xref: /openbsd/usr.bin/vi/ex/ex_global.c (revision 721c3ea3)
1*721c3ea3Smartijn /*	$OpenBSD: ex_global.c,v 1.17 2016/05/27 09:18:12 martijn Exp $	*/
2d4e7c603Sniklas 
3df930be7Sderaadt /*-
4df930be7Sderaadt  * Copyright (c) 1992, 1993, 1994
5df930be7Sderaadt  *	The Regents of the University of California.  All rights reserved.
645f2ab88Sderaadt  * Copyright (c) 1992, 1993, 1994, 1995, 1996
745f2ab88Sderaadt  *	Keith Bostic.  All rights reserved.
8df930be7Sderaadt  *
945f2ab88Sderaadt  * See the LICENSE file for redistribution information.
10df930be7Sderaadt  */
11df930be7Sderaadt 
1245f2ab88Sderaadt #include "config.h"
1345f2ab88Sderaadt 
14df930be7Sderaadt #include <sys/types.h>
15df930be7Sderaadt #include <sys/queue.h>
16df930be7Sderaadt 
17df930be7Sderaadt #include <bitstring.h>
18df930be7Sderaadt #include <ctype.h>
19df930be7Sderaadt #include <errno.h>
20df930be7Sderaadt #include <limits.h>
21df930be7Sderaadt #include <stdio.h>
22df930be7Sderaadt #include <stdlib.h>
23df930be7Sderaadt #include <string.h>
24df930be7Sderaadt #include <unistd.h>
25df930be7Sderaadt 
2645f2ab88Sderaadt #include "../common/common.h"
27df930be7Sderaadt 
2845f2ab88Sderaadt enum which {GLOBAL, V};
29df930be7Sderaadt 
30c72b5b24Smillert static int ex_g_setup(SCR *, EXCMD *, enum which);
31df930be7Sderaadt 
32df930be7Sderaadt /*
33df930be7Sderaadt  * ex_global -- [line [,line]] g[lobal][!] /pattern/ [commands]
34df930be7Sderaadt  *	Exec on lines matching a pattern.
3545f2ab88Sderaadt  *
36c72b5b24Smillert  * PUBLIC: int ex_global(SCR *, EXCMD *);
37df930be7Sderaadt  */
38df930be7Sderaadt int
ex_global(SCR * sp,EXCMD * cmdp)39486aa1f0Sbentley ex_global(SCR *sp, EXCMD *cmdp)
40df930be7Sderaadt {
4145f2ab88Sderaadt 	return (ex_g_setup(sp,
4245f2ab88Sderaadt 	    cmdp, FL_ISSET(cmdp->iflags, E_C_FORCE) ? V : GLOBAL));
43df930be7Sderaadt }
44df930be7Sderaadt 
45df930be7Sderaadt /*
4645f2ab88Sderaadt  * ex_v -- [line [,line]] v /pattern/ [commands]
47df930be7Sderaadt  *	Exec on lines not matching a pattern.
4845f2ab88Sderaadt  *
49c72b5b24Smillert  * PUBLIC: int ex_v(SCR *, EXCMD *);
50df930be7Sderaadt  */
51df930be7Sderaadt int
ex_v(SCR * sp,EXCMD * cmdp)52486aa1f0Sbentley ex_v(SCR *sp, EXCMD *cmdp)
53df930be7Sderaadt {
5445f2ab88Sderaadt 	return (ex_g_setup(sp, cmdp, V));
55df930be7Sderaadt }
56df930be7Sderaadt 
5745f2ab88Sderaadt /*
5845f2ab88Sderaadt  * ex_g_setup --
5945f2ab88Sderaadt  *	Ex global and v commands.
