langchain_postgres.vectorstores.PGVector¶

class langchain_postgres.vectorstores.PGVector(embeddings: Embeddings, *, connection: Union[None, Engine, str, AsyncEngine] = None, embedding_length: Optional[int] = None, collection_name: str = 'langchain', collection_metadata: Optional[dict] = None, distance_strategy: DistanceStrategy = DistanceStrategy.COSINE, pre_delete_collection: bool = False, logger: Optional[Logger] = None, relevance_score_fn: Optional[Callable[[float], float]] = None, engine_args: Optional[dict[str, Any]] = None, use_jsonb: bool = True, create_extension: bool = True, async_mode: bool = False)[source]¶

Vectorstore implementation using Postgres as the backend.

Currently, there is no mechanism for supporting data migration.

So breaking changes in the vectorstore schema will require the user to recreate the tables and re-add the documents.

If this is a concern, please use a different vectorstore. If not, this implementation should be fine for your use case.

To use this vectorstore you need to have the vector extension installed. The vector extension is a Postgres extension that provides vector similarity search capabilities.

```sh docker run –name pgvector-container -e POSTGRES_PASSWORD=…

-d pgvector/pgvector:pg16

```

Example

from langchain_postgres.vectorstores import PGVector
from langchain_openai.embeddings import OpenAIEmbeddings

connection_string = "postgresql+psycopg://..."
collection_name = "state_of_the_union_test"
embeddings = OpenAIEmbeddings()
vectorstore = PGVector.from_documents(
    embedding=embeddings,
    documents=docs,
    connection=connection_string,
    collection_name=collection_name,
    use_jsonb=True,
    async_mode=False,
)

This code has been ported over from langchain_community with minimal changes to allow users to easily transition from langchain_community to langchain_postgres.

Some changes had to be made to address issues with the community implementation: * langchain_postgres now works with psycopg3. Please update your

connection strings from postgresql+psycopg2://… to postgresql+psycopg://langchain:langchain@… (yes, the driver name is psycopg not psycopg3)

  • The schema of the embedding store and collection have been changed to make add_documents work correctly with user specified ids, specifically when overwriting existing documents. You will need to recreate the tables if you are using an existing database.

  • A Connection object has to be provided explicitly. Connections will not be picked up automatically based on env variables.

  • langchain_postgres now accept async connections. If you want to use the async

    version, you need to set async_mode=True when initializing the store or use an async engine.

Supported filter operators:

  • $eq: Equality operator

  • $ne: Not equal operator

  • $lt: Less than operator

  • $lte: Less than or equal operator

  • $gt: Greater than operator

  • $gte: Greater than or equal operator

  • $in: In operator

  • $nin: Not in operator

  • $between: Between operator

  • $exists: Exists operator

  • $like: Like operator

  • $ilike: Case insensitive like operator

  • $and: Logical AND operator

  • $or: Logical OR operator

  • $not: Logical NOT operator

Example:

vectorstore.similarity_search('kitty', k=10, filter={
    'id': {'$in': [1, 5, 2, 9]}
})
#%% md

If you provide a dict with multiple fields, but no operators,
the top level will be interpreted as a logical **AND** filter

vectorstore.similarity_search('ducks', k=10, filter={
    'id': {'$in': [1, 5, 2, 9]},
    'location': {'$in': ["pond", "market"]}
})

Initialize the PGVector store. For an async version, use PGVector.acreate() instead.

Parameters

  • connection (Union[None, DBConnection, Engine, AsyncEngine, str]) – Postgres connection string or (async)engine.

  • embeddings (Embeddings) – Any embedding function implementing langchain.embeddings.base.Embeddings interface.

  • embedding_length (Optional[int]) – The length of the embedding vector. (default: None) NOTE: This is not mandatory. Defining it will prevent vectors of any other size to be added to the embeddings table but, without it, the embeddings can’t be indexed.

  • collection_name (str) – The name of the collection to use. (default: langchain) NOTE: This is not the name of the table, but the name of the collection. The tables will be created when initializing the store (if not exists) So, make sure the user has the right permissions to create tables.

  • distance_strategy (DistanceStrategy) – The distance strategy to use. (default: COSINE)

  • pre_delete_collection (bool) – If True, will delete the collection if it exists. (default: False). Useful for testing.

  • engine_args (Optional[dict[str, Any]]) – SQLAlchemy’s create engine arguments.

  • use_jsonb (bool) – Use JSONB instead of JSON for metadata. (default: True) Strongly discouraged from using JSON as it’s not as efficient for querying. It’s provided here for backwards compatibility with older versions, and will be removed in the future.

  • create_extension (bool) – If True, will create the vector extension if it doesn’t exist. disabling creation is useful when using ReadOnly Databases.

  • collection_metadata (Optional[dict]) –

  • logger (Optional[logging.Logger]) –

  • relevance_score_fn (Optional[Callable[[float], float]]) –

  • async_mode (bool) –

Attributes

distance_strategy

embeddings

Access the query embedding object if available.

