Fix bug #9147 - winbind can't fetch user or group info from AD via LDAP
[Samba.git] / source3 / lib / ldb / common / ldb_parse.c
blob1412d5746c62bd2ce5362b9876583a5da2475b7d
1 /*
2 ldb database library
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
8 ** under the LGPL
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/>.
25 * Name: ldb
27 * Component: ldb expression parsing
29 * Description: parse LDAP-like search expressions
31 * Author: Andrew Tridgell
35 TODO:
36 - add RFC2254 binary string handling
37 - possibly add ~=, <= and >= handling
38 - expand the test suite
39 - add better parse error handling
43 #include "includes.h"
44 #include "ldb/include/includes.h"
45 #include "system/locale.h"
47 struct ldb_val ldb_binary_decode(void *mem_ctx, const char *str);
50 a filter is defined by:
51 <filter> ::= '(' <filtercomp> ')'
52 <filtercomp> ::= <and> | <or> | <not> | <simple>
53 <and> ::= '&' <filterlist>
54 <or> ::= '|' <filterlist>
55 <not> ::= '!' <filter>
56 <filterlist> ::= <filter> | <filter> <filterlist>
57 <simple> ::= <attributetype> <filtertype> <attributevalue>
58 <filtertype> ::= '=' | '~=' | '<=' | '>='
62 decode a RFC2254 binary string representation of a buffer.
63 Used in LDAP filters.
65 struct ldb_val ldb_binary_decode(void *mem_ctx, const char *str)
67 int i, j;
68 struct ldb_val ret;
69 int slen = str?strlen(str):0;
71 ret.data = (uint8_t *)talloc_size(mem_ctx, slen+1);
72 ret.length = 0;
73 if (ret.data == NULL) return ret;
75 for (i=j=0;i<slen;i++) {
76 if (str[i] == '\\') {
77 unsigned c;
78 if (sscanf(&str[i+1], "%02X", &c) != 1) {
79 talloc_free(ret.data);
80 memset(&ret, 0, sizeof(ret));
81 return ret;
83 ((uint8_t *)ret.data)[j++] = c;
84 i += 2;
85 } else {
86 ((uint8_t *)ret.data)[j++] = str[i];
89 ret.length = j;
90 ((uint8_t *)ret.data)[j] = 0;
92 return ret;
95 static bool need_encode(unsigned char cval)
97 if (cval < 0x20 || cval > 0x7E || strchr(" *()\\&|!\"", cval)) {
98 return true;
100 return false;
104 encode a blob as a RFC2254 binary string, escaping any
105 non-printable or '\' characters
107 char *ldb_binary_encode(void *mem_ctx, struct ldb_val val)
109 int i;
110 char *ret;
111 int len = val.length;
112 unsigned char *buf = val.data;
114 for (i=0;i<val.length;i++) {
115 if (need_encode(buf[i])) {
116 len += 2;
119 ret = talloc_array(mem_ctx, char, len+1);
120 if (ret == NULL) return NULL;
122 len = 0;
123 for (i=0;i<val.length;i++) {
124 if (need_encode(buf[i])) {
125 snprintf(ret+len, 4, "\\%02X", buf[i]);
126 len += 3;
127 } else {
128 ret[len++] = buf[i];
132 ret[len] = 0;
134 return ret;
138 encode a string as a RFC2254 binary string, escaping any
139 non-printable or '\' characters. This routine is suitable for use
140 in escaping user data in ldap filters.
