from __future__ import annotations
import enum
import logging
import os
from hashlib import md5
from typing import (
Any,
Callable,
Dict,
Iterable,
List,
Optional,
Tuple,
Type,
)
import numpy as np
from langchain_core.documents import Document
from langchain_core.embeddings import Embeddings
from langchain_core.utils import get_from_dict_or_env
from langchain_core.vectorstores import VectorStore
from langchain_community.graphs import Neo4jGraph
from langchain_community.vectorstores.utils import (
DistanceStrategy,
maximal_marginal_relevance,
)
DEFAULT_DISTANCE_STRATEGY = DistanceStrategy.COSINE
DISTANCE_MAPPING = {
DistanceStrategy.EUCLIDEAN_DISTANCE: "euclidean",
DistanceStrategy.COSINE: "cosine",
}
COMPARISONS_TO_NATIVE = {
"$eq": "=",
"$ne": "<>",
"$lt": "<",
"$lte": "<=",
"$gt": ">",
"$gte": ">=",
}
SPECIAL_CASED_OPERATORS = {
"$in",
"$nin",
"$between",
}
TEXT_OPERATORS = {
"$like",
"$ilike",
}
LOGICAL_OPERATORS = {"$and", "$or"}
SUPPORTED_OPERATORS = (
set(COMPARISONS_TO_NATIVE)
.union(TEXT_OPERATORS)
.union(LOGICAL_OPERATORS)
.union(SPECIAL_CASED_OPERATORS)
)
[docs]class SearchType(str, enum.Enum):
"""Enumerator of the Distance strategies."""
VECTOR = "vector"
HYBRID = "hybrid"
DEFAULT_SEARCH_TYPE = SearchType.VECTOR
[docs]class IndexType(str, enum.Enum):
"""Enumerator of the index types."""
NODE = "NODE"
RELATIONSHIP = "RELATIONSHIP"
DEFAULT_INDEX_TYPE = IndexType.NODE
def _get_search_index_query(
search_type: SearchType, index_type: IndexType = DEFAULT_INDEX_TYPE
) -> str:
if index_type == IndexType.NODE:
type_to_query_map = {
SearchType.VECTOR: (
"CALL db.index.vector.queryNodes($index, $k, $embedding) "
"YIELD node, score "
),
SearchType.HYBRID: (
"CALL { "
"CALL db.index.vector.queryNodes($index, $k, $embedding) "
"YIELD node, score "
"WITH collect({node:node, score:score}) AS nodes, max(score) AS max "
"UNWIND nodes AS n "
# We use 0 as min
"RETURN n.node AS node, (n.score / max) AS score UNION "
"CALL db.index.fulltext.queryNodes($keyword_index, $query, "
"{limit: $k}) YIELD node, score "
"WITH collect({node:node, score:score}) AS nodes, max(score) AS max "
"UNWIND nodes AS n "
# We use 0 as min
"RETURN n.node AS node, (n.score / max) AS score "
"} "
# dedup
"WITH node, max(score) AS score ORDER BY score DESC LIMIT $k "
),
}
return type_to_query_map[search_type]
else:
return (
"CALL db.index.vector.queryRelationships($index, $k, $embedding) "
"YIELD relationship, score "
)
[docs]def check_if_not_null(props: List[str], values: List[Any]) -> None:
"""Check if the values are not None or empty string"""
for prop, value in zip(props, values):
if not value:
raise ValueError(f"Parameter `{prop}` must not be None or empty string")
[docs]def sort_by_index_name(
lst: List[Dict[str, Any]], index_name: str
) -> List[Dict[str, Any]]:
"""Sort first element to match the index_name if exists"""
return sorted(lst, key=lambda x: x.get("name") != index_name)
[docs]def remove_lucene_chars(text: str) -> str:
"""Remove Lucene special characters"""
special_chars = [
"+",
"-",
"&",
"|",
"!",
"(",
")",
"{",
"}",
"[",
"]",
"^",
'"',
"~",
"*",
"?",
":",
"\\",
]
for char in special_chars:
if char in text:
text = text.replace(char, " ")
return text.strip()
[docs]def dict_to_yaml_str(input_dict: Dict, indent: int = 0) -> str:
"""
Convert a dictionary to a YAML-like string without using external libraries.
Parameters:
- input_dict (dict): The dictionary to convert.
- indent (int): The current indentation level.
Returns:
- str: The YAML-like string representation of the input dictionary.
"""
yaml_str = ""
for key, value in input_dict.items():
padding = " " * indent
if isinstance(value, dict):
yaml_str += f"{padding}{key}:\n{dict_to_yaml_str(value, indent + 1)}"
elif isinstance(value, list):
yaml_str += f"{padding}{key}:\n"
for item in value:
yaml_str += f"{padding}- {item}\n"
else:
yaml_str += f"{padding}{key}: {value}\n"
return yaml_str
[docs]def combine_queries(
input_queries: List[Tuple[str, Dict[str, Any]]], operator: str
) -> Tuple[str, Dict[str, Any]]:
"""Combine multiple queries with an operator."""
# Initialize variables to hold the combined query and parameters
combined_query: str = ""
combined_params: Dict = {}
param_counter: Dict = {}
for query, params in input_queries:
# Process each query fragment and its parameters
new_query = query
for param, value in params.items():
# Update the parameter name to ensure uniqueness
if param in param_counter:
param_counter[param] += 1
else:
param_counter[param] = 1
new_param_name = f"{param}_{param_counter[param]}"
# Replace the parameter in the query fragment
new_query = new_query.replace(f"${param}", f"${new_param_name}")
# Add the parameter to the combined parameters dictionary
combined_params[new_param_name] = value
# Combine the query fragments with an AND operator
if combined_query:
combined_query += f" {operator} "
combined_query += f"({new_query})"
return combined_query, combined_params
[docs]def collect_params(
input_data: List[Tuple[str, Dict[str, str]]],
) -> Tuple[List[str], Dict[str, Any]]:
"""Transform the input data into the desired format.