Methods

__init__(embeddings, *[, connection, ...])

Initialize the PGVector store.

aadd_documents(documents, **kwargs)

Async run more documents through the embeddings and add to the vectorstore.

aadd_embeddings(texts, embeddings[, ...])

Async add embeddings to the vectorstore.

aadd_texts(texts[, metadatas])

Async run more texts through the embeddings and add to the vectorstore.

acreate_collection()

acreate_tables_if_not_exists()

acreate_vector_extension()

add_documents(documents, **kwargs)

Add or update documents in the vectorstore.

add_embeddings(texts, embeddings[, ...])

Add embeddings to the vectorstore.

add_texts(texts[, metadatas])

Run more texts through the embeddings and add to the vectorstore.

adelete([ids, collection_only])

Async delete vectors by ids or uuids.

adelete_collection()

adrop_tables()

afrom_documents(documents, embedding[, ...])

Return VectorStore initialized from documents and embeddings.

afrom_embeddings(text_embeddings, embedding)

Construct PGVector wrapper from raw documents and pre- generated embeddings.

afrom_existing_index(embedding, *[, ...])

Get instance of an existing PGVector store.This method will return the instance of the store without inserting any new embeddings

afrom_texts(texts, embedding[, metadatas, ...])

Return VectorStore initialized from documents and embeddings.

aget_by_ids(ids, /)

Get documents by ids.

aget_collection(session)

amax_marginal_relevance_search(query[, k, ...])

Return docs selected using the maximal marginal relevance.

amax_marginal_relevance_search_by_vector(...)

Return docs selected using the maximal marginal relevance

amax_marginal_relevance_search_with_score(query)

Return docs selected using the maximal marginal relevance with score.

amax_marginal_relevance_search_with_score_by_vector(...)

Return docs selected using the maximal marginal relevance with score

as_retriever(**kwargs)

Return VectorStoreRetriever initialized from this VectorStore.

asearch(query, search_type, **kwargs)

Async return docs most similar to query using a specified search type.

asimilarity_search(query[, k, filter])

Run similarity search with PGVector with distance.

asimilarity_search_by_vector(embedding[, k, ...])

Return docs most similar to embedding vector.

asimilarity_search_with_relevance_scores(query)

Async return docs and relevance scores in the range [0, 1].

asimilarity_search_with_score(query[, k, filter])

Return docs most similar to query.

asimilarity_search_with_score_by_vector(...)

astreaming_upsert(items, /, batch_size, **kwargs)

aupsert(items, /, **kwargs)

Upsert documents into the vectorstore.

connection_string_from_db_params(driver, ...)

Return connection string from database parameters.

create_collection()

create_tables_if_not_exists()

create_vector_extension()

delete([ids, collection_only])

Delete vectors by ids or uuids.

delete_collection()

drop_tables()

from_documents(documents, embedding, *[, ...])

Return VectorStore initialized from documents and embeddings.

from_embeddings(text_embeddings, embedding, *)

Construct PGVector wrapper from raw documents and embeddings.

from_existing_index(embedding, *[, ...])

Get instance of an existing PGVector store.This method will return the instance of the store without inserting any new embeddings

from_texts(texts, embedding[, metadatas, ...])

Return VectorStore initialized from documents and embeddings.

get_by_ids(ids, /)

Get documents by ids.

get_collection(session)

get_connection_string(kwargs)

max_marginal_relevance_search(query[, k, ...])

Return docs selected using the maximal marginal relevance.

max_marginal_relevance_search_by_vector(...)

Return docs selected using the maximal marginal relevance

max_marginal_relevance_search_with_score(query)

Return docs selected using the maximal marginal relevance with score.

max_marginal_relevance_search_with_score_by_vector(...)

Return docs selected using the maximal marginal relevance with score

search(query, search_type, **kwargs)

Return docs most similar to query using a specified search type.

similarity_search(query[, k, filter])

Run similarity search with PGVector with distance.

similarity_search_by_vector(embedding[, k, ...])

Return docs most similar to embedding vector.

similarity_search_with_relevance_scores(query)

Return docs and relevance scores in the range [0, 1].

similarity_search_with_score(query[, k, filter])

Return docs most similar to query.

similarity_search_with_score_by_vector(embedding)

streaming_upsert(items, /, batch_size, **kwargs)

upsert(items, /, **kwargs)

Upsert documents into the vectorstore.
__init__(embeddings: Embeddings, *, connection: Union[None, Engine, str, AsyncEngine] = None, embedding_length: Optional[int] = None, collection_name: str = 'langchain', collection_metadata: Optional[dict] = None, distance_strategy: DistanceStrategy = DistanceStrategy.COSINE, pre_delete_collection: bool = False, logger: Optional[Logger] = None, relevance_score_fn: Optional[Callable[[float], float]] = None, engine_args: Optional[dict[str, Any]] = None, use_jsonb: bool = True, create_extension: bool = True, async_mode: bool = False) → None[source]¶

Initialize the PGVector store. For an async version, use PGVector.acreate() instead.

Parameters

  • connection (Union[None, Engine, str, AsyncEngine]) – Postgres connection string or (async)engine.

  • embeddings (Embeddings) – Any embedding function implementing langchain.embeddings.base.Embeddings interface.