6045f2ab88Sderaadt  */
61df930be7Sderaadt static int
ex_g_setup(SCR * sp,EXCMD * cmdp,enum which cmd)62486aa1f0Sbentley ex_g_setup(SCR *sp, EXCMD *cmdp, enum which cmd)
63df930be7Sderaadt {
64*721c3ea3Smartijn 	CHAR_T *ptrn, *p, *t;
6545f2ab88Sderaadt 	EXCMD *ecp;
6604626f32Sray 	MARK abs_mark;
67df930be7Sderaadt 	RANGE *rp;
6845f2ab88Sderaadt 	busy_t btype;
6945f2ab88Sderaadt 	recno_t start, end;
7045f2ab88Sderaadt 	regex_t *re;
71df930be7Sderaadt 	regmatch_t match[1];
7245f2ab88Sderaadt 	size_t len;
7345f2ab88Sderaadt 	int cnt, delim, eval;
7432ceefb1Sdownsj 	char *dbp;
7545f2ab88Sderaadt 
7645f2ab88Sderaadt 	NEEDFILE(sp, cmdp);
7745f2ab88Sderaadt 
7845f2ab88Sderaadt 	if (F_ISSET(sp, SC_EX_GLOBAL)) {
7945f2ab88Sderaadt 		msgq(sp, M_ERR,
80a0b15055Smillert 	"The %s command can't be used as part of a global or v command",
8145f2ab88Sderaadt 		    cmdp->cmd->name);
8245f2ab88Sderaadt 		return (1);
8345f2ab88Sderaadt 	}
84df930be7Sderaadt 
85df930be7Sderaadt 	/*
8645f2ab88Sderaadt 	 * Skip leading white space.  Historic vi allowed any non-alphanumeric
8745f2ab88Sderaadt 	 * to serve as the global command delimiter.
88df930be7Sderaadt 	 */
8945f2ab88Sderaadt 	if (cmdp->argc == 0)
9045f2ab88Sderaadt 		goto usage;
91df930be7Sderaadt 	for (p = cmdp->argv[0]->bp; isblank(*p); ++p);
9245f2ab88Sderaadt 	if (*p == '\0' || isalnum(*p) ||
9345f2ab88Sderaadt 	    *p == '\\' || *p == '|' || *p == '\n') {
9445f2ab88Sderaadt usage:		ex_emsg(sp, cmdp->cmd->usage, EXM_USAGE);
95df930be7Sderaadt 		return (1);
96df930be7Sderaadt 	}
97df930be7Sderaadt 	delim = *p++;
98df930be7Sderaadt 
99df930be7Sderaadt 	/*
100df930be7Sderaadt 	 * Get the pattern string, toss escaped characters.
101df930be7Sderaadt 	 *
102df930be7Sderaadt 	 * QUOTING NOTE:
103df930be7Sderaadt 	 * Only toss an escaped character if it escapes a delimiter.
104df930be7Sderaadt 	 */
105df930be7Sderaadt 	for (ptrn = t = p;;) {
106df930be7Sderaadt 		if (p[0] == '\0' || p[0] == delim) {
107df930be7Sderaadt 			if (p[0] == delim)
108df930be7Sderaadt 				++p;
109df930be7Sderaadt 			/*
110df930be7Sderaadt 			 * !!!
111df930be7Sderaadt 			 * Nul terminate the pattern string -- it's passed
112df930be7Sderaadt 			 * to regcomp which doesn't understand anything else.
113df930be7Sderaadt 			 */
114df930be7Sderaadt 			*t = '\0';
115df930be7Sderaadt 			break;
116df930be7Sderaadt 		}
117ddc53e66Smiod 		if (p[0] == '\\') {
11845f2ab88Sderaadt 			if (p[1] == delim)
119df930be7Sderaadt 				++p;
12045f2ab88Sderaadt 			else if (p[1] == '\\')
12145f2ab88Sderaadt 				*t++ = *p++;
122ddc53e66Smiod 		}
123df930be7Sderaadt 		*t++ = *p++;
124df930be7Sderaadt 	}
125df930be7Sderaadt 
126df930be7Sderaadt 	/* If the pattern string is empty, use the last one. */
127df930be7Sderaadt 	if (*ptrn == '\0') {
12845f2ab88Sderaadt 		if (sp->re == NULL) {
12945f2ab88Sderaadt 			ex_emsg(sp, NULL, EXM_NOPREVRE);
130df930be7Sderaadt 			return (1);
131df930be7Sderaadt 		}
13245f2ab88Sderaadt 
133e91eb1deSmickey 		/* Re-compile the RE if necessary. */
134e91eb1deSmickey 		if (!F_ISSET(sp, SC_RE_SEARCH) && re_compile(sp,
135e91eb1deSmickey 		    sp->re, sp->re_len, NULL, NULL, &sp->re_c, RE_C_SEARCH))
13645f2ab88Sderaadt 			return (1);
137df930be7Sderaadt 	} else {
138df930be7Sderaadt 		/* Compile the RE. */
139e91eb1deSmickey 		if (re_compile(sp, ptrn, t - ptrn,
140e91eb1deSmickey 		    &sp->re, &sp->re_len, &sp->re_c, RE_C_SEARCH))
141df930be7Sderaadt 			return (1);
142df930be7Sderaadt 
143df930be7Sderaadt 		/*
14445f2ab88Sderaadt 		 * Set saved RE.  Historic practice is that globals set
14545f2ab88Sderaadt 		 * direction as well as the RE.
