Merge remote-tracking branch 'origin/dev-add-lucene-regex-compliance'…

… into dev-add-lucene-regex-compliance

logprep/filter/expression/filter_expression.py

@@ -318,6 +318,22 @@ def does_match(self, document: dict) -> bool:
         return self._lower_bound <= value <= self._upper_bound
 
 
+class LuceneRegexExpression(KeyValueBasedFilterExpression):
+    """Lucene compliant filter expression that matches a value using regex."""
+
+    def __init__(self, key: List[str], regex: str):
+        self._regex = regex
+        self._matcher = re.compile(self._regex)
+        super().__init__(key, f"/{self._regex.strip('^$')}/")
+
+    def does_match(self, document: dict) -> bool:
+        value = self._get_value(self.key, document)
+
+        if isinstance(value, list):
+            return any(filter(self._matcher.match, value))
+        return self._matcher.match(str(value)) is not None
+
+
 class RegExFilterExpression(KeyValueBasedFilterExpression):
     """Filter expression that matches a value using regex."""
 

logprep/filter/lucene_filter.py

@@ -276,8 +276,13 @@ def _parse_tree(self, tree: luqum.tree) -> FilterExpression:
             if self._last_search_field:
                 return self._create_field_group_expression(tree)
             return self._create_value_expression(tree)
+        if isinstance(tree, Regex):
+            return self._create_regex_expression(tree)
         raise LuceneFilterError(f'The expression "{str(tree)}" is invalid!')
 
+    def _create_regex_expression(self, tree: luqum.tree) -> LuceneRegexExpression:
+        pass
+
     def _create_field_group_expression(self, tree: luqum.tree) -> FilterExpression:
         """Creates filter expression that is resulting from a field group.