4 Copyright (C) Andrew Tridgell 2004
6 ** NOTE! The following LGPL license applies to the ldb
7 ** library. This does NOT imply that all of Samba is released
10 This library is free software; you can redistribute it and/or
11 modify it under the terms of the GNU Lesser General Public
12 License as published by the Free Software Foundation; either
13 version 3 of the License, or (at your option) any later version.
15 This library is distributed in the hope that it will be useful,
16 but WITHOUT ANY WARRANTY; without even the implied warranty of
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 Lesser General Public License for more details.
20 You should have received a copy of the GNU Lesser General Public
21 License along with this library; if not, see <http://www.gnu.org/licenses/>.
27 * Component: ldb expression parsing
29 * Description: parse LDAP-like search expressions
31 * Author: Andrew Tridgell
36 - add RFC2254 binary string handling
37 - possibly add ~=, <= and >= handling
38 - expand the test suite
39 - add better parse error handling
43 #include "ldb_private.h"
44 #include "system/locale.h"
46 static int ldb_parse_hex2char(const char *x)
48 if (isxdigit(x[0]) && isxdigit(x[1])) {
49 const char h1 = x[0], h2 = x[1];
52 if (h1 >= 'a') c = h1 - (int)'a' + 10;
53 else if (h1 >= 'A') c = h1 - (int)'A' + 10;
54 else if (h1 >= '0') c = h1 - (int)'0';
56 if (h2 >= 'a') c += h2 - (int)'a' + 10;
57 else if (h1 >= 'A') c += h2 - (int)'A' + 10;
58 else if (h1 >= '0') c += h2 - (int)'0';
67 a filter is defined by:
68 <filter> ::= '(' <filtercomp> ')'
69 <filtercomp> ::= <and> | <or> | <not> | <simple>
70 <and> ::= '&' <filterlist>
71 <or> ::= '|' <filterlist>
72 <not> ::= '!' <filter>
73 <filterlist> ::= <filter> | <filter> <filterlist>
74 <simple> ::= <attributetype> <filtertype> <attributevalue>
75 <filtertype> ::= '=' | '~=' | '<=' | '>='
79 decode a RFC2254 binary string representation of a buffer.
82 struct ldb_val ldb_binary_decode(void *mem_ctx, const char *str)
86 size_t slen = str?strlen(str):0;
88 ret.data = (uint8_t *)talloc_size(mem_ctx, slen+1);
90 if (ret.data == NULL) return ret;
92 for (i=j=0;i<slen;i++) {
96 c = ldb_parse_hex2char(&str[i+1]);
98 talloc_free(ret.data);
99 memset(&ret, 0, sizeof(ret));
102 ((uint8_t *)ret.data)[j++] = c;
105 ((uint8_t *)ret.data)[j++] = str[i];
109 ((uint8_t *)ret.data)[j] = 0;
116 encode a blob as a RFC2254 binary string, escaping any
117 non-printable or '\' characters
119 char *ldb_binary_encode(void *mem_ctx, struct ldb_val val)
123 size_t len = val.length;
124 unsigned char *buf = val.data;
126 for (i=0;i<val.length;i++) {
127 if (!isprint(buf[i]) || strchr(" *()\\&|!\"", buf[i])) {
131 ret = talloc_array(mem_ctx, char, len+1);
132 if (ret == NULL) return NULL;
135 for (i=0;i<val.length;i++) {
136 if (!isprint(buf[i]) || strchr(" *()\\&|!\"", buf[i])) {
137 snprintf(ret+len, 4, "\\%02X", buf[i]);
150 encode a string as a RFC2254 binary string, escaping any
151 non-printable or '\' characters. This routine is suitable for use
152 in escaping user data in ldap filters.
