109ade360Sjoerg /*-
209ade360Sjoerg * Copyright (c) 2008 Joerg Sonnenberger
309ade360Sjoerg * All rights reserved.
409ade360Sjoerg *
509ade360Sjoerg * Redistribution and use in source and binary forms, with or without
609ade360Sjoerg * modification, are permitted provided that the following conditions
709ade360Sjoerg * are met:
809ade360Sjoerg * 1. Redistributions of source code must retain the above copyright
909ade360Sjoerg * notice, this list of conditions and the following disclaimer.
1009ade360Sjoerg * 2. Redistributions in binary form must reproduce the above copyright
1109ade360Sjoerg * notice, this list of conditions and the following disclaimer in the
1209ade360Sjoerg * documentation and/or other materials provided with the distribution.
1309ade360Sjoerg *
1409ade360Sjoerg * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
1509ade360Sjoerg * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
1609ade360Sjoerg * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
1709ade360Sjoerg * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
1809ade360Sjoerg * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
1909ade360Sjoerg * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
2009ade360Sjoerg * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
2109ade360Sjoerg * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
2209ade360Sjoerg * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
2309ade360Sjoerg * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
2409ade360Sjoerg */
2509ade360Sjoerg
2609ade360Sjoerg #include "bsdtar_platform.h"
2709ade360Sjoerg __FBSDID("$FreeBSD: src/usr.bin/tar/subst.c,v 1.4 2008/06/15 10:08:16 kientzle Exp $");
2809ade360Sjoerg
29*d05f0226Sjoerg #if defined(HAVE_REGEX_H) || defined(HAVE_PCREPOSIX_H)
3009ade360Sjoerg #include "bsdtar.h"
3109ade360Sjoerg
3209ade360Sjoerg #include <errno.h>
33*d05f0226Sjoerg #ifdef HAVE_PCREPOSIX_H
34*d05f0226Sjoerg #include <pcreposix.h>
35*d05f0226Sjoerg #else
3609ade360Sjoerg #include <regex.h>
37*d05f0226Sjoerg #endif
3809ade360Sjoerg #include <stdlib.h>
3909ade360Sjoerg #include <string.h>
4009ade360Sjoerg
4109ade360Sjoerg #ifndef REG_BASIC
4209ade360Sjoerg #define REG_BASIC 0
4309ade360Sjoerg #endif
4409ade360Sjoerg
459fde5391Sjoerg #include "err.h"
469fde5391Sjoerg
4709ade360Sjoerg struct subst_rule {
4809ade360Sjoerg struct subst_rule *next;
4909ade360Sjoerg regex_t re;
5009ade360Sjoerg char *result;
51*d05f0226Sjoerg unsigned int global:1, print:1, regular:1, symlink:1, hardlink:1;
5209ade360Sjoerg };
5309ade360Sjoerg
5409ade360Sjoerg struct substitution {
5509ade360Sjoerg struct subst_rule *first_rule, *last_rule;
5609ade360Sjoerg };
5709ade360Sjoerg
5809ade360Sjoerg static void
init_substitution(struct bsdtar * bsdtar)5909ade360Sjoerg init_substitution(struct bsdtar *bsdtar)
6009ade360Sjoerg {
6109ade360Sjoerg struct substitution *subst;
6209ade360Sjoerg
6309ade360Sjoerg bsdtar->substitution = subst = malloc(sizeof(*subst));
6409ade360Sjoerg if (subst == NULL)
659fde5391Sjoerg lafe_errc(1, errno, "Out of memory");
6609ade360Sjoerg subst->first_rule = subst->last_rule = NULL;
6709ade360Sjoerg }
6809ade360Sjoerg
6909ade360Sjoerg void
add_substitution(struct bsdtar * bsdtar,const char * rule_text)7009ade360Sjoerg add_substitution(struct bsdtar *bsdtar, const char *rule_text)
