BUGFIX: Fix bugs with Cypher validation (#13849)

Fixes https://github.com/langchain-ai/langchain/issues/13803. Thanks to
@sakusaku-rich

libs/langchain/langchain/chains/graph_qa/cypher_utils.py

@@ -14,7 +14,9 @@ class CypherQueryCorrector:
 
     property_pattern = re.compile(r"\{.+?\}")
     node_pattern = re.compile(r"\(.+?\)")
-    path_pattern = re.compile(r"\(.*\).*-.*-.*\(.*\)")
+    path_pattern = re.compile(
+        r"(\([^\,\(\)]*?(\{.+\})?[^\,\(\)]*?\))(<?-)(\[.*?\])?(->?)(\([^\,\(\)]*?(\{.+\})?[^\,\(\)]*?\))"
+    )
     node_relation_node_pattern = re.compile(
         r"(\()+(?P<left_node>[^()]*?)\)(?P<relation>.*?)\((?P<right_node>[^()]*?)(\))+"
     )
@@ -62,7 +64,17 @@ class CypherQueryCorrector:
         Args:
             query: cypher query
         """
-        return re.findall(self.path_pattern, query)
+        paths = []
+        idx = 0
+        while matched := self.path_pattern.findall(query[idx:]):
+            matched = matched[0]
+            matched = [
+                m for i, m in enumerate(matched) if i not in [1, len(matched) - 1]
+            ]
+            path = "".join(matched)
+            idx = query.find(path) + len(path) - len(matched[-1])
+            paths.append(path)
+        return paths
 
     def judge_direction(self, relation: str) -> str:
         """

libs/langchain/tests/unit_tests/data/cypher_corrector.csv

@@ -506,3 +506,7 @@ RETURN a.title AS title, c.text AS text, a.date AS date
 ORDER BY date DESC LIMIT 3"
 "MATCH (n:`Some Label`)-[:`SOME REL TYPE üäß`]->(m:`Sömé Øther Læbel`) RETURN n,m","(Some Label, SOME REL TYPE üäß, Sömé Øther Læbel)","MATCH (n:`Some Label`)-[:`SOME REL TYPE üäß`]->(m:`Sömé Øther Læbel`) RETURN n,m"
 "MATCH (n:`Some Label`)<-[:`SOME REL TYPE üäß`]-(m:`Sömé Øther Læbel`) RETURN n,m","(Some Label, SOME REL TYPE üäß, Sömé Øther Læbel)","MATCH (n:`Some Label`)-[:`SOME REL TYPE üäß`]->(m:`Sömé Øther Læbel`) RETURN n,m"
+"MATCH (a:Actor {name: 'Tom Hanks'})-[:ACTED_IN]->(m:Movie) RETURN count(m)","(Movie, IN_GENRE, Genre), (User, RATED, Movie), (Actor, ACTED_IN, Movie), (Actor, DIRECTED, Movie), (Director, DIRECTED, Movie), (Director, ACTED_IN, Movie), (Person, ACTED_IN, Movie), (Person, DIRECTED, Movie)","MATCH (a:Actor {name: 'Tom Hanks'})-[:ACTED_IN]->(m:Movie) RETURN count(m)"
+"MATCH (a:Actor)-[:ACTED_IN]->(:Movie)-[:IN_GENRE]->(g1:Genre), (a)-[:ACTED_IN]->(:Movie)-[:IN_GENRE]->(g2:Genre) WHERE g1.name = 'Comedy' AND g2.name = 'Action' RETURN DISTINCT a.name","(Movie, IN_GENRE, Genre), (User, RATED, Movie), (Actor, ACTED_IN, Movie), (Actor, DIRECTED, Movie), (Director, DIRECTED, Movie), (Director, ACTED_IN, Movie), (Person, ACTED_IN, Movie), (Person, DIRECTED, Movie)","MATCH (a:Actor)-[:ACTED_IN]->(:Movie)-[:IN_GENRE]->(g1:Genre), (a)-[:ACTED_IN]->(:Movie)-[:IN_GENRE]->(g2:Genre) WHERE g1.name = 'Comedy' AND g2.name = 'Action' RETURN DISTINCT a.name"
+"MATCH (a:Actor)-[:ACTED_IN]->(m:Movie) RETURN a.name, COUNT(m) AS movieCount ORDER BY movieCount DESC LIMIT 1","(Movie, IN_GENRE, Genre), (User, RATED, Movie), (Actor, ACTED_IN, Movie), (Actor, DIRECTED, Movie), (Director, DIRECTED, Movie), (Director, ACTED_IN, Movie), (Person, ACTED_IN, Movie), (Person, DIRECTED, Movie)","MATCH (a:Actor)-[:ACTED_IN]->(m:Movie) RETURN a.name, COUNT(m) AS movieCount ORDER BY movieCount DESC LIMIT 1"
+"MATCH (g:Genre)<-[:IN_GENRE]-(m:Movie) RETURN g.name, COUNT(m) AS movieCount","(Movie, IN_GENRE, Genre), (User, RATED, Movie), (Actor, ACTED_IN, Movie), (Actor, DIRECTED, Movie), (Director, DIRECTED, Movie), (Director, ACTED_IN, Movie), (Person, ACTED_IN, Movie), (Person, DIRECTED, Movie)","MATCH (g:Genre)<-[:IN_GENRE]-(m:Movie) RETURN g.name, COUNT(m) AS movieCount"