xref: /openbsd/usr.bin/vi/ex/ex_global.c (revision 404b540a)
1 /*	$OpenBSD: ex_global.c,v 1.10 2007/05/14 12:32:29 pyr 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 #ifndef lint
15 static const char sccsid[] = "@(#)ex_global.c	10.22 (Berkeley) 10/10/96";
16 #endif /* not lint */
17 
18 #include <sys/types.h>
19 #include <sys/queue.h>
20 
21 #include <bitstring.h>
22 #include <ctype.h>
23 #include <errno.h>
24 #include <limits.h>
25 #include <stdio.h>
26 #include <stdlib.h>
27 #include <string.h>
28 #include <unistd.h>
29 
30 #include "../common/common.h"
31 
32 enum which {GLOBAL, V};
33 
34 static int ex_g_setup(SCR *, EXCMD *, enum which);
35 
36 /*
37  * ex_global -- [line [,line]] g[lobal][!] /pattern/ [commands]
38  *	Exec on lines matching a pattern.
39  *
40  * PUBLIC: int ex_global(SCR *, EXCMD *);
41  */
42 int
43 ex_global(sp, cmdp)
44 	SCR *sp;
45 	EXCMD *cmdp;
46 {
47 	return (ex_g_setup(sp,
48 	    cmdp, FL_ISSET(cmdp->iflags, E_C_FORCE) ? V : GLOBAL));
49 }
50 
51 /*
52  * ex_v -- [line [,line]] v /pattern/ [commands]
53  *	Exec on lines not matching a pattern.
54  *
55  * PUBLIC: int ex_v(SCR *, EXCMD *);
56  */
57 int
58 ex_v(sp, cmdp)
59 	SCR *sp;
60 	EXCMD *cmdp;
61 {
62 	return (ex_g_setup(sp, cmdp, V));
63 }
64 
65 /*
66  * ex_g_setup --
67  *	Ex global and v commands.
68  */
69 static int
70 ex_g_setup(sp, cmdp, cmd)
71 	SCR *sp;
72 	EXCMD *cmdp;
73 	enum which cmd;
74 {
75 	CHAR_T *ptrn, *p, *t;
76 	EXCMD *ecp;
77 	MARK abs_mark;
78 	RANGE *rp;
79 	busy_t btype;
80 	recno_t start, end;
81 	regex_t *re;
82 	regmatch_t match[1];
83 	size_t len;
84 	int cnt, delim, eval;
85 	char *dbp;
86 
87 	NEEDFILE(sp, cmdp);
88 
89 	if (F_ISSET(sp, SC_EX_GLOBAL)) {
90 		msgq(sp, M_ERR,
91 	"124|The %s command can't be used as part of a global or v command",
92 		    cmdp->cmd->name);
93 		return (1);
94 	}
95 
96 	/*
97 	 * Skip leading white space.  Historic vi allowed any non-alphanumeric
98 	 * to serve as the global command delimiter.
99 	 */
100 	if (cmdp->argc == 0)
101 		goto usage;
102 	for (p = cmdp->argv[0]->bp; isblank(*p); ++p);
103 	if (*p == '\0' || isalnum(*p) ||
104 	    *p == '\\' || *p == '|' || *p == '\n') {
105 usage:		ex_emsg(sp, cmdp->cmd->usage, EXM_USAGE);
106 		return (1);
107 	}
108 	delim = *p++;
109 
110 	/*
111 	 * Get the pattern string, toss escaped characters.
112 	 *
113 	 * QUOTING NOTE:
114 	 * Only toss an escaped character if it escapes a delimiter.
115 	 */
116 	for (ptrn = t = p;;) {
117 		if (p[0] == '\0' || p[0] == delim) {
118 			if (p[0] == delim)
119 				++p;
120 			/*
121 			 * !!!
122 			 * Nul terminate the pattern string -- it's passed
123 			 * to regcomp which doesn't understand anything else.