142 char *ldb_binary_encode_string(void *mem_ctx, const char *string)
144 struct ldb_val val;
145 val.data = discard_const_p(uint8_t, string);
146 val.length = strlen(string);
147 return ldb_binary_encode(mem_ctx, val);
150 /* find the first matching wildcard */
151 static char *ldb_parse_find_wildcard(char *value)
153 while (*value) {
154 value = strpbrk(value, "\\*");
155 if (value == NULL) return NULL;
157 if (value[0] == '\\') {
158 if (value[1] == '\0') return NULL;
159 value += 2;
160 continue;
163 if (value[0] == '*') return value;
166 return NULL;
169 /* return a NULL terminated list of binary strings representing the value
170 chunks separated by wildcards that makes the value portion of the filter
172 static struct ldb_val **ldb_wildcard_decode(void *mem_ctx, const char *string)
174 struct ldb_val **ret = NULL;
175 int val = 0;
176 char *wc, *str;
178 wc = talloc_strdup(mem_ctx, string);
179 if (wc == NULL) return NULL;
181 while (wc && *wc) {
182 str = wc;
183 wc = ldb_parse_find_wildcard(str);
184 if (wc && *wc) {
185 if (wc == str) {
186 wc++;
187 continue;
189 *wc = 0;
190 wc++;
193 ret = talloc_realloc(mem_ctx, ret, struct ldb_val *, val + 2);
194 if (ret == NULL) return NULL;
196 ret[val] = talloc(mem_ctx, struct ldb_val);
197 if (ret[val] == NULL) return NULL;
199 *(ret[val]) = ldb_binary_decode(mem_ctx, str);
200 if ((ret[val])->data == NULL) return NULL;
202 val++;
205 if (ret != NULL) {
206 ret[val] = NULL;
209 return ret;
212 static struct ldb_parse_tree *ldb_parse_filter(void *mem_ctx, const char **s);
216 parse an extended match
218 possible forms:
219 (attr:oid:=value)
220 (attr:dn:oid:=value)
221 (attr:dn:=value)
222 (:dn:oid:=value)
224 the ':dn' part sets the dnAttributes boolean if present
225 the oid sets the rule_id string
228 static struct ldb_parse_tree *ldb_parse_extended(struct ldb_parse_tree *ret,
229 char *attr, char *value)
231 char *p1, *p2;
233 ret->operation = LDB_OP_EXTENDED;
234 ret->u.extended.value = ldb_binary_decode(ret, value);
235 if (ret->u.extended.value.data == NULL) goto failed;
237 p1 = strchr(attr, ':');
238 if (p1 == NULL) goto failed;
239 p2 = strchr(p1+1, ':');
241 *p1 = 0;
242 if (p2) *p2 = 0;
244 ret->u.extended.attr = attr;
245 if (strcmp(p1+1, "dn") == 0) {
246 ret->u.extended.dnAttributes = 1;
247 if (p2) {
248 ret->u.extended.rule_id = talloc_strdup(ret, p2+1);
249 if (ret->u.extended.rule_id == NULL) goto failed;
250 } else {
251 ret->u.extended.rule_id = NULL;
253 } else {
254 ret->u.extended.dnAttributes = 0;
255 ret->u.extended.rule_id = talloc_strdup(ret, p1+1);
256 if (ret->u.extended.rule_id == NULL) goto failed;
259 return ret;
261 failed:
262 talloc_free(ret);
263 return NULL;
266 static enum ldb_parse_op ldb_parse_filtertype(void *mem_ctx, char **type, char **value, const char **s)
268 enum ldb_parse_op filter = 0;
269 char *name, *val, *k;
270 const char *p = *s;
271 const char *t, *t1;
273 /* retrieve attributetype name */
274 t = p;
276 while ((isascii(*p) && isalnum((unsigned char)*p)) || (*p == '-')) { /* attribute names can only be alphanums */
277 p++;
280 if (*p == ':') { /* but extended searches have : and . chars too */
281 p = strstr(p, ":=");
282 if (p == NULL) { /* malformed attribute name */
283 return 0;
287 t1 = p;
289 while (isspace((unsigned char)*p)) p++;
291 if (!strchr("=<>~:", *p)) {
292 return 0;
295 /* save name */
296 name = (char *)talloc_memdup(mem_ctx, t, t1 - t + 1);
297 if (name == NULL) return 0;
298 name[t1 - t] = '\0';
300 /* retrieve filtertype */
302 if (*p == '=') {
303 filter = LDB_OP_EQUALITY;
304 } else if (*(p + 1) == '=') {
305 switch (*p) {
306 case '<':
307 filter = LDB_OP_LESS;