Args:
- input_data (list of tuples): Input data to transform.
Each tuple contains a string and a dictionary.
Returns:
- tuple: A tuple containing a list of strings and a dictionary.
"""
# Initialize variables to hold the output parts
query_parts = []
params = {}
# Loop through each item in the input data
for query_part, param in input_data:
# Append the query part to the list
query_parts.append(query_part)
# Update the params dictionary with the param dictionary
params.update(param)
# Return the transformed data
return (query_parts, params)
def _handle_field_filter(
field: str, value: Any, param_number: int = 1
) -> Tuple[str, Dict]:
"""Create a filter for a specific field.
Args:
field: name of field
value: value to filter
If provided as is then this will be an equality filter
If provided as a dictionary then this will be a filter, the key
will be the operator and the value will be the value to filter by
param_number: sequence number of parameters used to map between param
dict and Cypher snippet
Returns a tuple of
- Cypher filter snippet
- Dictionary with parameters used in filter snippet
"""
if not isinstance(field, str):
raise ValueError(
f"field should be a string but got: {type(field)} with value: {field}"
)
if field.startswith("$"):
raise ValueError(
f"Invalid filter condition. Expected a field but got an operator: "
f"{field}"
)
# Allow [a-zA-Z0-9_], disallow $ for now until we support escape characters
if not field.isidentifier():
raise ValueError(f"Invalid field name: {field}. Expected a valid identifier.")
if isinstance(value, dict):
# This is a filter specification
if len(value) != 1:
raise ValueError(
"Invalid filter condition. Expected a value which "
"is a dictionary with a single key that corresponds to an operator "
f"but got a dictionary with {len(value)} keys. The first few "
f"keys are: {list(value.keys())[:3]}"
)
operator, filter_value = list(value.items())[0]
# Verify that that operator is an operator
if operator not in SUPPORTED_OPERATORS:
raise ValueError(
f"Invalid operator: {operator}. "
f"Expected one of {SUPPORTED_OPERATORS}"
)
else: # Then we assume an equality operator
operator = "$eq"
filter_value = value
if operator in COMPARISONS_TO_NATIVE:
# Then we implement an equality filter
# native is trusted input
native = COMPARISONS_TO_NATIVE[operator]
query_snippet = f"n.`{field}` {native} $param_{param_number}"
query_param = {f"param_{param_number}": filter_value}
return (query_snippet, query_param)
elif operator == "$between":
low, high = filter_value
query_snippet = (
f"$param_{param_number}_low <= n.`{field}` <= $param_{param_number}_high"
)
query_param = {
f"param_{param_number}_low": low,
f"param_{param_number}_high": high,
}
return (query_snippet, query_param)
elif operator in {"$in", "$nin", "$like", "$ilike"}:
# We'll do force coercion to text
if operator in {"$in", "$nin"}:
for val in filter_value:
if not isinstance(val, (str, int, float)):
raise NotImplementedError(
f"Unsupported type: {type(val)} for value: {val}"
)
if operator in {"$in"}:
query_snippet = f"n.`{field}` IN $param_{param_number}"
query_param = {f"param_{param_number}": filter_value}
return (query_snippet, query_param)
elif operator in {"$nin"}:
query_snippet = f"n.`{field}` NOT IN $param_{param_number}"
query_param = {f"param_{param_number}": filter_value}
return (query_snippet, query_param)
elif operator in {"$like"}:
query_snippet = f"n.`{field}` CONTAINS $param_{param_number}"
query_param = {f"param_{param_number}": filter_value.rstrip("%")}
return (query_snippet, query_param)
elif operator in {"$ilike"}:
query_snippet = f"toLower(n.`{field}`) CONTAINS $param_{param_number}"
query_param = {f"param_{param_number}": filter_value.rstrip("%")}
return (query_snippet, query_param)
else:
raise NotImplementedError()
else:
raise NotImplementedError()
[docs]class Neo4jVector(VectorStore):
"""`Neo4j` vector index.
To use, you should have the ``neo4j`` python package installed.
Args:
url: Neo4j connection url
username: Neo4j username.
password: Neo4j password
database: Optionally provide Neo4j database
Defaults to "neo4j"
embedding: Any embedding function implementing
`langchain.embeddings.base.Embeddings` interface.
distance_strategy: The distance strategy to use. (default: COSINE)
pre_delete_collection: If True, will delete existing data if it exists.
(default: False). Useful for testing.
Example:
.. code-block:: python
from langchain_community.vectorstores.neo4j_vector import Neo4jVector
from langchain_community.embeddings.openai import OpenAIEmbeddings
url="bolt://localhost:7687"
username="neo4j"
password="pleaseletmein"
embeddings = OpenAIEmbeddings()
vectorestore = Neo4jVector.from_documents(
embedding=embeddings,
documents=docs,
url=url
username=username,
password=password,
)
"""
[docs] def __init__(
self,
embedding: Embeddings,
*,
search_type: SearchType = SearchType.VECTOR,
username: Optional[str] = None,
password: Optional[str] = None,
url: Optional[str] = None,
keyword_index_name: Optional[str] = "keyword",
database: Optional[str] = None,
index_name: str = "vector",
node_label: str = "Chunk",
embedding_node_property: str = "embedding",
text_node_property: str = "text",
distance_strategy: DistanceStrategy = DEFAULT_DISTANCE_STRATEGY,
logger: Optional[logging.Logger] = None,
pre_delete_collection: bool = False,
retrieval_query: str = "",
relevance_score_fn: Optional[Callable[[float], float]] = None,
index_type: IndexType = DEFAULT_INDEX_TYPE,
graph: Optional[Neo4jGraph] = None,
) -> None:
try:
import neo4j
except ImportError:
raise ImportError(
"Could not import neo4j python package. "
"Please install it with `pip install neo4j`."
