3 * Copyright (C) 1984-2005 Mark Nudelman
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.
8 * For more information about less, or for information on how to
9 * contact the author, see the README file.
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.
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) */
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 */
32 static int curr; /* Index into linebuf */
33 static int column; /* Printable length, accounting for
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 */
41 static POSITION pendpos;
42 static char *end_ansi_chars;
43 static char *mid_ansi_chars;
45 static int attr_swidth();
46 static int attr_ewidth();
47 static int do_append();
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;
63 extern POSITION start_attnpos;
64 extern POSITION end_attnpos;
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;
72 * Initialize from environment variables.
77 end_ansi_chars = lgetenv("LESSANSIENDCHARS");
78 if (end_ansi_chars == NULL || *end_ansi_chars == '\0')
81 mid_ansi_chars = lgetenv("LESSANSIMIDCHARS");
82 if (mid_ansi_chars == NULL || *mid_ansi_chars == '\0')
83 mid_ansi_chars = "0123456789;[?!\"'#%()*+ ";
85 linebuf = (char *) ecalloc(LINEBUF_SIZE, sizeof(char));
86 attr = (char *) ecalloc(LINEBUF_SIZE, sizeof(char));
87 size_linebuf = LINEBUF_SIZE;
91 * Expand the line buffer.
96 /* Double the size of the line buffer. */
97 int new_size = size_linebuf * 2;
99 /* Just realloc to expand the buffer, if we can. */
101 char *new_buf = (char *) realloc(linebuf, new_size);
102 char *new_attr = (char *) realloc(attr, new_size);
104 char *new_buf = (char *) calloc(new_size, sizeof(char));
105 char *new_attr = (char *) calloc(new_size, sizeof(char));
107 if (new_buf == NULL || new_attr == NULL)
109 if (new_attr != NULL)
117 * We realloc'd the buffers; they already have the old contents.
120 memset(new_buf + size_linebuf, 0, new_size - size_linebuf);
121 memset(new_attr + size_linebuf, 0, new_size - size_linebuf);
125 * We just calloc'd the buffers; copy the old contents.
127 memcpy(new_buf, linebuf, size_linebuf * sizeof(char));
128 memcpy(new_attr, attr, size_linebuf * sizeof(char));
134 size_linebuf = new_size;
139 * Is a character ASCII?
149 * Rewind the line buffer.
158 last_overstrike = AT_NORMAL;
171 * Insert the line number (of the given position) into the line buffer.
177 register LINENUM linenum = 0;
180 if (linenums == OPT_ONPLUS)
183 * Get the line number and put it in the current line.
184 * {{ Note: since find_linenum calls forw_raw_line,
185 * it may seek in the input file, requiring the caller
186 * of plinenum to re-seek if necessary. }}
187 * {{ Since forw_raw_line modifies linebuf, we must
188 * do this first, before storing anything in linebuf. }}
190 linenum = find_linenum(pos);
194 * Display a status column if the -J option is set.
199 if (start_attnpos != NULL_POSITION &&
200 pos >= start_attnpos && pos < end_attnpos)
201 attr[curr] = AT_NORMAL|AT_HILITE;
203 attr[curr] = AT_NORMAL;
208 * Display the line number at the start of each line
209 * if the -N option is set.
211 if (linenums == OPT_ONPLUS)
213 char buf[INT_STRLEN_BOUND(pos) + 2];
216 linenumtoa(linenum, buf);
218 if (n < MIN_LINENUM_WIDTH)
219 n = MIN_LINENUM_WIDTH;
220 sprintf(linebuf+curr, "%*s ", n, buf);
221 n++; /* One space after the line number. */
222 for (i = 0; i < n; i++)
223 attr[curr+i] = AT_NORMAL;
230 * Append enough spaces to bring us to the lmargin.
232 while (column < lmargin)
235 attr[curr++] = AT_NORMAL;
241 * Shift the input line left.
242 * This means discarding N printable chars at the start of the buffer.
258 if (shift > column - lmargin)
259 shift = column - lmargin;
260 if (shift > curr - lmargin)
261 shift = curr - lmargin;
265 * We keep on going when shifted == shift
266 * to get all combining chars.
