Annotation of src/usr.bin/less/line.c, Revision 1.25
1.1 etheisen 1: /*
1.12 shadchin 2: * Copyright (C) 1984-2012 Mark Nudelman
1.14 nicm 3: * Modified for use with illumos by Garrett D'Amore.
4: * Copyright 2014 Garrett D'Amore <garrett@damore.org>
1.1 etheisen 5: *
1.5 millert 6: * You may distribute under the terms of either the GNU General Public
7: * License or the Less License, as specified in the README file.
1.1 etheisen 8: *
1.12 shadchin 9: * For more information, see the README file.
1.13 nicm 10: */
1.1 etheisen 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:
1.23 schwarze 18: #include <wchar.h>
19:
1.19 mmcc 20: #include "charset.h"
1.1 etheisen 21: #include "less.h"
22:
1.10 shadchin 23: static char *linebuf = NULL; /* Buffer which holds the current output line */
1.5 millert 24: static char *attr = NULL; /* Extension of linebuf to hold attributes */
1.13 nicm 25: int size_linebuf = 0; /* Size of line buffer (and attr buffer) */
1.5 millert 26:
1.10 shadchin 27: static int cshift; /* Current left-shift of output line buffer */
1.13 nicm 28: int hshift; /* Desired left-shift of output line buffer */
29: int tabstops[TABSTOP_MAX] = { 0 }; /* Custom tabstops */
30: int ntabstops = 1; /* Number of tabstops */
31: int tabdefault = 8; /* Default repeated tabstops */
32: off_t highest_hilite; /* Pos of last hilite in file found so far */
1.1 etheisen 33:
34: static int curr; /* Index into linebuf */
1.13 nicm 35: static int column; /* Printable length, accounting for backspaces, etc. */
1.1 etheisen 36: static int overstrike; /* Next char should overstrike previous char */
37: static int is_null_line; /* There is no current line */
1.5 millert 38: static int lmargin; /* Left margin */
1.1 etheisen 39: static char pendc;
1.13 nicm 40: static off_t pendpos;
1.5 millert 41: static char *end_ansi_chars;
1.10 shadchin 42: static char *mid_ansi_chars;
1.1 etheisen 43:
1.13 nicm 44: static int attr_swidth(int);
45: static int attr_ewidth(int);
46: static int do_append(LWCHAR, char *, off_t);
1.1 etheisen 47:
1.11 millert 48: extern volatile sig_atomic_t sigs;
1.1 etheisen 49: extern int bs_mode;
50: extern int linenums;
51: extern int ctldisp;
52: extern int twiddle;
53: extern int binattr;
1.5 millert 54: extern int status_col;
1.1 etheisen 55: extern int auto_wrap, ignaw;
56: extern int bo_s_width, bo_e_width;
57: extern int ul_s_width, ul_e_width;
58: extern int bl_s_width, bl_e_width;
59: extern int so_s_width, so_e_width;
60: extern int sc_width, sc_height;
1.5 millert 61: extern int utf_mode;
1.13 nicm 62: extern off_t start_attnpos;
63: extern off_t end_attnpos;
1.5 millert 64:
1.10 shadchin 65: static char mbc_buf[MAX_UTF_CHAR_LEN];
66: static int mbc_buf_len = 0;
67: static int mbc_buf_index = 0;
1.13 nicm 68: static off_t mbc_pos;
1.10 shadchin 69:
1.5 millert 70: /*
71: * Initialize from environment variables.
72: */
1.13 nicm 73: void
74: init_line(void)
1.5 millert 75: {
76: end_ansi_chars = lgetenv("LESSANSIENDCHARS");
77: if (end_ansi_chars == NULL || *end_ansi_chars == '\0')
78: end_ansi_chars = "m";
1.10 shadchin 79:
80: mid_ansi_chars = lgetenv("LESSANSIMIDCHARS");
81: if (mid_ansi_chars == NULL || *mid_ansi_chars == '\0')
82: mid_ansi_chars = "0123456789;[?!\"'#%()*+ ";
83:
1.13 nicm 84: linebuf = ecalloc(LINEBUF_SIZE, sizeof (char));
85: attr = ecalloc(LINEBUF_SIZE, sizeof (char));
1.5 millert 86: size_linebuf = LINEBUF_SIZE;
87: }
88:
89: /*
90: * Expand the line buffer.
91: */
1.13 nicm 92: static int
93: expand_linebuf(void)
1.5 millert 94: {
1.10 shadchin 95: /* Double the size of the line buffer. */
1.7 millert 96: int new_size = size_linebuf * 2;
1.10 shadchin 97:
98: /* Just realloc to expand the buffer, if we can. */
1.22 deraadt 99: char *new_buf = recallocarray(linebuf, size_linebuf, new_size, 1);
100: char *new_attr = recallocarray(attr, size_linebuf, new_size, 1);
1.13 nicm 101: if (new_buf == NULL || new_attr == NULL) {
1.15 mmcc 102: free(new_attr);
103: free(new_buf);
1.13 nicm 104: return (1);
1.5 millert 105: }
106: linebuf = new_buf;
107: attr = new_attr;
108: size_linebuf = new_size;
1.13 nicm 109: return (0);
1.5 millert 110: }
1.1 etheisen 111:
112: /*
1.10 shadchin 113: * Is a character ASCII?
114: */
1.25 ! schwarze 115: static int
1.13 nicm 116: is_ascii_char(LWCHAR ch)
1.10 shadchin 117: {
118: return (ch <= 0x7F);
119: }
120:
121: /*
1.1 etheisen 122: * Rewind the line buffer.
