xref: /freebsd/contrib/nvi/ex/ex_txt.c (revision 1e413cf93298b5b97441a21d9a50fdcd0ee9945e)
1 /*-
2  * Copyright (c) 1992, 1993, 1994
3  *	The Regents of the University of California.  All rights reserved.
4  * Copyright (c) 1992, 1993, 1994, 1995, 1996
5  *	Keith Bostic.  All rights reserved.
6  *
7  * See the LICENSE file for redistribution information.
8  */
9 
10 #include "config.h"
11 
12 #ifndef lint
13 static const char sccsid[] = "@(#)ex_txt.c	10.17 (Berkeley) 10/10/96";
14 #endif /* not lint */
15 
16 #include <sys/types.h>
17 #include <sys/queue.h>
18 
19 #include <bitstring.h>
20 #include <ctype.h>
21 #include <limits.h>
22 #include <stdio.h>
23 #include <stdlib.h>
24 #include <string.h>
25 
26 #include "../common/common.h"
27 
28 /*
29  * !!!
30  * The backslash characters was special when it preceded a newline as part of
31  * a substitution replacement pattern.  For example, the input ":a\<cr>" would
32  * failed immediately with an error, as the <cr> wasn't part of a substitution
33  * replacement pattern.  This implies a frightening integration of the editor
34  * and the parser and/or the RE engine.  There's no way I'm going to reproduce
35  * those semantics.
36  *
37  * So, if backslashes are special, this code inserts the backslash and the next
38  * character into the string, without regard for the character or the command
39  * being entered.  Since "\<cr>" was illegal historically (except for the one
40  * special case), and the command will fail eventually, no historical scripts
41  * should break (presuming they didn't depend on the failure mode itself or the
42  * characters remaining when failure occurred.
43  */
44 
45 static int	txt_dent __P((SCR *, TEXT *));
46 static void	txt_prompt __P((SCR *, TEXT *, ARG_CHAR_T, u_int32_t));
47 
48 /*
49  * ex_txt --
50  *	Get lines from the terminal for ex.
51  *
52  * PUBLIC: int ex_txt __P((SCR *, TEXTH *, ARG_CHAR_T, u_int32_t));
53  */
54 int
55 ex_txt(sp, tiqh, prompt, flags)
56 	SCR *sp;
57 	TEXTH *tiqh;
58 	ARG_CHAR_T prompt;
59 	u_int32_t flags;
60 {
61 	EVENT ev;
62 	GS *gp;
63 	TEXT ait, *ntp, *tp;
64 	carat_t carat_st;
65 	size_t cnt;
66 	int rval;
67 
68 	rval = 0;
69 
70 	/*
71 	 * Get a TEXT structure with some initial buffer space, reusing the
72 	 * last one if it's big enough.  (All TEXT bookkeeping fields default
73 	 * to 0 -- text_init() handles this.)
74 	 */
75 	if (tiqh->cqh_first != (void *)tiqh) {
76 		tp = tiqh->cqh_first;
77 		if (tp->q.cqe_next != (void *)tiqh || tp->lb_len < 32) {
78 			text_lfree(tiqh);
79 			goto newtp;
80 		}
81 		tp->len = 0;
82 	} else {
83 newtp:		if ((tp = text_init(sp, NULL, 0, 32)) == NULL)
84 			goto err;
85 		CIRCLEQ_INSERT_HEAD(tiqh, tp, q);
86 	}
87 
88 	/* Set the starting line number. */
89 	tp->lno = sp->lno + 1;
90 
91 	/*
92 	 * If it's a terminal, set up autoindent, put out the prompt, and
93 	 * set it up so we know we were suspended.  Otherwise, turn off
94 	 * the autoindent flag, as that requires less special casing below.
95 	 *
96 	 * XXX
97 	 * Historic practice is that ^Z suspended command mode (but, because
98 	 * it ran in cooked mode, it was unaffected by the autowrite option.)
99 	 * On restart, any "current" input was discarded, whether in insert
100 	 * mode or not, and ex was in command mode.  This code matches historic
101 	 * practice, but not 'cause it's easier.