268 while (shifted <= shift && from < curr)
271 if (c == ESC && ctldisp == OPT_ONPLUS)
273 /* Keep cumulative effect. */
275 attr[to++] = attr[from++];
276 while (from < curr && linebuf[from])
278 linebuf[to] = linebuf[from];
279 attr[to++] = attr[from];
280 if (!is_ansi_middle(linebuf[from++]))
288 if (!IS_ASCII_OCTET(c) && utf_mode)
290 /* Assumes well-formedness validation already done. */
294 if (from + len > curr)
296 ch = get_wchar(linebuf + from);
297 if (!is_composing_char(ch) && !is_combining_char(prev_ch, ch))
298 width = is_wide_char(ch) ? 2 : 1;
304 /* XXX - Incorrect if several '\b' in a row. */
305 width = (utf_mode && is_wide_char(prev_ch)) ? -2 : -1;
306 else if (!control_char(c))
311 if (width == 2 && shift - shifted == 1) {
312 /* Should never happen when called by pshift_all(). */
313 attr[to] = attr[from];
315 * Assume a wide_char will never be the first half of a
316 * combining_char pair, so reset prev_ch in case we're
317 * followed by a '\b'.
319 prev_ch = linebuf[to++] = ' ';
325 /* Adjust width for magic cookies. */
326 prev_attr = (to > 0) ? attr[to-1] : AT_NORMAL;
327 next_attr = (from + len < curr) ? attr[from + len] : prev_attr;
328 if (!is_at_equiv(attr[from], prev_attr) &&
329 !is_at_equiv(attr[from], next_attr))
331 width += attr_swidth(attr[from]);
332 if (from + len < curr)
333 width += attr_ewidth(attr[from]);
334 if (is_at_equiv(prev_attr, next_attr))
336 width += attr_ewidth(prev_attr);
337 if (from + len < curr)
338 width += attr_swidth(next_attr);
342 if (shift - shifted < width)
351 linebuf[to] = linebuf[from];
352 attr[to++] = attr[from++];
369 * Return the printing width of the start (enter) sequence
370 * for a given character attribute.
378 a = apply_at_specials(a);
380 if (a & AT_UNDERLINE)
393 * Return the printing width of the end (exit) sequence
394 * for a given character attribute.
402 a = apply_at_specials(a);
404 if (a & AT_UNDERLINE)
417 * Return the printing width of a given character and attribute,
418 * if the character were added to the current position in the line buffer.
419 * Adding a character with a given attribute may cause an enter or exit
420 * attribute sequence to be inserted, so this must be taken into account.
423 pwidth(ch, a, prev_ch)
432 * Backspace moves backwards one or two positions.
433 * XXX - Incorrect if several '\b' in a row.
435 return (utf_mode && is_wide_char(prev_ch)) ? -2 : -1;
437 if (!utf_mode || is_ascii_char(ch))
439 if (control_char((char)ch))
442 * Control characters do unpredictable things,
443 * so we don't even try to guess; say it doesn't move.
444 * This can only happen if the -r flag is in effect.
450 if (is_composing_char(ch) || is_combining_char(prev_ch, ch))
453 * Composing and combining chars take up no space.
455 * Some terminals, upon failure to compose a
456 * composing character with the character(s) that
457 * precede(s) it will actually take up one column
458 * for the composing character; there isn't much
459 * we could do short of testing the (complex)
460 * composition process ourselves and printing
461 * a binary representation when it fails.
468 * Other characters take one or two columns,
469 * plus the width of any attribute enter/exit sequence.
472 if (is_wide_char(ch))
474 if (curr > 0 && !is_at_equiv(attr[curr-1], a))
475 w += attr_ewidth(attr[curr-1]);
476 if ((apply_at_specials(a) != AT_NORMAL) &&
477 (curr == 0 || !is_at_equiv(attr[curr-1], a)))
483 * Delete to the previous base character in the line buffer.
484 * Return 1 if one is found.