7109ade360Sjoerg {
7209ade360Sjoerg struct subst_rule *rule;
7309ade360Sjoerg struct substitution *subst;
7409ade360Sjoerg const char *end_pattern, *start_subst;
7509ade360Sjoerg char *pattern;
7609ade360Sjoerg int r;
7709ade360Sjoerg
7809ade360Sjoerg if ((subst = bsdtar->substitution) == NULL) {
7909ade360Sjoerg init_substitution(bsdtar);
8009ade360Sjoerg subst = bsdtar->substitution;
8109ade360Sjoerg }
8209ade360Sjoerg
8309ade360Sjoerg rule = malloc(sizeof(*rule));
8409ade360Sjoerg if (rule == NULL)
859fde5391Sjoerg lafe_errc(1, errno, "Out of memory");
8609ade360Sjoerg rule->next = NULL;
87*d05f0226Sjoerg rule->result = NULL;
8809ade360Sjoerg
8909ade360Sjoerg if (subst->last_rule == NULL)
9009ade360Sjoerg subst->first_rule = rule;
9109ade360Sjoerg else
9209ade360Sjoerg subst->last_rule->next = rule;
9309ade360Sjoerg subst->last_rule = rule;
9409ade360Sjoerg
9509ade360Sjoerg if (*rule_text == '\0')
969fde5391Sjoerg lafe_errc(1, 0, "Empty replacement string");
9709ade360Sjoerg end_pattern = strchr(rule_text + 1, *rule_text);
9809ade360Sjoerg if (end_pattern == NULL)
999fde5391Sjoerg lafe_errc(1, 0, "Invalid replacement string");
10009ade360Sjoerg
10109ade360Sjoerg pattern = malloc(end_pattern - rule_text);
10209ade360Sjoerg if (pattern == NULL)
1039fde5391Sjoerg lafe_errc(1, errno, "Out of memory");
10409ade360Sjoerg memcpy(pattern, rule_text + 1, end_pattern - rule_text - 1);
10509ade360Sjoerg pattern[end_pattern - rule_text - 1] = '\0';
10609ade360Sjoerg
10709ade360Sjoerg if ((r = regcomp(&rule->re, pattern, REG_BASIC)) != 0) {
10809ade360Sjoerg char buf[80];
10909ade360Sjoerg regerror(r, &rule->re, buf, sizeof(buf));
1109fde5391Sjoerg lafe_errc(1, 0, "Invalid regular expression: %s", buf);
11109ade360Sjoerg }
11209ade360Sjoerg free(pattern);
11309ade360Sjoerg
11409ade360Sjoerg start_subst = end_pattern + 1;
11509ade360Sjoerg end_pattern = strchr(start_subst, *rule_text);
11609ade360Sjoerg if (end_pattern == NULL)
1179fde5391Sjoerg lafe_errc(1, 0, "Invalid replacement string");
11809ade360Sjoerg
11909ade360Sjoerg rule->result = malloc(end_pattern - start_subst + 1);
12009ade360Sjoerg if (rule->result == NULL)
1219fde5391Sjoerg lafe_errc(1, errno, "Out of memory");
12209ade360Sjoerg memcpy(rule->result, start_subst, end_pattern - start_subst);
12309ade360Sjoerg rule->result[end_pattern - start_subst] = '\0';
12409ade360Sjoerg
125*d05f0226Sjoerg /* Defaults */
126*d05f0226Sjoerg rule->global = 0; /* Don't do multiple replacements. */
127*d05f0226Sjoerg rule->print = 0; /* Don't print. */
128*d05f0226Sjoerg rule->regular = 1; /* Rewrite regular filenames. */
129*d05f0226Sjoerg rule->symlink = 1; /* Rewrite symlink targets. */
130*d05f0226Sjoerg rule->hardlink = 1; /* Rewrite hardlink targets. */
13109ade360Sjoerg
13209ade360Sjoerg while (*++end_pattern) {
13309ade360Sjoerg switch (*end_pattern) {
13409ade360Sjoerg case 'g':
13509ade360Sjoerg case 'G':
13609ade360Sjoerg rule->global = 1;
13709ade360Sjoerg break;
138*d05f0226Sjoerg case 'h':
139*d05f0226Sjoerg rule->hardlink = 1;
140*d05f0226Sjoerg break;
141*d05f0226Sjoerg case 'H':
142*d05f0226Sjoerg rule->hardlink = 0;
143*d05f0226Sjoerg break;
14409ade360Sjoerg case 'p':
14509ade360Sjoerg case 'P':
14609ade360Sjoerg rule->print = 1;
14709ade360Sjoerg break;
148*d05f0226Sjoerg case 'r':
149*d05f0226Sjoerg rule->regular = 1;