123: */
1.13 nicm 124: void
125: prewind(void)
1.1 etheisen 126: {
127: curr = 0;
128: column = 0;
1.10 shadchin 129: cshift = 0;
1.1 etheisen 130: overstrike = 0;
1.10 shadchin 131: mbc_buf_len = 0;
1.1 etheisen 132: is_null_line = 0;
133: pendc = '\0';
1.5 millert 134: lmargin = 0;
135: if (status_col)
136: lmargin += 1;
1.1 etheisen 137: }
138:
139: /*
140: * Insert the line number (of the given position) into the line buffer.
141: */
1.13 nicm 142: void
143: plinenum(off_t pos)
1.1 etheisen 144: {
1.16 mmcc 145: off_t linenum = 0;
1.13 nicm 146: int i;
1.5 millert 147:
1.13 nicm 148: if (linenums == OPT_ONPLUS) {
1.5 millert 149: /*
150: * Get the line number and put it in the current line.
151: * {{ Note: since find_linenum calls forw_raw_line,
1.13 nicm 152: * it may seek in the input file, requiring the caller
1.5 millert 153: * of plinenum to re-seek if necessary. }}
154: * {{ Since forw_raw_line modifies linebuf, we must
155: * do this first, before storing anything in linebuf. }}
156: */
157: linenum = find_linenum(pos);
158: }
1.1 etheisen 159:
160: /*
1.5 millert 161: * Display a status column if the -J option is set.
1.1 etheisen 162: */
1.13 nicm 163: if (status_col) {
1.5 millert 164: linebuf[curr] = ' ';
1.13 nicm 165: if (start_attnpos != -1 &&
1.5 millert 166: pos >= start_attnpos && pos < end_attnpos)
1.10 shadchin 167: attr[curr] = AT_NORMAL|AT_HILITE;
1.5 millert 168: else
1.10 shadchin 169: attr[curr] = AT_NORMAL;
1.5 millert 170: curr++;
171: column++;
172: }
1.1 etheisen 173: /*
1.5 millert 174: * Display the line number at the start of each line
175: * if the -N option is set.
1.1 etheisen 176: */
1.13 nicm 177: if (linenums == OPT_ONPLUS) {
1.18 mmcc 178: char buf[23];
1.5 millert 179: int n;
1.1 etheisen 180:
1.17 mmcc 181: postoa(linenum, buf, sizeof(buf));
1.5 millert 182: n = strlen(buf);
183: if (n < MIN_LINENUM_WIDTH)
184: n = MIN_LINENUM_WIDTH;
1.6 millert 185: snprintf(linebuf+curr, size_linebuf-curr, "%*s ", n, buf);
1.21 deraadt 186: n++; /* One space after the line number. */
1.5 millert 187: for (i = 0; i < n; i++)
188: attr[curr+i] = AT_NORMAL;
189: curr += n;
190: column += n;
191: lmargin += n;
192: }
1.1 etheisen 193:
194: /*
1.5 millert 195: * Append enough spaces to bring us to the lmargin.
1.1 etheisen 196: */
1.13 nicm 197: while (column < lmargin) {
1.1 etheisen 198: linebuf[curr] = ' ';
199: attr[curr++] = AT_NORMAL;
200: column++;
1.5 millert 201: }
202: }
203:
204: /*
1.10 shadchin 205: * Shift the input line left.
206: * This means discarding N printable chars at the start of the buffer.
1.5 millert 207: */
1.13 nicm 208: static void
209: pshift(int shift)
1.10 shadchin 210: {
211: LWCHAR prev_ch = 0;
212: unsigned char c;
213: int shifted = 0;
214: int to;
215: int from;
1.5 millert 216: int len;
1.10 shadchin 217: int width;
218: int prev_attr;
219: int next_attr;
220:
221: if (shift > column - lmargin)
222: shift = column - lmargin;
223: if (shift > curr - lmargin)
224: shift = curr - lmargin;
1.5 millert 225:
1.10 shadchin 226: to = from = lmargin;
1.5 millert 227: /*
1.10 shadchin 228: * We keep on going when shifted == shift
229: * to get all combining chars.
1.5 millert 230: */
1.13 nicm 231: while (shifted <= shift && from < curr) {
1.10 shadchin 232: c = linebuf[from];
1.24 schwarze 233: if (ctldisp == OPT_ONPLUS && c == ESC) {
1.10 shadchin 234: /* Keep cumulative effect. */
235: linebuf[to] = c;
236: attr[to++] = attr[from++];
1.13 nicm 237: while (from < curr && linebuf[from]) {
1.10 shadchin 238: linebuf[to] = linebuf[from];
239: attr[to++] = attr[from];
240: if (!is_ansi_middle(linebuf[from++]))
241: break;
1.13 nicm 242: }
1.10 shadchin 243: continue;
244: }
245:
246: width = 0;
247:
1.13 nicm 248: if (!IS_ASCII_OCTET(c) && utf_mode) {
1.10 shadchin 249: /* Assumes well-formedness validation already done. */
250: LWCHAR ch;
251:
252: len = utf_len(c);
253: if (from + len > curr)
254: break;
255: ch = get_wchar(linebuf + from);
1.13 nicm 256: if (!is_composing_char(ch) &&
257: !is_combining_char(prev_ch, ch))
1.10 shadchin 258: width = is_wide_char(ch) ? 2 : 1;
259: prev_ch = ch;
1.13 nicm 260: } else {
1.10 shadchin 261: len = 1;
262: if (c == '\b')
263: /* XXX - Incorrect if several '\b' in a row. */
1.13 nicm 264: width = (utf_mode && is_wide_char(prev_ch)) ?