146df930be7Sderaadt 		 */
147df930be7Sderaadt 		sp->searchdir = FORWARD;
148df930be7Sderaadt 	}
14945f2ab88Sderaadt 	re = &sp->re_c;
150df930be7Sderaadt 
151df930be7Sderaadt 	/* The global commands always set the previous context mark. */
15204626f32Sray 	abs_mark.lno = sp->lno;
15304626f32Sray 	abs_mark.cno = sp->cno;
15404626f32Sray 	if (mark_set(sp, ABSMARK1, &abs_mark, 1))
15545f2ab88Sderaadt 		return (1);
15645f2ab88Sderaadt 
15745f2ab88Sderaadt 	/* Get an EXCMD structure. */
15854d5890cSmmcc 	CALLOC_RET(sp, ecp, 1, sizeof(EXCMD));
159c889a549Skrw 	TAILQ_INIT(&ecp->rq);
16045f2ab88Sderaadt 
16145f2ab88Sderaadt 	/*
16245f2ab88Sderaadt 	 * Get a copy of the command string; the default command is print.
16345f2ab88Sderaadt 	 * Don't worry about a set of <blank>s with no command, that will
16445f2ab88Sderaadt 	 * default to print in the ex parser.  We need to have two copies
16545f2ab88Sderaadt 	 * because the ex parser may step on the command string when it's
16645f2ab88Sderaadt 	 * parsing it.
16745f2ab88Sderaadt 	 */
16832ceefb1Sdownsj 	if ((len = cmdp->argv[0]->len - (p - cmdp->argv[0]->bp)) == 0) {
16945f2ab88Sderaadt 		p = "pp";
17045f2ab88Sderaadt 		len = 1;
17145f2ab88Sderaadt 	}
17245f2ab88Sderaadt 
17354d5890cSmmcc 	MALLOC_RET(sp, ecp->cp, len * 2);
17445f2ab88Sderaadt 	ecp->o_cp = ecp->cp;
17545f2ab88Sderaadt 	ecp->o_clen = len;
176e91eb1deSmickey 	memcpy(ecp->cp + len, p, len);
17745f2ab88Sderaadt 	ecp->range_lno = OOBLNO;
17845f2ab88Sderaadt 	FL_SET(ecp->agv_flags, cmd == GLOBAL ? AGV_GLOBAL : AGV_V);
17945f2ab88Sderaadt 	LIST_INSERT_HEAD(&sp->gp->ecq, ecp, q);
180df930be7Sderaadt 
181df930be7Sderaadt 	/*
182df930be7Sderaadt 	 * For each line...  The semantics of global matching are that we first
183df930be7Sderaadt 	 * have to decide which lines are going to get passed to the command,
184df930be7Sderaadt 	 * and then pass them to the command, ignoring other changes.  There's
185df930be7Sderaadt 	 * really no way to do this in a single pass, since arbitrary line
186df930be7Sderaadt 	 * creation, deletion and movement can be done in the ex command.  For
187df930be7Sderaadt 	 * example, a good vi clone test is ":g/X/mo.-3", or "g/X/.,.+1d".