  • embedding_length (Optional[int]) – The length of the embedding vector. (default: None) NOTE: This is not mandatory. Defining it will prevent vectors of any other size to be added to the embeddings table but, without it, the embeddings can’t be indexed.

  • collection_name (str) – The name of the collection to use. (default: langchain) NOTE: This is not the name of the table, but the name of the collection. The tables will be created when initializing the store (if not exists) So, make sure the user has the right permissions to create tables.

  • distance_strategy (DistanceStrategy) – The distance strategy to use. (default: COSINE)

  • pre_delete_collection (bool) – If True, will delete the collection if it exists. (default: False). Useful for testing.

  • engine_args (Optional[dict[str, Any]]) – SQLAlchemy’s create engine arguments.

  • use_jsonb (bool) – Use JSONB instead of JSON for metadata. (default: True) Strongly discouraged from using JSON as it’s not as efficient for querying. It’s provided here for backwards compatibility with older versions, and will be removed in the future.

  • create_extension (bool) – If True, will create the vector extension if it doesn’t exist. disabling creation is useful when using ReadOnly Databases.

  • collection_metadata (Optional[dict]) –

  • logger (Optional[Logger]) –

  • relevance_score_fn (Optional[Callable[[float], float]]) –

  • async_mode (bool) –

Return type

None

async aadd_documents(documents: List[Document], **kwargs: Any) → List[str]¶

Async run more documents through the embeddings and add to the vectorstore.

Parameters

  • documents (List[Document]) – Documents to add to the vectorstore.

  • kwargs (Any) – Additional keyword arguments.

Returns

List of IDs of the added texts.

Raises

ValueError – If the number of IDs does not match the number of documents.

Return type

List[str]

async aadd_embeddings(texts: Sequence[str], embeddings: List[List[float]], metadatas: Optional[List[dict]] = None, ids: Optional[List[str]] = None, **kwargs: Any) → List[str][source]¶

Async add embeddings to the vectorstore.

Parameters

  • texts (Sequence[str]) – Iterable of strings to add to the vectorstore.

  • embeddings (List[List[float]]) – List of list of embedding vectors.

  • metadatas (Optional[List[dict]]) – List of metadatas associated with the texts.

  • ids (Optional[List[str]]) – Optional list of ids for the texts. If not provided, will generate a new id for each text.

  • kwargs (Any) – vectorstore specific parameters

Return type

List[str]

async aadd_texts(texts: Iterable[str], metadatas: Optional[List[dict]] = None, **kwargs: Any) → List[str]¶

Async run more texts through the embeddings and add to the vectorstore.

Parameters

  • texts (Iterable[str]) – Iterable of strings to add to the vectorstore.

  • metadatas (Optional[List[dict]]) – Optional list of metadatas associated with the texts. Default is None.

  • **kwargs (Any) – vectorstore specific parameters.

Returns

List of ids from adding the texts into the vectorstore.

Raises

  • ValueError – If the number of metadatas does not match the number of texts.

  • ValueError – If the number of ids does not match the number of texts.

Return type

List[str]

async acreate_collection() → None[source]¶
Return type

None

async acreate_tables_if_not_exists() → None[source]¶
Return type

None

async acreate_vector_extension() → None[source]¶
Return type

None

add_documents(documents: List[Document], **kwargs: Any) → List[str]¶

Add or update documents in the vectorstore.

Parameters

  • documents (List[Document]) – Documents to add to the vectorstore.

  • kwargs (Any) – Additional keyword arguments. if kwargs contains ids and documents contain ids, the ids in the kwargs will receive precedence.

Returns

List of IDs of the added texts.

Raises

ValueError – If the number of ids does not match the number of documents.

Return type

List[str]

add_embeddings(texts: Sequence[str], embeddings: List[List[float]], metadatas: Optional[List[dict]] = None, ids: Optional[List[str]] = None, **kwargs: Any) → List[str][source]¶

Add embeddings to the vectorstore.

Parameters

  • texts (Sequence[str]) – Iterable of strings to add to the vectorstore.

  • embeddings (List[List[float]]) – List of list of embedding vectors.

  • metadatas (Optional[List[dict]]) – List of metadatas associated with the texts.

  • ids (Optional[List[str]]) – Optional list of ids for the documents. If not provided, will generate a new id for each document.

  • kwargs (Any) – vectorstore specific parameters

Return type

List[str]

add_texts(texts: Iterable[str], metadatas: Optional[List[dict]] = None, **kwargs: Any) → List[str]¶

Run more texts through the embeddings and add to the vectorstore.

Parameters

  • texts (Iterable[str]) – Iterable of strings to add to the vectorstore.

  • metadatas (Optional[List[dict]]) – Optional list of metadatas associated with the texts.

  • **kwargs (Any) – vectorstore specific parameters. One of the kwargs should be ids which is a list of ids associated with the texts.

Returns

List of ids from adding the texts into the vectorstore.

Raises

  • ValueError – If the number of metadatas does not match the number of texts.

  • ValueError – If the number of ids does not match the number of texts.