102 	 */
103 	gp = sp->gp;
104 	if (F_ISSET(gp, G_SCRIPTED))
105 		LF_CLR(TXT_AUTOINDENT);
106 	else {
107 		if (LF_ISSET(TXT_AUTOINDENT)) {
108 			LF_SET(TXT_EOFCHAR);
109 			if (v_txt_auto(sp, sp->lno, NULL, 0, tp))
110 				goto err;
111 		}
112 		txt_prompt(sp, tp, prompt, flags);
113 	}
114 
115 	for (carat_st = C_NOTSET;;) {
116 		if (v_event_get(sp, &ev, 0, 0))
117 			goto err;
118 
119 		/* Deal with all non-character events. */
120 		switch (ev.e_event) {
121 		case E_CHARACTER:
122 			break;
123 		case E_ERR:
124 			goto err;
125 		case E_REPAINT:
126 		case E_WRESIZE:
127 			continue;
128 		case E_EOF:
129 			rval = 1;
130 			/* FALLTHROUGH */
131 		case E_INTERRUPT:
132 			/*
133 			 * Handle EOF/SIGINT events by discarding partially
134 			 * entered text and returning.  EOF returns failure,
135 			 * E_INTERRUPT returns success.
136 			 */
137 			goto notlast;
138 		default:
139 			v_event_err(sp, &ev);
140 			goto notlast;
141 		}
142 
143 		/*
144 		 * Deal with character events.
145 		 *
146 		 * Check to see if the character fits into the input buffer.
147 		 * (Use tp->len, ignore overwrite and non-printable chars.)
148 		 */
149 		BINC_GOTO(sp, tp->lb, tp->lb_len, tp->len + 1);
150 
151 		switch (ev.e_value) {
152 		case K_CR:
153 			/*
154 			 * !!!
155 			 * Historically, <carriage-return>'s in the command
156 			 * weren't special, so the ex parser would return an
157 			 * unknown command error message.  However, if they
158 			 * terminated the command if they were in a map.  I'm
159 			 * pretty sure this still isn't right, but it handles
160 			 * what I've seen so far.
161 			 */
162 			if (!F_ISSET(&ev.e_ch, CH_MAPPED))
163 				goto ins_ch;
164 			/* FALLTHROUGH */
165 		case K_NL:
166 			/*
167 			 * '\' can escape <carriage-return>/<newline>.  We
168 			 * don't discard the backslash because we need it
169 			 * to get the <newline> through the ex parser.
170 			 */
171 			if (LF_ISSET(TXT_BACKSLASH) &&
172 			    tp->len != 0 && tp->lb[tp->len - 1] == '\\')
173 				goto ins_ch;
174 
175 			/*
176 			 * CR returns from the ex command line.
177 			 *
178 			 * XXX
179 			 * Terminate with a nul, needed by filter.
180 			 */
181 			if (LF_ISSET(TXT_CR)) {
182 				tp->lb[tp->len] = '\0';
183 				goto done;
184 			}
185 
186 			/*
187 			 * '.' may terminate text input mode; free the current
188 			 * TEXT.
189 			 */
190 			if (LF_ISSET(TXT_DOTTERM) && tp->len == tp->ai + 1 &&
191 			    tp->lb[tp->len - 1] == '.') {
192 notlast:			CIRCLEQ_REMOVE(tiqh, tp, q);
193 				text_free(tp);
194 				goto done;
195 			}
196 
197 			/* Set up bookkeeping for the new line. */
198 			if ((ntp = text_init(sp, NULL, 0, 32)) == NULL)
199 				goto err;
200 			ntp->lno = tp->lno + 1;
201 
202 			/*
203 			 * Reset the autoindent line value.  0^D keeps the ai
204 			 * line from changing, ^D changes the level, even if
205 			 * there were no characters in the old line.  Note, if
206 			 * using the current tp structure, use the cursor as
207 			 * the length, the autoindent characters may have been
208 			 * erased.
