Home | History | Annotate | Line # | Download | only in ed
      1 /*	$NetBSD: sub.c,v 1.8 2025/09/17 20:35:11 rillig Exp $	*/
      2 
      3 /* sub.c: This file contains the substitution routines for the ed
      4    line editor */
      5 /*-
      6  * Copyright (c) 1993 Andrew Moore, Talke Studio.
      7  * All rights reserved.
      8  *
      9  * Redistribution and use in source and binary forms, with or without
     10  * modification, are permitted provided that the following conditions
     11  * are met:
     12  * 1. Redistributions of source code must retain the above copyright
     13  *    notice, this list of conditions and the following disclaimer.
     14  * 2. Redistributions in binary form must reproduce the above copyright
     15  *    notice, this list of conditions and the following disclaimer in the
     16  *    documentation and/or other materials provided with the distribution.
     17  *
     18  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
     19  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     20  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     21  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
     22  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     23  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     24  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     25  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     26  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     27  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     28  * SUCH DAMAGE.
     29  */
     30 
     31 #include <sys/cdefs.h>
     32 #ifndef lint
     33 #if 0
     34 static char *rcsid = "@(#)sub.c,v 1.1 1994/02/01 00:34:44 alm Exp";
     35 #else
     36 __RCSID("$NetBSD: sub.c,v 1.8 2025/09/17 20:35:11 rillig Exp $");
     37 #endif
     38 #endif /* not lint */
     39 
     40 #include "ed.h"
     41 
     42 
     43 char *rhbuf;			/* rhs substitution buffer */
     44 int rhbufsz;			/* rhs substitution buffer size */
     45 int rhbufi;			/* rhs substitution buffer index */
     46 
     47 /* extract_subst_tail: extract substitution tail from the command buffer */
     48 int
     49 extract_subst_tail(int *flagp, long *np)
     50 {
     51 	char delimiter;
     52 
     53 	*flagp = *np = 0;
     54 	if ((delimiter = *ibufp) == '\n') {
     55 		rhbufi = 0;
     56 		*flagp = GPR;
     57 		return 0;
     58 	} else if (extract_subst_template() == NULL)
     59 		return  ERR;
     60 	else if (*ibufp == '\n') {
     61 		*flagp = GPR;
     62 		return 0;
     63 	} else if (*ibufp == delimiter)
     64 		ibufp++;
     65 	if ('1' <= *ibufp && *ibufp <= '9') {
     66 		STRTOL(*np, ibufp);
     67 		return 0;
     68 	} else if (*ibufp == 'g') {
     69 		ibufp++;
     70 		*flagp = GSG;
     71 		return 0;
     72 	}
     73 	return 0;
     74 }
     75 
     76 
     77 /* extract_subst_template: return pointer to copy of substitution template
     78    in the command buffer */
     79 char *
     80 extract_subst_template(void)
     81 {
     82 	int n = 0;
     83 	int i = 0;
     84 	char c;
     85 	char delimiter = *ibufp++;
     86 
     87 	if (*ibufp == '%' && *(ibufp + 1) == delimiter) {
     88 		ibufp++;
     89 		if (!rhbuf) {
     90 			seterrmsg("no previous substitution");
     91 		}
     92 		return rhbuf;
     93 	}
     94 	while (*ibufp != delimiter) {
     95 		REALLOC(rhbuf, rhbufsz, i + 2, NULL);
     96 		if ((c = rhbuf[i++] = *ibufp++) == '\n' && *ibufp == '\0') {
     97 			i--, ibufp--;
     98 			break;
     99 		} else if (c != '\\')
    100 			;
    101 		else if ((rhbuf[i++] = *ibufp++) != '\n')
    102 			;
    103 		else if (!isglobal) {
    104 			while ((n = get_tty_line()) == 0 ||
    105 			    (n > 0 && ibuf[n - 1] != '\n'))
    106 				clearerr(stdin);
    107 			if (n < 0)
    108 				return NULL;
    109 		}
    110 	}
    111 	REALLOC(rhbuf, rhbufsz, i + 1, NULL);
    112 	rhbuf[rhbufi = i] = '\0';
    113 	return  rhbuf;
    114 }
    115 
    116 
    117 char *rbuf;			/* substitute_matching_text buffer */
    118 int rbufsz;			/* substitute_matching_text buffer size */
    119 
    120 /* search_and_replace: for each line in a range, change text matching a pattern
    121    according to a substitution template; return status  */
    122 int
    123 search_and_replace(pattern_t *pat, int gflag, int kth)
    124 {