124 			 */
125 			*t = '\0';
126 			break;
127 		}
128 		if (p[0] == '\\') {
129 			if (p[1] == delim)
130 				++p;
131 			else if (p[1] == '\\')
132 				*t++ = *p++;
133 		}
134 		*t++ = *p++;
135 	}
136 
137 	/* If the pattern string is empty, use the last one. */
138 	if (*ptrn == '\0') {
139 		if (sp->re == NULL) {
140 			ex_emsg(sp, NULL, EXM_NOPREVRE);
141 			return (1);
142 		}
143 
144 		/* Re-compile the RE if necessary. */
145 		if (!F_ISSET(sp, SC_RE_SEARCH) && re_compile(sp,
146 		    sp->re, sp->re_len, NULL, NULL, &sp->re_c, RE_C_SEARCH))
147 			return (1);
148 	} else {
149 		/* Compile the RE. */
150 		if (re_compile(sp, ptrn, t - ptrn,
151 		    &sp->re, &sp->re_len, &sp->re_c, RE_C_SEARCH))
152 			return (1);
153 
154 		/*
155 		 * Set saved RE.  Historic practice is that globals set
156 		 * direction as well as the RE.
157 		 */
158 		sp->searchdir = FORWARD;
159 	}
160 	re = &sp->re_c;
161 
162 	/* The global commands always set the previous context mark. */
163 	abs_mark.lno = sp->lno;
164 	abs_mark.cno = sp->cno;
165 	if (mark_set(sp, ABSMARK1, &abs_mark, 1))
166 		return (1);
167 
168 	/* Get an EXCMD structure. */
169 	CALLOC_RET(sp, ecp, EXCMD *, 1, sizeof(EXCMD));
170 	CIRCLEQ_INIT(&ecp->rq);
171 
172 	/*
173 	 * Get a copy of the command string; the default command is print.
174 	 * Don't worry about a set of <blank>s with no command, that will
175 	 * default to print in the ex parser.  We need to have two copies
176 	 * because the ex parser may step on the command string when it's
177 	 * parsing it.
178 	 */
179 	if ((len = cmdp->argv[0]->len - (p - cmdp->argv[0]->bp)) == 0) {
180 		p = "pp";
181 		len = 1;
182 	}
183 
184 	MALLOC_RET(sp, ecp->cp, char *, len * 2);
185 	ecp->o_cp = ecp->cp;
186 	ecp->o_clen = len;
187 	memcpy(ecp->cp + len, p, len);
188 	ecp->range_lno = OOBLNO;
189 	FL_SET(ecp->agv_flags, cmd == GLOBAL ? AGV_GLOBAL : AGV_V);
190 	LIST_INSERT_HEAD(&sp->gp->ecq, ecp, q);
191 
192 	/*
193 	 * For each line...  The semantics of global matching are that we first
194 	 * have to decide which lines are going to get passed to the command,
195 	 * and then pass them to the command, ignoring other changes.  There's
196 	 * really no way to do this in a single pass, since arbitrary line
197 	 * creation, deletion and movement can be done in the ex command.  For
198 	 * example, a good vi clone test is ":g/X/mo.-3", or "g/X/.,.+1d".
199 	 * What we do is create linked list of lines that are tracked through
200 	 * each ex command.  There's a callback routine which the DB interface
201 	 * routines call when a line is created or deleted.  This doesn't help
202 	 * the layering much.