209 			 */
210 			if (LF_ISSET(TXT_AUTOINDENT)) {
211 				if (carat_st == C_NOCHANGE) {
212 					if (v_txt_auto(sp,
213 					    OOBLNO, &ait, ait.ai, ntp))
214 						goto err;
215 					free(ait.lb);
216 				} else
217 					if (v_txt_auto(sp,
218 					    OOBLNO, tp, tp->len, ntp))
219 						goto err;
220 				carat_st = C_NOTSET;
221 			}
222 			txt_prompt(sp, ntp, prompt, flags);
223 
224 			/*
225 			 * Swap old and new TEXT's, and insert the new TEXT
226 			 * into the queue.
227 			 */
228 			tp = ntp;
229 			CIRCLEQ_INSERT_TAIL(tiqh, tp, q);
230 			break;
231 		case K_CARAT:			/* Delete autoindent chars. */
232 			if (tp->len <= tp->ai && LF_ISSET(TXT_AUTOINDENT))
233 				carat_st = C_CARATSET;
234 			goto ins_ch;
235 		case K_ZERO:			/* Delete autoindent chars. */
236 			if (tp->len <= tp->ai && LF_ISSET(TXT_AUTOINDENT))
237 				carat_st = C_ZEROSET;
238 			goto ins_ch;
239 		case K_CNTRLD:			/* Delete autoindent char. */
240 			/*
241 			 * !!!
242 			 * Historically, the ^D command took (but then ignored)
243 			 * a count.  For simplicity, we don't return it unless
244 			 * it's the first character entered.  The check for len
245 			 * equal to 0 is okay, TXT_AUTOINDENT won't be set.
246 			 */
247 			if (LF_ISSET(TXT_CNTRLD)) {
248 				for (cnt = 0; cnt < tp->len; ++cnt)
249 					if (!isblank(tp->lb[cnt]))
250 						break;
251 				if (cnt == tp->len) {
252 					tp->len = 1;
253 					tp->lb[0] = ev.e_c;
254 					tp->lb[1] = '\0';
255 
256 					/*
257 					 * Put out a line separator, in case
258 					 * the command fails.
259 					 */
260 					(void)putchar('\n');
261 					goto done;
262 				}
263 			}
264 
265 			/*
266 			 * POSIX 1003.1b-1993, paragraph 7.1.1.9, states that
267 			 * the EOF characters are discarded if there are other
268 			 * characters to process in the line, i.e. if the EOF
269 			 * is not the first character in the line.  For this
270 			 * reason, historic ex discarded the EOF characters,
271 			 * even if occurring in the middle of the input line.
272 			 * We match that historic practice.
273 			 *
274 			 * !!!
275 			 * The test for discarding in the middle of the line is
276 			 * done in the switch, because the CARAT forms are N+1,
277 			 * not N.
278 			 *
279 			 * !!!
280 			 * There's considerable magic to make the terminal code
281 			 * return the EOF character at all.  See that code for
282 			 * details.
283 			 */
284 			if (!LF_ISSET(TXT_AUTOINDENT) || tp->len == 0)
285 				continue;
286 			switch (carat_st) {
287 			case C_CARATSET:		/* ^^D */
288 				if (tp->len > tp->ai + 1)
289 					continue;
290 
291 				/* Save the ai string for later. */
292 				ait.lb = NULL;
293 				ait.lb_len = 0;
294 				BINC_GOTO(sp, ait.lb, ait.lb_len, tp->ai);
295 				memcpy(ait.lb, tp->lb, tp->ai);
296 				ait.ai = ait.len = tp->ai;
297 
298 				carat_st = C_NOCHANGE;
299 				goto leftmargin;
300 			case C_ZEROSET:			/* 0^D */
301 				if (tp->len > tp->ai + 1)
302 					continue;
303 
304 				carat_st = C_NOTSET;
305 leftmargin:			(void)gp->scr_ex_adjust(sp, EX_TERM_CE);
306 				tp->ai = tp->len = 0;
307 				break;
308 			case C_NOTSET:			/* ^D */
309 				if (tp->len > tp->ai)
310 					continue;
311 
312 				if (txt_dent(sp, tp))
313 					goto err;
314 				break;
315 			default:
316 				abort();
317 			}
318 
319 			/* Clear and redisplay the line. */
320 			(void)gp->scr_ex_adjust(sp, EX_TERM_CE);
321 			txt_prompt(sp, tp, prompt, flags);
322 			break;
323 		default:
324 			/*
325 			 * See the TXT_BEAUTIFY comment in vi/v_txt_ev.c.
