i >= 0;  i--)
684 			if (to_tab >= tabstops[i])
685 				break;
686 		to_tab = tabstops[i+1] - to_tab;
687 	}
688 
689 	if (column + to_tab - 1 + pwidth(' ', attr, 0) + attr_ewidth(attr) > sc_width)
690 		return 1;
691 
692 	do {
693 		STORE_CHAR(' ', attr, " ", pos);
694 	} while (--to_tab > 0);
695 	return 0;
696 }
697 
698 #define STORE_PRCHAR(c, pos) \
699 	do { if (store_prchar((c), (pos))) return 1; } while (0)
700 
701 	static int
702 store_prchar(c, pos)
703 	char c;
704 	POSITION pos;
705 {
706 	char *s;
707 
708 	/*
709 	 * Convert to printable representation.
710 	 */
711 	s = prchar(c);
712 
713 	/*
714 	 * Make sure we can get the entire representation
715 	 * of the character on this line.
716 	 */
717 	if (column + (int) strlen(s) - 1 +
718             pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
719 		return 1;
720 
721 	for ( ;  *s != 0;  s++)
722 		STORE_CHAR(*s, AT_BINARY, NULL, pos);
723 
724 	return 0;
725 }
726 
727 #if !SMALL
728 	static int
729 flush_mbc_buf(pos)
730 	POSITION pos;
731 {
732 	int i;
733 
734 	for (i = 0; i < mbc_buf_index; i++)
735 		if (store_prchar(mbc_buf[i], pos))
736 			return mbc_buf_index - i;
737 
738 	return 0;
739 }
740 #endif /* !SMALL */
741 
742 /*
743  * Append a character to the line buffer.
744  * Expand tabs into spaces, handle underlining, boldfacing, etc.
745  * Returns 0 if ok, 1 if couldn't fit in buffer.
746  */
747 	public int
748 pappend(c, pos)
749 	char c;
750 	POSITION pos;
751 {
752 	int r;
753 
754 	if (pendc)
755 	{
756 		if (do_append(pendc, NULL, pendpos))
757 			/*
758 			 * Oops.  We've probably lost the char which
759 			 * was in pendc, since caller won't back up.
760 			 */
761 			return (1);
762 		pendc = '\0';
763 	}
764 
765 	if (c == '\r' && bs_mode == BS_SPECIAL)
766 	{
767 #if !SMALL
768 		if (mbc_buf_len > 0)  /* utf_mode must be on. */
769 		{
770 			/* Flush incomplete (truncated) sequence. */
771 			r = flush_mbc_buf(mbc_pos);
772 			mbc_buf_index = r + 1;
773 			mbc_buf_len = 0;
774 			if (r)
775 				return (mbc_buf_index);
776 		}
777 #endif /* !SMALL */
778 
779 		/*
780 		 * Don't put the CR into the buffer until we see
781 		 * the next char.  If the next char is a newline,
782 		 * discard the CR.
783 		 */
784 		pendc = c;
785 		pendpos = pos;
786 		return (0);
787 	}
788 
789 	if (!utf_mode)
790 	{
791 		r = do_append((LWCHAR) c, NULL, pos);
792 	}
793 #if !SMALL
794 	else
795 	{
796 		/* Perform strict validation in all possible cases. */
797 		if (mbc_buf_len == 0)
798 		{
799 		retry:
800 			mbc_buf_index = 1;
801 			*mbc_buf = c;
802 			if (IS_ASCII_OCTET(c))
803 				r = do_append((LWCHAR) c, NULL, pos);
804 			else if (IS_UTF8_LEAD(c))
805 			{
806 				mbc_buf_len = utf_len(c);
807 				mbc_pos = pos;
808 				return (0);
809 			} else
810 				/* UTF8_INVALID or stray UTF8_TRAIL */
811 				r = flush_mbc_buf(pos);
812 		} else if (IS_UTF8_TRAIL(c))
813 		{
814 			mbc_buf[mbc_buf_index++] = c;
815 			if (mbc_buf_index < mbc_buf_len)
816 				return (0);
817 			if (is_utf8_well_formed(mbc_buf))
818 				r = do_append(get_wchar(mbc_buf), mbc_buf, mbc_pos);
819 			else
820 				/* Complete, but not shortest form, sequence. */
821 				mbc_buf_index = r = flush_mbc_buf(mbc_pos);
822 			mbc_buf_len = 0;
823 		} else
824 		{
825 			/* Flush incomplete (truncated) sequence.  */
826 			r = flush_mbc_buf(mbc_pos);
827 			mbc_buf_index = r + 1;
828 			mbc_buf_len = 0;
829 			/* Handle new char.  */
830 			if (!r)
831 				goto retry;
832  		}
833 	}
834 #endif /* !SMALL */
835 
836 	/*
837 	 * If we need to shift the line, do it.