203 	 */
204 	btype = BUSY_ON;
205 	cnt = INTERRUPT_CHECK;
206 	for (start = cmdp->addr1.lno,
207 	    end = cmdp->addr2.lno; start <= end; ++start) {
208 		if (cnt-- == 0) {
209 			if (INTERRUPTED(sp)) {
210 				LIST_REMOVE(ecp, q);
211 				free(ecp->cp);
212 				free(ecp);
213 				break;
214 			}
215 			search_busy(sp, btype);
216 			btype = BUSY_UPDATE;
217 			cnt = INTERRUPT_CHECK;
218 		}
219 		if (db_get(sp, start, DBG_FATAL, &dbp, &len))
220 			return (1);
221 		match[0].rm_so = 0;
222 		match[0].rm_eo = len;
223 		switch (eval =
224 		    regexec(&sp->re_c, dbp, 0, match, REG_STARTEND)) {
225 		case 0:
226 			if (cmd == V)
227 				continue;
228 			break;
229 		case REG_NOMATCH:
230 			if (cmd == GLOBAL)
231 				continue;
232 			break;
233 		default:
234 			re_error(sp, eval, &sp->re_c);
235 			break;
236 		}
237 
238 		/* If follows the last entry, extend the last entry's range. */
239 		if ((rp = CIRCLEQ_LAST(&ecp->rq)) != CIRCLEQ_END(&ecp->rq) &&
240 		    rp->stop == start - 1) {
241 			++rp->stop;
242 			continue;
243 		}
244 
245 		/* Allocate a new range, and append it to the list. */
246 		CALLOC(sp, rp, RANGE *, 1, sizeof(RANGE));
247 		if (rp == NULL)
248 			return (1);
249 		rp->start = rp->stop = start;
250 		CIRCLEQ_INSERT_TAIL(&ecp->rq, rp, q);
251 	}
252 	search_busy(sp, BUSY_OFF);
253 	return (0);
254 }
255 
256 /*
257  * ex_g_insdel --
258  *	Update the ranges based on an insertion or deletion.
259  *
260  * PUBLIC: int ex_g_insdel(SCR *, lnop_t, recno_t);
261  */
262 int
263 ex_g_insdel(sp, op, lno)
264 	SCR *sp;
265 	lnop_t op;
266 	recno_t lno;
267 {
268 	EXCMD *ecp;
269 	RANGE *nrp, *rp;
270 
271 	/* All insert/append operations are done as inserts. */
272 	if (op == LINE_APPEND)
273 		abort();
274 
275 	if (op == LINE_RESET)
276 		return (0);
277 
278 	LIST_FOREACH(ecp, &sp->gp->ecq, q) {
279 		if (!FL_ISSET(ecp->agv_flags, AGV_AT | AGV_GLOBAL | AGV_V))
280 			continue;
281 		for (rp = CIRCLEQ_FIRST(&ecp->rq); rp != CIRCLEQ_END(&ecp->rq);
282 		    rp = nrp) {
283 			nrp = CIRCLEQ_NEXT(rp, q);
284 
285 			/* If range less than the line, ignore it. */
286 			if (rp->stop < lno)
287 				continue;
288 
289 			/*
290 			 * If range greater than the line, decrement or
291 			 * increment the range.
292 			 */
293 			if (rp->start > lno) {
294 				if (op == LINE_DELETE) {
295 					--rp->start;
296 					--rp->stop;
297 				} else {
298 					++rp->start;
299 					++rp->stop;
300 				}
301 				continue;
302 			}
303 
304 			/*
305 			 * Lno is inside the range, decrement the end point
306 			 * for deletion, and split the range for insertion.
307 			 * In the latter case, since we're inserting a new
308 			 * element, neither range can be exhausted.
309 			 */
310 			if (op == LINE_DELETE) {
311 				if (rp->start > --rp->stop) {
312 					CIRCLEQ_REMOVE(&ecp->rq, rp, q);
313 					free(rp);
314 				}
315 			} else {
316 				CALLOC_RET(sp, nrp, RANGE *, 1, sizeof(RANGE));
317 				nrp->start = lno + 1;
318 				nrp->stop = rp->stop + 1;
319 				rp->stop = lno - 1;
320 				CIRCLEQ_INSERT_AFTER(&ecp->rq, rp, nrp, q);
321 				rp = nrp;
322 			}
323 		}
324 
325 		/*
326 		 * If the command deleted/inserted lines, the cursor moves to
327 		 * the line after the deleted/inserted line.
328 		 */
329 		ecp->range_lno = lno;
330 	}
331 	return (0);
332 }
333