dissect.util 3.23.dev6__pp311-pypy311_pp73-manylinux_2_28_i686.whl → 3.23.dev8__pp311-pypy311_pp73-manylinux_2_28_i686.whl

dissect/util/ldap.py

@@ -0,0 +1,237 @@
+from __future__ import annotations
+
+import operator
+import re
+from enum import Enum
+
+from dissect.util.exceptions import Error
+
+
+class InvalidQueryError(Error):
+    pass
+
+
+class LogicalOperator(Enum):
+    AND = "&"
+    OR = "|"
+    NOT = "!"
+
+
+_LOGICAL_OPERATORS = tuple(op.value for op in LogicalOperator)
+
+
+class ComparisonOperator(Enum):
+    GE = ">="
+    LE = "<="
+    GT = ">"
+    LT = "<"
+    EQ = "="
+    APPROX = "~="
+    BIT = ":="
+    EXTENDED = ":"
+
+
+_NORMAL_COMPARISON_OPERATORS = [op for op in ComparisonOperator if op != ComparisonOperator.EXTENDED]
+_SORTED_COMPARISON_OPERATORS = sorted(_NORMAL_COMPARISON_OPERATORS, key=lambda op: len(op.value), reverse=True)
+
+_RE_EXTENDED = re.compile(r"(.+?):(.+?):=(.+)?")
+
+
+class SearchFilter:
+    """Represents an LDAP search filter (simple or nested).
+
+    Args:
+        query: The LDAP search filter string.
+    """
+
+    def __init__(self, query: str) -> None:
+        self.query: str = query
+
+        self.children: list[SearchFilter] = []
+        self.operator: LogicalOperator | ComparisonOperator | None = None
+        self.attribute: str | None = None
+        self.value: str | None = None
+        self._extended_rule: str | None = None
+
+        _validate_syntax(query)
+
+        if query[1:-1].startswith(_LOGICAL_OPERATORS):
+            self._parse_nested()
+        else:
+            self._parse_simple()
+
+    def __repr__(self) -> str:
+        if self.is_nested():
+            return f"<SearchFilter nested operator={self.operator.value!r} children={self.children}>"
+        return f"<SearchFilter attribute={self.attribute!r} operator={self.operator.value!r} value={self.value}>"
+
+    @classmethod
+    def parse(cls, query: str, optimize: bool = True) -> SearchFilter:
+        """Parse an LDAP query into a filter object, with optional optimization."""
+        result = cls(query)
+        if optimize:
+            return optimize_ldap_query(result)[0]
+        return result
+
+    def is_nested(self) -> bool:
+        """Return whether the filter is nested (i.e., contains logical operators and child filters)."""
+        return isinstance(self.operator, LogicalOperator)
+
+    def format(self) -> str:
+        """Format the search filter back into an LDAP query string."""
+        if self.is_nested():
+            childs = "".join([child.format() for child in self.children])
+            return f"({self.operator.value}{childs})"
+
+        if self.operator == ComparisonOperator.EXTENDED:
+            return f"({self.attribute}:{self._extended_rule}:={self.value})"
+
+        return f"({self.attribute}{self.operator.value}{self.value})"
+
+    def _parse_simple(self) -> None:
+        """Parse simple filter."""
+        query = self.query[1:-1]
+
+        # Check for extended matching rules first
+        if ":" in query and (match := _RE_EXTENDED.match(query)):
+            self.operator = ComparisonOperator.EXTENDED
+            self.attribute, self._extended_rule, self.value = match.groups()
+            return
+
+        # Regular operator parsing
+        test = query
+        operators: list[ComparisonOperator] = []
+        for op in _SORTED_COMPARISON_OPERATORS:
+            if op.value not in test:
+                continue
+
+            if test.count(op.value) > 1:
+                raise InvalidQueryError(f"Comparison operator {op.value} found multiple times in query: {self.query}")
+
+            operators.append(op)
+            test = test.replace(op.value, "")
+
+        if len(operators) == 0:
+            raise InvalidQueryError(
+                f"No comparison operator found in query: {self.query}. "