150*d05f0226Sjoerg break;
151*d05f0226Sjoerg case 'R':
152*d05f0226Sjoerg rule->regular = 0;
153*d05f0226Sjoerg break;
15409ade360Sjoerg case 's':
15509ade360Sjoerg rule->symlink = 1;
15609ade360Sjoerg break;
157*d05f0226Sjoerg case 'S':
158*d05f0226Sjoerg rule->symlink = 0;
159*d05f0226Sjoerg break;
16009ade360Sjoerg default:
1619fde5391Sjoerg lafe_errc(1, 0, "Invalid replacement flag %c", *end_pattern);
16209ade360Sjoerg }
16309ade360Sjoerg }
16409ade360Sjoerg }
16509ade360Sjoerg
16609ade360Sjoerg static void
realloc_strncat(char ** str,const char * append,size_t len)1679fde5391Sjoerg realloc_strncat(char **str, const char *append, size_t len)
16809ade360Sjoerg {
16909ade360Sjoerg char *new_str;
17009ade360Sjoerg size_t old_len;
17109ade360Sjoerg
17209ade360Sjoerg if (*str == NULL)
17309ade360Sjoerg old_len = 0;
17409ade360Sjoerg else
17509ade360Sjoerg old_len = strlen(*str);
17609ade360Sjoerg
17709ade360Sjoerg new_str = malloc(old_len + len + 1);
17809ade360Sjoerg if (new_str == NULL)
1799fde5391Sjoerg lafe_errc(1, errno, "Out of memory");
180*d05f0226Sjoerg if (*str != NULL)
18109ade360Sjoerg memcpy(new_str, *str, old_len);
18209ade360Sjoerg memcpy(new_str + old_len, append, len);
18309ade360Sjoerg new_str[old_len + len] = '\0';
18409ade360Sjoerg free(*str);
18509ade360Sjoerg *str = new_str;
18609ade360Sjoerg }
18709ade360Sjoerg
18809ade360Sjoerg static void
realloc_strcat(char ** str,const char * append)1899fde5391Sjoerg realloc_strcat(char **str, const char *append)
19009ade360Sjoerg {
19109ade360Sjoerg char *new_str;
19209ade360Sjoerg size_t old_len;
19309ade360Sjoerg
19409ade360Sjoerg if (*str == NULL)
19509ade360Sjoerg old_len = 0;
19609ade360Sjoerg else
19709ade360Sjoerg old_len = strlen(*str);
19809ade360Sjoerg
19909ade360Sjoerg new_str = malloc(old_len + strlen(append) + 1);
20009ade360Sjoerg if (new_str == NULL)
2019fde5391Sjoerg lafe_errc(1, errno, "Out of memory");
202*d05f0226Sjoerg if (*str != NULL)
20309ade360Sjoerg memcpy(new_str, *str, old_len);
20409ade360Sjoerg strcpy(new_str + old_len, append);
20509ade360Sjoerg free(*str);
20609ade360Sjoerg *str = new_str;
20709ade360Sjoerg }
20809ade360Sjoerg
20909ade360Sjoerg int
apply_substitution(struct bsdtar * bsdtar,const char * name,char ** result,int symlink_target,int hardlink_target)210*d05f0226Sjoerg apply_substitution(struct bsdtar *bsdtar, const char *name, char **result,
211*d05f0226Sjoerg int symlink_target, int hardlink_target)
21209ade360Sjoerg {
21309ade360Sjoerg const char *path = name;
21409ade360Sjoerg regmatch_t matches[10];
21509ade360Sjoerg size_t i, j;
21609ade360Sjoerg struct subst_rule *rule;
21709ade360Sjoerg struct substitution *subst;
21809ade360Sjoerg int c, got_match, print_match;
21909ade360Sjoerg
22009ade360Sjoerg *result = NULL;
22109ade360Sjoerg
22209ade360Sjoerg if ((subst = bsdtar->substitution) == NULL)
22309ade360Sjoerg return 0;
22409ade360Sjoerg
22509ade360Sjoerg got_match = 0;
22609ade360Sjoerg print_match = 0;
22709ade360Sjoerg
22809ade360Sjoerg for (rule = subst->first_rule; rule != NULL; rule = rule->next) {
229*d05f0226Sjoerg if (symlink_target) {
230*d05f0226Sjoerg if (!rule->symlink)
23109ade360Sjoerg continue;
232*d05f0226Sjoerg } else if (hardlink_target) {
233*d05f0226Sjoerg if (!rule->hardlink)
234*d05f0226Sjoerg continue;