265: -2 : -1;
1.10 shadchin 266: else if (!control_char(c))
267: width = 1;
268: prev_ch = 0;
269: }
270:
271: if (width == 2 && shift - shifted == 1) {
272: /* Should never happen when called by pshift_all(). */
273: attr[to] = attr[from];
274: /*
275: * Assume a wide_char will never be the first half of a
276: * combining_char pair, so reset prev_ch in case we're
277: * followed by a '\b'.
278: */
279: prev_ch = linebuf[to++] = ' ';
280: from += len;
281: shifted++;
282: continue;
283: }
284:
285: /* Adjust width for magic cookies. */
286: prev_attr = (to > 0) ? attr[to-1] : AT_NORMAL;
287: next_attr = (from + len < curr) ? attr[from + len] : prev_attr;
1.13 nicm 288: if (!is_at_equiv(attr[from], prev_attr) &&
289: !is_at_equiv(attr[from], next_attr)) {
1.10 shadchin 290: width += attr_swidth(attr[from]);
291: if (from + len < curr)
292: width += attr_ewidth(attr[from]);
1.13 nicm 293: if (is_at_equiv(prev_attr, next_attr)) {
1.10 shadchin 294: width += attr_ewidth(prev_attr);
295: if (from + len < curr)
296: width += attr_swidth(next_attr);
1.5 millert 297: }
298: }
299:
1.10 shadchin 300: if (shift - shifted < width)
301: break;
302: from += len;
303: shifted += width;
304: if (shifted < 0)
305: shifted = 0;
306: }
1.13 nicm 307: while (from < curr) {
1.10 shadchin 308: linebuf[to] = linebuf[from];
309: attr[to++] = attr[from++];
310: }
311: curr = to;
312: column -= shifted;
313: cshift += shifted;
1.5 millert 314: }
315:
316: /*
1.10 shadchin 317: *
1.5 millert 318: */
1.13 nicm 319: void
320: pshift_all(void)
1.5 millert 321: {
1.10 shadchin 322: pshift(column);
1.1 etheisen 323: }
324:
325: /*
326: * Return the printing width of the start (enter) sequence
327: * for a given character attribute.
328: */
1.13 nicm 329: static int
330: attr_swidth(int a)
1.1 etheisen 331: {
1.10 shadchin 332: int w = 0;
333:
334: a = apply_at_specials(a);
335:
336: if (a & AT_UNDERLINE)
337: w += ul_s_width;
338: if (a & AT_BOLD)
339: w += bo_s_width;
340: if (a & AT_BLINK)
341: w += bl_s_width;
342: if (a & AT_STANDOUT)
343: w += so_s_width;
344:
1.13 nicm 345: return (w);
1.1 etheisen 346: }
347:
348: /*
349: * Return the printing width of the end (exit) sequence
350: * for a given character attribute.
351: */
1.13 nicm 352: static int
353: attr_ewidth(int a)
1.1 etheisen 354: {
1.10 shadchin 355: int w = 0;
356:
357: a = apply_at_specials(a);
358:
359: if (a & AT_UNDERLINE)
360: w += ul_e_width;
361: if (a & AT_BOLD)
362: w += bo_e_width;
363: if (a & AT_BLINK)
364: w += bl_e_width;
365: if (a & AT_STANDOUT)
366: w += so_e_width;
367:
1.13 nicm 368: return (w);
1.1 etheisen 369: }
370:
371: /*
372: * Return the printing width of a given character and attribute,
373: * if the character were added to the current position in the line buffer.
374: * Adding a character with a given attribute may cause an enter or exit
375: * attribute sequence to be inserted, so this must be taken into account.
376: */
1.13 nicm 377: static int
1.23 schwarze 378: pwidth(wchar_t ch, int a, wchar_t prev_ch)
1.1 etheisen 379: {
1.10 shadchin 380: int w;
1.1 etheisen 381:
1.23 schwarze 382: /*
383: * In case of a backspace, back up by the width of the previous
384: * character. If that is non-printable (for example another
385: * backspace) or zero width (for example a combining accent),
386: * the terminal may actually back up to a character even further
387: * back, but we no longer know how wide that may have been.
388: * The best guess possible at this point is that it was
389: * hopefully width one.
390: */
391: if (ch == L'\b') {
392: w = wcwidth(prev_ch);
393: if (w <= 0)
394: w = 1;
395: return (-w);
396: }
397:
398: w = wcwidth(ch);
399:
400: /*
401: * Non-printable characters can get here if the -r flag is in
402: * effect, and possibly in some other situations (XXX check that!).
403: * Treat them as zero width.
404: * That may not always match their actual behaviour,
405: * but there is no reasonable way to be more exact.
406: */
407: if (w == -1)
408: w = 0;
1.13 nicm 409:
1.23 schwarze 410: /*
411: * Combining accents take up no space.
412: * Some terminals, upon failure to compose them with the
413: * characters that precede them, will actually take up one column
414: * for the combining accent; there isn't much we could do short
415: * of testing the (complex) composition process ourselves and
416: * printing a binary representation when it fails.
417: */
418: if (w == 0)
419: return (0);
1.1 etheisen 420:
421: /*
1.10 shadchin 422: * Other characters take one or two columns,
1.1 etheisen 423: * plus the width of any attribute enter/exit sequence.
424: */
1.10 shadchin 425: if (curr > 0 && !is_at_equiv(attr[curr-1], a))
1.1 etheisen 426: w += attr_ewidth(attr[curr-1]);
1.10 shadchin 427: if ((apply_at_specials(a) != AT_NORMAL) &&
428: (curr == 0 || !is_at_equiv(attr[curr-1], a)))
1.1 etheisen 429: w += attr_swidth(a);
430: return (w);
431: }
432:
433: /*
1.10 shadchin 434: * Delete to the previous base character in the line buffer.