838 	 * But wait until we get to at least the middle of the screen,
839 	 * so shifting it doesn't affect the chars we're currently
840 	 * pappending.  (Bold & underline can get messed up otherwise.)
841 	 */
842 	if (cshift < hshift && column > sc_width / 2)
843 	{
844 		linebuf[curr] = '\0';
845 		pshift(hshift - cshift);
846 	}
847 	if (r)
848 	{
849 		/* How many chars should caller back up? */
850 		r = (!utf_mode) ? 1 : mbc_buf_index;
851 	}
852 	return (r);
853 }
854 
855 	static int
856 do_append(ch, rep, pos)
857 	LWCHAR ch;
858 	char *rep;
859 	POSITION pos;
860 {
861 	register int a;
862 	LWCHAR prev_ch;
863 
864 	a = AT_NORMAL;
865 
866 	if (ch == '\b')
867 	{
868 		if (bs_mode == BS_CONTROL)
869 			goto do_control_char;
870 
871 		/*
872 		 * A better test is needed here so we don't
873 		 * backspace over part of the printed
874 		 * representation of a binary character.
875 		 */
876 		if (   curr <= lmargin
877 		    || column <= lmargin
878 		    || (attr[curr - 1] & (AT_ANSI|AT_BINARY)))
879 			STORE_PRCHAR('\b', pos);
880 		else if (bs_mode == BS_NORMAL)
881 			STORE_CHAR(ch, AT_NORMAL, NULL, pos);
882 		else if (bs_mode == BS_SPECIAL)
883 			overstrike = backc();
884 
885 		return 0;
886 	}
887 
888 	if (overstrike > 0)
889 	{
890 		/*
891 		 * Overstrike the character at the current position
892 		 * in the line buffer.  This will cause either
893 		 * underline (if a "_" is overstruck),
894 		 * bold (if an identical character is overstruck),
895 		 * or just deletion of the character in the buffer.
896 		 */
897 		overstrike = utf_mode ? -1 : 0;
898 		/* To be correct, this must be a base character.  */
899 #if !SMALL
900 		prev_ch = get_wchar(linebuf + curr);
901 #else
902 		prev_ch = (LWCHAR)((char)(linebuf + curr)[0] & 0xFF);
903 #endif /* !SMALL */
904 		a = attr[curr];
905 		if (ch == prev_ch)
906 		{
907 			/*
908 			 * Overstriking a char with itself means make it bold.
909 			 * But overstriking an underscore with itself is
910 			 * ambiguous.  It could mean make it bold, or
911 			 * it could mean make it underlined.
912 			 * Use the previous overstrike to resolve it.
913 			 */
914 			if (ch == '_')
915 			{
916 				if ((a & (AT_BOLD|AT_UNDERLINE)) != AT_NORMAL)
917 					a |= (AT_BOLD|AT_UNDERLINE);
918 				else if (last_overstrike != AT_NORMAL)
919 					a |= last_overstrike;
920 				else
921 					a |= AT_BOLD;
922 			} else
923 				a |= AT_BOLD;
924 		} else if (ch == '_')
925 		{
926 			a |= AT_UNDERLINE;
927 			ch = prev_ch;
928 			rep = linebuf + curr;
929 		} else if (prev_ch == '_')
930 		{
931 			a |= AT_UNDERLINE;
932 		}
933 		/* Else we replace prev_ch, but we keep its attributes.  */
934 	} else if (overstrike < 0)
935 	{
936 #if !SMALL
937 		if (   is_composing_char(ch)
938 		    || is_combining_char(get_wchar(linebuf + curr), ch))
939 			/* Continuation of the same overstrike.  */
940 			a = last_overstrike;
941 		else
942 #endif /* !SMALL */
943 			overstrike = 0;
944 	}
945 
946 	if (ch == '\t')
947 	{
948 		/*
949 		 * Expand a tab into spaces.
950 		 */
951 		switch (bs_mode)
952 		{
953 		case BS_CONTROL:
954 			goto do_control_char;
955 		case BS_NORMAL:
956 		case BS_SPECIAL:
957 			STORE_TAB(a, pos);
958 			break;
959 		}
960 	} else if ((!utf_mode || is_ascii_char(ch)) && control_char((char)ch))
961 	{
962 	do_control_char:
963 		if (ctldisp == OPT_ON || (ctldisp == OPT_ONPLUS && IS_CSI_START(ch)))
964 		{
965 			/*
966 			 * Output as a normal character.