188df930be7Sderaadt 	 * What we do is create linked list of lines that are tracked through
189df930be7Sderaadt 	 * each ex command.  There's a callback routine which the DB interface
190df930be7Sderaadt 	 * routines call when a line is created or deleted.  This doesn't help
191df930be7Sderaadt 	 * the layering much.
192df930be7Sderaadt 	 */
19345f2ab88Sderaadt 	btype = BUSY_ON;
19445f2ab88Sderaadt 	cnt = INTERRUPT_CHECK;
19545f2ab88Sderaadt 	for (start = cmdp->addr1.lno,
19645f2ab88Sderaadt 	    end = cmdp->addr2.lno; start <= end; ++start) {
19745f2ab88Sderaadt 		if (cnt-- == 0) {
19845f2ab88Sderaadt 			if (INTERRUPTED(sp)) {
19945f2ab88Sderaadt 				LIST_REMOVE(ecp, q);
20045f2ab88Sderaadt 				free(ecp->cp);
20145f2ab88Sderaadt 				free(ecp);
20245f2ab88Sderaadt 				break;
203df930be7Sderaadt 			}
20445f2ab88Sderaadt 			search_busy(sp, btype);
20545f2ab88Sderaadt 			btype = BUSY_UPDATE;
20645f2ab88Sderaadt 			cnt = INTERRUPT_CHECK;
20745f2ab88Sderaadt 		}
20832ceefb1Sdownsj 		if (db_get(sp, start, DBG_FATAL, &dbp, &len))
20945f2ab88Sderaadt 			return (1);
210df930be7Sderaadt 		match[0].rm_so = 0;
211df930be7Sderaadt 		match[0].rm_eo = len;
21232ceefb1Sdownsj 		switch (eval =
21332ceefb1Sdownsj 		    regexec(&sp->re_c, dbp, 0, match, REG_STARTEND)) {
214df930be7Sderaadt 		case 0:
21545f2ab88Sderaadt 			if (cmd == V)
216df930be7Sderaadt 				continue;
217df930be7Sderaadt 			break;
218df930be7Sderaadt 		case REG_NOMATCH:
219df930be7Sderaadt 			if (cmd == GLOBAL)
220df930be7Sderaadt 				continue;
221df930be7Sderaadt 			break;
222df930be7Sderaadt 		default:
22345f2ab88Sderaadt 			re_error(sp, eval, &sp->re_c);
22445f2ab88Sderaadt 			break;
225df930be7Sderaadt 		}
226df930be7Sderaadt 
227df930be7Sderaadt 		/* If follows the last entry, extend the last entry's range. */
228c889a549Skrw 		if ((rp = TAILQ_LAST(&ecp->rq, _rh)) && rp->stop == start - 1) {
229df930be7Sderaadt 			++rp->stop;
230df930be7Sderaadt 			continue;
231df930be7Sderaadt 		}
232df930be7Sderaadt 
233df930be7Sderaadt 		/* Allocate a new range, and append it to the list. */
23454d5890cSmmcc 		CALLOC(sp, rp, 1, sizeof(RANGE));
235df930be7Sderaadt 		if (rp == NULL)
236df930be7Sderaadt 			return (1);
23745f2ab88Sderaadt 		rp->start = rp->stop = start;
238c889a549Skrw 		TAILQ_INSERT_TAIL(&ecp->rq, rp, q);
239df930be7Sderaadt 	}
24045f2ab88Sderaadt 	search_busy(sp, BUSY_OFF);
24145f2ab88Sderaadt 	return (0);
242df930be7Sderaadt }
243df930be7Sderaadt 
244df930be7Sderaadt /*
24545f2ab88Sderaadt  * ex_g_insdel --
246df930be7Sderaadt  *	Update the ranges based on an insertion or deletion.