435: * Return 1 if one is found.
1.1 etheisen 436: */
1.13 nicm 437: static int
438: backc(void)
1.1 etheisen 439: {
1.10 shadchin 440: LWCHAR prev_ch;
441: char *p = linebuf + curr;
442: LWCHAR ch = step_char(&p, -1, linebuf + lmargin);
443: int width;
444:
445: /* This assumes that there is no '\b' in linebuf. */
1.13 nicm 446: while (curr > lmargin && column > lmargin &&
447: (!(attr[curr - 1] & (AT_ANSI|AT_BINARY)))) {
1.10 shadchin 448: curr = p - linebuf;
449: prev_ch = step_char(&p, -1, linebuf + lmargin);
450: width = pwidth(ch, attr[curr], prev_ch);
451: column -= width;
452: if (width > 0)
1.13 nicm 453: return (1);
1.10 shadchin 454: ch = prev_ch;
455: }
456:
1.13 nicm 457: return (0);
1.1 etheisen 458: }
459:
460: /*
1.25 ! schwarze 461: * Is a character the end of an ANSI escape sequence?
1.5 millert 462: */
1.13 nicm 463: static int
464: is_ansi_end(LWCHAR ch)
1.10 shadchin 465: {
466: if (!is_ascii_char(ch))
467: return (0);
1.13 nicm 468: return (strchr(end_ansi_chars, (char)ch) != NULL);
1.10 shadchin 469: }
470:
471: /*
472: *
473: */
1.13 nicm 474: int
475: is_ansi_middle(LWCHAR ch)
1.5 millert 476: {
1.10 shadchin 477: if (!is_ascii_char(ch))
478: return (0);
479: if (is_ansi_end(ch))
480: return (0);
1.13 nicm 481: return (strchr(mid_ansi_chars, (char)ch) != NULL);
1.5 millert 482: }
483:
484: /*
1.1 etheisen 485: * Append a character and attribute to the line buffer.
486: */
1.13 nicm 487: #define STORE_CHAR(ch, a, rep, pos) \
488: if (store_char((ch), (a), (rep), (pos))) \
489: return (1)
490:
491: static int
492: store_char(LWCHAR ch, char a, char *rep, off_t pos)
1.1 etheisen 493: {
1.25 ! schwarze 494: int i;
1.10 shadchin 495: int w;
496: int replen;
497: char cs;
1.13 nicm 498: int matches;
1.10 shadchin 499:
1.13 nicm 500: if (is_hilited(pos, pos+1, 0, &matches)) {
501: /*
502: * This character should be highlighted.
503: * Override the attribute passed in.
504: */
505: if (a != AT_ANSI) {
506: if (highest_hilite != -1 && pos > highest_hilite)
507: highest_hilite = pos;
508: a |= AT_HILITE;
1.10 shadchin 509: }
1.5 millert 510: }
1.10 shadchin 511:
1.25 ! schwarze 512: w = -1;
! 513: if (ctldisp == OPT_ONPLUS) {
! 514: /*
! 515: * Set i to the beginning of an ANSI escape sequence
! 516: * that was begun and not yet ended, or to -1 otherwise.
! 517: */
! 518: for (i = curr - 1; i >= 0; i--) {
! 519: if (linebuf[i] == ESC)
! 520: break;
! 521: if (!is_ansi_middle(linebuf[i]))
! 522: i = 0;
! 523: }
! 524: if (i >= 0 && !is_ansi_end(ch) && !is_ansi_middle(ch)) {
1.10 shadchin 525: /* Remove whole unrecognized sequence. */
1.25 ! schwarze 526: curr = i;
1.13 nicm 527: return (0);
1.10 shadchin 528: }
1.25 ! schwarze 529: if (i >= 0 || ch == ESC) {
! 530: a = AT_ANSI; /* Will force re-AT_'ing around it. */
! 531: w = 0;
! 532: }
! 533: }
! 534: if (w == -1) {
! 535: wchar_t prev_ch;
! 536:
! 537: if (utf_mode) {
! 538: for (i = curr - 1; i >= 0; i--)
! 539: if (!IS_UTF8_TRAIL(linebuf[i]))
! 540: break;
! 541: if (i >= 0) {
! 542: w = mbtowc(&prev_ch, linebuf + i, curr - i);
! 543: if (w == -1 || i + w < curr)
! 544: prev_ch = L' ';
! 545: } else
! 546: prev_ch = L' ';
! 547: } else
! 548: prev_ch = curr > 0 ? linebuf[curr - 1] : L' ';
1.10 shadchin 549: w = pwidth(ch, a, prev_ch);
550: }
551:
1.5 millert 552: if (ctldisp != OPT_ON && column + w + attr_ewidth(a) > sc_width)
1.1 etheisen 553: /*
554: * Won't fit on screen.
555: */
556: return (1);
557:
1.13 nicm 558: if (rep == NULL) {
559: cs = (char)ch;
1.10 shadchin 560: rep = &cs;
561: replen = 1;
1.13 nicm 562: } else {
1.10 shadchin 563: replen = utf_len(rep[0]);
564: }
1.13 nicm 565: if (curr + replen >= size_linebuf-6) {
1.1 etheisen 566: /*
567: * Won't fit in line buffer.
1.5 millert 568: * Try to expand it.