)
# Allow only cosine and euclidean distance strategies
if distance_strategy not in [
DistanceStrategy.EUCLIDEAN_DISTANCE,
DistanceStrategy.COSINE,
]:
raise ValueError(
"distance_strategy must be either 'EUCLIDEAN_DISTANCE' or 'COSINE'"
)
# Graph object takes precedent over env or input params
if graph:
self._driver = graph._driver
self._database = graph._database
else:
# Handle if the credentials are environment variables
# Support URL for backwards compatibility
if not url:
url = os.environ.get("NEO4J_URL")
url = get_from_dict_or_env({"url": url}, "url", "NEO4J_URI")
username = get_from_dict_or_env(
{"username": username}, "username", "NEO4J_USERNAME"
)
password = get_from_dict_or_env(
{"password": password}, "password", "NEO4J_PASSWORD"
)
database = get_from_dict_or_env(
{"database": database}, "database", "NEO4J_DATABASE", "neo4j"
)
self._driver = neo4j.GraphDatabase.driver(url, auth=(username, password))
self._database = database
# Verify connection
try:
self._driver.verify_connectivity()
except neo4j.exceptions.ServiceUnavailable:
raise ValueError(
"Could not connect to Neo4j database. "
"Please ensure that the url is correct"
)
except neo4j.exceptions.AuthError:
raise ValueError(
"Could not connect to Neo4j database. "
"Please ensure that the username and password are correct"
)
self.schema = ""
# Verify if the version support vector index
self._is_enterprise = False
self.verify_version()
# Verify that required values are not null
check_if_not_null(
[
"index_name",
"node_label",
"embedding_node_property",
"text_node_property",
],
[index_name, node_label, embedding_node_property, text_node_property],
)
self.embedding = embedding
self._distance_strategy = distance_strategy
self.index_name = index_name
self.keyword_index_name = keyword_index_name
self.node_label = node_label
self.embedding_node_property = embedding_node_property
self.text_node_property = text_node_property
self.logger = logger or logging.getLogger(__name__)
self.override_relevance_score_fn = relevance_score_fn
self.retrieval_query = retrieval_query
self.search_type = search_type
self._index_type = index_type
# Calculate embedding dimension
self.embedding_dimension = len(embedding.embed_query("foo"))
# Delete existing data if flagged
if pre_delete_collection:
from neo4j.exceptions import DatabaseError
self.query(
f"MATCH (n:`{self.node_label}`) "
"CALL { WITH n DETACH DELETE n } "
"IN TRANSACTIONS OF 10000 ROWS;"
)
# Delete index
try:
self.query(f"DROP INDEX {self.index_name}")
except DatabaseError: # Index didn't exist yet
pass
[docs] def query(
self,
query: str,
*,
params: Optional[dict] = None,
retry_on_session_expired: bool = True,
) -> List[Dict[str, Any]]:
"""
This method sends a Cypher query to the connected Neo4j database
and returns the results as a list of dictionaries.
Args:
query (str): The Cypher query to execute.
params (dict, optional): Dictionary of query parameters. Defaults to {}.
Returns:
List[Dict[str, Any]]: List of dictionaries containing the query results.
"""
from neo4j.exceptions import CypherSyntaxError, SessionExpired
params = params or {}
with self._driver.session(database=self._database) as session:
try:
data = session.run(query, params)
return [r.data() for r in data]
except CypherSyntaxError as e:
raise ValueError(f"Cypher Statement is not valid\n{e}")
except (
SessionExpired
) as e: # Session expired is a transient error that can be retried
if retry_on_session_expired:
return self.query(
query, params=params, retry_on_session_expired=False
)
else:
raise e
[docs] def verify_version(self) -> None:
"""
Check if the connected Neo4j database version supports vector indexing.
Queries the Neo4j database to retrieve its version and compares it
against a target version (5.11.0) that is known to support vector
indexing. Raises a ValueError if the connected Neo4j version is
not supported.
"""
db_data = self.query("CALL dbms.components()")
version = db_data[0]["versions"][0]
if "aura" in version:
version_tuple = tuple(map(int, version.split("-")[0].split("."))) + (0,)
else:
version_tuple = tuple(map(int, version.split(".")))
target_version = (5, 11, 0)
if version_tuple < target_version:
raise ValueError(
"Version index is only supported in Neo4j version 5.11 or greater"
)
# Flag for metadata filtering
metadata_target_version = (5, 18, 0)
if version_tuple < metadata_target_version:
self.support_metadata_filter = False
else:
self.support_metadata_filter = True
# Flag for enterprise
self._is_enterprise = True if db_data[0]["edition"] == "enterprise" else False
[docs] def retrieve_existing_index(self) -> Tuple[Optional[int], Optional[str]]:
"""
Check if the vector index exists in the Neo4j database
and returns its embedding dimension.
This method queries the Neo4j database for existing indexes
and attempts to retrieve the dimension of the vector index
with the specified name. If the index exists, its dimension is returned.
If the index doesn't exist, `None` is returned.
Returns:
int or None: The embedding dimension of the existing index if found.
"""
index_information = self.query(
"SHOW INDEXES YIELD name, type, entityType, labelsOrTypes, "
"properties, options WHERE type = 'VECTOR' AND (name = $index_name "
"OR (labelsOrTypes[0] = $node_label AND "
"properties[0] = $embedding_node_property)) "
"RETURN name, entityType, labelsOrTypes, properties, options ",
params={
"index_name": self.index_name,
"node_label": self.node_label,
"embedding_node_property": self.embedding_node_property,
},
)
# sort by index_name
index_information = sort_by_index_name(index_information, self.index_name)
try:
self.index_name = index_information[0]["name"]
self.node_label = index_information[0]["labelsOrTypes"][0]
self.embedding_node_property = index_information[0]["properties"][0]
self._index_type = index_information[0]["entityType"]
embedding_dimension = None
index_config = index_information[0]["options"]["indexConfig"]
if "vector.dimensions" in index_config:
embedding_dimension = index_config["vector.dimensions"]
return embedding_dimension, index_information[0]["entityType"]
except IndexError:
return None, None
[docs] def retrieve_existing_fts_index(
self, text_node_properties: List[str] = []
) -> Optional[str]:
"""
Check if the fulltext index exists in the Neo4j database
This method queries the Neo4j database for existing fts indexes
with the specified name.
Returns:
(Tuple): keyword index information
"""
index_information = self.query(
"SHOW INDEXES YIELD name, type, labelsOrTypes, properties, options "
"WHERE type = 'FULLTEXT' AND (name = $keyword_index_name "
"OR (labelsOrTypes = [$node_label] AND "
"properties = $text_node_property)) "
"RETURN name, labelsOrTypes, properties, options ",
params={
"keyword_index_name": self.keyword_index_name,
"node_label": self.node_label,
"text_node_property": text_node_properties or [self.text_node_property],
},
)
# sort by index_name
index_information = sort_by_index_name(index_information, self.index_name)
try:
self.keyword_index_name = index_information[0]["name"]
self.text_node_property = index_information[0]["properties"][0]
node_label = index_information[0]["labelsOrTypes"][0]
return node_label
except IndexError:
return None
[docs] def create_new_index(self) -> None:
"""
This method constructs a Cypher query and executes it
to create a new vector index in Neo4j.
"""
index_query = (
"CALL db.index.vector.createNodeIndex("
"$index_name,"
"$node_label,"
"$embedding_node_property,"
"toInteger($embedding_dimension),"
"$similarity_metric )"
)
parameters = {
"index_name": self.index_name,
"node_label": self.node_label,
"embedding_node_property": self.embedding_node_property,
"embedding_dimension": self.embedding_dimension,
"similarity_metric": DISTANCE_MAPPING[self._distance_strategy],
}
self.query(index_query, params=parameters)
[docs] def create_new_keyword_index(self, text_node_properties: List[str] = []) -> None:
"""
This method constructs a Cypher query and executes it
to create a new full text index in Neo4j.
"""
node_props = text_node_properties or [self.text_node_property]
fts_index_query = (
f"CREATE FULLTEXT INDEX {self.keyword_index_name} "
f"FOR (n:`{self.node_label}`) ON EACH "
f"[{', '.join(['n.`' + el + '`' for el in node_props])}]"
)
self.query(fts_index_query)
@property
def embeddings(self) -> Embeddings:
return self.embedding
@classmethod
def __from(
cls,
texts: List[str],
embeddings: List[List[float]],
embedding: Embeddings,
metadatas: Optional[List[dict]] = None,
ids: Optional[List[str]] = None,
create_id_index: bool = True,
search_type: SearchType = SearchType.VECTOR,
**kwargs: Any,
) -> Neo4jVector:
if ids is None:
ids = [md5(text.encode("utf-8")).hexdigest() for text in texts]
if not metadatas:
metadatas = [{} for _ in texts]
store = cls(
embedding=embedding,
search_type=search_type,
**kwargs,
)
# Check if the vector index already exists
embedding_dimension, index_type = store.retrieve_existing_index()
# Raise error if relationship index type
if index_type == "RELATIONSHIP":
raise ValueError(
"Data ingestion is not supported with relationship vector index."
)
# If the vector index doesn't exist yet
if not index_type:
store.create_new_index()
# If the index already exists, check if embedding dimensions match
elif (
embedding_dimension and not store.embedding_dimension == embedding_dimension
):
raise ValueError(
f"Index with name {store.index_name} already exists."
"The provided embedding function and vector index "
"dimensions do not match.\n"
f"Embedding function dimension: {store.embedding_dimension}\n"
f"Vector index dimension: {embedding_dimension}"
)
if search_type == SearchType.HYBRID:
fts_node_label = store.retrieve_existing_fts_index()
# If the FTS index doesn't exist yet
if not fts_node_label:
store.create_new_keyword_index()
else: # Validate that FTS and Vector index use the same information
if not fts_node_label == store.node_label:
raise ValueError(
"Vector and keyword index don't index the same node label"
)
# Create unique constraint for faster import
if create_id_index:
store.query(
"CREATE CONSTRAINT IF NOT EXISTS "
f"FOR (n:`{store.node_label}`) REQUIRE n.id IS UNIQUE;"
)
store.add_embeddings(
texts=texts, embeddings=embeddings, metadatas=metadatas, ids=ids, **kwargs
)
return store
[docs] def add_embeddings(
self,
texts: Iterable[str],
embeddings: List[List[float]],
metadatas: Optional[List[dict]] = None,
ids: Optional[List[str]] = None,
**kwargs: Any,
) -> List[str]:
"""Add embeddings to the vectorstore.
Args:
texts: Iterable of strings to add to the vectorstore.
embeddings: List of list of embedding vectors.
metadatas: List of metadatas associated with the texts.
kwargs: vectorstore specific parameters
"""
if ids is None:
ids = [md5(text.encode("utf-8")).hexdigest() for text in texts]
if not metadatas:
metadatas = [{} for _ in texts]
import_query = (
"UNWIND $data AS row "
"CALL { WITH row "
f"MERGE (c:`{self.node_label}` {{id: row.id}}) "
"WITH c, row "
f"CALL db.create.setNodeVectorProperty(c, "
f"'{self.embedding_node_property}', row.embedding) "
f"SET c.`{self.text_node_property}` = row.text "
"SET c += row.metadata "
"} IN TRANSACTIONS OF 1000 ROWS "
)
parameters = {
"data": [
{"text": text, "metadata": metadata, "embedding": embedding, "id": id}
for text, metadata, embedding, id in zip(
texts, metadatas, embeddings, ids
)
]
}
self.query(import_query, params=parameters)
return ids
[docs] def add_texts(
self,
texts: Iterable[str],
metadatas: Optional[List[dict]] = None,
ids: Optional[List[str]] = None,
**kwargs: Any,
) -> List[str]:
"""Run more texts through the embeddings and add to the vectorstore.
Args:
texts: Iterable of strings to add to the vectorstore.
metadatas: Optional list of metadatas associated with the texts.
kwargs: vectorstore specific parameters
Returns:
List of ids from adding the texts into the vectorstore.
"""
embeddings = self.embedding.embed_documents(list(texts))
return self.add_embeddings(
texts=texts, embeddings=embeddings, metadatas=metadatas, ids=ids, **kwargs
)
[docs] def similarity_search(
self,
query: str,
k: int = 4,
params: Dict[str, Any] = {},
filter: Optional[Dict[str, Any]] = None,
**kwargs: Any,
) -> List[Document]:
"""Run similarity search with Neo4jVector.
Args:
query (str): Query text to search for.
k (int): Number of results to return. Defaults to 4.
params (Dict[str, Any]): The search params for the index type.
Defaults to empty dict.
filter (Optional[Dict[str, Any]]): Dictionary of argument(s) to
filter on metadata.
Defaults to None.
Returns:
List of Documents most similar to the query.
"""
embedding = self.embedding.embed_query(text=query)
return self.similarity_search_by_vector(
embedding=embedding,
k=k,
query=query,
params=params,
filter=filter,
**kwargs,
)
[docs] def similarity_search_with_score(
self,
query: str,
k: int = 4,
params: Dict[str, Any] = {},
filter: Optional[Dict[str, Any]] = None,
**kwargs: Any,
) -> List[Tuple[Document, float]]:
"""Return docs most similar to query.
Args:
query: Text to look up documents similar to.
k: Number of Documents to return. Defaults to 4.
params (Dict[str, Any]): The search params for the index type.
Defaults to empty dict.
filter (Optional[Dict[str, Any]]): Dictionary of argument(s) to
filter on metadata.
Defaults to None.
Returns:
List of Documents most similar to the query and score for each
"""
embedding = self.embedding.embed_query(query)
docs = self.similarity_search_with_score_by_vector(
embedding=embedding,
k=k,
query=query,
params=params,
filter=filter,
**kwargs,
)
return docs
[docs] def similarity_search_with_score_by_vector(
self,
embedding: List[float],
k: int = 4,
filter: Optional[Dict[str, Any]] = None,
params: Dict[str, Any] = {},
**kwargs: Any,
) -> List[Tuple[Document, float]]:
"""
Perform a similarity search in the Neo4j database using a
given vector and return the top k similar documents with their scores.
This method uses a Cypher query to find the top k documents that
are most similar to a given embedding. The similarity is measured
using a vector index in the Neo4j database. The results are returned
as a list of tuples, each containing a Document object and
its similarity score.
Args:
embedding (List[float]): The embedding vector to compare against.
k (int, optional): The number of top similar documents to retrieve.
filter (Optional[Dict[str, Any]]): Dictionary of argument(s) to
filter on metadata.
Defaults to None.
params (Dict[str, Any]): The search params for the index type.
Defaults to empty dict.
Returns:
List[Tuple[Document, float]]: A list of tuples, each containing
a Document object and its similarity score.
"""
if filter:
# Verify that 5.18 or later is used
if not self.support_metadata_filter:
raise ValueError(
"Metadata filtering is only supported in "
"Neo4j version 5.18 or greater"
)
# Metadata filtering and hybrid doesn't work
if self.search_type == SearchType.HYBRID:
raise ValueError(
"Metadata filtering can't be use in combination with "
"a hybrid search approach"
)
parallel_query = (
"CYPHER runtime = parallel parallelRuntimeSupport=all "
if self._is_enterprise
else ""
)
base_index_query = parallel_query + (
f"MATCH (n:`{self.node_label}`) WHERE "
f"n.`{self.embedding_node_property}` IS NOT NULL AND "
f"size(n.`{self.embedding_node_property}`) = "
f"toInteger({self.embedding_dimension}) AND "
)
base_cosine_query = (
" WITH n as node, vector.similarity.cosine("
f"n.`{self.embedding_node_property}`, "
"$embedding) AS score ORDER BY score DESC LIMIT toInteger($k) "
)
filter_snippets, filter_params = construct_metadata_filter(filter)
index_query = base_index_query + filter_snippets + base_cosine_query
else:
index_query = _get_search_index_query(self.search_type, self._index_type)
filter_params = {}
if self._index_type == IndexType.RELATIONSHIP:
if kwargs.get("return_embeddings"):
default_retrieval = (
f"RETURN relationship.`{self.text_node_property}` AS text, score, "
f"relationship {{.*, `{self.text_node_property}`: Null, "
f"`{self.embedding_node_property}`: Null, id: Null, "
f"_embedding_: relationship.`{self.embedding_node_property}`}} "
"AS metadata"
)
else:
default_retrieval = (
f"RETURN relationship.`{self.text_node_property}` AS text, score, "
f"relationship {{.*, `{self.text_node_property}`: Null, "
f"`{self.embedding_node_property}`: Null, id: Null }} AS metadata"
)
else:
if kwargs.get("return_embeddings"):
default_retrieval = (
f"RETURN node.`{self.text_node_property}` AS text, score, "
f"node {{.*, `{self.text_node_property}`: Null, "
f"`{self.embedding_node_property}`: Null, id: Null, "
f"_embedding_: node.`{self.embedding_node_property}`}} AS metadata"
)
else:
default_retrieval = (
f"RETURN node.`{self.text_node_property}` AS text, score, "
f"node {{.*, `{self.text_node_property}`: Null, "
f"`{self.embedding_node_property}`: Null, id: Null }} AS metadata"
)
retrieval_query = (
self.retrieval_query if self.retrieval_query else default_retrieval
)
read_query = index_query + retrieval_query
parameters = {
"index": self.index_name,
"k": k,
"embedding": embedding,
"keyword_index": self.keyword_index_name,
"query": remove_lucene_chars(kwargs["query"]),
**params,
**filter_params,
}
results = self.query(read_query, params=parameters)
if any(result["text"] is None for result in results):
if not self.retrieval_query:
raise ValueError(
f"Make sure that none of the `{self.text_node_property}` "
f"properties on nodes with label `{self.node_label}` "
"are missing or empty"
)
else:
raise ValueError(
"Inspect the `retrieval_query` and ensure it doesn't "
"return None for the `text` column"
)
if kwargs.get("return_embeddings") and any(
result["metadata"]["_embedding_"] is None for result in results
):
if not self.retrieval_query:
raise ValueError(
f"Make sure that none of the `{self.embedding_node_property}` "
f"properties on nodes with label `{self.node_label}` "
"are missing or empty"
)
else:
raise ValueError(
"Inspect the `retrieval_query` and ensure it doesn't "
"return None for the `_embedding_` metadata column"
)
docs = [
(
Document(
page_content=dict_to_yaml_str(result["text"])
if isinstance(result["text"], dict)
else result["text"],
metadata={
k: v for k, v in result["metadata"].items() if v is not None
},
),
result["score"],
)
for result in results
]
return docs
[docs] def similarity_search_by_vector(
self,
embedding: List[float],
k: int = 4,
filter: Optional[Dict[str, Any]] = None,
params: Dict[str, Any] = {},
**kwargs: Any,
) -> List[Document]:
"""Return docs most similar to embedding vector.
Args:
embedding: Embedding to look up documents similar to.
k: Number of Documents to return. Defaults to 4.
filter (Optional[Dict[str, Any]]): Dictionary of argument(s) to
filter on metadata.
Defaults to None.
params (Dict[str, Any]): The search params for the index type.
Defaults to empty dict.
Returns:
List of Documents most similar to the query vector.
"""
docs_and_scores = self.similarity_search_with_score_by_vector(
embedding=embedding, k=k, filter=filter, params=params, **kwargs
)
return [doc for doc, _ in docs_and_scores]
[docs] @classmethod
def from_texts(
cls: Type[Neo4jVector],
texts: List[str],
embedding: Embeddings,
metadatas: Optional[List[dict]] = None,
distance_strategy: DistanceStrategy = DEFAULT_DISTANCE_STRATEGY,
ids: Optional[List[str]] = None,
**kwargs: Any,
) -> Neo4jVector:
"""
Return Neo4jVector initialized from texts and embeddings.
Neo4j credentials are required in the form of `url`, `username`,
and `password` and optional `database` parameters.
"""
embeddings = embedding.embed_documents(list(texts))
return cls.__from(
texts,
embeddings,
embedding,
metadatas=metadatas,
ids=ids,
distance_strategy=distance_strategy,
**kwargs,
)
[docs] @classmethod
def from_embeddings(
cls,
text_embeddings: List[Tuple[str, List[float]]],
embedding: Embeddings,
metadatas: Optional[List[dict]] = None,
distance_strategy: DistanceStrategy = DEFAULT_DISTANCE_STRATEGY,
ids: Optional[List[str]] = None,
pre_delete_collection: bool = False,
**kwargs: Any,
) -> Neo4jVector:
"""Construct Neo4jVector wrapper from raw documents and pre-
generated embeddings.
Return Neo4jVector initialized from documents and embeddings.
Neo4j credentials are required in the form of `url`, `username`,
and `password` and optional `database` parameters.
Example:
.. code-block:: python
from langchain_community.vectorstores.neo4j_vector import Neo4jVector
from langchain_community.embeddings import OpenAIEmbeddings
embeddings = OpenAIEmbeddings()
text_embeddings = embeddings.embed_documents(texts)
text_embedding_pairs = list(zip(texts, text_embeddings))
vectorstore = Neo4jVector.from_embeddings(
text_embedding_pairs, embeddings)
"""
texts = [t[0] for t in text_embeddings]
embeddings = [t[1] for t in text_embeddings]
return cls.__from(
texts,
embeddings,
embedding,
metadatas=metadatas,
ids=ids,
distance_strategy=distance_strategy,
pre_delete_collection=pre_delete_collection,
**kwargs,
)
[docs] @classmethod
def from_existing_index(
cls: Type[Neo4jVector],
embedding: Embeddings,
index_name: str,
search_type: SearchType = DEFAULT_SEARCH_TYPE,
keyword_index_name: Optional[str] = None,
**kwargs: Any,
) -> Neo4jVector:
"""
Get instance of an existing Neo4j vector index. This method will
return the instance of the store without inserting any new
embeddings.
Neo4j credentials are required in the form of `url`, `username`,
and `password` and optional `database` parameters along with
the `index_name` definition.
"""
if search_type == SearchType.HYBRID and not keyword_index_name:
raise ValueError(
"keyword_index name has to be specified "
"when using hybrid search option"
)
store = cls(
embedding=embedding,
index_name=index_name,
keyword_index_name=keyword_index_name,
search_type=search_type,
**kwargs,
)
embedding_dimension, index_type = store.retrieve_existing_index()
# Raise error if relationship index type
if index_type == "RELATIONSHIP":
raise ValueError(
"Relationship vector index is not supported with "
"`from_existing_index` method. Please use the "
"`from_existing_relationship_index` method."
)
if not index_type:
raise ValueError(
"The specified vector index name does not exist. "
"Make sure to check if you spelled it correctly"
)
# Check if embedding function and vector index dimensions match
if embedding_dimension and not store.embedding_dimension == embedding_dimension:
raise ValueError(
"The provided embedding function and vector index "
"dimensions do not match.\n"
f"Embedding function dimension: {store.embedding_dimension}\n"
f"Vector index dimension: {embedding_dimension}"
)
if search_type == SearchType.HYBRID:
fts_node_label = store.retrieve_existing_fts_index()
# If the FTS index doesn't exist yet
if not fts_node_label:
raise ValueError(
"The specified keyword index name does not exist. "
"Make sure to check if you spelled it correctly"
)
else: # Validate that FTS and Vector index use the same information
if not fts_node_label == store.node_label:
raise ValueError(
"Vector and keyword index don't index the same node label"
)
return store
[docs] @classmethod
def from_existing_relationship_index(
cls: Type[Neo4jVector],
embedding: Embeddings,
index_name: str,
search_type: SearchType = DEFAULT_SEARCH_TYPE,
**kwargs: Any,
) -> Neo4jVector:
"""
Get instance of an existing Neo4j relationship vector index.
This method will return the instance of the store without
inserting any new embeddings.
Neo4j credentials are required in the form of `url`, `username`,
and `password` and optional `database` parameters along with
the `index_name` definition.
"""
if search_type == SearchType.HYBRID:
raise ValueError(
"Hybrid search is not supported in combination "
"with relationship vector index"
)
store = cls(
embedding=embedding,
index_name=index_name,
**kwargs,
)
embedding_dimension, index_type = store.retrieve_existing_index()
if not index_type:
raise ValueError(
"The specified vector index name does not exist. "
"Make sure to check if you spelled it correctly"
)
# Raise error if relationship index type
if index_type == "NODE":
raise ValueError(
"Node vector index is not supported with "
"`from_existing_relationship_index` method. Please use the "
"`from_existing_index` method."
)
# Check if embedding function and vector index dimensions match
if embedding_dimension and not store.embedding_dimension == embedding_dimension:
raise ValueError(
"The provided embedding function and vector index "
"dimensions do not match.\n"
f"Embedding function dimension: {store.embedding_dimension}\n"
f"Vector index dimension: {embedding_dimension}"
)
return store
[docs] @classmethod
def from_documents(
cls: Type[Neo4jVector],
documents: List[Document],
embedding: Embeddings,
distance_strategy: DistanceStrategy = DEFAULT_DISTANCE_STRATEGY,
ids: Optional[List[str]] = None,
**kwargs: Any,
) -> Neo4jVector:
"""
Return Neo4jVector initialized from documents and embeddings.
Neo4j credentials are required in the form of `url`, `username`,
and `password` and optional `database` parameters.
"""
texts = [d.page_content for d in documents]
metadatas = [d.metadata for d in documents]
return cls.from_texts(
texts=texts,
embedding=embedding,
distance_strategy=distance_strategy,
metadatas=metadatas,
ids=ids,
**kwargs,
)
[docs] @classmethod
def from_existing_graph(
cls: Type[Neo4jVector],
embedding: Embeddings,
node_label: str,
embedding_node_property: str,
text_node_properties: List[str],
*,
keyword_index_name: Optional[str] = "keyword",
index_name: str = "vector",
search_type: SearchType = DEFAULT_SEARCH_TYPE,
retrieval_query: str = "",
**kwargs: Any,
) -> Neo4jVector:
"""
Initialize and return a Neo4jVector instance from an existing graph.
This method initializes a Neo4jVector instance using the provided
parameters and the existing graph. It validates the existence of
the indices and creates new ones if they don't exist.
Returns:
Neo4jVector: An instance of Neo4jVector initialized with the provided parameters
and existing graph.
Example:
>>> neo4j_vector = Neo4jVector.from_existing_graph(
... embedding=my_embedding,
... node_label="Document",
... embedding_node_property="embedding",
... text_node_properties=["title", "content"]
... )
Note:
Neo4j credentials are required in the form of `url`, `username`, and `password`,
and optional `database` parameters passed as additional keyword arguments.