154 char *ldb_binary_encode_string(void *mem_ctx, const char *string)
157 val.data = discard_const_p(uint8_t, string);
158 val.length = strlen(string);
159 return ldb_binary_encode(mem_ctx, val);
162 /* find the first matching wildcard */
163 static char *ldb_parse_find_wildcard(char *value)
166 value = strpbrk(value, "\\*");
167 if (value == NULL) return NULL;
169 if (value[0] == '\\') {
170 if (value[1] == '\0') return NULL;
175 if (value[0] == '*') return value;
181 /* return a NULL terminated list of binary strings representing the value
182 chunks separated by wildcards that makes the value portion of the filter
184 static struct ldb_val **ldb_wildcard_decode(void *mem_ctx, const char *string)
186 struct ldb_val **ret = NULL;
187 unsigned int val = 0;
190 wc = talloc_strdup(mem_ctx, string);
191 if (wc == NULL) return NULL;
195 wc = ldb_parse_find_wildcard(str);
205 ret = talloc_realloc(mem_ctx, ret, struct ldb_val *, val + 2);
206 if (ret == NULL) return NULL;
208 ret[val] = talloc(mem_ctx, struct ldb_val);
209 if (ret[val] == NULL) return NULL;
211 *(ret[val]) = ldb_binary_decode(mem_ctx, str);
212 if ((ret[val])->data == NULL) return NULL;
224 static struct ldb_parse_tree *ldb_parse_filter(void *mem_ctx, const char **s);
228 parse an extended match
236 the ':dn' part sets the dnAttributes boolean if present
237 the oid sets the rule_id string
240 static struct ldb_parse_tree *ldb_parse_extended(struct ldb_parse_tree *ret,
241 char *attr, char *value)
245 ret->operation = LDB_OP_EXTENDED;
246 ret->u.extended.value = ldb_binary_decode(ret, value);
247 if (ret->u.extended.value.data == NULL) goto failed;
249 p1 = strchr(attr, ':');
250 if (p1 == NULL) goto failed;
251 p2 = strchr(p1+1, ':');
256 ret->u.extended.attr = attr;
257 if (strcmp(p1+1, "dn") == 0) {
258 ret->u.extended.dnAttributes = 1;
260 ret->u.extended.rule_id = talloc_strdup(ret, p2+1);
261 if (ret->u.extended.rule_id == NULL) goto failed;
263 ret->u.extended.rule_id = NULL;
266 ret->u.extended.dnAttributes = 0;
267 ret->u.extended.rule_id = talloc_strdup(ret, p1+1);
268 if (ret->u.extended.rule_id == NULL) goto failed;
278 static enum ldb_parse_op ldb_parse_filtertype(void *mem_ctx, char **type, char **value, const char **s)
280 enum ldb_parse_op filter = 0;
281 char *name, *val, *k;
285 /* retrieve attributetype name */
288 if (*p == '@') { /* for internal attributes the first char can be @ */
292 while ((isascii(*p) && isalnum((unsigned char)*p)) || (*p == '-') || (*p == '.')) {
293 /* attribute names can only be alphanums */
297 if (*p == ':') { /* but extended searches have : and . chars too */
299 if (p == NULL) { /* malformed attribute name */
306 while (isspace((unsigned char)*p)) p++;
308 if (!strchr("=<>~:", *p)) {
313 name = (char *)talloc_memdup(mem_ctx, t, t1 - t + 1);
314 if (name == NULL) return 0;
317 /* retrieve filtertype */
320 filter = LDB_OP_EQUALITY;
321 } else if (*(p + 1) == '=') {
324 filter = LDB_OP_LESS;
328 filter = LDB_OP_GREATER;
332 filter = LDB_OP_APPROX;
336 filter = LDB_OP_EXTENDED;
347 while (isspace((unsigned char)*p)) p++;
352 while (*p && ((*p != ')') || ((*p == ')') && (*(p - 1) == '\\')))) p++;
354 val = (char *)talloc_memdup(mem_ctx, t, p - t + 1);
363 /* remove trailing spaces from value */
364 while ((k > val) && (isspace((unsigned char)*(k - 1)))) k--;
374 <simple> ::= <attributetype> <filtertype> <attributevalue>
376 static struct ldb_parse_tree *ldb_parse_simple(void *mem_ctx, const char **s)
379 struct ldb_parse_tree *ret;
380 enum ldb_parse_op filtertype;
382 ret = talloc(mem_ctx, struct ldb_parse_tree);
388 filtertype = ldb_parse_filtertype(ret, &attr, &value, s);
394 switch (filtertype) {
397 ret->operation = LDB_OP_PRESENT;
398 ret->u.present.attr = attr;
401 case LDB_OP_EQUALITY:
403 if (strcmp(value, "*") == 0) {