1.1 etheisen 569: */
1.5 millert 570: if (expand_linebuf())
571: return (1);
572: }
1.1 etheisen 573:
1.13 nicm 574: while (replen-- > 0) {
1.10 shadchin 575: linebuf[curr] = *rep++;
576: attr[curr] = a;
577: curr++;
1.1 etheisen 578: }
579: column += w;
580: return (0);
581: }
582:
583: /*
1.5 millert 584: * Append a tab to the line buffer.
585: * Store spaces to represent the tab.
586: */
1.13 nicm 587: #define STORE_TAB(a, pos) \
588: if (store_tab((a), (pos))) \
589: return (1)
1.5 millert 590:
1.13 nicm 591: static int
592: store_tab(int attr, off_t pos)
1.5 millert 593: {
594: int to_tab = column + cshift - lmargin;
595: int i;
596:
597: if (ntabstops < 2 || to_tab >= tabstops[ntabstops-1])
598: to_tab = tabdefault -
1.13 nicm 599: ((to_tab - tabstops[ntabstops-1]) % tabdefault);
600: else {
1.21 deraadt 601: for (i = ntabstops - 2; i >= 0; i--)
1.5 millert 602: if (to_tab >= tabstops[i])
603: break;
604: to_tab = tabstops[i+1] - to_tab;
605: }
606:
1.13 nicm 607: if (column + to_tab - 1 + pwidth(' ', attr, 0) +
608: attr_ewidth(attr) > sc_width)
609: return (1);
1.10 shadchin 610:
1.5 millert 611: do {
1.10 shadchin 612: STORE_CHAR(' ', attr, " ", pos);
1.5 millert 613: } while (--to_tab > 0);
1.13 nicm 614: return (0);
1.5 millert 615: }
616:
1.13 nicm 617: #define STORE_PRCHAR(c, pos) \
618: if (store_prchar((c), (pos))) \
619: return (1)
1.10 shadchin 620:
1.13 nicm 621: static int
622: store_prchar(char c, off_t pos)
1.10 shadchin 623: {
624: char *s;
625:
626: /*
627: * Convert to printable representation.
628: */
629: s = prchar(c);
630:
631: /*
632: * Make sure we can get the entire representation
633: * of the character on this line.
634: */
1.13 nicm 635: if (column + (int)strlen(s) - 1 +
636: pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
637: return (1);
1.10 shadchin 638:
1.13 nicm 639: for (; *s != 0; s++) {
1.10 shadchin 640: STORE_CHAR(*s, AT_BINARY, NULL, pos);
1.13 nicm 641: }
642: return (0);
1.10 shadchin 643: }
644:
1.13 nicm 645: static int
646: flush_mbc_buf(off_t pos)
1.10 shadchin 647: {
648: int i;
649:
650: for (i = 0; i < mbc_buf_index; i++)
651: if (store_prchar(mbc_buf[i], pos))
1.13 nicm 652: return (mbc_buf_index - i);
1.10 shadchin 653:
1.13 nicm 654: return (0);
1.10 shadchin 655: }
656:
1.5 millert 657: /*
1.1 etheisen 658: * Append a character to the line buffer.
659: * Expand tabs into spaces, handle underlining, boldfacing, etc.
660: * Returns 0 if ok, 1 if couldn't fit in buffer.
661: */
1.13 nicm 662: int
663: pappend(char c, off_t pos)
1.1 etheisen 664: {
1.5 millert 665: int r;
666:
1.13 nicm 667: if (pendc) {
1.10 shadchin 668: if (do_append(pendc, NULL, pendpos))
1.1 etheisen 669: /*
670: * Oops. We've probably lost the char which
671: * was in pendc, since caller won't back up.
672: */
673: return (1);
674: pendc = '\0';
675: }
676:
1.13 nicm 677: if (c == '\r' && bs_mode == BS_SPECIAL) {
678: if (mbc_buf_len > 0) /* utf_mode must be on. */ {
1.10 shadchin 679: /* Flush incomplete (truncated) sequence. */
680: r = flush_mbc_buf(mbc_pos);
681: mbc_buf_index = r + 1;
682: mbc_buf_len = 0;
683: if (r)
684: return (mbc_buf_index);
685: }
686:
1.1 etheisen 687: /*
1.13 nicm 688: * Don't put the CR into the buffer until we see
1.1 etheisen 689: * the next char. If the next char is a newline,
690: * discard the CR.
691: */
692: pendc = c;
693: pendpos = pos;
694: return (0);
695: }
696:
1.13 nicm 697: if (!utf_mode) {
1.10 shadchin 698: r = do_append((LWCHAR) c, NULL, pos);
1.13 nicm 699: } else {
1.10 shadchin 700: /* Perform strict validation in all possible cases. */
1.13 nicm 701: if (mbc_buf_len == 0) {
702: retry:
1.10 shadchin 703: mbc_buf_index = 1;
704: *mbc_buf = c;
1.13 nicm 705: if (IS_ASCII_OCTET(c)) {
1.10 shadchin 706: r = do_append((LWCHAR) c, NULL, pos);
1.13 nicm 707: } else if (IS_UTF8_LEAD(c)) {
1.10 shadchin 708: mbc_buf_len = utf_len(c);
709: mbc_pos = pos;
710: return (0);
1.13 nicm 711: } else {
1.10 shadchin 712: /* UTF8_INVALID or stray UTF8_TRAIL */
713: r = flush_mbc_buf(pos);
1.13 nicm 714: }
715: } else if (IS_UTF8_TRAIL(c)) {
1.10 shadchin 716: mbc_buf[mbc_buf_index++] = c;
717: if (mbc_buf_index < mbc_buf_len)
718: return (0);
719: if (is_utf8_well_formed(mbc_buf))
1.13 nicm 720: r = do_append(get_wchar(mbc_buf), mbc_buf,
721: mbc_pos);
1.10 shadchin 722: else
723: /* Complete, but not shortest form, sequence. */
724: mbc_buf_index = r = flush_mbc_buf(mbc_pos);
725: mbc_buf_len = 0;
1.13 nicm 726: } else {
1.10 shadchin 727: /* Flush incomplete (truncated) sequence. */
728: r = flush_mbc_buf(mbc_pos);
729: mbc_buf_index = r + 1;
730: mbc_buf_len = 0;
731: /* Handle new char. */
732: if (!r)
733: goto retry;
1.13 nicm 734: }
1.10 shadchin 735: }
736:
1.5 millert 737: /*
738: * If we need to shift the line, do it.