967 			 */
968 			STORE_CHAR(ch, AT_NORMAL, rep, pos);
969 		} else
970 		{
971 			STORE_PRCHAR((char) ch, pos);
972 		}
973 	}
974 #if !SMALL
975 	else if (utf_mode && ctldisp != OPT_ON && is_ubin_char(ch))
976 	{
977 		char *s;
978 
979 		s = prutfchar(ch);
980 
981 		if (column + (int) strlen(s) - 1 +
982 		    pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
983 			return (1);
984 
985 		for ( ;  *s != 0;  s++)
986 			STORE_CHAR(*s, AT_BINARY, NULL, pos);
987  	}
988 #endif /* !SMALL */
989 	else
990 	{
991 		STORE_CHAR(ch, a, rep, pos);
992 	}
993  	return (0);
994 }
995 
996 /*
997  *
998  */
999 	public int
1000 pflushmbc()
1001 {
1002 	int r = 0;
1003 
1004 #if !SMALL
1005 	if (mbc_buf_len > 0)
1006 	{
1007 		/* Flush incomplete (truncated) sequence.  */
1008 		r = flush_mbc_buf(mbc_pos);
1009 		mbc_buf_len = 0;
1010 	}
1011 #endif /* !SMALL */
1012 	return r;
1013 }
1014 
1015 /*
1016  * Terminate the line in the line buffer.
1017  */
1018 	public void
1019 pdone(endline, forw)
1020 	int endline;
1021 	int forw;
1022 {
1023 	(void) pflushmbc();
1024 
1025 	if (pendc && (pendc != '\r' || !endline))
1026 		/*
1027 		 * If we had a pending character, put it in the buffer.
1028 		 * But discard a pending CR if we are at end of line
1029 		 * (that is, discard the CR in a CR/LF sequence).
1030 		 */
1031 		(void) do_append(pendc, NULL, pendpos);
1032 
1033 	/*
1034 	 * Make sure we've shifted the line, if we need to.
1035 	 */
1036 	if (cshift < hshift)
1037 		pshift(hshift - cshift);
1038 
1039 	if (ctldisp == OPT_ONPLUS && is_ansi_end('m'))
1040 	{
1041 		/* Switch to normal attribute at end of line. */
1042 		char *p = "\033[m";
1043 		for ( ;  *p != '\0';  p++)
1044 		{
1045 			linebuf[curr] = *p;
1046 			attr[curr++] = AT_ANSI;
1047 		}
1048 	}
1049 
1050 	/*
1051 	 * Add a newline if necessary,
1052 	 * and append a '\0' to the end of the line.
1053 	 * We output a newline if we're not at the right edge of the screen,
1054 	 * or if the terminal doesn't auto wrap,
1055 	 * or if this is really the end of the line AND the terminal ignores
1056 	 * a newline at the right edge.
1057 	 * (In the last case we don't want to output a newline if the terminal
1058 	 * doesn't ignore it since that would produce an extra blank line.
1059 	 * But we do want to output a newline if the terminal ignores it in case
1060 	 * the next line is blank.  In that case the single newline output for
1061 	 * that blank line would be ignored!)
1062 	 */
1063 	if (column < sc_width || !auto_wrap || (endline && ignaw) || ctldisp == OPT_ON)
1064 	{
1065 		linebuf[curr] = '\n';
1066 		attr[curr] = AT_NORMAL;
1067 		curr++;
1068 	}
1069 	else if (ignaw && column >= sc_width && forw)
1070 	{
1071 		/*
1072 		 * Terminals with "ignaw" don't wrap until they *really* need
1073 		 * to, i.e. when the character *after* the last one to fit on a
1074 		 * line is output. But they are too hard to deal with when they
1075 		 * get in the state where a full screen width of characters
1076 		 * have been output but the cursor is sitting on the right edge
1077 		 * instead of at the start of the next line.
1078 		 * So we nudge them into wrapping by outputting a space
1079 		 * character plus a backspace.  But do this only if moving
1080 		 * forward; if we're moving backward and drawing this line at
1081 		 * the top of the screen, the space would overwrite the first
1082 		 * char on the next line.  We don't need to do this "nudge"
1083 		 * at the top of the screen anyway.
1084 		 */
1085 		linebuf[curr] = ' ';
1086 		attr[curr++] = AT_NORMAL;
1087 		linebuf[curr] = '\b';
1088 		attr[curr++] = AT_NORMAL;
1089 	}
1090 	linebuf[curr] = '\0';
1091 	attr[curr] = AT_NORMAL;
1092 }
1093 
1094 /*
1095  *
1096  */
1097 	public void
1098 set_status_col(c)
1099 	char c;
1100 {
1101 	linebuf[0] = c;
1102 	attr[0] = AT_NORMAL|AT_HILITE;
1103 }
1104 
1105 /*
1106  * Get a character from the current line.
1107  * Return the character as the function return value,
1108  * and the character attribute in *ap.