308 p++;
309 break;
310 case '>':
311 filter = LDB_OP_GREATER;
312 p++;
313 break;
314 case '~':
315 filter = LDB_OP_APPROX;
316 p++;
317 break;
318 case ':':
319 filter = LDB_OP_EXTENDED;
320 p++;
321 break;
324 if (!filter) {
325 talloc_free(name);
326 return filter;
328 p++;
330 while (isspace((unsigned char)*p)) p++;
332 /* retrieve value */
333 t = p;
335 while (*p && ((*p != ')') || ((*p == ')') && (*(p - 1) == '\\')))) p++;
337 val = (char *)talloc_memdup(mem_ctx, t, p - t + 1);
338 if (val == NULL) {
339 talloc_free(name);
340 return 0;
342 val[p - t] = '\0';
344 k = &(val[p - t]);
346 /* remove trailing spaces from value */
347 while ((k > val) && (isspace((unsigned char)*(k - 1)))) k--;
348 *k = '\0';
350 *type = name;
351 *value = val;
352 *s = p;
353 return filter;
357 <simple> ::= <attributetype> <filtertype> <attributevalue>
359 static struct ldb_parse_tree *ldb_parse_simple(void *mem_ctx, const char **s)
361 char *attr, *value;
362 struct ldb_parse_tree *ret;
363 enum ldb_parse_op filtertype;
365 ret = talloc(mem_ctx, struct ldb_parse_tree);
366 if (!ret) {
367 errno = ENOMEM;
368 return NULL;
371 filtertype = ldb_parse_filtertype(ret, &attr, &value, s);
372 if (!filtertype) {
373 talloc_free(ret);
374 return NULL;
377 switch (filtertype) {
379 case LDB_OP_PRESENT:
380 ret->operation = LDB_OP_PRESENT;
381 ret->u.present.attr = attr;
382 break;
384 case LDB_OP_EQUALITY:
386 if (strcmp(value, "*") == 0) {
387 ret->operation = LDB_OP_PRESENT;
388 ret->u.present.attr = attr;
389 break;
392 if (ldb_parse_find_wildcard(value) != NULL) {
393 ret->operation = LDB_OP_SUBSTRING;
394 ret->u.substring.attr = attr;
395 ret->u.substring.start_with_wildcard = 0;
396 ret->u.substring.end_with_wildcard = 0;
397 ret->u.substring.chunks = ldb_wildcard_decode(ret, value);
398 if (ret->u.substring.chunks == NULL){
399 talloc_free(ret);
400 return NULL;
402 if (value[0] == '*')
403 ret->u.substring.start_with_wildcard = 1;
404 if (value[strlen(value) - 1] == '*')
405 ret->u.substring.end_with_wildcard = 1;
406 talloc_free(value);
408 break;
411 ret->operation = LDB_OP_EQUALITY;
412 ret->u.equality.attr = attr;
413 ret->u.equality.value = ldb_binary_decode(ret, value);
414 if (ret->u.equality.value.data == NULL) {
415 talloc_free(ret);
416 return NULL;
418 talloc_free(value);
419 break;
421 case LDB_OP_GREATER:
422 ret->operation = LDB_OP_GREATER;
423 ret->u.comparison.attr = attr;
424 ret->u.comparison.value = ldb_binary_decode(ret, value);
425 if (ret->u.comparison.value.data == NULL) {
426 talloc_free(ret);
427 return NULL;
429 talloc_free(value);
430 break;
432 case LDB_OP_LESS:
433 ret->operation = LDB_OP_LESS;
434 ret->u.comparison.attr = attr;
435 ret->u.comparison.value = ldb_binary_decode(ret, value);
436 if (ret->u.comparison.value.data == NULL) {
437 talloc_free(ret);
438 return NULL;
440 talloc_free(value);
441 break;
443 case LDB_OP_APPROX:
444 ret->operation = LDB_OP_APPROX;
445 ret->u.comparison.attr = attr;
446 ret->u.comparison.value = ldb_binary_decode(ret, value);
447 if (ret->u.comparison.value.data == NULL) {
448 talloc_free(ret);
449 return NULL;
451 talloc_free(value);
452 break;
454 case LDB_OP_EXTENDED:
456 ret = ldb_parse_extended(ret, attr, value);
457 break;
459 default:
460 talloc_free(ret);
461 return NULL;
464 return ret;
469 parse a filterlist
470 <and> ::= '&' <filterlist>
471 <or> ::= '|' <filterlist>
472 <filterlist> ::= <filter> | <filter> <filterlist>
474 static struct ldb_parse_tree *ldb_parse_filterlist(void *mem_ctx, const char **s)
476 struct ldb_parse_tree *ret, *next;
477 enum ldb_parse_op op;
478 const char *p = *s;
480 switch (*p) {
481 case '&':
