2 * Copyright (C) 1984-2011 Mark Nudelman
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.
7 * For more information about less, or for information on how to
8 * contact the author, see the README file.
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.
21 static char *linebuf = NULL; /* Buffer which holds the current output line */
22 static char *attr = NULL; /* Extension of linebuf to hold attributes */
23 public int size_linebuf = 0; /* Size of line buffer (and attr buffer) */
25 static int cshift; /* Current left-shift of output line buffer */
26 public int hshift; /* Desired left-shift of output line buffer */
27 public int tabstops[TABSTOP_MAX] = { 0 }; /* Custom tabstops */
28 public int ntabstops = 1; /* Number of tabstops */
29 public int tabdefault = 8; /* Default repeated tabstops */
31 static int curr; /* Index into linebuf */
32 static int column; /* Printable length, accounting for
34 static int overstrike; /* Next char should overstrike previous char */
35 static int last_overstrike = AT_NORMAL;
36 static int is_null_line; /* There is no current line */
37 static int lmargin; /* Left margin */
39 static POSITION pendpos;
40 static char *end_ansi_chars;
41 static char *mid_ansi_chars;
43 static int attr_swidth();
44 static int attr_ewidth();
45 static int do_append();
53 extern int status_col;
54 extern int auto_wrap, ignaw;
55 extern int bo_s_width, bo_e_width;
56 extern int ul_s_width, ul_e_width;
57 extern int bl_s_width, bl_e_width;
58 extern int so_s_width, so_e_width;
59 extern int sc_width, sc_height;
61 extern POSITION start_attnpos;
62 extern POSITION end_attnpos;
64 static char mbc_buf[MAX_UTF_CHAR_LEN];
65 static int mbc_buf_len = 0;
66 static int mbc_buf_index = 0;
67 static POSITION mbc_pos;
70 * Initialize from environment variables.
75 end_ansi_chars = lgetenv("LESSANSIENDCHARS");
76 if (end_ansi_chars == NULL || *end_ansi_chars == '\0')
79 mid_ansi_chars = lgetenv("LESSANSIMIDCHARS");
80 if (mid_ansi_chars == NULL || *mid_ansi_chars == '\0')
81 mid_ansi_chars = "0123456789;[?!\"'#%()*+ ";
83 linebuf = (char *) ecalloc(LINEBUF_SIZE, sizeof(char));
84 attr = (char *) ecalloc(LINEBUF_SIZE, sizeof(char));
85 size_linebuf = LINEBUF_SIZE;
89 * Expand the line buffer.
94 /* Double the size of the line buffer. */
95 int new_size = size_linebuf * 2;
97 /* Just realloc to expand the buffer, if we can. */
99 char *new_buf = (char *) realloc(linebuf, new_size);
100 char *new_attr = (char *) realloc(attr, new_size);
102 char *new_buf = (char *) calloc(new_size, sizeof(char));
103 char *new_attr = (char *) calloc(new_size, sizeof(char));
105 if (new_buf == NULL || new_attr == NULL)
107 if (new_attr != NULL)
115 * We realloc'd the buffers; they already have the old contents.
118 memset(new_buf + size_linebuf, 0, new_size - size_linebuf);
119 memset(new_attr + size_linebuf, 0, new_size - size_linebuf);
123 * We just calloc'd the buffers; copy the old contents.
125 memcpy(new_buf, linebuf, size_linebuf * sizeof(char));
126 memcpy(new_attr, attr, size_linebuf * sizeof(char));
132 size_linebuf = new_size;
137 * Is a character ASCII?
147 * Rewind the line buffer.
156 last_overstrike = AT_NORMAL;
166 * Insert the line number (of the given position) into the line buffer.
172 register LINENUM linenum = 0;
175 if (linenums == OPT_ONPLUS)
178 * Get the line number and put it in the current line.