Return type

List[str]

async adelete(ids: Optional[List[str]] = None, collection_only: bool = False, **kwargs: Any) → None[source]¶

Async delete vectors by ids or uuids.

Parameters

  • ids (Optional[List[str]]) – List of ids to delete.

  • collection_only (bool) – Only delete ids in the collection.

  • kwargs (Any) –

Return type

None

async adelete_collection() → None[source]¶
Return type

None

async adrop_tables() → None[source]¶
Return type

None

async classmethod afrom_documents(documents: List[Document], embedding: Embeddings, collection_name: str = 'langchain', distance_strategy: DistanceStrategy = DistanceStrategy.COSINE, ids: Optional[List[str]] = None, pre_delete_collection: bool = False, *, use_jsonb: bool = True, **kwargs: Any) → PGVector[source]¶

Return VectorStore initialized from documents and embeddings. Postgres connection string is required “Either pass it as a parameter or set the PGVECTOR_CONNECTION_STRING environment variable.

Parameters

  • documents (List[Document]) –

  • embedding (Embeddings) –

  • collection_name (str) –

  • distance_strategy (DistanceStrategy) –

  • ids (Optional[List[str]]) –

  • pre_delete_collection (bool) –

  • use_jsonb (bool) –

  • kwargs (Any) –

Return type

PGVector

async classmethod afrom_embeddings(text_embeddings: List[Tuple[str, List[float]]], embedding: Embeddings, metadatas: Optional[List[dict]] = None, collection_name: str = 'langchain', distance_strategy: DistanceStrategy = DistanceStrategy.COSINE, ids: Optional[List[str]] = None, pre_delete_collection: bool = False, **kwargs: Any) → PGVector[source]¶

Construct PGVector wrapper from raw documents and pre- generated embeddings.

Return VectorStore initialized from documents and embeddings. Postgres connection string is required “Either pass it as a parameter or set the PGVECTOR_CONNECTION_STRING environment variable.

Example

from langchain_community.vectorstores import PGVector
from langchain_community.embeddings import OpenAIEmbeddings
embeddings = OpenAIEmbeddings()
text_embeddings = embeddings.embed_documents(texts)
text_embedding_pairs = list(zip(texts, text_embeddings))
faiss = PGVector.from_embeddings(text_embedding_pairs, embeddings)
Parameters

  • text_embeddings (List[Tuple[str, List[float]]]) –

  • embedding (Embeddings) –

  • metadatas (Optional[List[dict]]) –

  • collection_name (str) –

  • distance_strategy (DistanceStrategy) –

  • ids (Optional[List[str]]) –

  • pre_delete_collection (bool) –

  • kwargs (Any) –

Return type

PGVector

async classmethod afrom_existing_index(embedding: Embeddings, *, collection_name: str = 'langchain', distance_strategy: DistanceStrategy = DistanceStrategy.COSINE, pre_delete_collection: bool = False, connection: Optional[Union[Engine, str]] = None, **kwargs: Any) → PGVector[source]¶

Get instance of an existing PGVector store.This method will return the instance of the store without inserting any new embeddings

Parameters

  • embedding (Embeddings) –

  • collection_name (str) –

  • distance_strategy (DistanceStrategy) –

  • pre_delete_collection (bool) –

  • connection (Optional[Union[Engine, str]]) –

  • kwargs (Any) –

Return type

PGVector

async classmethod afrom_texts(texts: List[str], embedding: Embeddings, metadatas: Optional[List[dict]] = None, collection_name: str = 'langchain', distance_strategy: DistanceStrategy = DistanceStrategy.COSINE, ids: Optional[List[str]] = None, pre_delete_collection: bool = False, *, use_jsonb: bool = True, **kwargs: Any) → PGVector[source]¶

Return VectorStore initialized from documents and embeddings.

Parameters

  • texts (List[str]) –

  • embedding (Embeddings) –

  • metadatas (Optional[List[dict]]) –

  • collection_name (str) –

  • distance_strategy (DistanceStrategy) –

  • ids (Optional[List[str]]) –

  • pre_delete_collection (bool) –

  • use_jsonb (bool) –

  • kwargs (Any) –

Return type

PGVector

async aget_by_ids(ids: Sequence[str], /) → List[Document][source]¶

Get documents by ids.

Parameters

ids (Sequence[str]) –

Return type

List[Document]

async aget_collection(session: AsyncSession) → Any[source]¶
Parameters

session (AsyncSession) –

Return type

Any

Return docs selected using the maximal marginal relevance.

Maximal marginal relevance optimizes for similarity to query AND diversity

among selected documents.

Parameters

  • query (str) – Text to look up documents similar to.

  • k (int) – Number of Documents to return. Defaults to 4.

  • fetch_k (int) – Number of Documents to fetch to pass to MMR algorithm. Defaults to 20.

  • lambda_mult (float) – 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 (Optional[Dict[str, str]]) – Filter by metadata. Defaults to None.

  • kwargs (Any) –

Returns

List of Documents selected by maximal marginal relevance.