490 char *p = linebuf + curr;
491 LWCHAR ch = step_char(&p, -1, linebuf + lmargin);
494 /* This assumes that there is no '\b' in linebuf. */
495 while ( curr > lmargin
497 && (!(attr[curr - 1] & (AT_ANSI|AT_BINARY))))
500 prev_ch = step_char(&p, -1, linebuf + lmargin);
501 width = pwidth(ch, attr[curr], prev_ch);
512 * Are we currently within a recognized ANSI escape sequence?
520 * Search backwards for either an ESC (which means we ARE in a seq);
521 * or an end char (which means we're NOT in a seq).
523 for (p = &linebuf[curr]; p > linebuf; )
525 LWCHAR ch = step_char(&p, -1, linebuf);
528 if (!is_ansi_middle(ch))
535 * Is a character the end of an ANSI escape sequence?
541 if (!is_ascii_char(ch))
543 return (strchr(end_ansi_chars, (char) ch) != NULL);
553 if (!is_ascii_char(ch))
557 return (strchr(mid_ansi_chars, (char) ch) != NULL);
561 * Append a character and attribute to the line buffer.
563 #define STORE_CHAR(ch,a,rep,pos) \
565 if (store_char((ch),(a),(rep),(pos))) return (1); \
569 store_char(ch, a, rep, pos)
579 w = (a & (AT_UNDERLINE|AT_BOLD)); /* Pre-use w. */
586 if (is_hilited(pos, pos+1, 0, &matches))
589 * This character should be highlighted.
590 * Override the attribute passed in.
595 line_matches += matches;
599 if (ctldisp == OPT_ONPLUS && in_ansi_esc_seq())
601 if (!is_ansi_end(ch) && !is_ansi_middle(ch)) {
602 /* Remove whole unrecognized sequence. */
605 } while (linebuf[curr] != ESC);
608 a = AT_ANSI; /* Will force re-AT_'ing around it. */
611 else if (ctldisp == OPT_ONPLUS && ch == ESC)
613 a = AT_ANSI; /* Will force re-AT_'ing around it. */
618 char *p = &linebuf[curr];
619 LWCHAR prev_ch = step_char(&p, -1, linebuf);
620 w = pwidth(ch, a, prev_ch);
623 if (ctldisp != OPT_ON && column + w + attr_ewidth(a) > sc_width)
625 * Won't fit on screen.
636 replen = utf_len(rep[0]);
638 if (curr + replen >= size_linebuf-6)
641 * Won't fit in line buffer.
644 if (expand_linebuf())
650 linebuf[curr] = *rep++;
659 * Append a tab to the line buffer.
660 * Store spaces to represent the tab.
662 #define STORE_TAB(a,pos) \
663 do { if (store_tab((a),(pos))) return (1); } while (0)
670 int to_tab = column + cshift - lmargin;
673 if (ntabstops < 2 || to_tab >= tabstops[ntabstops-1])
674 to_tab = tabdefault -
675 ((to_tab - tabstops[ntabstops-1]) % tabdefault);
678 for (i = ntabstops - 2; i >= 0; i--)
679 if (to_tab >= tabstops[i])
681 to_tab = tabstops[i+1] - to_tab;
684 if (column + to_tab - 1 + pwidth(' ', attr, 0) + attr_ewidth(attr) > sc_width)
688 STORE_CHAR(' ', attr, " ", pos);
689 } while (--to_tab > 0);
693 #define STORE_PRCHAR(c, pos) \
694 do { if (store_prchar((c), (pos))) return 1; } while (0)
704 * Convert to printable representation.
709 * Make sure we can get the entire representation
710 * of the character on this line.
712 if (column + (int) strlen(s) - 1 +
713 pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
716 for ( ; *s != 0; s++)
717 STORE_CHAR(*s, AT_BINARY, NULL, pos);
728 for (i = 0; i < mbc_buf_index; i++)
729 if (store_prchar(mbc_buf[i], pos))
730 return mbc_buf_index - i;
736 * Append a character to the line buffer.
737 * Expand tabs into spaces, handle underlining, boldfacing, etc.
738 * Returns 0 if ok, 1 if couldn't fit in buffer.
749 if (do_append(pendc, NULL, pendpos))
751 * Oops. We've probably lost the char which
752 * was in pendc, since caller won't back up.