+                f"Expected one of {[op.value for op in _NORMAL_COMPARISON_OPERATORS]}."
+            )
+
+        if len(operators) > 1:
+            raise InvalidQueryError(
+                f"Multiple comparison operators found in query: {self.query} -> {[o.value for o in operators]} "
+                f"Expected only one of {[op.value for op in _NORMAL_COMPARISON_OPERATORS]}."
+            )
+
+        self.operator = operators[0]
+        self.attribute, _, self.value = query.partition(self.operator.value)
+
+    def _parse_nested(self) -> None:
+        """Parse nested filter."""
+        query = self.query[1:-1]
+        self.operator = LogicalOperator(query[0])
+
+        start = 1
+        while start < len(query):
+            end = start + 1
+            depth = 1
+
+            while end < len(query) and depth > 0:
+                if query[end] == "(":
+                    depth += 1
+                elif query[end] == ")":
+                    depth -= 1
+                end += 1
+
+            self.children.append(SearchFilter(query[start:end]))
+            start = end
+
+
+_ATTRIBUTE_WEIGHTS = {
+    "objectGUID": 1,
+    "distinguishedName": 1,
+    "sAMAccountName": 2,
+    "userPrincipalName": 2,
+    "mail": 2,
+    "sAMAccountType": 3,
+    "servicePrincipalName": 3,
+    "userAccountControl": 4,
+    "memberOf": 5,
+    "member": 5,
+    "pwdLastSet": 5,
+    "primaryGroupID": 6,
+    "whenCreated": 6,
+    "ou": 6,
+    "lastLogonTimestamp": 6,
+    "cn": 7,
+    "givenName": 7,
+    "name": 7,
+    "telephoneNumber": 7,
+    "objectCategory": 8,
+    "description": 9,
+    "objectClass": 10,
+}
+
+
+def optimize_ldap_query(query: SearchFilter) -> tuple[SearchFilter, int]:
+    """Optimize an LDAP query in-place.
+
+    Removes redundant conditions and sorts filters and conditions based on how specific they are.
+
+    Args:
+        query: The LDAP query to optimize.
+
+    Returns:
+        A tuple containing the optimized LDAP query and its weight.
+    """
+    # Simplify single-child AND/OR
+    if query.is_nested() and len(query.children) == 1 and query.operator in (LogicalOperator.AND, LogicalOperator.OR):
+        return optimize_ldap_query(query.children[0])
+
+    # Sort nested children by weight
+    if query.is_nested() and len(query.children) > 1:
+        children = sorted((optimize_ldap_query(child) for child in query.children), key=operator.itemgetter(1))
+
+        query.children = [child for child, _ in children]
+        query.query = query.format()
+
+        return query, max(weight for _, weight in children)
+
+    # Handle NOT
+    if query.is_nested() and len(query.children) == 1 and query.operator == LogicalOperator.NOT:
+        child, weight = optimize_ldap_query(query.children[0])
+
+        query.children[0] = child
+        query.query = query.format()
+
+        return query, weight
+
+    # Base case: simple filter
+    if not query.is_nested():
+        return query, _ATTRIBUTE_WEIGHTS.get(query.attribute, max(_ATTRIBUTE_WEIGHTS.values()))
+
+    return query, max(_ATTRIBUTE_WEIGHTS.values())
+
+
+def _validate_syntax(query: str) -> None:
+    """Validate basic LDAP query syntax.
+
+    Args:
+        query: The LDAP query to validate.
+    """
+    if not query:
+        raise InvalidQueryError("Empty query")
+
+    if not query.startswith("(") or not query.endswith(")"):
+        raise InvalidQueryError(f"Query must be wrapped in parentheses: {query}")
+
+    if query.count("(") != query.count(")"):
+        raise InvalidQueryError(f"Unbalanced parentheses in query: {query}")
+
+    # Check for empty parentheses
+    if "()" in query:
+        raise InvalidQueryError(f"Empty parentheses found in query: {query}")
+
+    # Check for queries that start with double opening parentheses
+    if query.startswith("(("):
+        raise InvalidQueryError(f"Invalid query structure: {query}")

