Annotation of src/usr.bin/less/line.c, Revision 1.24
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: */
1.13 nicm 115: int
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: /*
1.5 millert 461: * Are we currently within a recognized ANSI escape sequence?
462: */
1.13 nicm 463: static int
464: in_ansi_esc_seq(void)
1.5 millert 465: {
1.24 ! schwarze 466: int i;
1.5 millert 467:
468: /*
469: * Search backwards for either an ESC (which means we ARE in a seq);
470: * or an end char (which means we're NOT in a seq).
471: */
1.24 ! schwarze 472: for (i = curr - 1; i >= 0; i--) {
! 473: if (linebuf[i] == ESC)
1.5 millert 474: return (1);
1.24 ! schwarze 475: if (!is_ansi_middle(linebuf[i]))
1.5 millert 476: return (0);
477: }
478: return (0);
479: }
480:
481: /*
482: * Is a character the end of an ANSI escape sequence?
483: */
1.13 nicm 484: int
485: is_ansi_end(LWCHAR ch)
1.10 shadchin 486: {
487: if (!is_ascii_char(ch))
488: return (0);
1.13 nicm 489: return (strchr(end_ansi_chars, (char)ch) != NULL);
1.10 shadchin 490: }
491:
492: /*
493: *
494: */
1.13 nicm 495: int
496: is_ansi_middle(LWCHAR ch)
1.5 millert 497: {
1.10 shadchin 498: if (!is_ascii_char(ch))
499: return (0);
500: if (is_ansi_end(ch))
501: return (0);
1.13 nicm 502: return (strchr(mid_ansi_chars, (char)ch) != NULL);
1.5 millert 503: }
504:
505: /*
1.1 etheisen 506: * Append a character and attribute to the line buffer.
507: */
1.13 nicm 508: #define STORE_CHAR(ch, a, rep, pos) \
509: if (store_char((ch), (a), (rep), (pos))) \
510: return (1)
511:
512: static int
513: store_char(LWCHAR ch, char a, char *rep, off_t pos)
1.1 etheisen 514: {
1.10 shadchin 515: int w;
516: int replen;
517: char cs;
1.13 nicm 518: int matches;
1.10 shadchin 519:
1.13 nicm 520: if (is_hilited(pos, pos+1, 0, &matches)) {
521: /*
522: * This character should be highlighted.
523: * Override the attribute passed in.
524: */
525: if (a != AT_ANSI) {
526: if (highest_hilite != -1 && pos > highest_hilite)
527: highest_hilite = pos;
528: a |= AT_HILITE;
1.10 shadchin 529: }
1.5 millert 530: }
1.10 shadchin 531:
1.13 nicm 532: if (ctldisp == OPT_ONPLUS && in_ansi_esc_seq()) {
1.10 shadchin 533: if (!is_ansi_end(ch) && !is_ansi_middle(ch)) {
534: /* Remove whole unrecognized sequence. */
535: do {
1.24 ! schwarze 536: curr--;
! 537: } while (curr > 0 && linebuf[curr] != ESC);
1.13 nicm 538: return (0);
1.10 shadchin 539: }
540: a = AT_ANSI; /* Will force re-AT_'ing around it. */
1.5 millert 541: w = 0;
1.24 ! schwarze 542: } else if (ctldisp == OPT_ONPLUS && ch == ESC) {
1.10 shadchin 543: a = AT_ANSI; /* Will force re-AT_'ing around it. */
544: w = 0;
1.13 nicm 545: } else {
1.10 shadchin 546: char *p = &linebuf[curr];
547: LWCHAR prev_ch = step_char(&p, -1, linebuf);
548: w = pwidth(ch, a, prev_ch);
549: }
550:
1.5 millert 551: if (ctldisp != OPT_ON && column + w + attr_ewidth(a) > sc_width)
1.1 etheisen 552: /*
553: * Won't fit on screen.