758 if (c == '\r' && bs_mode == BS_SPECIAL)
760 if (mbc_buf_len > 0) /* utf_mode must be on. */
762 /* Flush incomplete (truncated) sequence. */
763 r = flush_mbc_buf(mbc_pos);
764 mbc_buf_index = r + 1;
767 return (mbc_buf_index);
771 * Don't put the CR into the buffer until we see
772 * the next char. If the next char is a newline,
782 r = do_append((LWCHAR) c, NULL, pos);
785 /* Perform strict validation in all possible cases. */
786 if (mbc_buf_len == 0)
791 if (IS_ASCII_OCTET(c))
792 r = do_append((LWCHAR) c, NULL, pos);
793 else if (IS_UTF8_LEAD(c))
795 mbc_buf_len = utf_len(c);
799 /* UTF8_INVALID or stray UTF8_TRAIL */
800 r = flush_mbc_buf(pos);
801 } else if (IS_UTF8_TRAIL(c))
803 mbc_buf[mbc_buf_index++] = c;
804 if (mbc_buf_index < mbc_buf_len)
806 if (is_utf8_well_formed(mbc_buf))
807 r = do_append(get_wchar(mbc_buf), mbc_buf, mbc_pos);
809 /* Complete, but not shortest form, sequence. */
810 mbc_buf_index = r = flush_mbc_buf(mbc_pos);
814 /* Flush incomplete (truncated) sequence. */
815 r = flush_mbc_buf(mbc_pos);
816 mbc_buf_index = r + 1;
818 /* Handle new char. */
825 * If we need to shift the line, do it.
826 * But wait until we get to at least the middle of the screen,
827 * so shifting it doesn't affect the chars we're currently
828 * pappending. (Bold & underline can get messed up otherwise.)
830 if (cshift < hshift && column > sc_width / 2)
832 linebuf[curr] = '\0';
833 pshift(hshift - cshift);
837 /* How many chars should caller back up? */
838 r = (!utf_mode) ? 1 : mbc_buf_index;
844 do_append(ch, rep, pos)
856 if (bs_mode == BS_CONTROL)
857 goto do_control_char;
860 * A better test is needed here so we don't
861 * backspace over part of the printed
862 * representation of a binary character.
866 || (attr[curr - 1] & (AT_ANSI|AT_BINARY)))
867 STORE_PRCHAR('\b', pos);
868 else if (bs_mode == BS_NORMAL)
869 STORE_CHAR(ch, AT_NORMAL, NULL, pos);
870 else if (bs_mode == BS_SPECIAL)
871 overstrike = backc();
879 * Overstrike the character at the current position
880 * in the line buffer. This will cause either
881 * underline (if a "_" is overstruck),
882 * bold (if an identical character is overstruck),
883 * or just deletion of the character in the buffer.
885 overstrike = utf_mode ? -1 : 0;
886 /* To be correct, this must be a base character. */
887 prev_ch = get_wchar(linebuf + curr);
892 * Overstriking a char with itself means make it bold.
893 * But overstriking an underscore with itself is
894 * ambiguous. It could mean make it bold, or
895 * it could mean make it underlined.
896 * Use the previous overstrike to resolve it.
900 if ((a & (AT_BOLD|AT_UNDERLINE)) != AT_NORMAL)
901 a |= (AT_BOLD|AT_UNDERLINE);
902 else if (last_overstrike != AT_NORMAL)
903 a |= last_overstrike;
908 } else if (ch == '_')
912 rep = linebuf + curr;
913 } else if (prev_ch == '_')
917 /* Else we replace prev_ch, but we keep its attributes. */
918 } else if (overstrike < 0)
920 if ( is_composing_char(ch)
921 || is_combining_char(get_wchar(linebuf + curr), ch))
922 /* Continuation of the same overstrike. */
931 * Expand a tab into spaces.
936 goto do_control_char;
942 } else if ((!utf_mode || is_ascii_char(ch)) && control_char((char)ch))
945 if (ctldisp == OPT_ON || (ctldisp == OPT_ONPLUS && ch == ESC))
948 * Output as a normal character.