{dissect_util-3.23.dev6.dist-info → dissect_util-3.23.dev8.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dissect.util
-Version: 3.23.dev6
+Version: 3.23.dev8
 Summary: A Dissect module implementing various utility functions for the other Dissect modules
 Author-email: Dissect Team <dissect@fox-it.com>
 License-Expression: Apache-2.0

{dissect_util-3.23.dev6.dist-info → dissect_util-3.23.dev8.dist-info}/RECORD

@@ -4,6 +4,7 @@ dissect/util/_native.pypy311-pp73-x86-linux-gnu.so,sha256=V5nOSzAA_40AcZizatJggq
 dissect/util/cpio.py,sha256=epeCGfqeCglP1VVFjKvupzuSvI1DPjZdkgbGAo1R8-8,7576
 dissect/util/exceptions.py,sha256=vkoy9zXcx8U6K4prI5Cqu1mFndizIeIihAfR6DDj44Q,75
 dissect/util/feature.py,sha256=6uciTovTG97AMG5EPy1-xlvkbnTML7Mw2wzhyCwr6qc,2062
+dissect/util/ldap.py,sha256=gykOupKh_MqtcXBGyYfhs7Ye3gYdxc9rhtngbXexeKo,7364
 dissect/util/plist.py,sha256=cuOP1GQLpF2qC71jYGBXVxCWSWwi3LL92sHbn4AEBaA,4551
 dissect/util/sid.py,sha256=ojvKL6vaHHKukfkvMMKqJfpC40UypVGc-ppkqPnj2Co,1558
 dissect/util/stream.py,sha256=M7H2Fg6tQtqhaVVaAkOpve7cbHdzW_g4tGOt6cZx_EA,23113
@@ -33,10 +34,10 @@ dissect/util/hash/crc32c.py,sha256=ncXaXxEVxVomOnF4y0Do_lOXlB-qzXsqS-SDNq4Gels,3
 dissect/util/hash/jenkins.py,sha256=QYgEy9heN-IEirfTcdyi6s0U2jm9n7XuXl_3D_OmOQQ,3527
 dissect/util/tools/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 dissect/util/tools/dump_nskeyedarchiver.py,sha256=ZMSM4TPL8pRn685tX3SAKackgE6EFr2bExPU6Lm8jWk,1454
-dissect_util-3.23.dev6.dist-info/METADATA,sha256=V8gArEcF0_3Z6swYTAXaQjyOV1q--BTyLUFgtQYOlqs,3614
-dissect_util-3.23.dev6.dist-info/WHEEL,sha256=u40QNX8TBeea9UTf3BA0Qi-K-KIQG4GovVWcq-m2DAg,118
-dissect_util-3.23.dev6.dist-info/entry_points.txt,sha256=3mwSDD1MUdj0nnTcl2qYFMeu6KYMTycUQS0rlp5CQHI,86
-dissect_util-3.23.dev6.dist-info/top_level.txt,sha256=Mn-CQzEYsAbkxrUI0TnplHuXnGVKzxpDw_po_sXpvv4,8
-dissect_util-3.23.dev6.dist-info/RECORD,,
-dissect_util-3.23.dev6.dist-info/licenses/COPYRIGHT,sha256=I5uKS6nwQl9xC2xgk-8jvw0gNrpFhvZC7FDTALpj18k,313
-dissect_util-3.23.dev6.dist-info/licenses/LICENSE,sha256=PhUqiw6jAh2KbBdVRPBq_hfAvfcTBin7nZ3CK7NQbTM,11341
+dissect_util-3.23.dev8.dist-info/METADATA,sha256=WnpK47O3EawR3as0r2Yqe9rBXyEIBFNFsBMJ3-fWJ38,3614
+dissect_util-3.23.dev8.dist-info/WHEEL,sha256=u40QNX8TBeea9UTf3BA0Qi-K-KIQG4GovVWcq-m2DAg,118
+dissect_util-3.23.dev8.dist-info/entry_points.txt,sha256=3mwSDD1MUdj0nnTcl2qYFMeu6KYMTycUQS0rlp5CQHI,86
+dissect_util-3.23.dev8.dist-info/top_level.txt,sha256=Mn-CQzEYsAbkxrUI0TnplHuXnGVKzxpDw_po_sXpvv4,8
+dissect_util-3.23.dev8.dist-info/RECORD,,
+dissect_util-3.23.dev8.dist-info/licenses/COPYRIGHT,sha256=I5uKS6nwQl9xC2xgk-8jvw0gNrpFhvZC7FDTALpj18k,313
+dissect_util-3.23.dev8.dist-info/licenses/LICENSE,sha256=PhUqiw6jAh2KbBdVRPBq_hfAvfcTBin7nZ3CK7NQbTM,11341