24745f2ab88Sderaadt  *
248c72b5b24Smillert  * PUBLIC: int ex_g_insdel(SCR *, lnop_t, recno_t);
249df930be7Sderaadt  */
25045f2ab88Sderaadt int
ex_g_insdel(SCR * sp,lnop_t op,recno_t lno)251486aa1f0Sbentley ex_g_insdel(SCR *sp, lnop_t op, recno_t lno)
252df930be7Sderaadt {
25345f2ab88Sderaadt 	EXCMD *ecp;
254df930be7Sderaadt 	RANGE *nrp, *rp;
255df930be7Sderaadt 
25645f2ab88Sderaadt 	/* All insert/append operations are done as inserts. */
25745f2ab88Sderaadt 	if (op == LINE_APPEND)
25845f2ab88Sderaadt 		abort();
259df930be7Sderaadt 
26045f2ab88Sderaadt 	if (op == LINE_RESET)
26145f2ab88Sderaadt 		return (0);
26245f2ab88Sderaadt 
263ce7de1e9Sotto 	LIST_FOREACH(ecp, &sp->gp->ecq, q) {
26445f2ab88Sderaadt 		if (!FL_ISSET(ecp->agv_flags, AGV_AT | AGV_GLOBAL | AGV_V))
26545f2ab88Sderaadt 			continue;
266c889a549Skrw 		for (rp = TAILQ_FIRST(&ecp->rq); rp != NULL; rp = nrp) {
267c889a549Skrw 			nrp = TAILQ_NEXT(rp, q);
26845f2ab88Sderaadt 
269df930be7Sderaadt 			/* If range less than the line, ignore it. */
270df930be7Sderaadt 			if (rp->stop < lno)
271df930be7Sderaadt 				continue;
27245f2ab88Sderaadt 
27345f2ab88Sderaadt 			/*
27445f2ab88Sderaadt 			 * If range greater than the line, decrement or
27545f2ab88Sderaadt 			 * increment the range.
27645f2ab88Sderaadt 			 */
277df930be7Sderaadt 			if (rp->start > lno) {
27845f2ab88Sderaadt 				if (op == LINE_DELETE) {
279df930be7Sderaadt 					--rp->start;
280df930be7Sderaadt 					--rp->stop;
28145f2ab88Sderaadt 				} else {
282df930be7Sderaadt 					++rp->start;
283df930be7Sderaadt 					++rp->stop;
28445f2ab88Sderaadt 				}
285df930be7Sderaadt 				continue;
286df930be7Sderaadt 			}
28745f2ab88Sderaadt 
288df930be7Sderaadt 			/*
28945f2ab88Sderaadt 			 * Lno is inside the range, decrement the end point
29045f2ab88Sderaadt 			 * for deletion, and split the range for insertion.
29145f2ab88Sderaadt 			 * In the latter case, since we're inserting a new
29245f2ab88Sderaadt 			 * element, neither range can be exhausted.
293df930be7Sderaadt 			 */
29445f2ab88Sderaadt 			if (op == LINE_DELETE) {
29545f2ab88Sderaadt 				if (rp->start > --rp->stop) {
296c889a549Skrw 					TAILQ_REMOVE(&ecp->rq, rp, q);
29745f2ab88Sderaadt 					free(rp);
298df930be7Sderaadt 				}
29945f2ab88Sderaadt 			} else {
30054d5890cSmmcc 				CALLOC_RET(sp, nrp, 1, sizeof(RANGE));
301df930be7Sderaadt 				nrp->start = lno + 1;
302df930be7Sderaadt 				nrp->stop = rp->stop + 1;
303df930be7Sderaadt 				rp->stop = lno - 1;
304c889a549Skrw 				TAILQ_INSERT_AFTER(&ecp->rq, rp, nrp, q);
305df930be7Sderaadt 				rp = nrp;
306df930be7Sderaadt 			}
307df930be7Sderaadt 		}
30845f2ab88Sderaadt 
309df930be7Sderaadt 		/*
310df930be7Sderaadt 		 * If the command deleted/inserted lines, the cursor moves to
311df930be7Sderaadt 		 * the line after the deleted/inserted line.
312df930be7Sderaadt 		 */
31345f2ab88Sderaadt 		ecp->range_lno = lno;
31445f2ab88Sderaadt 	}
31545f2ab88Sderaadt 	return (0);
316df930be7Sderaadt }
317