950 STORE_CHAR(ch, AT_NORMAL, rep, pos);
953 STORE_PRCHAR((char) ch, pos);
955 } else if (utf_mode && ctldisp != OPT_ON && is_ubin_char(ch))
961 if (column + (int) strlen(s) - 1 +
962 pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
965 for ( ; *s != 0; s++)
966 STORE_CHAR(*s, AT_BINARY, NULL, pos);
969 STORE_CHAR(ch, a, rep, pos);
984 /* Flush incomplete (truncated) sequence. */
985 r = flush_mbc_buf(mbc_pos);
992 * Terminate the line in the line buffer.
1000 if (pendc && (pendc != '\r' || !endline))
1002 * If we had a pending character, put it in the buffer.
1003 * But discard a pending CR if we are at end of line
1004 * (that is, discard the CR in a CR/LF sequence).
1006 (void) do_append(pendc, NULL, pendpos);
1009 * Make sure we've shifted the line, if we need to.
1011 if (cshift < hshift)
1012 pshift(hshift - cshift);
1014 if (ctldisp == OPT_ONPLUS && is_ansi_end('m'))
1016 /* Switch to normal attribute at end of line. */
1018 for ( ; *p != '\0'; p++)
1021 attr[curr++] = AT_ANSI;
1026 * Add a newline if necessary,
1027 * and append a '\0' to the end of the line.
1029 if (column < sc_width || !auto_wrap || ignaw || ctldisp == OPT_ON)
1031 linebuf[curr] = '\n';
1032 attr[curr] = AT_NORMAL;
1035 linebuf[curr] = '\0';
1036 attr[curr] = AT_NORMAL;
1039 if (status_col && line_matches > 0)
1042 attr[0] = AT_NORMAL|AT_HILITE;
1048 * Get a character from the current line.
1049 * Return the character as the function return value,
1050 * and the character attribute in *ap.
1060 * If there is no current line, we pretend the line is
1061 * either "~" or "", depending on the "twiddle" flag.
1072 /* Make sure we're back to AT_NORMAL before the '\n'. */
1074 return i ? '\0' : '\n';
1078 return (linebuf[i] & 0xFF);
1082 * Indicate that there is no current line.
1092 * Analogous to forw_line(), but deals with "raw lines":
1093 * lines which are not split for screen width.
1094 * {{ This is supposed to be more efficient than forw_line(). }}
1097 forw_raw_line(curr_pos, linep)
1105 if (curr_pos == NULL_POSITION || ch_seek(curr_pos) ||
1106 (c = ch_forw_get()) == EOI)
1107 return (NULL_POSITION);
1112 if (c == '\n' || c == EOI || ABORT_SIGS())
1114 new_pos = ch_tell();
1117 if (n >= size_linebuf-1)
1119 if (expand_linebuf())
1122 * Overflowed the input buffer.
1123 * Pretend the line ended here.
1125 new_pos = ch_tell() - 1;
1139 * Analogous to back_line(), but deals with "raw lines".
1140 * {{ This is supposed to be more efficient than back_line(). }}
1143 back_raw_line(curr_pos, linep)
1151 if (curr_pos == NULL_POSITION || curr_pos <= ch_zero() ||
1152 ch_seek(curr_pos-1))
1153 return (NULL_POSITION);
1156 linebuf[--n] = '\0';
1160 if (c == '\n' || ABORT_SIGS())
1163 * This is the newline ending the previous line.
1164 * We have hit the beginning of the line.
1166 new_pos = ch_tell() + 1;
1172 * We have hit the beginning of the file.
1173 * This must be the first line in the file.
1174 * This must, of course, be the beginning of the line.
1176 new_pos = ch_zero();
1181 int old_size_linebuf = size_linebuf;
1184 if (expand_linebuf())
1187 * Overflowed the input buffer.
1188 * Pretend the line ended here.
1190 new_pos = ch_tell() + 1;
1194 * Shift the data to the end of the new linebuf.
1196 for (fm = linebuf + old_size_linebuf - 1,
1197 to = linebuf + size_linebuf - 1;
1198 fm >= linebuf; fm--, to--)
1200 n = size_linebuf - old_size_linebuf;
1205 *linep = &linebuf[n];