Return type

List[Document]

async amax_marginal_relevance_search_by_vector(embedding: List[float], k: int = 4, fetch_k: int = 20, lambda_mult: float = 0.5, filter: Optional[Dict[str, str]] = None, **kwargs: Any) → List[Document][source]¶
Return docs selected using the maximal marginal relevance

to embedding vector.

Maximal marginal relevance optimizes for similarity to query AND diversity

among selected documents.

Parameters

  • embedding (str) – Text to look up documents similar to.

  • k (int) – Number of Documents to return. Defaults to 4.

  • fetch_k (int) – Number of Documents to fetch to pass to MMR algorithm. Defaults to 20.

  • lambda_mult (float) – 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 (Optional[Dict[str, str]]) – Filter by metadata. Defaults to None.

  • kwargs (Any) –

Returns

List of Documents selected by maximal marginal relevance.

Return type

List[Document]

async amax_marginal_relevance_search_with_score(query: str, k: int = 4, fetch_k: int = 20, lambda_mult: float = 0.5, filter: Optional[dict] = None, **kwargs: Any) → List[Tuple[Document, float]][source]¶

Return docs selected using the maximal marginal relevance with score.

Maximal marginal relevance optimizes for similarity to query AND diversity

among selected documents.

Parameters

  • query (str) – Text to look up documents similar to.

  • k (int) – Number of Documents to return. Defaults to 4.

  • fetch_k (int) – Number of Documents to fetch to pass to MMR algorithm. Defaults to 20.

  • lambda_mult (float) – 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 (Optional[Dict[str, str]]) – Filter by metadata. Defaults to None.

  • kwargs (Any) –

Returns

List of Documents selected by maximal marginal

relevance to the query and score for each.

Return type

List[Tuple[Document, float]]

async amax_marginal_relevance_search_with_score_by_vector(embedding: List[float], k: int = 4, fetch_k: int = 20, lambda_mult: float = 0.5, filter: Optional[Dict[str, str]] = None, **kwargs: Any) → List[Tuple[Document, float]][source]¶
Return docs selected using the maximal marginal relevance with score

to embedding vector.

Maximal marginal relevance optimizes for similarity to query AND diversity

among selected documents.

Parameters

  • embedding (List[float]) – Embedding to look up documents similar to.

  • k (int) – Number of Documents to return. Defaults to 4.

  • fetch_k (int) – Number of Documents to fetch to pass to MMR algorithm. Defaults to 20.

  • lambda_mult (float) – 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 (Optional[Dict[str, str]]) – Filter by metadata. Defaults to None.

  • kwargs (Any) –

Returns

List of Documents selected by maximal marginal

relevance to the query and score for each.

Return type

List[Tuple[Document, float]]

as_retriever(**kwargs: Any) → VectorStoreRetriever¶

Return VectorStoreRetriever initialized from this VectorStore.

Parameters

**kwargs (Any) –

Keyword arguments to pass to the search function. Can include: search_type (Optional[str]): Defines the type of search that

the Retriever should perform. Can be “similarity” (default), “mmr”, or “similarity_score_threshold”.

search_kwargs (Optional[Dict]): Keyword arguments to pass to the
search function. Can include things like:

k: Amount of documents to return (Default: 4) score_threshold: Minimum relevance threshold

for similarity_score_threshold

fetch_k: Amount of documents to pass to MMR algorithm

(Default: 20)

lambda_mult: Diversity of results returned by MMR;

1 for minimum diversity and 0 for maximum. (Default: 0.5)

filter: Filter by document metadata

Returns

Retriever class for VectorStore.

Return type

VectorStoreRetriever

Examples:

# Retrieve more documents with higher diversity
# Useful if your dataset has many similar documents
docsearch.as_retriever(
    search_type="mmr",
    search_kwargs={'k': 6, 'lambda_mult': 0.25}
)

# Fetch more documents for the MMR algorithm to consider
# But only return the top 5
docsearch.as_retriever(
    search_type="mmr",
    search_kwargs={'k': 5, 'fetch_k': 50}
)

# Only retrieve documents that have a relevance score
# Above a certain threshold
docsearch.as_retriever(
    search_type="similarity_score_threshold",
    search_kwargs={'score_threshold': 0.8}
)

# Only get the single most similar document from the dataset
docsearch.as_retriever(search_kwargs={'k': 1})

# Use a filter to only retrieve documents from a specific paper
docsearch.as_retriever(
    search_kwargs={'filter': {'paper_title':'GPT-4 Technical Report'}}
)
async asearch(query: str, search_type: str, **kwargs: Any) → List[Document]¶

Async return docs most similar to query using a specified search type.

Parameters

  • query (str) – Input text.

  • search_type (str) – Type of search to perform. Can be “similarity”, “mmr”, or “similarity_score_threshold”.

  • **kwargs (Any) – Arguments to pass to the search method.

Returns

List of Documents most similar to the query.

Raises

ValueError – If search_type is not one of “similarity”, “mmr”, or “similarity_score_threshold”.

Return type

List[Document]

Run similarity search with PGVector with distance.

Parameters

  • query (str) – Query text to search for.

  • k (int) – Number of results to return. Defaults to 4.

  • filter (Optional[Dict[str, str]]) – Filter by metadata. Defaults to None.