554: */
555: return (1);
556:
1.13 nicm 557: if (rep == NULL) {
558: cs = (char)ch;
1.10 shadchin 559: rep = &cs;
560: replen = 1;
1.13 nicm 561: } else {
1.10 shadchin 562: replen = utf_len(rep[0]);
563: }
1.13 nicm 564: if (curr + replen >= size_linebuf-6) {
1.1 etheisen 565: /*
566: * Won't fit in line buffer.
1.5 millert 567: * Try to expand it.
1.1 etheisen 568: */
1.5 millert 569: if (expand_linebuf())
570: return (1);
571: }
1.1 etheisen 572:
1.13 nicm 573: while (replen-- > 0) {
1.10 shadchin 574: linebuf[curr] = *rep++;
575: attr[curr] = a;
576: curr++;
1.1 etheisen 577: }
578: column += w;
579: return (0);
580: }
581:
582: /*
1.5 millert 583: * Append a tab to the line buffer.
584: * Store spaces to represent the tab.
585: */
1.13 nicm 586: #define STORE_TAB(a, pos) \
587: if (store_tab((a), (pos))) \
588: return (1)
1.5 millert 589:
1.13 nicm 590: static int
591: store_tab(int attr, off_t pos)
1.5 millert 592: {
593: int to_tab = column + cshift - lmargin;
594: int i;
595:
596: if (ntabstops < 2 || to_tab >= tabstops[ntabstops-1])
597: to_tab = tabdefault -
1.13 nicm 598: ((to_tab - tabstops[ntabstops-1]) % tabdefault);
599: else {
1.21 deraadt 600: for (i = ntabstops - 2; i >= 0; i--)
1.5 millert 601: if (to_tab >= tabstops[i])
602: break;
603: to_tab = tabstops[i+1] - to_tab;
604: }
605:
1.13 nicm 606: if (column + to_tab - 1 + pwidth(' ', attr, 0) +
607: attr_ewidth(attr) > sc_width)
608: return (1);
1.10 shadchin 609:
1.5 millert 610: do {
1.10 shadchin 611: STORE_CHAR(' ', attr, " ", pos);
1.5 millert 612: } while (--to_tab > 0);
1.13 nicm 613: return (0);
1.5 millert 614: }
615:
1.13 nicm 616: #define STORE_PRCHAR(c, pos) \
617: if (store_prchar((c), (pos))) \
618: return (1)
1.10 shadchin 619:
1.13 nicm 620: static int
621: store_prchar(char c, off_t pos)
1.10 shadchin 622: {
623: char *s;
624:
625: /*
626: * Convert to printable representation.
627: */
628: s = prchar(c);
629:
630: /*
631: * Make sure we can get the entire representation
632: * of the character on this line.
633: */
1.13 nicm 634: if (column + (int)strlen(s) - 1 +
635: pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
636: return (1);
1.10 shadchin 637:
1.13 nicm 638: for (; *s != 0; s++) {
1.10 shadchin 639: STORE_CHAR(*s, AT_BINARY, NULL, pos);
1.13 nicm 640: }
641: return (0);
1.10 shadchin 642: }
643:
1.13 nicm 644: static int
645: flush_mbc_buf(off_t pos)
1.10 shadchin 646: {
647: int i;
648:
649: for (i = 0; i < mbc_buf_index; i++)
650: if (store_prchar(mbc_buf[i], pos))
1.13 nicm 651: return (mbc_buf_index - i);
1.10 shadchin 652:
1.13 nicm 653: return (0);
1.10 shadchin 654: }
655:
1.5 millert 656: /*
1.1 etheisen 657: * Append a character to the line buffer.
658: * Expand tabs into spaces, handle underlining, boldfacing, etc.
659: * Returns 0 if ok, 1 if couldn't fit in buffer.