739: * But wait until we get to at least the middle of the screen,
740: * so shifting it doesn't affect the chars we're currently
741: * pappending. (Bold & underline can get messed up otherwise.)
742: */
1.13 nicm 743: if (cshift < hshift && column > sc_width / 2) {
1.5 millert 744: linebuf[curr] = '\0';
745: pshift(hshift - cshift);
746: }
1.13 nicm 747: if (r) {
1.10 shadchin 748: /* How many chars should caller back up? */
749: r = (!utf_mode) ? 1 : mbc_buf_index;
750: }
1.5 millert 751: return (r);
1.1 etheisen 752: }
753:
1.13 nicm 754: static int
755: do_append(LWCHAR ch, char *rep, off_t pos)
1.1 etheisen 756: {
1.13 nicm 757: int a;
1.10 shadchin 758: LWCHAR prev_ch;
1.1 etheisen 759:
1.10 shadchin 760: a = AT_NORMAL;
1.1 etheisen 761:
1.13 nicm 762: if (ch == '\b') {
1.10 shadchin 763: if (bs_mode == BS_CONTROL)
1.5 millert 764: goto do_control_char;
1.10 shadchin 765:
766: /*
767: * A better test is needed here so we don't
768: * backspace over part of the printed
769: * representation of a binary character.
770: */
1.13 nicm 771: if (curr <= lmargin ||
772: column <= lmargin ||
773: (attr[curr - 1] & (AT_ANSI|AT_BINARY))) {
1.10 shadchin 774: STORE_PRCHAR('\b', pos);
1.13 nicm 775: } else if (bs_mode == BS_NORMAL) {
1.10 shadchin 776: STORE_CHAR(ch, AT_NORMAL, NULL, pos);
1.13 nicm 777: } else if (bs_mode == BS_SPECIAL) {
1.10 shadchin 778: overstrike = backc();
1.13 nicm 779: }
1.10 shadchin 780:
1.13 nicm 781: return (0);
1.10 shadchin 782: }
783:
1.13 nicm 784: if (overstrike > 0) {
1.1 etheisen 785: /*
786: * Overstrike the character at the current position
1.13 nicm 787: * in the line buffer. This will cause either
788: * underline (if a "_" is overstruck),
1.1 etheisen 789: * bold (if an identical character is overstruck),
790: * or just deletion of the character in the buffer.
791: */
1.10 shadchin 792: overstrike = utf_mode ? -1 : 0;
793: /* To be correct, this must be a base character. */
794: prev_ch = get_wchar(linebuf + curr);
795: a = attr[curr];
1.13 nicm 796: if (ch == prev_ch) {
1.5 millert 797: /*
798: * Overstriking a char with itself means make it bold.
799: * But overstriking an underscore with itself is
800: * ambiguous. It could mean make it bold, or
801: * it could mean make it underlined.
802: * Use the previous overstrike to resolve it.
803: */
1.13 nicm 804: if (ch == '_') {
1.10 shadchin 805: if ((a & (AT_BOLD|AT_UNDERLINE)) != AT_NORMAL)
806: a |= (AT_BOLD|AT_UNDERLINE);
1.20 natano 807: else if (curr > 0 && attr[curr - 1] & AT_UNDERLINE)
808: a |= AT_UNDERLINE;
809: else if (curr > 0 && attr[curr - 1] & AT_BOLD)
810: a |= AT_BOLD;
1.10 shadchin 811: else
1.20 natano 812: a |= AT_INDET;
1.13 nicm 813: } else {
1.10 shadchin 814: a |= AT_BOLD;
1.13 nicm 815: }
816: } else if (ch == '_') {
1.10 shadchin 817: a |= AT_UNDERLINE;
818: ch = prev_ch;
819: rep = linebuf + curr;
1.13 nicm 820: } else if (prev_ch == '_') {
1.10 shadchin 821: a |= AT_UNDERLINE;
822: }
823: /* Else we replace prev_ch, but we keep its attributes. */
1.13 nicm 824: } else if (overstrike < 0) {
825: if (is_composing_char(ch) ||
1.20 natano 826: is_combining_char(get_wchar(linebuf + curr), ch)) {
1.10 shadchin 827: /* Continuation of the same overstrike. */
1.20 natano 828: if (curr > 0)
829: a = attr[curr - 1] & (AT_UNDERLINE | AT_BOLD);
830: else
831: a = AT_NORMAL;
832: } else
1.10 shadchin 833: overstrike = 0;
834: }
835:
1.13 nicm 836: if (ch == '\t') {
1.5 millert 837: /*
838: * Expand a tab into spaces.
839: */
1.13 nicm 840: switch (bs_mode) {
1.1 etheisen 841: case BS_CONTROL:
842: goto do_control_char;
1.5 millert 843: case BS_NORMAL:
1.1 etheisen 844: case BS_SPECIAL:
1.10 shadchin 845: STORE_TAB(a, pos);
1.1 etheisen 846: break;
847: }
1.13 nicm 848: } else if ((!utf_mode || is_ascii_char(ch)) && control_char((char)ch)) {
849: do_control_char:
850: if (ctldisp == OPT_ON ||
1.24 schwarze 851: (ctldisp == OPT_ONPLUS && ch == ESC)) {
1.1 etheisen 852: /*
853: * Output as a normal character.