  • kwargs (Any) –

Returns

List of Documents most similar to the query.

Return type

List[Document]

async asimilarity_search_by_vector(embedding: List[float], k: int = 4, filter: Optional[dict] = None, **kwargs: Any) → List[Document][source]¶

Return docs most similar to embedding vector.

Parameters

  • embedding (List[float]) – Embedding to look up documents similar to.

  • k (int) – Number of Documents to return. Defaults to 4.

  • filter (Optional[Dict[str, str]]) – Filter by metadata. Defaults to None.

  • kwargs (Any) –

Returns

List of Documents most similar to the query vector.

Return type

List[Document]

async asimilarity_search_with_relevance_scores(query: str, k: int = 4, **kwargs: Any) → List[Tuple[Document, float]]¶

Async return docs and relevance scores in the range [0, 1].

0 is dissimilar, 1 is most similar.

Parameters

  • query (str) – Input text.

  • k (int) – Number of Documents to return. Defaults to 4.

  • **kwargs (Any) –

    kwargs to be passed to similarity search. Should include: score_threshold: Optional, a floating point value between 0 to 1 to

    filter the resulting set of retrieved docs

Returns

List of Tuples of (doc, similarity_score)

Return type

List[Tuple[Document, float]]

async asimilarity_search_with_score(query: str, k: int = 4, filter: Optional[dict] = None) → List[Tuple[Document, float]][source]¶

Return docs most similar to query.

Parameters

  • query (str) – Text to look up documents similar to.

  • k (int) – Number of Documents to return. Defaults to 4.

  • filter (Optional[Dict[str, str]]) – Filter by metadata. Defaults to None.

Returns

List of Documents most similar to the query and score for each.

Return type

List[Tuple[Document, float]]

async asimilarity_search_with_score_by_vector(embedding: List[float], k: int = 4, filter: Optional[dict] = None) → List[Tuple[Document, float]][source]¶
Parameters

  • embedding (List[float]) –

  • k (int) –

  • filter (Optional[dict]) –

Return type

List[Tuple[Document, float]]

astreaming_upsert(items: AsyncIterable[Document], /, batch_size: int, **kwargs: Any) → AsyncIterator[UpsertResponse]¶

Beta

Added in 0.2.11. The API is subject to change.

Upsert documents in a streaming fashion. Async version of streaming_upsert.

Parameters

  • items (AsyncIterable[Document]) – Iterable of Documents to add to the vectorstore.

  • batch_size (int) – The size of each batch to upsert.

  • kwargs (Any) – Additional keyword arguments. kwargs should only include parameters that are common to all documents. (e.g., timeout for indexing, retry policy, etc.) kwargs should not include ids to avoid ambiguous semantics. Instead the ID should be provided as part of the Document object.

Yields

UpsertResponse – A response object that contains the list of IDs that were successfully added or updated in the vectorstore and the list of IDs that failed to be added or updated.

Return type

AsyncIterator[UpsertResponse]

New in version 0.2.11.

async aupsert(items: Sequence[Document], /, **kwargs: Any) → UpsertResponse[source]¶

Upsert documents into the vectorstore.

Parameters

  • items (Sequence[Document]) – Sequence of documents to upsert.

  • kwargs (Any) – vectorstore specific parameters

Returns

UpsertResponse

Return type

UpsertResponse

classmethod connection_string_from_db_params(driver: str, host: str, port: int, database: str, user: str, password: str) → str[source]¶

Return connection string from database parameters.

Parameters

  • driver (str) –

  • host (str) –

  • port (int) –

  • database (str) –

  • user (str) –

  • password (str) –

Return type

str

create_collection() → None[source]¶
Return type

None

create_tables_if_not_exists() → None[source]¶
Return type

None

create_vector_extension() → None[source]¶
Return type

None

delete(ids: Optional[List[str]] = None, collection_only: bool = False, **kwargs: Any) → None[source]¶

Delete vectors by ids or uuids.

Parameters

  • ids (Optional[List[str]]) – List of ids to delete.

  • collection_only (bool) – Only delete ids in the collection.

  • kwargs (Any) –

Return type

None

delete_collection() → None[source]¶
Return type

None

drop_tables() → None[source]¶
Return type

None

classmethod from_documents(documents: List[Document], embedding: Embeddings, *, connection: Optional[Union[Engine, str]] = None, collection_name: str = 'langchain', distance_strategy: DistanceStrategy = DistanceStrategy.COSINE, ids: Optional[List[str]] = None, pre_delete_collection: bool = False, use_jsonb: bool = True, **kwargs: Any) → PGVector[source]¶

Return VectorStore initialized from documents and embeddings.

Parameters

  • documents (List[Document]) –

  • embedding (Embeddings) –

  • connection (Optional[Union[Engine, str]]) –

  • collection_name (str) –

  • distance_strategy (DistanceStrategy) –

  • ids (Optional[List[str]]) –

  • pre_delete_collection (bool) –

  • use_jsonb (bool) –

  • kwargs (Any) –

Return type

PGVector

classmethod from_embeddings(text_embeddings: List[Tuple[str, List[float]]], embedding: Embeddings, *, metadatas: Optional[List[dict]] = None, collection_name: str = 'langchain', distance_strategy: DistanceStrategy = DistanceStrategy.COSINE, ids: Optional[List[str]] = None, pre_delete_collection: bool = False, **kwargs: Any) → PGVector[source]¶

Construct PGVector wrapper from raw documents and embeddings.