660: */
1.13 nicm 661: int
662: pappend(char c, off_t pos)
1.1 etheisen 663: {
1.5 millert 664: int r;
665:
1.13 nicm 666: if (pendc) {
1.10 shadchin 667: if (do_append(pendc, NULL, pendpos))
1.1 etheisen 668: /*
669: * Oops. We've probably lost the char which
670: * was in pendc, since caller won't back up.
671: */
672: return (1);
673: pendc = '\0';
674: }
675:
1.13 nicm 676: if (c == '\r' && bs_mode == BS_SPECIAL) {
677: if (mbc_buf_len > 0) /* utf_mode must be on. */ {
1.10 shadchin 678: /* Flush incomplete (truncated) sequence. */
679: r = flush_mbc_buf(mbc_pos);
680: mbc_buf_index = r + 1;
681: mbc_buf_len = 0;
682: if (r)
683: return (mbc_buf_index);
684: }
685:
1.1 etheisen 686: /*
1.13 nicm 687: * Don't put the CR into the buffer until we see
1.1 etheisen 688: * the next char. If the next char is a newline,
689: * discard the CR.
690: */
691: pendc = c;
692: pendpos = pos;
693: return (0);
694: }
695:
1.13 nicm 696: if (!utf_mode) {
1.10 shadchin 697: r = do_append((LWCHAR) c, NULL, pos);
1.13 nicm 698: } else {
1.10 shadchin 699: /* Perform strict validation in all possible cases. */
1.13 nicm 700: if (mbc_buf_len == 0) {
701: retry:
1.10 shadchin 702: mbc_buf_index = 1;
703: *mbc_buf = c;
1.13 nicm 704: if (IS_ASCII_OCTET(c)) {
1.10 shadchin 705: r = do_append((LWCHAR) c, NULL, pos);
1.13 nicm 706: } else if (IS_UTF8_LEAD(c)) {
1.10 shadchin 707: mbc_buf_len = utf_len(c);
708: mbc_pos = pos;
709: return (0);
1.13 nicm 710: } else {
1.10 shadchin 711: /* UTF8_INVALID or stray UTF8_TRAIL */
712: r = flush_mbc_buf(pos);
1.13 nicm 713: }
714: } else if (IS_UTF8_TRAIL(c)) {
1.10 shadchin 715: mbc_buf[mbc_buf_index++] = c;
716: if (mbc_buf_index < mbc_buf_len)
717: return (0);
718: if (is_utf8_well_formed(mbc_buf))
1.13 nicm 719: r = do_append(get_wchar(mbc_buf), mbc_buf,
720: mbc_pos);
1.10 shadchin 721: else
722: /* Complete, but not shortest form, sequence. */
723: mbc_buf_index = r = flush_mbc_buf(mbc_pos);
724: mbc_buf_len = 0;
1.13 nicm 725: } else {
1.10 shadchin 726: /* Flush incomplete (truncated) sequence. */
727: r = flush_mbc_buf(mbc_pos);
728: mbc_buf_index = r + 1;
729: mbc_buf_len = 0;
730: /* Handle new char. */
731: if (!r)
732: goto retry;
1.13 nicm 733: }
1.10 shadchin 734: }
735:
1.5 millert 736: /*
737: * If we need to shift the line, do it.
738: * But wait until we get to at least the middle of the screen,
739: * so shifting it doesn't affect the chars we're currently
740: * pappending. (Bold & underline can get messed up otherwise.)
741: */
1.13 nicm 742: if (cshift < hshift && column > sc_width / 2) {
1.5 millert 743: linebuf[curr] = '\0';
744: pshift(hshift - cshift);
745: }
1.13 nicm 746: if (r) {
1.10 shadchin 747: /* How many chars should caller back up? */
748: r = (!utf_mode) ? 1 : mbc_buf_index;
749: }
1.5 millert 750: return (r);
1.1 etheisen 751: }
752:
1.13 nicm 753: static int
754: do_append(LWCHAR ch, char *rep, off_t pos)
1.1 etheisen 755: {
1.13 nicm 756: int a;
1.10 shadchin 757: LWCHAR prev_ch;
1.1 etheisen 758:
1.10 shadchin 759: a = AT_NORMAL;
1.1 etheisen 760:
1.13 nicm 761: if (ch == '\b') {
1.10 shadchin 762: if (bs_mode == BS_CONTROL)
1.5 millert 763: goto do_control_char;
1.10 shadchin 764:
765: /*
766: * A better test is needed here so we don't
767: * backspace over part of the printed
768: * representation of a binary character.
769: */
1.13 nicm 770: if (curr <= lmargin ||
771: column <= lmargin ||
772: (attr[curr - 1] & (AT_ANSI|AT_BINARY))) {
1.10 shadchin 773: STORE_PRCHAR('\b', pos);
1.13 nicm 774: } else if (bs_mode == BS_NORMAL) {
1.10 shadchin 775: STORE_CHAR(ch, AT_NORMAL, NULL, pos);
1.13 nicm 776: } else if (bs_mode == BS_SPECIAL) {
1.10 shadchin 777: overstrike = backc();
1.13 nicm 778: }
1.10 shadchin 779:
1.13 nicm 780: return (0);
1.10 shadchin 781: }
782:
1.13 nicm 783: if (overstrike > 0) {
1.1 etheisen 784: /*
785: * Overstrike the character at the current position
1.13 nicm 786: * in the line buffer. This will cause either
787: * underline (if a "_" is overstruck),
1.1 etheisen 788: * bold (if an identical character is overstruck),
789: * or just deletion of the character in the buffer.
790: */
1.10 shadchin 791: overstrike = utf_mode ? -1 : 0;
792: /* To be correct, this must be a base character. */
793: prev_ch = get_wchar(linebuf + curr);
794: a = attr[curr];
1.13 nicm 795: if (ch == prev_ch) {
1.5 millert 796: /*
797: * Overstriking a char with itself means make it bold.
798: * But overstriking an underscore with itself is
799: * ambiguous. It could mean make it bold, or
800: * it could mean make it underlined.
801: * Use the previous overstrike to resolve it.
802: */
1.13 nicm 803: if (ch == '_') {
1.10 shadchin 804: if ((a & (AT_BOLD|AT_UNDERLINE)) != AT_NORMAL)
805: a |= (AT_BOLD|AT_UNDERLINE);
1.20 natano 806: else if (curr > 0 && attr[curr - 1] & AT_UNDERLINE)
807: a |= AT_UNDERLINE;
808: else if (curr > 0 && attr[curr - 1] & AT_BOLD)
809: a |= AT_BOLD;
1.10 shadchin 810: else
1.20 natano 811: a |= AT_INDET;
1.13 nicm 812: } else {
1.10 shadchin 813: a |= AT_BOLD;
1.13 nicm 814: }
815: } else if (ch == '_') {
1.10 shadchin 816: a |= AT_UNDERLINE;
817: ch = prev_ch;
818: rep = linebuf + curr;
1.13 nicm 819: } else if (prev_ch == '_') {
1.10 shadchin 820: a |= AT_UNDERLINE;
821: }
822: /* Else we replace prev_ch, but we keep its attributes. */
1.13 nicm 823: } else if (overstrike < 0) {
824: if (is_composing_char(ch) ||
1.20 natano 825: is_combining_char(get_wchar(linebuf + curr), ch)) {
1.10 shadchin 826: /* Continuation of the same overstrike. */
1.20 natano 827: if (curr > 0)
828: a = attr[curr - 1] & (AT_UNDERLINE | AT_BOLD);
829: else
830: a = AT_NORMAL;
831: } else
1.10 shadchin 832: overstrike = 0;
833: }
834:
1.13 nicm 835: if (ch == '\t') {
1.5 millert 836: /*
837: * Expand a tab into spaces.
838: */
1.13 nicm 839: switch (bs_mode) {
1.1 etheisen 840: case BS_CONTROL:
841: goto do_control_char;
1.5 millert 842: case BS_NORMAL:
1.1 etheisen 843: case BS_SPECIAL:
1.10 shadchin 844: STORE_TAB(a, pos);
1.1 etheisen 845: break;
846: }
1.13 nicm 847: } else if ((!utf_mode || is_ascii_char(ch)) && control_char((char)ch)) {
848: do_control_char:
849: if (ctldisp == OPT_ON ||
1.24 ! schwarze 850: (ctldisp == OPT_ONPLUS && ch == ESC)) {
1.1 etheisen 851: /*
852: * Output as a normal character.
853: */
1.10 shadchin 854: STORE_CHAR(ch, AT_NORMAL, rep, pos);
1.13 nicm 855: } else {
856: STORE_PRCHAR((char)ch, pos);
1.10 shadchin 857: }
1.13 nicm 858: } else if (utf_mode && ctldisp != OPT_ON && is_ubin_char(ch)) {
1.10 shadchin 859: char *s;
860:
861: s = prutfchar(ch);
1.1 etheisen 862:
1.13 nicm 863: if (column + (int)strlen(s) - 1 +
1.10 shadchin 864: pwidth(' ', binattr, 0) + attr_ewidth(binattr) > sc_width)
865: return (1);
1.1 etheisen 866:
1.13 nicm 867: for (; *s != 0; s++)
1.10 shadchin 868: STORE_CHAR(*s, AT_BINARY, NULL, pos);
1.13 nicm 869: } else {
1.10 shadchin 870: STORE_CHAR(ch, a, rep, pos);
1.1 etheisen 871: }
1.13 nicm 872: return (0);
1.10 shadchin 873: }
874:
875: /*
876: *
877: */
1.13 nicm 878: int
879: pflushmbc(void)
1.10 shadchin 880: {
881: int r = 0;
1.1 etheisen 882:
1.13 nicm 883: if (mbc_buf_len > 0) {
1.10 shadchin 884: /* Flush incomplete (truncated) sequence. */
885: r = flush_mbc_buf(mbc_pos);
886: mbc_buf_len = 0;
887: }
1.13 nicm 888: return (r);
1.1 etheisen 889: }
890:
891: /*
892: * Terminate the line in the line buffer.
893: */
1.13 nicm 894: void
895: pdone(int endline, int forw)
1.1 etheisen 896: {
1.20 natano 897: int i;
898:
1.10 shadchin 899: (void) pflushmbc();
900:
1.1 etheisen 901: if (pendc && (pendc != '\r' || !endline))
902: /*
903: * If we had a pending character, put it in the buffer.
904: * But discard a pending CR if we are at end of line
905: * (that is, discard the CR in a CR/LF sequence).
906: */
1.10 shadchin 907: (void) do_append(pendc, NULL, pendpos);
1.20 natano 908:
909: for (i = curr - 1; i >= 0; i--) {
910: if (attr[i] & AT_INDET) {
911: attr[i] &= ~AT_INDET;
912: if (i < curr - 1 && attr[i + 1] & AT_BOLD)
913: attr[i] |= AT_BOLD;
914: else
915: attr[i] |= AT_UNDERLINE;
916: }
917: }
1.1 etheisen 918:
919: /*
1.5 millert 920: * Make sure we've shifted the line, if we need to.
921: */
922: if (cshift < hshift)
923: pshift(hshift - cshift);
924:
1.13 nicm 925: if (ctldisp == OPT_ONPLUS && is_ansi_end('m')) {
1.10 shadchin 926: /* Switch to normal attribute at end of line. */
927: char *p = "\033[m";
1.13 nicm 928: for (; *p != '\0'; p++) {
1.10 shadchin 929: linebuf[curr] = *p;
930: attr[curr++] = AT_ANSI;
931: }
932: }
933:
1.5 millert 934: /*
1.1 etheisen 935: * Add a newline if necessary,
936: * and append a '\0' to the end of the line.
1.10 shadchin 937: * We output a newline if we're not at the right edge of the screen,
938: * or if the terminal doesn't auto wrap,
939: * or if this is really the end of the line AND the terminal ignores
940: * a newline at the right edge.
1.13 nicm 941: * (In the last case we don't want to output a newline if the terminal
1.10 shadchin 942: * doesn't ignore it since that would produce an extra blank line.
943: * But we do want to output a newline if the terminal ignores it in case
944: * the next line is blank. In that case the single newline output for
945: * that blank line would be ignored!)
1.1 etheisen 946: */
1.13 nicm 947: if (column < sc_width || !auto_wrap || (endline && ignaw) ||
948: ctldisp == OPT_ON) {
1.1 etheisen 949: linebuf[curr] = '\n';
950: attr[curr] = AT_NORMAL;
951: curr++;
1.13 nicm 952: } else if (ignaw && column >= sc_width && forw) {
1.10 shadchin 953: /*
954: * Terminals with "ignaw" don't wrap until they *really* need
955: * to, i.e. when the character *after* the last one to fit on a
956: * line is output. But they are too hard to deal with when they
957: * get in the state where a full screen width of characters
958: * have been output but the cursor is sitting on the right edge
959: * instead of at the start of the next line.
1.13 nicm 960: * So we nudge them into wrapping by outputting a space
961: * character plus a backspace. But do this only if moving
1.10 shadchin 962: * forward; if we're moving backward and drawing this line at
963: * the top of the screen, the space would overwrite the first
1.13 nicm 964: * char on the next line. We don't need to do this "nudge"
1.10 shadchin 965: * at the top of the screen anyway.
966: */
967: linebuf[curr] = ' ';
968: attr[curr++] = AT_NORMAL;
1.13 nicm 969: linebuf[curr] = '\b';
1.10 shadchin 970: attr[curr++] = AT_NORMAL;
1.1 etheisen 971: }
972: linebuf[curr] = '\0';
973: attr[curr] = AT_NORMAL;
1.10 shadchin 974: }
1.5 millert 975:
1.10 shadchin 976: /*
977: *
978: */
1.13 nicm 979: void
980: set_status_col(char c)
1.10 shadchin 981: {
982: linebuf[0] = c;
983: attr[0] = AT_NORMAL|AT_HILITE;
1.1 etheisen 984: }
985:
986: /*
987: * Get a character from the current line.
988: * Return the character as the function return value,
989: * and the character attribute in *ap.
990: */
1.13 nicm 991: int
992: gline(int i, int *ap)
1.1 etheisen 993: {
1.13 nicm 994: if (is_null_line) {
1.1 etheisen 995: /*
996: * If there is no current line, we pretend the line is
997: * either "~" or "", depending on the "twiddle" flag.
998: */
1.13 nicm 999: if (twiddle) {
1000: if (i == 0) {
1.10 shadchin 1001: *ap = AT_BOLD;
1.13 nicm 1002: return ('~');
1.10 shadchin 1003: }
1004: --i;
1005: }
1006: /* Make sure we're back to AT_NORMAL before the '\n'. */
1007: *ap = AT_NORMAL;
1.13 nicm 1008: return (i ? '\0' : '\n');
1.1 etheisen 1009: }
1010:
1011: *ap = attr[i];
1.10 shadchin 1012: return (linebuf[i] & 0xFF);
1.1 etheisen 1013: }
1014:
1015: /*
1016: * Indicate that there is no current line.
1017: */
1.13 nicm 1018: void
1019: null_line(void)
1.1 etheisen 1020: {
1021: is_null_line = 1;
1.5 millert 1022: cshift = 0;
1.1 etheisen 1023: }
1024:
1025: /*
1026: * Analogous to forw_line(), but deals with "raw lines":
1027: * lines which are not split for screen width.