404 ret->operation = LDB_OP_PRESENT;
405 ret->u.present.attr = attr;
409 if (ldb_parse_find_wildcard(value) != NULL) {
410 ret->operation = LDB_OP_SUBSTRING;
411 ret->u.substring.attr = attr;
412 ret->u.substring.start_with_wildcard = 0;
413 ret->u.substring.end_with_wildcard = 0;
414 ret->u.substring.chunks = ldb_wildcard_decode(ret, value);
415 if (ret->u.substring.chunks == NULL){
420 ret->u.substring.start_with_wildcard = 1;
421 if (value[strlen(value) - 1] == '*')
422 ret->u.substring.end_with_wildcard = 1;
428 ret->operation = LDB_OP_EQUALITY;
429 ret->u.equality.attr = attr;
430 ret->u.equality.value = ldb_binary_decode(ret, value);
431 if (ret->u.equality.value.data == NULL) {
439 ret->operation = LDB_OP_GREATER;
440 ret->u.comparison.attr = attr;
441 ret->u.comparison.value = ldb_binary_decode(ret, value);
442 if (ret->u.comparison.value.data == NULL) {
450 ret->operation = LDB_OP_LESS;
451 ret->u.comparison.attr = attr;
452 ret->u.comparison.value = ldb_binary_decode(ret, value);
453 if (ret->u.comparison.value.data == NULL) {
461 ret->operation = LDB_OP_APPROX;
462 ret->u.comparison.attr = attr;
463 ret->u.comparison.value = ldb_binary_decode(ret, value);
464 if (ret->u.comparison.value.data == NULL) {
471 case LDB_OP_EXTENDED:
473 ret = ldb_parse_extended(ret, attr, value);
487 <and> ::= '&' <filterlist>
488 <or> ::= '|' <filterlist>
489 <filterlist> ::= <filter> | <filter> <filterlist>
491 static struct ldb_parse_tree *ldb_parse_filterlist(void *mem_ctx, const char **s)
493 struct ldb_parse_tree *ret, *next;
494 enum ldb_parse_op op;
509 while (isspace((unsigned char)*p)) p++;
511 ret = talloc(mem_ctx, struct ldb_parse_tree);
518 ret->u.list.num_elements = 1;
519 ret->u.list.elements = talloc(ret, struct ldb_parse_tree *);
520 if (!ret->u.list.elements) {
526 ret->u.list.elements[0] = ldb_parse_filter(ret->u.list.elements, &p);
527 if (!ret->u.list.elements[0]) {
532 while (isspace((unsigned char)*p)) p++;
534 while (*p && (next = ldb_parse_filter(ret->u.list.elements, &p))) {
535 struct ldb_parse_tree **e;
536 e = talloc_realloc(ret, ret->u.list.elements,
537 struct ldb_parse_tree *,
538 ret->u.list.num_elements + 1);
544 ret->u.list.elements = e;
545 ret->u.list.elements[ret->u.list.num_elements] = next;
546 ret->u.list.num_elements++;
547 while (isspace((unsigned char)*p)) p++;
557 <not> ::= '!' <filter>
559 static struct ldb_parse_tree *ldb_parse_not(void *mem_ctx, const char **s)
561 struct ldb_parse_tree *ret;
569 ret = talloc(mem_ctx, struct ldb_parse_tree);
575 ret->operation = LDB_OP_NOT;
576 ret->u.isnot.child = ldb_parse_filter(ret, &p);
577 if (!ret->u.isnot.child) {
589 <filtercomp> ::= <and> | <or> | <not> | <simple>
591 static struct ldb_parse_tree *ldb_parse_filtercomp(void *mem_ctx, const char **s)
593 struct ldb_parse_tree *ret;
596 while (isspace((unsigned char)*p)) p++;
600 ret = ldb_parse_filterlist(mem_ctx, &p);
604 ret = ldb_parse_filterlist(mem_ctx, &p);
608 ret = ldb_parse_not(mem_ctx, &p);
616 ret = ldb_parse_simple(mem_ctx, &p);
626 <filter> ::= '(' <filtercomp> ')'
628 static struct ldb_parse_tree *ldb_parse_filter(void *mem_ctx, const char **s)
630 struct ldb_parse_tree *ret;
638 ret = ldb_parse_filtercomp(mem_ctx, &p);
645 while (isspace((unsigned char)*p)) {
656 main parser entry point. Takes a search string and returns a parse tree
658 expression ::= <simple> | <filter>
660 struct ldb_parse_tree *ldb_parse_tree(void *mem_ctx, const char *s)
662 if (s == NULL || *s == 0) {
663 s = "(|(objectClass=*)(distinguishedName=*))";
666 while (isspace((unsigned char)*s)) s++;
669 return ldb_parse_filter(mem_ctx, &s);
672 return ldb_parse_simple(mem_ctx, &s);
677 construct a ldap parse filter given a parse tree
679 char *ldb_filter_from_tree(void *mem_ctx, struct ldb_parse_tree *tree)
688 switch (tree->operation) {