"""
# Validate the list is not empty
if not text_node_properties:
raise ValueError(
"Parameter `text_node_properties` must not be an empty list"
)
# Prefer retrieval query from params, otherwise construct it
if not retrieval_query:
retrieval_query = (
f"RETURN reduce(str='', k IN {text_node_properties} |"
" str + '\\n' + k + ': ' + coalesce(node[k], '')) AS text, "
"node {.*, `"
+ embedding_node_property
+ "`: Null, id: Null, "
+ ", ".join([f"`{prop}`: Null" for prop in text_node_properties])
+ "} AS metadata, score"
)
store = cls(
embedding=embedding,
index_name=index_name,
keyword_index_name=keyword_index_name,
search_type=search_type,
retrieval_query=retrieval_query,
node_label=node_label,
embedding_node_property=embedding_node_property,
**kwargs,
)
# Check if the vector index already exists
embedding_dimension, index_type = store.retrieve_existing_index()
# Raise error if relationship index type
if index_type == "RELATIONSHIP":
raise ValueError(
"`from_existing_graph` method does not support "
" existing relationship vector index. "
"Please use `from_existing_relationship_index` method"
)
# If the vector index doesn't exist yet
if not index_type:
store.create_new_index()
# If the index already exists, check if embedding dimensions match
elif (
embedding_dimension and not store.embedding_dimension == embedding_dimension
):
raise ValueError(
f"Index with name {store.index_name} already exists."
"The provided embedding function and vector index "
"dimensions do not match.\n"
f"Embedding function dimension: {store.embedding_dimension}\n"
f"Vector index dimension: {embedding_dimension}"
)
# FTS index for Hybrid search
if search_type == SearchType.HYBRID:
fts_node_label = store.retrieve_existing_fts_index(text_node_properties)
# If the FTS index doesn't exist yet
if not fts_node_label:
store.create_new_keyword_index(text_node_properties)
else: # Validate that FTS and Vector index use the same information
if not fts_node_label == store.node_label:
raise ValueError(
"Vector and keyword index don't index the same node label"
)
# Populate embeddings
while True:
fetch_query = (
f"MATCH (n:`{node_label}`) "
f"WHERE n.{embedding_node_property} IS null "
"AND any(k in $props WHERE n[k] IS NOT null) "
f"RETURN elementId(n) AS id, reduce(str='',"
"k IN $props | str + '\\n' + k + ':' + coalesce(n[k], '')) AS text "
"LIMIT 1000"
)
data = store.query(fetch_query, params={"props": text_node_properties})
if not data:
break
text_embeddings = embedding.embed_documents([el["text"] for el in data])
params = {
"data": [
{"id": el["id"], "embedding": embedding}
for el, embedding in zip(data, text_embeddings)
]
}
store.query(
"UNWIND $data AS row "
f"MATCH (n:`{node_label}`) "
"WHERE elementId(n) = row.id "
f"CALL db.create.setNodeVectorProperty(n, "
f"'{embedding_node_property}', row.embedding) "
"RETURN count(*)",
params=params,
)
# If embedding calculation should be stopped
if len(data) < 1000:
break
return store
[docs] def max_marginal_relevance_search(
self,
query: str,
k: int = 4,
fetch_k: int = 20,
lambda_mult: float = 0.5,
filter: Optional[dict] = None,
**kwargs: Any,
) -> List[Document]:
"""Return docs selected using the maximal marginal relevance.
Maximal marginal relevance optimizes for similarity to query AND diversity
among selected documents.
Args:
query: search query text.
k: Number of Documents to return. Defaults to 4.
fetch_k: Number of Documents to fetch to pass to MMR algorithm.
lambda_mult: Number between 0 and 1 that determines the degree
of diversity among the results with 0 corresponding
to maximum diversity and 1 to minimum diversity.
Defaults to 0.5.
filter: Filter on metadata properties, e.g.
{
"str_property": "foo",
"int_property": 123
}
Returns:
List of Documents selected by maximal marginal relevance.
"""
# Embed the query
query_embedding = self.embedding.embed_query(query)
# Fetch the initial documents
got_docs = self.similarity_search_with_score_by_vector(
embedding=query_embedding,
query=query,
k=fetch_k,
return_embeddings=True,
filter=filter,
**kwargs,
)
# Get the embeddings for the fetched documents
got_embeddings = [doc.metadata["_embedding_"] for doc, _ in got_docs]
# Select documents using maximal marginal relevance
selected_indices = maximal_marginal_relevance(
np.array(query_embedding), got_embeddings, lambda_mult=lambda_mult, k=k
)
selected_docs = [got_docs[i][0] for i in selected_indices]
# Remove embedding values from metadata
for doc in selected_docs:
del doc.metadata["_embedding_"]
return selected_docs
def _select_relevance_score_fn(self) -> Callable[[float], float]:
"""
The 'correct' relevance function
may differ depending on a few things, including:
- the distance / similarity metric used by the VectorStore
- the scale of your embeddings (OpenAI's are unit normed. Many others are not!)
- embedding dimensionality
- etc.
"""
if self.override_relevance_score_fn is not None:
return self.override_relevance_score_fn
# Default strategy is to rely on distance strategy provided
# in vectorstore constructor
if self._distance_strategy == DistanceStrategy.COSINE:
return lambda x: x
elif self._distance_strategy == DistanceStrategy.EUCLIDEAN_DISTANCE:
return lambda x: x
else:
raise ValueError(
"No supported normalization function"
f" for distance_strategy of {self._distance_strategy}."
"Consider providing relevance_score_fn to PGVector constructor."
)