482 op = LDB_OP_AND;
483 break;
484 case '|':
485 op = LDB_OP_OR;
486 break;
487 default:
488 return NULL;
490 p++;
492 while (isspace((unsigned char)*p)) p++;
494 ret = talloc(mem_ctx, struct ldb_parse_tree);
495 if (!ret) {
496 errno = ENOMEM;
497 return NULL;
500 ret->operation = op;
501 ret->u.list.num_elements = 1;
502 ret->u.list.elements = talloc(ret, struct ldb_parse_tree *);
503 if (!ret->u.list.elements) {
504 errno = ENOMEM;
505 talloc_free(ret);
506 return NULL;
509 ret->u.list.elements[0] = ldb_parse_filter(ret->u.list.elements, &p);
510 if (!ret->u.list.elements[0]) {
511 talloc_free(ret);
512 return NULL;
515 while (isspace((unsigned char)*p)) p++;
517 while (*p && (next = ldb_parse_filter(ret->u.list.elements, &p))) {
518 struct ldb_parse_tree **e;
519 e = talloc_realloc(ret, ret->u.list.elements,
520 struct ldb_parse_tree *,
521 ret->u.list.num_elements + 1);
522 if (!e) {
523 errno = ENOMEM;
524 talloc_free(ret);
525 return NULL;
527 ret->u.list.elements = e;
528 ret->u.list.elements[ret->u.list.num_elements] = next;
529 ret->u.list.num_elements++;
530 while (isspace((unsigned char)*p)) p++;
533 *s = p;
535 return ret;
540 <not> ::= '!' <filter>
542 static struct ldb_parse_tree *ldb_parse_not(void *mem_ctx, const char **s)
544 struct ldb_parse_tree *ret;
545 const char *p = *s;
547 if (*p != '!') {
548 return NULL;
550 p++;
552 ret = talloc(mem_ctx, struct ldb_parse_tree);
553 if (!ret) {
554 errno = ENOMEM;
555 return NULL;
558 ret->operation = LDB_OP_NOT;
559 ret->u.isnot.child = ldb_parse_filter(ret, &p);
560 if (!ret->u.isnot.child) {
561 talloc_free(ret);
562 return NULL;
565 *s = p;
567 return ret;
571 parse a filtercomp
572 <filtercomp> ::= <and> | <or> | <not> | <simple>
574 static struct ldb_parse_tree *ldb_parse_filtercomp(void *mem_ctx, const char **s)
576 struct ldb_parse_tree *ret;
577 const char *p = *s;
579 while (isspace((unsigned char)*p)) p++;
581 switch (*p) {
582 case '&':
583 ret = ldb_parse_filterlist(mem_ctx, &p);
584 break;
586 case '|':
587 ret = ldb_parse_filterlist(mem_ctx, &p);
588 break;
590 case '!':
591 ret = ldb_parse_not(mem_ctx, &p);
592 break;
594 case '(':
595 case ')':
596 return NULL;
598 default:
599 ret = ldb_parse_simple(mem_ctx, &p);
603 *s = p;
604 return ret;
609 <filter> ::= '(' <filtercomp> ')'
611 static struct ldb_parse_tree *ldb_parse_filter(void *mem_ctx, const char **s)
613 struct ldb_parse_tree *ret;
614 const char *p = *s;
616 if (*p != '(') {
617 return NULL;
619 p++;
621 ret = ldb_parse_filtercomp(mem_ctx, &p);
623 if (*p != ')') {
624 return NULL;
626 p++;
628 while (isspace((unsigned char)*p)) {
629 p++;
632 *s = p;
634 return ret;
639 main parser entry point. Takes a search string and returns a parse tree
641 expression ::= <simple> | <filter>
643 struct ldb_parse_tree *ldb_parse_tree(void *mem_ctx, const char *s)
645 if (s == NULL || *s == 0) {
646 s = "(|(objectClass=*)(distinguishedName=*))";
649 while (isspace((unsigned char)*s)) s++;
651 if (*s == '(') {
652 return ldb_parse_filter(mem_ctx, &s);
655 return ldb_parse_simple(mem_ctx, &s);
660 construct a ldap parse filter given a parse tree
662 char *ldb_filter_from_tree(void *mem_ctx, struct ldb_parse_tree *tree)
664 char *s, *s2, *ret;
665 int i;
667 if (tree == NULL) {
668 return NULL;
671 switch (tree->operation) {
672 case LDB_OP_AND:
673 case LDB_OP_OR:
674 ret = talloc_asprintf(mem_ctx, "(%c", tree->operation==LDB_OP_AND?'&':'|');
675 if (ret == NULL) return NULL;
676 for (i=0;i<tree->u.list.num_elements;i++) {
677 s = ldb_filter_from_tree(mem_ctx, tree->u.list.elements[i]);
678 if (s == NULL) {
679 talloc_free(ret);
680 return NULL;
682 s2 = talloc_asprintf_append(ret, "%s", s);