691 ret = talloc_asprintf(mem_ctx, "(%c", tree->operation==LDB_OP_AND?'&':'|');
692 if (ret == NULL) return NULL;
693 for (i=0;i<tree->u.list.num_elements;i++) {
694 s = ldb_filter_from_tree(mem_ctx, tree->u.list.elements[i]);
699 s2 = talloc_asprintf_append(ret, "%s", s);
707 s = talloc_asprintf_append(ret, ")");
714 s = ldb_filter_from_tree(mem_ctx, tree->u.isnot.child);
715 if (s == NULL) return NULL;
717 ret = talloc_asprintf(mem_ctx, "(!%s)", s);
720 case LDB_OP_EQUALITY:
721 s = ldb_binary_encode(mem_ctx, tree->u.equality.value);
722 if (s == NULL) return NULL;
723 ret = talloc_asprintf(mem_ctx, "(%s=%s)",
724 tree->u.equality.attr, s);
727 case LDB_OP_SUBSTRING:
728 ret = talloc_asprintf(mem_ctx, "(%s=%s", tree->u.substring.attr,
729 tree->u.substring.start_with_wildcard?"*":"");
730 if (ret == NULL) return NULL;
731 for (i = 0; tree->u.substring.chunks[i]; i++) {
732 s2 = ldb_binary_encode(mem_ctx, *(tree->u.substring.chunks[i]));
737 if (tree->u.substring.chunks[i+1] ||
738 tree->u.substring.end_with_wildcard) {
739 s = talloc_asprintf_append(ret, "%s*", s2);
741 s = talloc_asprintf_append(ret, "%s", s2);
749 s = talloc_asprintf_append(ret, ")");
757 s = ldb_binary_encode(mem_ctx, tree->u.equality.value);
758 if (s == NULL) return NULL;
759 ret = talloc_asprintf(mem_ctx, "(%s>=%s)",
760 tree->u.equality.attr, s);
764 s = ldb_binary_encode(mem_ctx, tree->u.equality.value);
765 if (s == NULL) return NULL;
766 ret = talloc_asprintf(mem_ctx, "(%s<=%s)",
767 tree->u.equality.attr, s);
771 ret = talloc_asprintf(mem_ctx, "(%s=*)", tree->u.present.attr);
774 s = ldb_binary_encode(mem_ctx, tree->u.equality.value);
775 if (s == NULL) return NULL;
776 ret = talloc_asprintf(mem_ctx, "(%s~=%s)",
777 tree->u.equality.attr, s);
780 case LDB_OP_EXTENDED:
781 s = ldb_binary_encode(mem_ctx, tree->u.extended.value);
782 if (s == NULL) return NULL;
783 ret = talloc_asprintf(mem_ctx, "(%s%s%s%s:=%s)",
784 tree->u.extended.attr?tree->u.extended.attr:"",
785 tree->u.extended.dnAttributes?":dn":"",
786 tree->u.extended.rule_id?":":"",
787 tree->u.extended.rule_id?tree->u.extended.rule_id:"",
798 replace any occurrences of an attribute name in the parse tree with a
801 void ldb_parse_tree_attr_replace(struct ldb_parse_tree *tree,
806 switch (tree->operation) {
809 for (i=0;i<tree->u.list.num_elements;i++) {
810 ldb_parse_tree_attr_replace(tree->u.list.elements[i],
815 ldb_parse_tree_attr_replace(tree->u.isnot.child, attr, replace);
817 case LDB_OP_EQUALITY:
821 if (ldb_attr_cmp(tree->u.equality.attr, attr) == 0) {
822 tree->u.equality.attr = replace;
825 case LDB_OP_SUBSTRING:
826 if (ldb_attr_cmp(tree->u.substring.attr, attr) == 0) {
827 tree->u.substring.attr = replace;
831 if (ldb_attr_cmp(tree->u.present.attr, attr) == 0) {
832 tree->u.present.attr = replace;
835 case LDB_OP_EXTENDED:
836 if (tree->u.extended.attr &&
837 ldb_attr_cmp(tree->u.extended.attr, attr) == 0) {
838 tree->u.extended.attr = replace;
845 shallow copy a tree - copying only the elements array so that the caller
846 can safely add new elements without changing the message
848 struct ldb_parse_tree *ldb_parse_tree_copy_shallow(TALLOC_CTX *mem_ctx,
849 const struct ldb_parse_tree *ot)
852 struct ldb_parse_tree *nt;
854 nt = talloc(mem_ctx, struct ldb_parse_tree);
861 switch (ot->operation) {
864 nt->u.list.elements = talloc_array(nt, struct ldb_parse_tree *,
865 ot->u.list.num_elements);
866 if (!nt->u.list.elements) {
871 for (i=0;i<ot->u.list.num_elements;i++) {
872 nt->u.list.elements[i] =
873 ldb_parse_tree_copy_shallow(nt->u.list.elements,
874 ot->u.list.elements[i]);
875 if (!nt->u.list.elements[i]) {
882 nt->u.isnot.child = ldb_parse_tree_copy_shallow(nt,
884 if (!nt->u.isnot.child) {
889 case LDB_OP_EQUALITY:
893 case LDB_OP_SUBSTRING:
895 case LDB_OP_EXTENDED: