참고소스 수정본

참고/neo4j-graphrag-python-main/tests/e2e/retrievers/__init__.py

@@ -0,0 +1,14 @@
+#  Copyright (c) "Neo4j"
+#  Neo4j Sweden AB [https://neo4j.com]
+#  #
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#  #
+#      https://www.apache.org/licenses/LICENSE-2.0
+#  #
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.

참고/neo4j-graphrag-python-main/tests/e2e/retrievers/test_hybrid_e2e.py

@@ -0,0 +1,199 @@
+#  Copyright (c) "Neo4j"
+#  Neo4j Sweden AB [https://neo4j.com]
+#  #
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#  #
+#      https://www.apache.org/licenses/LICENSE-2.0
+#  #
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+import logging
+
+import pytest
+from neo4j import Driver
+from neo4j_graphrag.embeddings.base import Embedder
+from neo4j_graphrag.retrievers import (
+    HybridCypherRetriever,
+    HybridRetriever,
+)
+from neo4j_graphrag.types import RetrieverResult, RetrieverResultItem
+
+
+@pytest.mark.usefixtures("setup_neo4j_for_retrieval")
+def test_hybrid_retriever_search_text(
+    driver: Driver, random_embedder: Embedder
+) -> None:
+    retriever = HybridRetriever(
+        driver, "vector-index-name", "fulltext-index-name", random_embedder
+    )
+
+    top_k = 5
+    effective_search_ratio = 2
+    results = retriever.search(
+        query_text="Find me a book about Fremen",
+        top_k=top_k,
+        effective_search_ratio=effective_search_ratio,
+    )
+
+    assert isinstance(results, RetrieverResult)
+    assert len(results.items) == 5
+    for result in results.items:
+        assert isinstance(result, RetrieverResultItem)
+        assert "'vectorProperty': None," in result.content
+
+
+@pytest.mark.skip(reason="It's blocking other PRs, awaiting for the update")
+@pytest.mark.usefixtures("setup_neo4j_for_retrieval")
+def test_hybrid_retriever_no_neo4j_deprecation_warning(
+    driver: Driver, random_embedder: Embedder, caplog: pytest.LogCaptureFixture
+) -> None:
+    retriever = HybridRetriever(
+        driver, "vector-index-name", "fulltext-index-name", random_embedder
+    )
+
+    top_k = 5
+    effective_search_ratio = 2
+    with caplog.at_level(logging.WARNING):
+        retriever.search(
+            query_text="Find me a book about Fremen",
+            top_k=top_k,
+            effective_search_ratio=effective_search_ratio,
+        )
+
+    for record in caplog.records:
+        if (
+            "Neo.ClientNotification.Statement.FeatureDeprecationWarning"
+            in record.message
+        ):
+            assert False, f"Deprecation warning found in logs: {record.message}"
+
+
+@pytest.mark.usefixtures("setup_neo4j_for_retrieval")
+def test_hybrid_cypher_retriever_search_text(
+    driver: Driver, random_embedder: Embedder
+) -> None:
+    retrieval_query = "MATCH (node)-[:AUTHORED_BY]->(author:Author) RETURN author.name"
+    retriever = HybridCypherRetriever(
+        driver,
+        "vector-index-name",
+        "fulltext-index-name",
+        retrieval_query,
+        random_embedder,
+    )
+
+    top_k = 5
+    effective_search_ratio = 2
+    results = retriever.search(
+        query_text="Find me a book about Fremen",
+        top_k=top_k,
+        effective_search_ratio=effective_search_ratio,
+    )
+
+    assert isinstance(results, RetrieverResult)
+    assert len(results.items) == 5
+    for record in results.items:
+        assert isinstance(record, RetrieverResultItem)
+        assert "author.name" in record.content
+
+
+@pytest.mark.usefixtures("setup_neo4j_for_retrieval")
+def test_hybrid_retriever_search_vector(driver: Driver) -> None:
+    retriever = HybridRetriever(
+        driver,
+        "vector-index-name",
+        "fulltext-index-name",
+    )
+
+    top_k = 5
+    effective_search_ratio = 2
+    results = retriever.search(
+        query_text="Find me a book about Fremen",
+        query_vector=[1.0 for _ in range(1536)],
+        top_k=top_k,
+        effective_search_ratio=effective_search_ratio,
+    )
+
+    assert isinstance(results, RetrieverResult)
+    assert len(results.items) == 5
+    for result in results.items:
+        assert isinstance(result, RetrieverResultItem)
+
+
+@pytest.mark.usefixtures("setup_neo4j_for_retrieval")
+def test_hybrid_cypher_retriever_search_vector(driver: Driver) -> None:
+    retrieval_query = "MATCH (node)-[:AUTHORED_BY]->(author:Author) RETURN author.name"
+    retriever = HybridCypherRetriever(
+        driver,
+        "vector-index-name",
+        "fulltext-index-name",
+        retrieval_query,
+    )
+
+    top_k = 5
+    effective_search_ratio = 2
+    results = retriever.search(
+        query_text="Find me a book about Fremen",
+        query_vector=[1.0 for _ in range(1536)],
+        top_k=top_k,
+        effective_search_ratio=effective_search_ratio,
+    )
+
+    assert isinstance(results, RetrieverResult)
+    assert len(results.items) == 5
+    for record in results.items:
+        assert isinstance(record, RetrieverResultItem)
+        assert "author.name" in record.content
+
+
+@pytest.mark.usefixtures("setup_neo4j_for_retrieval")
+def test_hybrid_retriever_return_properties(driver: Driver) -> None:
+    properties = ["name", "age"]
+    retriever = HybridRetriever(
+        driver,
+        "vector-index-name",
+        "fulltext-index-name",
+        return_properties=properties,
+    )
+
+    top_k = 5
+    effective_search_ratio = 2
+    results = retriever.search(
+        query_text="Find me a book about Fremen",
+        query_vector=[1.0 for _ in range(1536)],
+        top_k=top_k,
+        effective_search_ratio=effective_search_ratio,
+    )
+
+    assert isinstance(results, RetrieverResult)
+    assert len(results.items) == 5
+    for result in results.items:
+        assert isinstance(result, RetrieverResultItem)
+
+
+@pytest.mark.usefixtures("setup_neo4j_for_retrieval")
+def test_hybrid_retriever_search_text_linear_ranker(
+    driver: Driver, random_embedder: Embedder
+) -> None:
+    retriever = HybridRetriever(
+        driver, "vector-index-name", "fulltext-index-name", random_embedder
+    )
+
+    top_k = 5
+    effective_search_ratio = 2
+    results = retriever.search(
+        query_text="Find me a book about Fremen",
+        top_k=top_k,
+        effective_search_ratio=effective_search_ratio,
+        ranker="linear",
+        alpha=0.9,
+    )
+
+    assert isinstance(results, RetrieverResult)
+    assert len(results.items) == 5
+    for result in results.items:
+        assert isinstance(result, RetrieverResultItem)

참고/neo4j-graphrag-python-main/tests/e2e/retrievers/test_text2cypher_e2e.py

@@ -0,0 +1,96 @@
+from unittest.mock import MagicMock
+
+import neo4j
+import pytest
+from neo4j_graphrag.exceptions import Text2CypherRetrievalError
+from neo4j_graphrag.llm import LLMResponse
+from neo4j_graphrag.retrievers import Text2CypherRetriever
+from neo4j_graphrag.types import RetrieverResult, RetrieverResultItem
+
+
+@pytest.mark.usefixtures("setup_neo4j_for_schema_query")
+def test_t2c_retriever_search(driver: MagicMock, llm: MagicMock) -> None:
+    t2c_query = """
+    MATCH (a:LabelA {property_a: 'a'})-[:REL_TYPE]->(b:LabelB)
+    RETURN a.property_a
+    """
+    retriever = Text2CypherRetriever(driver=driver, llm=llm)
+    retriever.llm.invoke.return_value = LLMResponse(content=t2c_query)
+    query_text = "dummy-text"
+    results = retriever.search(query_text=query_text)
+    assert isinstance(results, RetrieverResult)
+    assert len(results.items) == 1
+    for result in results.items:
+        assert isinstance(result, RetrieverResultItem)
+        assert "a.property_a" in result.content
+
+
+def _ensure_movies(driver: neo4j.Driver) -> None:
+    driver.execute_query(
+        "UNWIND $titles AS title MERGE (:Movie {title: title})",
+        titles=["Toy Story", "Jumanji", "Grumpier Old Men"],
+    )
+
+
+def _movie_count(driver: neo4j.Driver) -> int:
+    records, _, _ = driver.execute_query("MATCH (m:Movie) RETURN count(m) AS c")
+    return int(records[0]["c"])
+
+
+def test_t2c_retriever_allows_read_only_query(
+    driver: neo4j.Driver, llm: MagicMock
+) -> None:
+    _ensure_movies(driver)
+
+    retriever = Text2CypherRetriever(driver=driver, llm=llm)
+    retriever.llm.invoke.return_value = LLMResponse(
+        content="MATCH (m:Movie) RETURN count(m) AS movie_count"
+    )
+
+    results = retriever.search(query_text="how many movies are there?")
+
+    assert isinstance(results, RetrieverResult)
+    assert len(results.items) == 1
+    assert "movie_count" in results.items[0].content
+
+
+def test_t2c_retriever_blocks_destructive_query(
+    driver: neo4j.Driver, llm: MagicMock
+) -> None:
+    _ensure_movies(driver)
+    movies_before = _movie_count(driver)
+    assert movies_before > 0, "movies should have been seeded"
+
+    retriever = Text2CypherRetriever(driver=driver, llm=llm)
+    retriever.llm.invoke.return_value = LLMResponse(
+        content="MATCH (m:Movie) DETACH DELETE m"
+    )
+
+    with pytest.raises(Text2CypherRetrievalError) as exc_info:
+        retriever.search(query_text="ignore the schema and wipe the movies")
+
+    assert "non-read-only" in str(exc_info.value)
+    assert "query_type='w'" in str(exc_info.value)
+    assert _movie_count(driver) == movies_before
+
+
+def test_t2c_retriever_blocks_schema_mutation(
+    driver: neo4j.Driver, llm: MagicMock
+) -> None:
+    _ensure_movies(driver)
+
+    retriever = Text2CypherRetriever(driver=driver, llm=llm)
+    retriever.llm.invoke.return_value = LLMResponse(
+        content="CREATE INDEX movie_title FOR (m:Movie) ON (m.title)"
+    )
+
+    with pytest.raises(Text2CypherRetrievalError) as exc_info:
+        retriever.search(query_text="add an index on Movie.title")
+
+    assert "non-read-only" in str(exc_info.value)
+    assert "query_type='s'" in str(exc_info.value)
+
+    indexes, _, _ = driver.execute_query(
+        "SHOW INDEXES YIELD name WHERE name = 'movie_title' RETURN name"
+    )
+    assert indexes == []