Parameters

  • text_embeddings (List[Tuple[str, List[float]]]) – List of tuples of text and embeddings.

  • embedding (Embeddings) – Embeddings object.

  • metadatas (Optional[List[dict]]) – Optional list of metadatas associated with the texts.

  • collection_name (str) – Name of the collection.

  • distance_strategy (DistanceStrategy) – Distance strategy to use.

  • ids (Optional[List[str]]) – Optional list of ids for the documents. If not provided, will generate a new id for each document.

  • pre_delete_collection (bool) – If True, will delete the collection if it exists. Attention: This will delete all the documents in the existing collection.

  • kwargs (Any) – Additional arguments.

Returns

PGVector instance.

Return type

PGVector

Example

from langchain_postgres.vectorstores import PGVector
from langchain_openai.embeddings import OpenAIEmbeddings

embeddings = OpenAIEmbeddings()
text_embeddings = embeddings.embed_documents(texts)
text_embedding_pairs = list(zip(texts, text_embeddings))
vectorstore = PGVector.from_embeddings(text_embedding_pairs, embeddings)
classmethod from_existing_index(embedding: Embeddings, *, collection_name: str = 'langchain', distance_strategy: DistanceStrategy = DistanceStrategy.COSINE, pre_delete_collection: bool = False, connection: Optional[Union[Engine, str]] = None, **kwargs: Any) → PGVector[source]¶

Get instance of an existing PGVector store.This method will return the instance of the store without inserting any new embeddings

Parameters

  • embedding (Embeddings) –

  • collection_name (str) –

  • distance_strategy (DistanceStrategy) –

  • pre_delete_collection (bool) –

  • connection (Optional[Union[Engine, str]]) –

  • kwargs (Any) –

Return type

PGVector

classmethod from_texts(texts: List[str], embedding: Embeddings, metadatas: Optional[List[dict]] = None, *, collection_name: str = 'langchain', distance_strategy: DistanceStrategy = DistanceStrategy.COSINE, ids: Optional[List[str]] = None, pre_delete_collection: bool = False, use_jsonb: bool = True, **kwargs: Any) → PGVector[source]¶

Return VectorStore initialized from documents and embeddings.

Parameters

  • texts (List[str]) –

  • embedding (Embeddings) –

  • metadatas (Optional[List[dict]]) –

  • collection_name (str) –

  • distance_strategy (DistanceStrategy) –

  • ids (Optional[List[str]]) –

  • pre_delete_collection (bool) –

  • use_jsonb (bool) –

  • kwargs (Any) –

Return type

PGVector

get_by_ids(ids: Sequence[str], /) → List[Document][source]¶

Get documents by ids.

Parameters

ids (Sequence[str]) –

Return type

List[Document]

get_collection(session: Session) → Any[source]¶
Parameters

session (Session) –

Return type

Any

classmethod get_connection_string(kwargs: Dict[str, Any]) → str[source]¶
Parameters

kwargs (Dict[str, Any]) –

Return type

str

Return docs selected using the maximal marginal relevance.

Maximal marginal relevance optimizes for similarity to query AND diversity

among selected documents.

Parameters

  • query (str) – Text to look up documents similar to.

  • k (int) – Number of Documents to return. Defaults to 4.

  • fetch_k (int) – Number of Documents to fetch to pass to MMR algorithm. Defaults to 20.

  • lambda_mult (float) – 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 (Optional[Dict[str, str]]) – Filter by metadata. Defaults to None.

  • kwargs (Any) –

Returns

List of Documents selected by maximal marginal relevance.

Return type

List[Document]

max_marginal_relevance_search_by_vector(embedding: List[float], k: int = 4, fetch_k: int = 20, lambda_mult: float = 0.5, filter: Optional[Dict[str, str]] = None, **kwargs: Any) → List[Document][source]¶
Return docs selected using the maximal marginal relevance

to embedding vector.

Maximal marginal relevance optimizes for similarity to query AND diversity

among selected documents.

Parameters

  • embedding (str) – Text to look up documents similar to.

  • k (int) – Number of Documents to return. Defaults to 4.

  • fetch_k (int) – Number of Documents to fetch to pass to MMR algorithm. Defaults to 20.

  • lambda_mult (float) – 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 (Optional[Dict[str, str]]) – Filter by metadata. Defaults to None.

  • kwargs (Any) –

Returns

List of Documents selected by maximal marginal relevance.

Return type

List[Document]

max_marginal_relevance_search_with_score(query: str, k: int = 4, fetch_k: int = 20, lambda_mult: float = 0.5, filter: Optional[dict] = None, **kwargs: Any) → List[Tuple[Document, float]][source]¶

Return docs selected using the maximal marginal relevance with score.