235*d05f0226Sjoerg } else { /* Regular filename. */
236*d05f0226Sjoerg if (!rule->regular)
237*d05f0226Sjoerg continue;
238*d05f0226Sjoerg }
239*d05f0226Sjoerg
240*d05f0226Sjoerg while (1) {
24109ade360Sjoerg if (regexec(&rule->re, name, 10, matches, 0))
242*d05f0226Sjoerg break;
24309ade360Sjoerg
24409ade360Sjoerg got_match = 1;
24509ade360Sjoerg print_match |= rule->print;
2469fde5391Sjoerg realloc_strncat(result, name, matches[0].rm_so);
24709ade360Sjoerg
24809ade360Sjoerg for (i = 0, j = 0; rule->result[i] != '\0'; ++i) {
24909ade360Sjoerg if (rule->result[i] == '~') {
2509fde5391Sjoerg realloc_strncat(result, rule->result + j, i - j);
251*d05f0226Sjoerg realloc_strncat(result,
252*d05f0226Sjoerg name + matches[0].rm_so,
253*d05f0226Sjoerg matches[0].rm_eo - matches[0].rm_so);
25409ade360Sjoerg j = i + 1;
25509ade360Sjoerg continue;
25609ade360Sjoerg }
25709ade360Sjoerg if (rule->result[i] != '\\')
25809ade360Sjoerg continue;
25909ade360Sjoerg
26009ade360Sjoerg ++i;
26109ade360Sjoerg c = rule->result[i];
26209ade360Sjoerg switch (c) {
26309ade360Sjoerg case '~':
26409ade360Sjoerg case '\\':
2659fde5391Sjoerg realloc_strncat(result, rule->result + j, i - j - 1);
26609ade360Sjoerg j = i;
26709ade360Sjoerg break;
26809ade360Sjoerg case '1':
26909ade360Sjoerg case '2':
27009ade360Sjoerg case '3':
27109ade360Sjoerg case '4':
27209ade360Sjoerg case '5':
27309ade360Sjoerg case '6':
27409ade360Sjoerg case '7':
27509ade360Sjoerg case '8':
27609ade360Sjoerg case '9':
2779fde5391Sjoerg realloc_strncat(result, rule->result + j, i - j - 1);
27809ade360Sjoerg if ((size_t)(c - '0') > (size_t)(rule->re.re_nsub)) {
27909ade360Sjoerg free(*result);
28009ade360Sjoerg *result = NULL;
28109ade360Sjoerg return -1;
28209ade360Sjoerg }
2839fde5391Sjoerg realloc_strncat(result, name + matches[c - '0'].rm_so, matches[c - '0'].rm_eo - matches[c - '0'].rm_so);
28409ade360Sjoerg j = i + 1;
28509ade360Sjoerg break;
28609ade360Sjoerg default:
28709ade360Sjoerg /* Just continue; */
28809ade360Sjoerg break;
28909ade360Sjoerg }
29009ade360Sjoerg
29109ade360Sjoerg }
29209ade360Sjoerg
2939fde5391Sjoerg realloc_strcat(result, rule->result + j);
29409ade360Sjoerg
29509ade360Sjoerg name += matches[0].rm_eo;
29609ade360Sjoerg
29709ade360Sjoerg if (!rule->global)
29809ade360Sjoerg break;
29909ade360Sjoerg }
300*d05f0226Sjoerg }
30109ade360Sjoerg
30209ade360Sjoerg if (got_match)
3039fde5391Sjoerg realloc_strcat(result, name);
30409ade360Sjoerg
30509ade360Sjoerg if (print_match)
30609ade360Sjoerg fprintf(stderr, "%s >> %s\n", path, *result);
30709ade360Sjoerg
30809ade360Sjoerg return got_match;
30909ade360Sjoerg }
31009ade360Sjoerg
31109ade360Sjoerg void
cleanup_substitution(struct bsdtar * bsdtar)31209ade360Sjoerg cleanup_substitution(struct bsdtar *bsdtar)
31309ade360Sjoerg {
31409ade360Sjoerg struct subst_rule *rule;
31509ade360Sjoerg struct substitution *subst;
31609ade360Sjoerg
31709ade360Sjoerg if ((subst = bsdtar->substitution) == NULL)
31809ade360Sjoerg return;
31909ade360Sjoerg
32009ade360Sjoerg while ((rule = subst->first_rule) != NULL) {
32109ade360Sjoerg subst->first_rule = rule->next;
32209ade360Sjoerg free(rule->result);
32309ade360Sjoerg free(rule);
32409ade360Sjoerg }
32509ade360Sjoerg free(subst);
32609ade360Sjoerg }
327*d05f0226Sjoerg #endif /* defined(HAVE_REGEX_H) || defined(HAVE_PCREPOSIX_H) */
328