179 * {{ Note: since find_linenum calls forw_raw_line,
180 * it may seek in the input file, requiring the caller
181 * of plinenum to re-seek if necessary. }}
182 * {{ Since forw_raw_line modifies linebuf, we must
183 * do this first, before storing anything in linebuf. }}
185 linenum = find_linenum(pos);
189 * Display a status column if the -J option is set.
194 if (start_attnpos != NULL_POSITION &&
195 pos >= start_attnpos && pos < end_attnpos)
196 attr[curr] = AT_NORMAL|AT_HILITE;
198 attr[curr] = AT_NORMAL;
203 * Display the line number at the start of each line
204 * if the -N option is set.
206 if (linenums == OPT_ONPLUS)
208 char buf[INT_STRLEN_BOUND(pos) + 2];
211 linenumtoa(linenum, buf);
213 if (n < MIN_LINENUM_WIDTH)
214 n = MIN_LINENUM_WIDTH;
215 sprintf(linebuf+curr, "%*s ", n, buf);
216 n++; /* One space after the line number. */
217 for (i = 0; i < n; i++)
218 attr[curr+i] = AT_NORMAL;
225 * Append enough spaces to bring us to the lmargin.
227 while (column < lmargin)
230 attr[curr++] = AT_NORMAL;
236 * Shift the input line left.
237 * This means discarding N printable chars at the start of the buffer.
253 if (shift > column - lmargin)
254 shift = column - lmargin;
255 if (shift > curr - lmargin)
256 shift = curr - lmargin;
260 * We keep on going when shifted == shift
261 * to get all combining chars.
263 while (shifted <= shift && from < curr)
266 if (ctldisp == OPT_ONPLUS && IS_CSI_START(c))
268 /* Keep cumulative effect. */
270 attr[to++] = attr[from++];
271 while (from < curr && linebuf[from])
273 linebuf[to] = linebuf[from];
274 attr[to++] = attr[from];
275 if (!is_ansi_middle(linebuf[from++]))
283 if (!IS_ASCII_OCTET(c) && utf_mode)
285 /* Assumes well-formedness validation already done. */
289 if (from + len > curr)
291 ch = get_wchar(linebuf + from);
292 if (!is_composing_char(ch) && !is_combining_char(prev_ch, ch))
293 width = is_wide_char(ch) ? 2 : 1;
299 /* XXX - Incorrect if several '\b' in a row. */
300 width = (utf_mode && is_wide_char(prev_ch)) ? -2 : -1;
301 else if (!control_char(c))
306 if (width == 2 && shift - shifted == 1) {
307 /* Should never happen when called by pshift_all(). */
308 attr[to] = attr[from];
310 * Assume a wide_char will never be the first half of a
311 * combining_char pair, so reset prev_ch in case we're
312 * followed by a '\b'.
314 prev_ch = linebuf[to++] = ' ';
320 /* Adjust width for magic cookies. */
321 prev_attr = (to > 0) ? attr[to-1] : AT_NORMAL;
322 next_attr = (from + len < curr) ? attr[from + len] : prev_attr;
323 if (!is_at_equiv(attr[from], prev_attr) &&
324 !is_at_equiv(attr[from], next_attr))
326 width += attr_swidth(attr[from]);
327 if (from + len < curr)
328 width += attr_ewidth(attr[from]);
329 if (is_at_equiv(prev_attr, next_attr))
331 width += attr_ewidth(prev_attr);
332 if (from + len < curr)
333 width += attr_swidth(next_attr);
337 if (shift - shifted < width)
346 linebuf[to] = linebuf[from];
347 attr[to++] = attr[from++];
364 * Return the printing width of the start (enter) sequence
365 * for a given character attribute.
373 a = apply_at_specials(a);
375 if (a & AT_UNDERLINE)
388 * Return the printing width of the end (exit) sequence
389 * for a given character attribute.
397 a = apply_at_specials(a);
399 if (a & AT_UNDERLINE)
412 * Return the printing width of a given character and attribute,
413 * if the character were added to the current position in the line buffer.
