xref: /openbsd/usr.bin/vi/ex/ex_txt.c (revision 17df1aa7)
1 /*	$OpenBSD: ex_txt.c,v 1.10 2009/10/27 23:59:47 deraadt Exp $	*/
2 
3 /*-
4  * Copyright (c) 1992, 1993, 1994
5  *	The Regents of the University of California.  All rights reserved.
6  * Copyright (c) 1992, 1993, 1994, 1995, 1996
7  *	Keith Bostic.  All rights reserved.
8  *
9  * See the LICENSE file for redistribution information.
10  */
11 
12 #include "config.h"
13 
14 #include <sys/types.h>
15 #include <sys/queue.h>
16 
17 #include <bitstring.h>
18 #include <ctype.h>
19 #include <limits.h>
20 #include <stdio.h>
21 #include <stdlib.h>
22 #include <string.h>
23 
24 #include "../common/common.h"
25 
26 /*
27  * !!!
28  * The backslash characters was special when it preceded a newline as part of
29  * a substitution replacement pattern.  For example, the input ":a\<cr>" would
30  * failed immediately with an error, as the <cr> wasn't part of a substitution
31  * replacement pattern.  This implies a frightening integration of the editor
32  * and the parser and/or the RE engine.  There's no way I'm going to reproduce
33  * those semantics.
34  *
35  * So, if backslashes are special, this code inserts the backslash and the next
36  * character into the string, without regard for the character or the command
37  * being entered.  Since "\<cr>" was illegal historically (except for the one
38  * special case), and the command will fail eventually, no historical scripts
39  * should break (presuming they didn't depend on the failure mode itself or the
40  * characters remaining when failure occurred.
41  */
42 
43 static int	txt_dent(SCR *, TEXT *);
44 static void	txt_prompt(SCR *, TEXT *, ARG_CHAR_T, u_int32_t);
45 
46 /*
47  * ex_txt --
48  *	Get lines from the terminal for ex.
49  *
50  * PUBLIC: int ex_txt(SCR *, TEXTH *, ARG_CHAR_T, u_int32_t);
51  */
52 int
53 ex_txt(sp, tiqh, prompt, flags)
54 	SCR *sp;
55 	TEXTH *tiqh;
56 	ARG_CHAR_T prompt;
57 	u_int32_t flags;
58 {
59 	EVENT ev;
60 	GS *gp;
61 	TEXT ait, *ntp, *tp;
62 	carat_t carat_st;
63 	size_t cnt;
64 	int rval;
65 
66 	rval = 0;
67 
68 	/*
69 	 * Get a TEXT structure with some initial buffer space, reusing the
70 	 * last one if it's big enough.  (All TEXT bookkeeping fields default
71 	 * to 0 -- text_init() handles this.)
72 	 */
73 	if (CIRCLEQ_FIRST(tiqh) != CIRCLEQ_END(tiqh)) {
74 		tp = CIRCLEQ_FIRST(tiqh);
75 		if (CIRCLEQ_NEXT(tp, q) != CIRCLEQ_END(tiqh) || tp->lb_len < 32) {
76 			text_lfree(tiqh);
77 			goto newtp;
78 		}
79 		tp->len = 0;
80 	} else {
81 newtp:		if ((tp = text_init(sp, NULL, 0, 32)) == NULL)
82 			goto err;
83 		CIRCLEQ_INSERT_HEAD(tiqh, tp, q);
84 	}
85 
86 	/* Set the starting line number. */
87 	tp->lno = sp->lno + 1;
88 
89 	/*
90 	 * If it's a terminal, set up autoindent, put out the prompt, and
91 	 * set it up so we know we were suspended.  Otherwise, turn off
92 	 * the autoindent flag, as that requires less special casing below.
93 	 *
94 	 * XXX
95 	 * Historic practice is that ^Z suspended command mode (but, because
96 	 * it ran in cooked mode, it was unaffected by the autowrite option.)
97 	 * On restart, any "current" input was discarded, whether in insert
98 	 * mode or not, and ex was in command mode.  This code matches historic
99 	 * practice, but not 'cause it's easier.