1109  */
1110 	public int
1111 gline(i, ap)
1112 	register int i;
1113 	register int *ap;
1114 {
1115 	if (is_null_line)
1116 	{
1117 		/*
1118 		 * If there is no current line, we pretend the line is
1119 		 * either "~" or "", depending on the "twiddle" flag.
1120 		 */
1121 		if (twiddle)
1122 		{
1123 			if (i == 0)
1124 			{
1125 				*ap = AT_BOLD;
1126 				return '~';
1127 			}
1128 			--i;
1129 		}
1130 		/* Make sure we're back to AT_NORMAL before the '\n'.  */
1131 		*ap = AT_NORMAL;
1132 		return i ? '\0' : '\n';
1133 	}
1134 
1135 	*ap = attr[i];
1136 	return (linebuf[i] & 0xFF);
1137 }
1138 
1139 /*
1140  * Indicate that there is no current line.
1141  */
1142 	public void
1143 null_line()
1144 {
1145 	is_null_line = 1;
1146 	cshift = 0;
1147 }
1148 
1149 /*
1150  * Analogous to forw_line(), but deals with "raw lines":
1151  * lines which are not split for screen width.
1152  * {{ This is supposed to be more efficient than forw_line(). }}
1153  */
1154 	public POSITION
1155 forw_raw_line(curr_pos, linep, line_lenp)
1156 	POSITION curr_pos;
1157 	char **linep;
1158 	int *line_lenp;
1159 {
1160 	register int n;
1161 	register int c;
1162 	POSITION new_pos;
1163 
1164 	if (curr_pos == NULL_POSITION || ch_seek(curr_pos) ||
1165 		(c = ch_forw_get()) == EOI)
1166 		return (NULL_POSITION);
1167 
1168 	n = 0;
1169 	for (;;)
1170 	{
1171 		if (c == '\n' || c == EOI || ABORT_SIGS())
1172 		{
1173 			new_pos = ch_tell();
1174 			break;
1175 		}
1176 		if (n >= size_linebuf-1)
1177 		{
1178 			if (expand_linebuf())
1179 			{
1180 				/*
1181 				 * Overflowed the input buffer.
1182 				 * Pretend the line ended here.
1183 				 */
1184 				new_pos = ch_tell() - 1;
1185 				break;
1186 			}
1187 		}
1188 		linebuf[n++] = c;
1189 		c = ch_forw_get();
1190 	}
1191 	linebuf[n] = '\0';
1192 	if (linep != NULL)
1193 		*linep = linebuf;
1194 	if (line_lenp != NULL)
1195 		*line_lenp = n;
1196 	return (new_pos);
1197 }
1198 
1199 /*
1200  * Analogous to back_line(), but deals with "raw lines".
1201  * {{ This is supposed to be more efficient than back_line(). }}
1202  */
1203 	public POSITION
1204 back_raw_line(curr_pos, linep, line_lenp)
1205 	POSITION curr_pos;
1206 	char **linep;
1207 	int *line_lenp;
1208 {
1209 	register int n;
1210 	register int c;
1211 	POSITION new_pos;
1212 
1213 	if (curr_pos == NULL_POSITION || curr_pos <= ch_zero() ||
1214 		ch_seek(curr_pos-1))
1215 		return (NULL_POSITION);
1216 
1217 	n = size_linebuf;
1218 	linebuf[--n] = '\0';
1219 	for (;;)
1220 	{
1221 		c = ch_back_get();
1222 		if (c == '\n' || ABORT_SIGS())
1223 		{
1224 			/*
1225 			 * This is the newline ending the previous line.
1226 			 * We have hit the beginning of the line.
1227 			 */
1228 			new_pos = ch_tell() + 1;
1229 			break;
1230 		}
1231 		if (c == EOI)
1232 		{
1233 			/*
1234 			 * We have hit the beginning of the file.
1235 			 * This must be the first line in the file.
1236 			 * This must, of course, be the beginning of the line.
1237 			 */
1238 			new_pos = ch_zero();
1239 			break;
1240 		}
1241 		if (n <= 0)
1242 		{
1243 			int old_size_linebuf = size_linebuf;
1244 			if (expand_linebuf())
1245 			{
1246 				/*
1247 				 * Overflowed the input buffer.
1248 				 * Pretend the line ended here.
1249 				 */
1250 				new_pos = ch_tell() + 1;
1251 				break;
1252 			}
1253 			/*
1254 			 * Shift the data to the end of the new linebuf.
1255 			 */
1256 			n = size_linebuf - old_size_linebuf;
1257 			memmove(linebuf + n, linebuf, old_size_linebuf);
1258 		}
1259 		linebuf[--n] = c;
1260 	}
1261 	if (linep != NULL)
1262 		*linep = &linebuf[n];
1263 	if (line_lenp != NULL)
1264 		*line_lenp = size_linebuf - 1 - n;
1265 	return (new_pos);
1266 }
1267