414 * Adding a character with a given attribute may cause an enter or exit
415 * attribute sequence to be inserted, so this must be taken into account.
418 pwidth(ch, a, prev_ch)
427 * Backspace moves backwards one or two positions.
428 * XXX - Incorrect if several '\b' in a row.
430 return (utf_mode && is_wide_char(prev_ch)) ? -2 : -1;
432 if (!utf_mode || is_ascii_char(ch))
434 if (control_char((char)ch))
437 * Control characters do unpredictable things,
438 * so we don't even try to guess; say it doesn't move.
439 * This can only happen if the -r flag is in effect.
445 if (is_composing_char(ch) || is_combining_char(prev_ch, ch))
448 * Composing and combining chars take up no space.
450 * Some terminals, upon failure to compose a
451 * composing character with the character(s) that
452 * precede(s) it will actually take up one column
453 * for the composing character; there isn't much
454 * we could do short of testing the (complex)
455 * composition process ourselves and printing
456 * a binary representation when it fails.
463 * Other characters take one or two columns,
464 * plus the width of any attribute enter/exit sequence.
467 if (is_wide_char(ch))
469 if (curr > 0 && !is_at_equiv(attr[curr-1], a))
470 w += attr_ewidth(attr[curr-1]);
471 if ((apply_at_specials(a) != AT_NORMAL) &&
472 (curr == 0 || !is_at_equiv(attr[curr-1], a)))
478 * Delete to the previous base character in the line buffer.
479 * Return 1 if one is found.
485 char *p = linebuf + curr;
486 LWCHAR ch = step_char(&p, -1, linebuf + lmargin);
489 /* This assumes that there is no '\b' in linebuf. */
490 while ( curr > lmargin
492 && (!(attr[curr - 1] & (AT_ANSI|AT_BINARY))))
495 prev_ch = step_char(&p, -1, linebuf + lmargin);
496 width = pwidth(ch, attr[curr], prev_ch);
507 * Are we currently within a recognized ANSI escape sequence?
515 * Search backwards for either an ESC (which means we ARE in a seq);
516 * or an end char (which means we're NOT in a seq).
518 for (p = &linebuf[curr]; p > linebuf; )
520 LWCHAR ch = step_char(&p, -1, linebuf);
521 if (IS_CSI_START(ch))
523 if (!is_ansi_middle(ch))
530 * Is a character the end of an ANSI escape sequence?
536 if (!is_ascii_char(ch))
538 return (strchr(end_ansi_chars, (char) ch) != NULL);
548 if (!is_ascii_char(ch))
552 return (strchr(mid_ansi_chars, (char) ch) != NULL);
556 * Append a character and attribute to the line buffer.
558 #define STORE_CHAR(ch,a,rep,pos) \
560 if (store_char((ch),(a),(rep),(pos))) return (1); \
564 store_char(ch, a, rep, pos)
574 w = (a & (AT_UNDERLINE|AT_BOLD)); /* Pre-use w. */
581 if (is_hilited(pos, pos+1, 0, &matches))
584 * This character should be highlighted.
585 * Override the attribute passed in.
593 if (ctldisp == OPT_ONPLUS && in_ansi_esc_seq())
595 if (!is_ansi_end(ch) && !is_ansi_middle(ch)) {
596 /* Remove whole unrecognized sequence. */
597 char *p = &linebuf[curr];
600 bch = step_char(&p, -1, linebuf);
601 } while (p > linebuf && !IS_CSI_START(bch));
605 a = AT_ANSI; /* Will force re-AT_'ing around it. */
608 else if (ctldisp == OPT_ONPLUS && IS_CSI_START(ch))
610 a = AT_ANSI; /* Will force re-AT_'ing around it. */
615 char *p = &linebuf[curr];
616 LWCHAR prev_ch = step_char(&p, -1, linebuf);
617 w = pwidth(ch, a, prev_ch);
620 if (ctldisp != OPT_ON && column + w + attr_ewidth(a) > sc_width)
622 * Won't fit on screen.