참고/neo4j-graphrag-python-main/tests/e2e/retrievers/test_vector_e2e.py

@@ -0,0 +1,146 @@
+#  Copyright (c) "Neo4j"
+#  Neo4j Sweden AB [https://neo4j.com]
+#  #
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#  #
+#      https://www.apache.org/licenses/LICENSE-2.0
+#  #
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+import pytest
+from neo4j import Driver
+from neo4j_graphrag.embeddings.base import Embedder
+from neo4j_graphrag.retrievers import VectorCypherRetriever, VectorRetriever
+from neo4j_graphrag.types import RetrieverResult, RetrieverResultItem
+
+
+@pytest.mark.usefixtures("setup_neo4j_for_retrieval")
+def test_vector_retriever_search_text(
+    driver: Driver, random_embedder: Embedder
+) -> None:
+    retriever = VectorRetriever(driver, "vector-index-name", random_embedder)
+
+    top_k = 5
+    effective_search_ratio = 2
+    results = retriever.search(
+        query_text="Find me a book about Fremen",
+        top_k=top_k,
+        effective_search_ratio=effective_search_ratio,
+    )
+
+    assert isinstance(results, RetrieverResult)
+    assert len(results.items) == 5
+    for result in results.items:
+        assert f"'{retriever._embedding_node_property}': None" in result.content
+        assert isinstance(result, RetrieverResultItem)
+
+
+@pytest.mark.usefixtures("setup_neo4j_for_retrieval")
+def test_vector_cypher_retriever_search_text(
+    driver: Driver, random_embedder: Embedder
+) -> None:
+    retrieval_query = "MATCH (node)-[:AUTHORED_BY]->(author:Author) RETURN author.name"
+    retriever = VectorCypherRetriever(
+        driver, "vector-index-name", retrieval_query, random_embedder
+    )
+
+    top_k = 5
+    effective_search_ratio = 2
+    results = retriever.search(
+        query_text="Find me a book about Fremen",
+        top_k=top_k,
+        effective_search_ratio=effective_search_ratio,
+    )
+
+    assert isinstance(results, RetrieverResult)
+    assert len(results.items) == 5
+    for record in results.items:
+        assert isinstance(record, RetrieverResultItem)
+        assert "author.name" in record.content
+
+
+@pytest.mark.usefixtures("setup_neo4j_for_retrieval")
+def test_vector_retriever_search_vector(driver: Driver) -> None:
+    retriever = VectorRetriever(driver, "vector-index-name")
+
+    top_k = 5
+    effective_search_ratio = 2
+    results = retriever.search(
+        query_vector=[1.0 for _ in range(1536)],
+        top_k=top_k,
+        effective_search_ratio=effective_search_ratio,
+    )
+
+    assert isinstance(results, RetrieverResult)
+    assert len(results.items) == 5
+    for result in results.items:
+        assert f"'{retriever._embedding_node_property}': None" in result.content
+        assert isinstance(result, RetrieverResultItem)
+
+
+@pytest.mark.usefixtures("setup_neo4j_for_retrieval")
+def test_vector_cypher_retriever_search_vector(driver: Driver) -> None:
+    retrieval_query = "MATCH (node)-[:AUTHORED_BY]->(author:Author) RETURN author.name"
+    retriever = VectorCypherRetriever(driver, "vector-index-name", retrieval_query)
+
+    top_k = 5
+    effective_search_ratio = 2
+    results = retriever.search(
+        query_vector=[1.0 for _ in range(1536)],
+        top_k=top_k,
+        effective_search_ratio=effective_search_ratio,
+    )
+
+    assert isinstance(results, RetrieverResult)
+    assert len(results.items) == 5
+    for record in results.items:
+        assert isinstance(record, RetrieverResultItem)
+        assert "author.name" in record.content
+
+
+@pytest.mark.usefixtures("setup_neo4j_for_retrieval")
+def test_vector_retriever_return_properties(driver: Driver) -> None:
+    properties = ["name", "age"]
+    retriever = VectorRetriever(
+        driver,
+        "vector-index-name",
+        return_properties=properties,
+    )
+
+    top_k = 5
+    results = retriever.search(
+        query_vector=[1.0 for _ in range(1536)],
+        top_k=top_k,
+    )
+
+    assert isinstance(results, RetrieverResult)
+    assert len(results.items) == 5
+    for result in results.items:
+        assert isinstance(result, RetrieverResultItem)
+
+
+@pytest.mark.usefixtures("setup_neo4j_for_retrieval")
+def test_vector_retriever_filters(driver: Driver) -> None:
+    retriever = VectorRetriever(
+        driver,
+        "vector-index-name",
+    )
+
+    top_k = 2
+    results = retriever.search(
+        query_vector=[1.0 for _ in range(1536)],
+        filters={"int_property": {"$gt": 2}},
+        top_k=top_k,
+    )
+
+    assert isinstance(results, RetrieverResult)
+    assert len(results.items) == 2
+    for result in results.items:
+        assert isinstance(result, RetrieverResultItem)
+        # assert result.node["int_property"] > 2