683 talloc_free(s);
684 if (s2 == NULL) {
685 talloc_free(ret);
686 return NULL;
688 ret = s2;
690 s = talloc_asprintf_append(ret, ")");
691 if (s == NULL) {
692 talloc_free(ret);
693 return NULL;
695 return s;
696 case LDB_OP_NOT:
697 s = ldb_filter_from_tree(mem_ctx, tree->u.isnot.child);
698 if (s == NULL) return NULL;
700 ret = talloc_asprintf(mem_ctx, "(!%s)", s);
701 talloc_free(s);
702 return ret;
703 case LDB_OP_EQUALITY:
704 s = ldb_binary_encode(mem_ctx, tree->u.equality.value);
705 if (s == NULL) return NULL;
706 ret = talloc_asprintf(mem_ctx, "(%s=%s)",
707 tree->u.equality.attr, s);
708 talloc_free(s);
709 return ret;
710 case LDB_OP_SUBSTRING:
711 ret = talloc_asprintf(mem_ctx, "(%s=%s", tree->u.substring.attr,
712 tree->u.substring.start_with_wildcard?"*":"");
713 if (ret == NULL) return NULL;
714 for (i = 0; tree->u.substring.chunks[i]; i++) {
715 s2 = ldb_binary_encode(mem_ctx, *(tree->u.substring.chunks[i]));
716 if (s2 == NULL) {
717 talloc_free(ret);
718 return NULL;
720 if (tree->u.substring.chunks[i+1] ||
721 tree->u.substring.end_with_wildcard) {
722 s = talloc_asprintf_append(ret, "%s*", s2);
723 } else {
724 s = talloc_asprintf_append(ret, "%s", s2);
726 if (s == NULL) {
727 talloc_free(ret);
728 return NULL;
730 ret = s;
732 s = talloc_asprintf_append(ret, ")");
733 if (s == NULL) {
734 talloc_free(ret);
735 return NULL;
737 ret = s;
738 return ret;
739 case LDB_OP_GREATER:
740 s = ldb_binary_encode(mem_ctx, tree->u.equality.value);
741 if (s == NULL) return NULL;
742 ret = talloc_asprintf(mem_ctx, "(%s>=%s)",
743 tree->u.equality.attr, s);
744 talloc_free(s);
745 return ret;
746 case LDB_OP_LESS:
747 s = ldb_binary_encode(mem_ctx, tree->u.equality.value);
748 if (s == NULL) return NULL;
749 ret = talloc_asprintf(mem_ctx, "(%s<=%s)",
750 tree->u.equality.attr, s);
751 talloc_free(s);
752 return ret;
753 case LDB_OP_PRESENT:
754 ret = talloc_asprintf(mem_ctx, "(%s=*)", tree->u.present.attr);
755 return ret;
756 case LDB_OP_APPROX:
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);
761 talloc_free(s);
762 return ret;
763 case LDB_OP_EXTENDED:
764 s = ldb_binary_encode(mem_ctx, tree->u.extended.value);
765 if (s == NULL) return NULL;
766 ret = talloc_asprintf(mem_ctx, "(%s%s%s%s:=%s)",
767 tree->u.extended.attr?tree->u.extended.attr:"",
768 tree->u.extended.dnAttributes?":dn":"",
769 tree->u.extended.rule_id?":":"",
770 tree->u.extended.rule_id?tree->u.extended.rule_id:"",
772 talloc_free(s);
773 return ret;
776 return NULL;
781 replace any occurances of an attribute name in the parse tree with a
782 new name
784 void ldb_parse_tree_attr_replace(struct ldb_parse_tree *tree,
785 const char *attr,
786 const char *replace)
788 int i;
789 switch (tree->operation) {
790 case LDB_OP_AND:
791 case LDB_OP_OR:
792 for (i=0;i<tree->u.list.num_elements;i++) {
793 ldb_parse_tree_attr_replace(tree->u.list.elements[i],
794 attr, replace);
796 break;
797 case LDB_OP_NOT:
798 ldb_parse_tree_attr_replace(tree->u.isnot.child, attr, replace);
799 break;
800 case LDB_OP_EQUALITY:
801 case LDB_OP_GREATER:
802 case LDB_OP_LESS:
803 case LDB_OP_APPROX:
804 if (ldb_attr_cmp(tree->u.equality.attr, attr) == 0) {
805 tree->u.equality.attr = replace;
807 break;
808 case LDB_OP_SUBSTRING:
809 if (ldb_attr_cmp(tree->u.substring.attr, attr) == 0) {
810 tree->u.substring.attr = replace;
812 break;
813 case LDB_OP_PRESENT:
814 if (ldb_attr_cmp(tree->u.present.attr, attr) == 0) {
815 tree->u.present.attr = replace;
817 break;
818 case LDB_OP_EXTENDED:
819 if (tree->u.extended.attr &&
820 ldb_attr_cmp(tree->u.extended.attr, attr) == 0) {
821 tree->u.extended.attr = replace;
823 break;