100 	 */
101 	gp = sp->gp;
102 	if (F_ISSET(gp, G_SCRIPTED))
103 		LF_CLR(TXT_AUTOINDENT);
104 	else {
105 		if (LF_ISSET(TXT_AUTOINDENT)) {
106 			LF_SET(TXT_EOFCHAR);
107 			if (v_txt_auto(sp, sp->lno, NULL, 0, tp))
108 				goto err;
109 		}
110 		txt_prompt(sp, tp, prompt, flags);
111 	}
112 
113 	for (carat_st = C_NOTSET;;) {
114 		if (v_event_get(sp, &ev, 0, 0))
115 			goto err;
116 
117 		/* Deal with all non-character events. */
118 		switch (ev.e_event) {
119 		case E_CHARACTER:
120 			break;
121 		case E_ERR:
122 			goto err;
123 		case E_REPAINT:
124 		case E_WRESIZE:
125 			continue;
126 		case E_EOF:
127 			rval = 1;
128 			/* FALLTHROUGH */
129 		case E_INTERRUPT:
130 			/*
131 			 * Handle EOF/SIGINT events by discarding partially
132 			 * entered text and returning.  EOF returns failure,
133 			 * E_INTERRUPT returns success.
134 			 */
135 			goto notlast;
136 		default:
137 			v_event_err(sp, &ev);
138 			goto notlast;
139 		}
140 
141 		/*
142 		 * Deal with character events.
143 		 *
144 		 * Check to see if the character fits into the input buffer.
145 		 * (Use tp->len, ignore overwrite and non-printable chars.)
146 		 */
147 		BINC_GOTO(sp, tp->lb, tp->lb_len, tp->len + 1);
148 
149 		switch (ev.e_value) {
150 		case K_CR:
151 			/*
152 			 * !!!
153 			 * Historically, <carriage-return>'s in the command
154 			 * weren't special, so the ex parser would return an
155 			 * unknown command error message.  However, if they
156 			 * terminated the command if they were in a map.  I'm
157 			 * pretty sure this still isn't right, but it handles
158 			 * what I've seen so far.
159 			 */
160 			if (!F_ISSET(&ev.e_ch, CH_MAPPED))
161 				goto ins_ch;
162 			/* FALLTHROUGH */
163 		case K_NL:
164 			/*
165 			 * '\' can escape <carriage-return>/<newline>.  We
166 			 * don't discard the backslash because we need it
167 			 * to get the <newline> through the ex parser.
168 			 */
169 			if (LF_ISSET(TXT_BACKSLASH) &&
170 			    tp->len != 0 && tp->lb[tp->len - 1] == '\\')
171 				goto ins_ch;
172 
173 			/*
174 			 * CR returns from the ex command line.
175 			 *
176 			 * XXX
177 			 * Terminate with a nul, needed by filter.
178 			 */
179 			if (LF_ISSET(TXT_CR)) {
180 				tp->lb[tp->len] = '\0';
181 				goto done;
182 			}
183 
184 			/*
185 			 * '.' may terminate text input mode; free the current
186 			 * TEXT.
187 			 */
188 			if (LF_ISSET(TXT_DOTTERM) && tp->len == tp->ai + 1 &&
189 			    tp->lb[tp->len - 1] == '.') {
190 notlast:			CIRCLEQ_REMOVE(tiqh, tp, q);
191 				text_free(tp);
192 				goto done;
193 			}
194 
195 			/* Set up bookkeeping for the new line. */
196 			if ((ntp = text_init(sp, NULL, 0, 32)) == NULL)
197 				goto err;
198 			ntp->lno = tp->lno + 1;
199 
200 			/*
201 			 * Reset the autoindent line value.  0^D keeps the ai
202 			 * line from changing, ^D changes the level, even if
203 			 * there were no characters in the old line.  Note, if
204 			 * using the current tp structure, use the cursor as
205 			 * the length, the autoindent characters may have been
206 			 * erased.
207 			 */
208 			if (LF_ISSET(TXT_AUTOINDENT)) {
209 				if (carat_st == C_NOCHANGE) {
210 					if (v_txt_auto(sp,
211 					    OOBLNO, &ait, ait.ai, ntp))
212 						goto err;
213 					free(ait.lb);
214 				} else
215 					if (v_txt_auto(sp,
216 					    OOBLNO, tp, tp->len, ntp))
217 						goto err;
218 				carat_st = C_NOTSET;
219 			}
220 			txt_prompt(sp, ntp, prompt, flags);
221 
222 			/*
223 			 * Swap old and new TEXT's, and insert the new TEXT
224 			 * into the queue.
225 			 */
226 			tp = ntp;
227 			CIRCLEQ_INSERT_TAIL(tiqh, tp, q);
228 			break;
229 		case K_CARAT:			/* Delete autoindent chars. */
230 			if (tp->len <= tp->ai && LF_ISSET(TXT_AUTOINDENT))
231 				carat_st = C_CARATSET;
232 			goto ins_ch;
233 		case K_ZERO:			/* Delete autoindent chars. */
234 			if (tp->len <= tp->ai && LF_ISSET(TXT_AUTOINDENT))
235 				carat_st = C_ZEROSET;
236 			goto ins_ch;
237 		case K_CNTRLD:			/* Delete autoindent char. */
238 			/*
239 			 * !!!
240 			 * Historically, the ^D command took (but then ignored)
241 			 * a count.  For simplicity, we don't return it unless
242 			 * it's the first character entered.  The check for len
243 			 * equal to 0 is okay, TXT_AUTOINDENT won't be set.
244 			 */
245 			if (LF_ISSET(TXT_CNTRLD)) {
246 				for (cnt = 0; cnt < tp->len; ++cnt)
247 					if (!isblank(tp->lb[cnt]))
248 						break;
249 				if (cnt == tp->len) {
250 					tp->len = 1;
251 					tp->lb[0] = ev.e_c;
252 					tp->lb[1] = '\0';
253 
254 					/*
255 					 * Put out a line separator, in case
256 					 * the command fails.
257 					 */
258 					(void)putchar('\n');
259 					goto done;
260 				}
261 			}
262 
263 			/*
264 			 * POSIX 1003.1b-1993, paragraph 7.1.1.9, states that
265 			 * the EOF characters are discarded if there are other
266 			 * characters to process in the line, i.e. if the EOF
267 			 * is not the first character in the line.  For this
268 			 * reason, historic ex discarded the EOF characters,
269 			 * even if occurring in the middle of the input line.
270 			 * We match that historic practice.
271 			 *
272 			 * !!!
273 			 * The test for discarding in the middle of the line is
274 			 * done in the switch, because the CARAT forms are N+1,
275 			 * not N.
276 			 *
277 			 * !!!
278 			 * There's considerable magic to make the terminal code
279 			 * return the EOF character at all.  See that code for
280 			 * details.
281 			 */
282 			if (!LF_ISSET(TXT_AUTOINDENT) || tp->len == 0)
283 				continue;
284 			switch (carat_st) {
285 			case C_CARATSET:		/* ^^D */
286 				if (tp->len > tp->ai + 1)
287 					continue;
288 
289 				/* Save the ai string for later. */
290 				ait.lb = NULL;
291 				ait.lb_len = 0;
292 				BINC_GOTO(sp, ait.lb, ait.lb_len, tp->ai);
293 				memcpy(ait.lb, tp->lb, tp->ai);
294 				ait.ai = ait.len = tp->ai;
295 
296 				carat_st = C_NOCHANGE;
297 				goto leftmargin;
298 			case C_ZEROSET:			/* 0^D */
299 				if (tp->len > tp->ai + 1)
300 					continue;
301 
302 				carat_st = C_NOTSET;
303 leftmargin:			(void)gp->scr_ex_adjust(sp, EX_TERM_CE);
304 				tp->ai = tp->len = 0;
305 				break;
306 			case C_NOTSET:			/* ^D */
307 				if (tp->len > tp->ai)
308 					continue;
309 
310 				if (txt_dent(sp, tp))
311 					goto err;
312 				break;
313 			default:
314 				abort();
315 			}
316 
317 			/* Clear and redisplay the line. */
318 			(void)gp->scr_ex_adjust(sp, EX_TERM_CE);
319 			txt_prompt(sp, tp, prompt, flags);
320 			break;
321 		default:
322 			/*
323 			 * See the TXT_BEAUTIFY comment in vi/v_txt_ev.c.