633 replen = utf_len(rep[0]);
635 if (curr + replen >= size_linebuf-6)
638 * Won't fit in line buffer.
641 if (expand_linebuf())
647 linebuf[curr] = *rep++;
656 * Append a tab to the line buffer.
657 * Store spaces to represent the tab.
659 #define STORE_TAB(a,pos) \
660 do { if (store_tab((a),(pos))) return (1); } while (0)
667 int to_tab = column + cshift - lmargin;
670 if (ntabstops < 2 || to_tab >= tabstops[ntabstops-1])
671 to_tab = tabdefault -
672 ((to_tab - tabstops[ntabstops-1]) % tabdefault);
675 for (i = ntabstops - 2; i >= 0; i--)
676 if (to_tab >= tabstops[i])
678 to_tab = tabstops[i+1] - to_tab;
681 if (column + to_tab - 1 + pwidth(' ', attr, 0) + attr_ewidth(attr) > sc_width)
685 STORE_CHAR(' ', attr, " ", pos);
686 } while (--to_tab > 0);
690 #define STORE_PRCHAR(c, pos) \
691 do { if (store_prchar((c), (pos))) return 1; } while (0)
701 * Convert to printable representation.
706 * Make sure we can get the entire representation
707 * of the character on this line.
709 if (column + (int) strlen(s) - 1 +
710 pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
713 for ( ; *s != 0; s++)
714 STORE_CHAR(*s, AT_BINARY, NULL, pos);
725 for (i = 0; i < mbc_buf_index; i++)
726 if (store_prchar(mbc_buf[i], pos))
727 return mbc_buf_index - i;
733 * Append a character to the line buffer.
734 * Expand tabs into spaces, handle underlining, boldfacing, etc.
735 * Returns 0 if ok, 1 if couldn't fit in buffer.
746 if (do_append(pendc, NULL, pendpos))
748 * Oops. We've probably lost the char which
749 * was in pendc, since caller won't back up.
755 if (c == '\r' && bs_mode == BS_SPECIAL)
757 if (mbc_buf_len > 0) /* utf_mode must be on. */
759 /* Flush incomplete (truncated) sequence. */
760 r = flush_mbc_buf(mbc_pos);
761 mbc_buf_index = r + 1;
764 return (mbc_buf_index);
768 * Don't put the CR into the buffer until we see
769 * the next char. If the next char is a newline,
779 r = do_append((LWCHAR) c, NULL, pos);
782 /* Perform strict validation in all possible cases. */
783 if (mbc_buf_len == 0)
788 if (IS_ASCII_OCTET(c))
789 r = do_append((LWCHAR) c, NULL, pos);
790 else if (IS_UTF8_LEAD(c))
792 mbc_buf_len = utf_len(c);
796 /* UTF8_INVALID or stray UTF8_TRAIL */
797 r = flush_mbc_buf(pos);
798 } else if (IS_UTF8_TRAIL(c))
800 mbc_buf[mbc_buf_index++] = c;
801 if (mbc_buf_index < mbc_buf_len)
803 if (is_utf8_well_formed(mbc_buf))
804 r = do_append(get_wchar(mbc_buf), mbc_buf, mbc_pos);
806 /* Complete, but not shortest form, sequence. */
807 mbc_buf_index = r = flush_mbc_buf(mbc_pos);
811 /* Flush incomplete (truncated) sequence. */
812 r = flush_mbc_buf(mbc_pos);
813 mbc_buf_index = r + 1;
815 /* Handle new char. */
822 * If we need to shift the line, do it.
823 * But wait until we get to at least the middle of the screen,
824 * so shifting it doesn't affect the chars we're currently
825 * pappending. (Bold & underline can get messed up otherwise.)