854: */
1.10 shadchin 855: STORE_CHAR(ch, AT_NORMAL, rep, pos);
1.13 nicm 856: } else {
857: STORE_PRCHAR((char)ch, pos);
1.10 shadchin 858: }
1.13 nicm 859: } else if (utf_mode && ctldisp != OPT_ON && is_ubin_char(ch)) {
1.10 shadchin 860: char *s;
861:
862: s = prutfchar(ch);
1.1 etheisen 863:
1.13 nicm 864: if (column + (int)strlen(s) - 1 +
1.10 shadchin 865: pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
866: return (1);
1.1 etheisen 867:
1.13 nicm 868: for (; *s != 0; s++)
1.10 shadchin 869: STORE_CHAR(*s, AT_BINARY, NULL, pos);
1.13 nicm 870: } else {
1.10 shadchin 871: STORE_CHAR(ch, a, rep, pos);
1.1 etheisen 872: }
1.13 nicm 873: return (0);
1.10 shadchin 874: }
875:
876: /*
877: *
878: */
1.13 nicm 879: int
880: pflushmbc(void)
1.10 shadchin 881: {
882: int r = 0;
1.1 etheisen 883:
1.13 nicm 884: if (mbc_buf_len > 0) {
1.10 shadchin 885: /* Flush incomplete (truncated) sequence. */
886: r = flush_mbc_buf(mbc_pos);
887: mbc_buf_len = 0;
888: }
1.13 nicm 889: return (r);
1.1 etheisen 890: }
891:
892: /*
893: * Terminate the line in the line buffer.
894: */
1.13 nicm 895: void
896: pdone(int endline, int forw)
1.1 etheisen 897: {
1.20 natano 898: int i;
899:
1.10 shadchin 900: (void) pflushmbc();
901:
1.1 etheisen 902: if (pendc && (pendc != '\r' || !endline))
903: /*
904: * If we had a pending character, put it in the buffer.
905: * But discard a pending CR if we are at end of line
906: * (that is, discard the CR in a CR/LF sequence).
907: */
1.10 shadchin 908: (void) do_append(pendc, NULL, pendpos);
1.20 natano 909:
910: for (i = curr - 1; i >= 0; i--) {
911: if (attr[i] & AT_INDET) {
912: attr[i] &= ~AT_INDET;
913: if (i < curr - 1 && attr[i + 1] & AT_BOLD)
914: attr[i] |= AT_BOLD;
915: else
916: attr[i] |= AT_UNDERLINE;
917: }
918: }
1.1 etheisen 919:
920: /*
1.5 millert 921: * Make sure we've shifted the line, if we need to.
922: */
923: if (cshift < hshift)
924: pshift(hshift - cshift);
925:
1.13 nicm 926: if (ctldisp == OPT_ONPLUS && is_ansi_end('m')) {
1.10 shadchin 927: /* Switch to normal attribute at end of line. */
928: char *p = "\033[m";
1.13 nicm 929: for (; *p != '\0'; p++) {
1.10 shadchin 930: linebuf[curr] = *p;
931: attr[curr++] = AT_ANSI;
932: }
933: }
934:
1.5 millert 935: /*
1.1 etheisen 936: * Add a newline if necessary,
937: * and append a '\0' to the end of the line.
1.10 shadchin 938: * We output a newline if we're not at the right edge of the screen,
939: * or if the terminal doesn't auto wrap,
940: * or if this is really the end of the line AND the terminal ignores
941: * a newline at the right edge.
1.13 nicm 942: * (In the last case we don't want to output a newline if the terminal
1.10 shadchin 943: * doesn't ignore it since that would produce an extra blank line.
944: * But we do want to output a newline if the terminal ignores it in case
945: * the next line is blank. In that case the single newline output for
946: * that blank line would be ignored!)
1.1 etheisen 947: */
1.13 nicm 948: if (column < sc_width || !auto_wrap || (endline && ignaw) ||
949: ctldisp == OPT_ON) {
1.1 etheisen 950: linebuf[curr] = '\n';
951: attr[curr] = AT_NORMAL;
952: curr++;
1.13 nicm 953: } else if (ignaw && column >= sc_width && forw) {
1.10 shadchin 954: /*
955: * Terminals with "ignaw" don't wrap until they *really* need
956: * to, i.e. when the character *after* the last one to fit on a
957: * line is output. But they are too hard to deal with when they
958: * get in the state where a full screen width of characters
959: * have been output but the cursor is sitting on the right edge
960: * instead of at the start of the next line.
1.13 nicm 961: * So we nudge them into wrapping by outputting a space
962: * character plus a backspace. But do this only if moving
1.10 shadchin 963: * forward; if we're moving backward and drawing this line at
964: * the top of the screen, the space would overwrite the first
1.13 nicm 965: * char on the next line. We don't need to do this "nudge"
1.10 shadchin 966: * at the top of the screen anyway.
967: */
968: linebuf[curr] = ' ';
969: attr[curr++] = AT_NORMAL;
1.13 nicm 970: linebuf[curr] = '\b';
1.10 shadchin 971: attr[curr++] = AT_NORMAL;
1.1 etheisen 972: }
973: linebuf[curr] = '\0';
974: attr[curr] = AT_NORMAL;
1.10 shadchin 975: }
1.5 millert 976:
1.10 shadchin 977: /*
978: *
979: */
1.13 nicm 980: void
981: set_status_col(char c)
1.10 shadchin 982: {
983: linebuf[0] = c;
984: attr[0] = AT_NORMAL|AT_HILITE;
1.1 etheisen 985: }
986:
987: /*
988: * Get a character from the current line.