324 			 *
325 			 * Silently eliminate any iscntrl() character that was
326 			 * not already handled specially, except for <tab> and
327 			 * <ff>.
328 			 */
329 ins_ch:			if (LF_ISSET(TXT_BEAUTIFY) && iscntrl(ev.e_c) &&
330 			    ev.e_value != K_FORMFEED && ev.e_value != K_TAB)
331 				break;
332 
333 			tp->lb[tp->len++] = ev.e_c;
334 			break;
335 		}
336 	}
337 	/* NOTREACHED */
338 
339 done:	return (rval);
340 
341 err:
342 alloc_err:
343 	return (1);
344 }
345 
346 /*
347  * txt_prompt --
348  *	Display the ex prompt, line number, ai characters.  Characters had
349  *	better be printable by the terminal driver, but that's its problem,
350  *	not ours.
351  */
352 static void
353 txt_prompt(sp, tp, prompt, flags)
354 	SCR *sp;
355 	TEXT *tp;
356 	ARG_CHAR_T prompt;
357 	u_int32_t flags;
358 {
359 	/* Display the prompt. */
360 	if (LF_ISSET(TXT_PROMPT))
361 		(void)printf("%c", prompt);
362 
363 	/* Display the line number. */
364 	if (LF_ISSET(TXT_NUMBER) && O_ISSET(sp, O_NUMBER))
365 		(void)printf("%6lu  ", (u_long)tp->lno);
366 
367 	/* Print out autoindent string. */
368 	if (LF_ISSET(TXT_AUTOINDENT))
369 		(void)printf("%.*s", (int)tp->ai, tp->lb);
370 	(void)fflush(stdout);
371 }
372 
373 /*
374  * txt_dent --
375  *	Handle ^D outdents.
376  *
377  * Ex version of vi/v_ntext.c:txt_dent().  See that code for the (usual)
378  * ranting and raving.  This is a fair bit simpler as ^T isn't special.
379  */
380 static int
381 txt_dent(sp, tp)
382 	SCR *sp;
383 	TEXT *tp;
384 {
385 	u_long sw, ts;
386 	size_t cno, off, scno, spaces, tabs;
387 
388 	ts = O_VAL(sp, O_TABSTOP);
389 	sw = O_VAL(sp, O_SHIFTWIDTH);
390 
391 	/* Get the current screen column. */
392 	for (off = scno = 0; off < tp->len; ++off)
393 		if (tp->lb[off] == '\t')
394 			scno += COL_OFF(scno, ts);
395 		else
396 			++scno;
397 
398 	/* Get the previous shiftwidth column. */
399 	cno = scno--;
400 	scno -= scno % sw;
401 
402 	/*
403 	 * Since we don't know what comes before the character(s) being
404 	 * deleted, we have to resolve the autoindent characters .  The
405 	 * example is a <tab>, which doesn't take up a full shiftwidth
406 	 * number of columns because it's preceded by <space>s.  This is
407 	 * easy to get if the user sets shiftwidth to a value less than
408 	 * tabstop, and then uses ^T to indent, and ^D to outdent.
409 	 *
410 	 * Count up spaces/tabs needed to get to the target.
411 	 */
412 	for (cno = 0, tabs = 0; cno + COL_OFF(cno, ts) <= scno; ++tabs)
413 		cno += COL_OFF(cno, ts);
414 	spaces = scno - cno;
415 
416 	/* Make sure there's enough room. */
417 	BINC_RET(sp, tp->lb, tp->lb_len, tabs + spaces + 1);
418 
419 	/* Adjust the final ai character count. */
420 	tp->ai = tabs + spaces;
421 
422 	/* Enter the replacement characters. */
423 	for (tp->len = 0; tabs > 0; --tabs)
424 		tp->lb[tp->len++] = '\t';
425 	for (; spaces > 0; --spaces)
426 		tp->lb[tp->len++] = ' ';
427 	return (0);
428 }
429