827 if (cshift < hshift && column > sc_width / 2)
829 linebuf[curr] = '\0';
830 pshift(hshift - cshift);
834 /* How many chars should caller back up? */
835 r = (!utf_mode) ? 1 : mbc_buf_index;
841 do_append(ch, rep, pos)
853 if (bs_mode == BS_CONTROL)
854 goto do_control_char;
857 * A better test is needed here so we don't
858 * backspace over part of the printed
859 * representation of a binary character.
863 || (attr[curr - 1] & (AT_ANSI|AT_BINARY)))
864 STORE_PRCHAR('\b', pos);
865 else if (bs_mode == BS_NORMAL)
866 STORE_CHAR(ch, AT_NORMAL, NULL, pos);
867 else if (bs_mode == BS_SPECIAL)
868 overstrike = backc();
876 * Overstrike the character at the current position
877 * in the line buffer. This will cause either
878 * underline (if a "_" is overstruck),
879 * bold (if an identical character is overstruck),
880 * or just deletion of the character in the buffer.
882 overstrike = utf_mode ? -1 : 0;
883 /* To be correct, this must be a base character. */
884 prev_ch = get_wchar(linebuf + curr);
889 * Overstriking a char with itself means make it bold.
890 * But overstriking an underscore with itself is
891 * ambiguous. It could mean make it bold, or
892 * it could mean make it underlined.
893 * Use the previous overstrike to resolve it.
897 if ((a & (AT_BOLD|AT_UNDERLINE)) != AT_NORMAL)
898 a |= (AT_BOLD|AT_UNDERLINE);
899 else if (last_overstrike != AT_NORMAL)
900 a |= last_overstrike;
905 } else if (ch == '_')
909 rep = linebuf + curr;
910 } else if (prev_ch == '_')
914 /* Else we replace prev_ch, but we keep its attributes. */
915 } else if (overstrike < 0)
917 if ( is_composing_char(ch)
918 || is_combining_char(get_wchar(linebuf + curr), ch))
919 /* Continuation of the same overstrike. */
928 * Expand a tab into spaces.
933 goto do_control_char;
939 } else if ((!utf_mode || is_ascii_char(ch)) && control_char((char)ch))
942 if (ctldisp == OPT_ON || (ctldisp == OPT_ONPLUS && IS_CSI_START(ch)))
945 * Output as a normal character.
947 STORE_CHAR(ch, AT_NORMAL, rep, pos);
950 STORE_PRCHAR((char) ch, pos);
952 } else if (utf_mode && ctldisp != OPT_ON && is_ubin_char(ch))
958 if (column + (int) strlen(s) - 1 +
959 pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
962 for ( ; *s != 0; s++)
963 STORE_CHAR(*s, AT_BINARY, NULL, pos);
966 STORE_CHAR(ch, a, rep, pos);
981 /* Flush incomplete (truncated) sequence. */
982 r = flush_mbc_buf(mbc_pos);
989 * Terminate the line in the line buffer.
998 if (pendc && (pendc != '\r' || !endline))
1000 * If we had a pending character, put it in the buffer.
1001 * But discard a pending CR if we are at end of line
1002 * (that is, discard the CR in a CR/LF sequence).
1004 (void) do_append(pendc, NULL, pendpos);
1007 * Make sure we've shifted the line, if we need to.
1009 if (cshift < hshift)
1010 pshift(hshift - cshift);
1012 if (ctldisp == OPT_ONPLUS && is_ansi_end('m'))
1014 /* Switch to normal attribute at end of line. */
1016 for ( ; *p != '\0'; p++)
1019 attr[curr++] = AT_ANSI;
1024 * Add a newline if necessary,
1025 * and append a '\0' to the end of the line.
1026 * We output a newline if we're not at the right edge of the screen,
1027 * or if the terminal doesn't auto wrap,
1028 * or if this is really the end of the line AND the terminal ignores
1029 * a newline at the right edge.
1030 * (In the last case we don't want to output a newline if the terminal
1031 * doesn't ignore it since that would produce an extra blank line.
1032 * But we do want to output a newline if the terminal ignores it in case
1033 * the next line is blank. In that case the single newline output for
1034 * that blank line would be ignored!)