1028: * {{ This is supposed to be more efficient than forw_line(). }}
1029: */
1.13 nicm 1030: off_t
1031: forw_raw_line(off_t curr_pos, char **linep, int *line_lenp)
1032: {
1033: int n;
1034: int c;
1035: off_t new_pos;
1036:
1037: if (curr_pos == -1 || ch_seek(curr_pos) ||
1038: (c = ch_forw_get()) == EOI)
1039: return (-1);
1.1 etheisen 1040:
1.5 millert 1041: n = 0;
1.13 nicm 1042: for (;;) {
1043: if (c == '\n' || c == EOI || ABORT_SIGS()) {
1.1 etheisen 1044: new_pos = ch_tell();
1045: break;
1046: }
1.13 nicm 1047: if (n >= size_linebuf-1) {
1048: if (expand_linebuf()) {
1.5 millert 1049: /*
1050: * Overflowed the input buffer.
1051: * Pretend the line ended here.
1052: */
1053: new_pos = ch_tell() - 1;
1054: break;
1055: }
1.1 etheisen 1056: }
1.13 nicm 1057: linebuf[n++] = (char)c;
1.1 etheisen 1058: c = ch_forw_get();
1059: }
1.5 millert 1060: linebuf[n] = '\0';
1.1 etheisen 1061: if (linep != NULL)
1062: *linep = linebuf;
1.10 shadchin 1063: if (line_lenp != NULL)
1064: *line_lenp = n;
1.1 etheisen 1065: return (new_pos);
1066: }
1067:
1068: /*
1069: * Analogous to back_line(), but deals with "raw lines".
1070: * {{ This is supposed to be more efficient than back_line(). }}
1071: */
1.13 nicm 1072: off_t
1073: back_raw_line(off_t curr_pos, char **linep, int *line_lenp)
1074: {
1075: int n;
1076: int c;
1077: off_t new_pos;
1078:
1079: if (curr_pos == -1 || curr_pos <= ch_zero() || ch_seek(curr_pos - 1))
1080: return (-1);
1.1 etheisen 1081:
1.5 millert 1082: n = size_linebuf;
1083: linebuf[--n] = '\0';
1.13 nicm 1084: for (;;) {
1.1 etheisen 1085: c = ch_back_get();
1.13 nicm 1086: if (c == '\n' || ABORT_SIGS()) {
1.1 etheisen 1087: /*
1088: * This is the newline ending the previous line.
1089: * We have hit the beginning of the line.
1090: */
1091: new_pos = ch_tell() + 1;
1092: break;
1093: }
1.13 nicm 1094: if (c == EOI) {
1.1 etheisen 1095: /*
1096: * We have hit the beginning of the file.
1097: * This must be the first line in the file.
1098: * This must, of course, be the beginning of the line.
1099: */
1100: new_pos = ch_zero();
1101: break;
1102: }
1.13 nicm 1103: if (n <= 0) {
1.5 millert 1104: int old_size_linebuf = size_linebuf;
1.13 nicm 1105: if (expand_linebuf()) {
1.5 millert 1106: /*
1107: * Overflowed the input buffer.
1108: * Pretend the line ended here.
1109: */
1110: new_pos = ch_tell() + 1;
1111: break;
1112: }
1.1 etheisen 1113: /*
1.5 millert 1114: * Shift the data to the end of the new linebuf.
1.1 etheisen 1115: */
1.5 millert 1116: n = size_linebuf - old_size_linebuf;
1.8 millert 1117: memmove(linebuf + n, linebuf, old_size_linebuf);
1.1 etheisen 1118: }
1.5 millert 1119: linebuf[--n] = c;
1.1 etheisen 1120: }
1121: if (linep != NULL)
1.5 millert 1122: *linep = &linebuf[n];
1.10 shadchin 1123: if (line_lenp != NULL)
1124: *line_lenp = size_linebuf - 1 - n;
1.1 etheisen 1125: return (new_pos);
1126: }