326 			 *
327 			 * Silently eliminate any iscntrl() character that was
328 			 * not already handled specially, except for <tab> and
329 			 * <ff>.
330 			 */
331 ins_ch:			if (LF_ISSET(TXT_BEAUTIFY) && iscntrl(ev.e_c) &&
332 			    ev.e_value != K_FORMFEED && ev.e_value != K_TAB)
333 				break;
334 
335 			tp->lb[tp->len++] = ev.e_c;
336 			break;
337 		}
338 	}
339 	/* NOTREACHED */
340 
341 done:	return (rval);
342 
343 err:
344 alloc_err:
345 	return (1);
346 }
347 
348 /*
349  * txt_prompt --
350  *	Display the ex prompt, line number, ai characters.  Characters had
351  *	better be printable by the terminal driver, but that's its problem,
352  *	not ours.
353  */
354 static void
355 txt_prompt(sp, tp, prompt, flags)
356 	SCR *sp;
357 	TEXT *tp;
358 	ARG_CHAR_T prompt;
359 	u_int32_t flags;
360 {
361 	/* Display the prompt. */
362 	if (LF_ISSET(TXT_PROMPT))
363 		(void)printf("%c", prompt);
364 
365 	/* Display the line number. */
366 	if (LF_ISSET(TXT_NUMBER) && O_ISSET(sp, O_NUMBER))
367 		(void)printf("%6lu  ", (u_long)tp->lno);
368 
369 	/* Print out autoindent string. */
370 	if (LF_ISSET(TXT_AUTOINDENT))
371 		(void)printf("%.*s", (int)tp->ai, tp->lb);
372 	(void)fflush(stdout);
373 }
374 
375 /*
376  * txt_dent --
377  *	Handle ^D outdents.
378  *
379  * Ex version of vi/v_ntext.c:txt_dent().  See that code for the (usual)
380  * ranting and raving.  This is a fair bit simpler as ^T isn't special.
381  */
382 static int
383 txt_dent(sp, tp)
384 	SCR *sp;
385 	TEXT *tp;
386 {
387 	u_long sw, ts;
388 	size_t cno, off, scno, spaces, tabs;
389 
390 	ts = O_VAL(sp, O_TABSTOP);
391 	sw = O_VAL(sp, O_SHIFTWIDTH);
392 
393 	/* Get the current screen column. */
394 	for (off = scno = 0; off < tp->len; ++off)
395 		if (tp->lb[off] == '\t')
396 			scno += COL_OFF(scno, ts);
397 		else
398 			++scno;
399 
400 	/* Get the previous shiftwidth column. */
401 	cno = scno;
402 	scno -= --scno % sw;
403 
404 	/*
405 	 * Since we don't know what comes before the character(s) being
406 	 * deleted, we have to resolve the autoindent characters .  The
407 	 * example is a <tab>, which doesn't take up a full shiftwidth
408 	 * number of columns because it's preceded by <space>s.  This is
409 	 * easy to get if the user sets shiftwidth to a value less than
410 	 * tabstop, and then uses ^T to indent, and ^D to outdent.
411 	 *
412 	 * Count up spaces/tabs needed to get to the target.
413 	 */
414 	for (cno = 0, tabs = 0; cno + COL_OFF(cno, ts) <= scno; ++tabs)
415 		cno += COL_OFF(cno, ts);
416 	spaces = scno - cno;
417 
418 	/* Make sure there's enough room. */
419 	BINC_RET(sp, tp->lb, tp->lb_len, tabs + spaces + 1);
420 
421 	/* Adjust the final ai character count. */
422 	tp->ai = tabs + spaces;
423 
424 	/* Enter the replacement characters. */
425 	for (tp->len = 0; tabs > 0; --tabs)
426 		tp->lb[tp->len++] = '\t';
427 	for (; spaces > 0; --spaces)
428 		tp->lb[tp->len++] = ' ';
429 	return (0);
430 }
431