1036 if (column < sc_width || !auto_wrap || (endline && ignaw) || ctldisp == OPT_ON)
1038 linebuf[curr] = '\n';
1039 attr[curr] = AT_NORMAL;
1042 else if (ignaw && column >= sc_width && forw)
1045 * Terminals with "ignaw" don't wrap until they *really* need
1046 * to, i.e. when the character *after* the last one to fit on a
1047 * line is output. But they are too hard to deal with when they
1048 * get in the state where a full screen width of characters
1049 * have been output but the cursor is sitting on the right edge
1050 * instead of at the start of the next line.
1051 * So we nudge them into wrapping by outputting a space
1052 * character plus a backspace. But do this only if moving
1053 * forward; if we're moving backward and drawing this line at
1054 * the top of the screen, the space would overwrite the first
1055 * char on the next line. We don't need to do this "nudge"
1056 * at the top of the screen anyway.
1058 linebuf[curr] = ' ';
1059 attr[curr++] = AT_NORMAL;
1060 linebuf[curr] = '\b';
1061 attr[curr++] = AT_NORMAL;
1063 linebuf[curr] = '\0';
1064 attr[curr] = AT_NORMAL;
1075 attr[0] = AT_NORMAL|AT_HILITE;
1079 * Get a character from the current line.
1080 * Return the character as the function return value,
1081 * and the character attribute in *ap.
1091 * If there is no current line, we pretend the line is
1092 * either "~" or "", depending on the "twiddle" flag.
1103 /* Make sure we're back to AT_NORMAL before the '\n'. */
1105 return i ? '\0' : '\n';
1109 return (linebuf[i] & 0xFF);
1113 * Indicate that there is no current line.
1123 * Analogous to forw_line(), but deals with "raw lines":
1124 * lines which are not split for screen width.
1125 * {{ This is supposed to be more efficient than forw_line(). }}
1128 forw_raw_line(curr_pos, linep, line_lenp)
1137 if (curr_pos == NULL_POSITION || ch_seek(curr_pos) ||
1138 (c = ch_forw_get()) == EOI)
1139 return (NULL_POSITION);
1144 if (c == '\n' || c == EOI || ABORT_SIGS())
1146 new_pos = ch_tell();
1149 if (n >= size_linebuf-1)
1151 if (expand_linebuf())
1154 * Overflowed the input buffer.
1155 * Pretend the line ended here.
1157 new_pos = ch_tell() - 1;
1167 if (line_lenp != NULL)
1173 * Analogous to back_line(), but deals with "raw lines".
1174 * {{ This is supposed to be more efficient than back_line(). }}
1177 back_raw_line(curr_pos, linep, line_lenp)
1186 if (curr_pos == NULL_POSITION || curr_pos <= ch_zero() ||
1187 ch_seek(curr_pos-1))
1188 return (NULL_POSITION);
1191 linebuf[--n] = '\0';
1195 if (c == '\n' || ABORT_SIGS())
1198 * This is the newline ending the previous line.
1199 * We have hit the beginning of the line.
1201 new_pos = ch_tell() + 1;
1207 * We have hit the beginning of the file.
1208 * This must be the first line in the file.
1209 * This must, of course, be the beginning of the line.
1211 new_pos = ch_zero();
1216 int old_size_linebuf = size_linebuf;
1219 if (expand_linebuf())
1222 * Overflowed the input buffer.
1223 * Pretend the line ended here.
1225 new_pos = ch_tell() + 1;
1229 * Shift the data to the end of the new linebuf.
1231 for (fm = linebuf + old_size_linebuf - 1,
1232 to = linebuf + size_linebuf - 1;
1233 fm >= linebuf; fm--, to--)
1235 n = size_linebuf - old_size_linebuf;
1240 *linep = &linebuf[n];
1241 if (line_lenp != NULL)
1242 *line_lenp = size_linebuf - 1 - n;