989: * Return the character as the function return value,
990: * and the character attribute in *ap.
991: */
1.13 nicm 992: int
993: gline(int i, int *ap)
1.1 etheisen 994: {
1.13 nicm 995: if (is_null_line) {
1.1 etheisen 996: /*
997: * If there is no current line, we pretend the line is
998: * either "~" or "", depending on the "twiddle" flag.
999: */
1.13 nicm 1000: if (twiddle) {
1001: if (i == 0) {
1.10 shadchin 1002: *ap = AT_BOLD;
1.13 nicm 1003: return ('~');
1.10 shadchin 1004: }
1005: --i;
1006: }
1007: /* Make sure we're back to AT_NORMAL before the '\n'. */
1008: *ap = AT_NORMAL;
1.13 nicm 1009: return (i ? '\0' : '\n');
1.1 etheisen 1010: }
1011:
1012: *ap = attr[i];
1.10 shadchin 1013: return (linebuf[i] & 0xFF);
1.1 etheisen 1014: }
1015:
1016: /*
1017: * Indicate that there is no current line.
1018: */
1.13 nicm 1019: void
1020: null_line(void)
1.1 etheisen 1021: {
1022: is_null_line = 1;
1.5 millert 1023: cshift = 0;
1.1 etheisen 1024: }
1025:
1026: /*
1027: * Analogous to forw_line(), but deals with "raw lines":
1028: * lines which are not split for screen width.
1029: * {{ This is supposed to be more efficient than forw_line(). }}
1030: */
1.13 nicm 1031: off_t
1032: forw_raw_line(off_t curr_pos, char **linep, int *line_lenp)
1033: {
1034: int n;
1035: int c;
1036: off_t new_pos;
1037:
1038: if (curr_pos == -1 || ch_seek(curr_pos) ||
1039: (c = ch_forw_get()) == EOI)
1040: return (-1);
1.1 etheisen 1041:
1.5 millert 1042: n = 0;
1.13 nicm 1043: for (;;) {
1044: if (c == '\n' || c == EOI || ABORT_SIGS()) {
1.1 etheisen 1045: new_pos = ch_tell();
1046: break;
1047: }
1.13 nicm 1048: if (n >= size_linebuf-1) {
1049: if (expand_linebuf()) {
1.5 millert 1050: /*
1051: * Overflowed the input buffer.
1052: * Pretend the line ended here.
1053: */
1054: new_pos = ch_tell() - 1;
1055: break;
1056: }
1.1 etheisen 1057: }
1.13 nicm 1058: linebuf[n++] = (char)c;
1.1 etheisen 1059: c = ch_forw_get();
1060: }
1.5 millert 1061: linebuf[n] = '\0';
1.1 etheisen 1062: if (linep != NULL)
1063: *linep = linebuf;
1.10 shadchin 1064: if (line_lenp != NULL)
1065: *line_lenp = n;
1.1 etheisen 1066: return (new_pos);
1067: }
1068:
1069: /*
1070: * Analogous to back_line(), but deals with "raw lines".
1071: * {{ This is supposed to be more efficient than back_line(). }}
1072: */
1.13 nicm 1073: off_t
1074: back_raw_line(off_t curr_pos, char **linep, int *line_lenp)
1075: {
1076: int n;
1077: int c;
1078: off_t new_pos;
1079:
1080: if (curr_pos == -1 || curr_pos <= ch_zero() || ch_seek(curr_pos - 1))
1081: return (-1);
1.1 etheisen 1082:
1.5 millert 1083: n = size_linebuf;
1084: linebuf[--n] = '\0';
1.13 nicm 1085: for (;;) {
1.1 etheisen 1086: c = ch_back_get();
1.13 nicm 1087: if (c == '\n' || ABORT_SIGS()) {
1.1 etheisen 1088: /*
1089: * This is the newline ending the previous line.
1090: * We have hit the beginning of the line.
1091: */
1092: new_pos = ch_tell() + 1;
1093: break;
1094: }
1.13 nicm 1095: if (c == EOI) {
1.1 etheisen 1096: /*
1097: * We have hit the beginning of the file.
1098: * This must be the first line in the file.
1099: * This must, of course, be the beginning of the line.
1100: */
1101: new_pos = ch_zero();
1102: break;
1103: }
1.13 nicm 1104: if (n <= 0) {
1.5 millert 1105: int old_size_linebuf = size_linebuf;
1.13 nicm 1106: if (expand_linebuf()) {
1.5 millert 1107: /*
1108: * Overflowed the input buffer.
1109: * Pretend the line ended here.
1110: */
1111: new_pos = ch_tell() + 1;
1112: break;
1113: }
1.1 etheisen 1114: /*
1.5 millert 1115: * Shift the data to the end of the new linebuf.
1.1 etheisen 1116: */
1.5 millert 1117: n = size_linebuf - old_size_linebuf;
1.8 millert 1118: memmove(linebuf + n, linebuf, old_size_linebuf);
1.1 etheisen 1119: }
1.5 millert 1120: linebuf[--n] = c;
1.1 etheisen 1121: }
1122: if (linep != NULL)
1.5 millert 1123: *linep = &linebuf[n];
1.10 shadchin 1124: if (line_lenp != NULL)
1125: *line_lenp = size_linebuf - 1 - n;
1.1 etheisen 1126: return (new_pos);
1127: }