    125 	undo_t *up;
    126 	char *txt;
    127 	char *eot;
    128 	long lc;
    129 	long xa = current_addr;
    130 	int nsubs = 0;
    131 	line_t *lp;
    132 	int len;
    133 
    134 	current_addr = first_addr - 1;
    135 	for (lc = 0; lc <= second_addr - first_addr; lc++) {
    136 		lp = get_addressed_line_node(++current_addr);
    137 		if ((len = substitute_matching_text(pat, lp, gflag, kth)) < 0)
    138 			return ERR;
    139 		else if (len) {
    140 			up = NULL;
    141 			if (delete_lines(current_addr, current_addr) < 0)
    142 				return ERR;
    143 			txt = rbuf;
    144 			eot = rbuf + len;
    145 			SPL1();
    146 			do {
    147 				if ((txt = put_sbuf_line(txt)) == NULL) {
    148 					SPL0();
    149 					return ERR;
    150 				} else if (up)
    151 					up->t = get_addressed_line_node(current_addr);
    152 				else if ((up = push_undo_stack(UADD,
    153 				    current_addr, current_addr)) == NULL) {
    154 					SPL0();
    155 					return ERR;
    156 				}
    157 			} while (txt != eot);
    158 			SPL0();
    159 			nsubs++;
    160 			xa = current_addr;
    161 		}
    162 	}
    163 	current_addr = xa;
    164 	if  (nsubs == 0 && !(gflag & GLB)) {
    165 		seterrmsg("no match");
    166 		return ERR;
    167 	} else if ((gflag & (GPR | GLS | GNP)) &&
    168 	    display_lines(current_addr, current_addr, gflag) < 0)
    169 		return ERR;
    170 	return 0;
    171 }
    172 
    173 
    174 /* substitute_matching_text: replace text matched by a pattern according to
    175    a substitution template; return pointer to the modified text */
    176 int
    177 substitute_matching_text(pattern_t *pat, line_t *lp, int gflag, int kth)
    178 {
    179 	int off = 0;
    180 	int changed = 0;
    181 	int matchno = 0;
    182 	int i = 0;
    183 	regmatch_t rm[SE_MAX];
    184 	char *txt;
    185 	char *eot;
    186 
    187 	if ((txt = get_sbuf_line(lp)) == NULL)
    188 		return ERR;
    189 	if (isbinary)
    190 		NUL_TO_NEWLINE(txt, lp->len);
    191 	eot = txt + lp->len;
    192 	if (!regexec(pat, txt, SE_MAX, rm, 0)) {
    193 		do {
    194 			if (!kth || kth == ++matchno) {
    195 				changed++;
    196 				i = rm[0].rm_so;
    197 				REALLOC(rbuf, rbufsz, off + i, ERR);
    198 				if (isbinary)
    199 					NEWLINE_TO_NUL(txt, rm[0].rm_eo);
    200 				memcpy(rbuf + off, txt, i);
    201 				off += i;
    202 				if ((off = apply_subst_template(txt, rm, off,
    203 				    pat->re_nsub)) < 0)
    204 					return ERR;
    205 			} else {
    206 				i = rm[0].rm_eo;
    207 				REALLOC(rbuf, rbufsz, off + i, ERR);
    208 				if (isbinary)
    209 					NEWLINE_TO_NUL(txt, i);
    210 				memcpy(rbuf + off, txt, i);
    211 				off += i;
    212 			}
    213 			txt += rm[0].rm_eo;
    214 		} while (*txt && (!changed || ((gflag & GSG) && rm[0].rm_eo))
    215 		    && !regexec(pat, txt, SE_MAX, rm, REG_NOTBOL));
    216 		i = eot - txt;
    217 		REALLOC(rbuf, rbufsz, off + i + 2, ERR);
    218 		if (i > 0 && !rm[0].rm_eo && (gflag & GSG)) {
    219 			seterrmsg("infinite substitution loop");
    220 			return  ERR;
    221 		}
    222 		if (isbinary)
    223 			NEWLINE_TO_NUL(txt, i);
    224 		memcpy(rbuf + off, txt, i);
    225 		memcpy(rbuf + off + i, "\n", 2);
    226 	}
    227 	return changed ? off + i + 1 : 0;
    228 }
    229 
    230 
    231 /* apply_subst_template: modify text according to a substitution template;
    232    return offset to end of modified text */
    233 int
    234 apply_subst_template(char *boln, regmatch_t *rm, int off, int re_nsub)
    235 {
    236 	int j = 0;
    237 	int k = 0;
    238 	int n;
    239 	char *sub = rhbuf;
    240 
    241 	for (; sub - rhbuf < rhbufi; sub++)
    242 		if (*sub == '&') {
    243 			j = rm[0].rm_so;
    244 			k = rm[0].rm_eo;
    245 			REALLOC(rbuf, rbufsz, off + k - j, ERR);
    246 			while (j < k)
    247 				rbuf[off++] = boln[j++];
    248 		} else if (*sub == '\\' && '1' <= *++sub && *sub <= '9' &&
    249 		    (n = *sub - '0') <= re_nsub) {
    250 			j = rm[n].rm_so;
    251 			k = rm[n].rm_eo;
    252 			REALLOC(rbuf, rbufsz, off + k - j, ERR);
    253 			while (j < k)
    254 				rbuf[off++] = boln[j++];
    255 		} else {
    256 			REALLOC(rbuf, rbufsz, off + 1, ERR);
    257 			rbuf[off++] = *sub;
    258 		}
    259 	REALLOC(rbuf, rbufsz, off + 1, ERR);
    260 	rbuf[off] = '\0';
    261 	return off;
    262 }
    263