Maximal marginal relevance optimizes for similarity to query AND diversity

among selected documents.

Parameters

  • query (str) – Text to look up documents similar to.

  • k (int) – Number of Documents to return. Defaults to 4.

  • fetch_k (int) – Number of Documents to fetch to pass to MMR algorithm. Defaults to 20.

  • lambda_mult (float) – 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 (Optional[Dict[str, str]]) – Filter by metadata. Defaults to None.

  • kwargs (Any) –

Returns

List of Documents selected by maximal marginal

relevance to the query and score for each.

Return type

List[Tuple[Document, float]]

max_marginal_relevance_search_with_score_by_vector(embedding: List[float], k: int = 4, fetch_k: int = 20, lambda_mult: float = 0.5, filter: Optional[Dict[str, str]] = None, **kwargs: Any) → List[Tuple[Document, float]][source]¶
Return docs selected using the maximal marginal relevance with score

to embedding vector.

Maximal marginal relevance optimizes for similarity to query AND diversity

among selected documents.

Parameters

  • embedding (List[float]) – Embedding to look up documents similar to.

  • k (int) – Number of Documents to return. Defaults to 4.

  • fetch_k (int) – Number of Documents to fetch to pass to MMR algorithm. Defaults to 20.

  • lambda_mult (float) – 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 (Optional[Dict[str, str]]) – Filter by metadata. Defaults to None.

  • kwargs (Any) –

Returns

List of Documents selected by maximal marginal

relevance to the query and score for each.

Return type

List[Tuple[Document, float]]

search(query: str, search_type: str, **kwargs: Any) → List[Document]¶

Return docs most similar to query using a specified search type.

Parameters

  • query (str) – Input text

  • search_type (str) – Type of search to perform. Can be “similarity”, “mmr”, or “similarity_score_threshold”.

  • **kwargs (Any) – Arguments to pass to the search method.

Returns

List of Documents most similar to the query.

Raises

ValueError – If search_type is not one of “similarity”, “mmr”, or “similarity_score_threshold”.

Return type

List[Document]

Run similarity search with PGVector with distance.

Parameters

  • query (str) – Query text to search for.

  • k (int) – Number of results to return. Defaults to 4.

  • filter (Optional[Dict[str, str]]) – Filter by metadata. Defaults to None.

  • kwargs (Any) –

Returns

List of Documents most similar to the query.

Return type

List[Document]

similarity_search_by_vector(embedding: List[float], k: int = 4, filter: Optional[dict] = None, **kwargs: Any) → List[Document][source]¶

Return docs most similar to embedding vector.

Parameters

  • embedding (List[float]) – Embedding to look up documents similar to.

  • k (int) – Number of Documents to return. Defaults to 4.

  • filter (Optional[Dict[str, str]]) – Filter by metadata. Defaults to None.

  • kwargs (Any) –

Returns

List of Documents most similar to the query vector.

Return type

List[Document]

similarity_search_with_relevance_scores(query: str, k: int = 4, **kwargs: Any) → List[Tuple[Document, float]]¶

Return docs and relevance scores in the range [0, 1].

0 is dissimilar, 1 is most similar.

Parameters

  • query (str) – Input text.

  • k (int) – Number of Documents to return. Defaults to 4.

  • **kwargs (Any) –

    kwargs to be passed to similarity search. Should include: score_threshold: Optional, a floating point value between 0 to 1 to

    filter the resulting set of retrieved docs.

Returns

List of Tuples of (doc, similarity_score).

Return type

List[Tuple[Document, float]]

similarity_search_with_score(query: str, k: int = 4, filter: Optional[dict] = None) → List[Tuple[Document, float]][source]¶

Return docs most similar to query.

Parameters

  • query (str) – Text to look up documents similar to.

  • k (int) – Number of Documents to return. Defaults to 4.

  • filter (Optional[Dict[str, str]]) – Filter by metadata. Defaults to None.

Returns

List of Documents most similar to the query and score for each.

Return type

List[Tuple[Document, float]]

similarity_search_with_score_by_vector(embedding: List[float], k: int = 4, filter: Optional[dict] = None) → List[Tuple[Document, float]][source]¶
Parameters

  • embedding (List[float]) –

  • k (int) –

  • filter (Optional[dict]) –

Return type

List[Tuple[Document, float]]

streaming_upsert(items: Iterable[Document], /, batch_size: int, **kwargs: Any) → Iterator[UpsertResponse]¶

Beta

Added in 0.2.11. The API is subject to change.

Upsert documents in a streaming fashion.

Parameters

  • items (Iterable[Document]) – Iterable of Documents to add to the vectorstore.

  • batch_size (int) – The size of each batch to upsert.

  • kwargs (Any) – Additional keyword arguments. kwargs should only include parameters that are common to all documents. (e.g., timeout for indexing, retry policy, etc.) kwargs should not include ids to avoid ambiguous semantics. Instead, the ID should be provided as part of the Document object.

Yields

UpsertResponse – A response object that contains the list of IDs that were successfully added or updated in the vectorstore and the list of IDs that failed to be added or updated.

Return type

Iterator[UpsertResponse]

New in version 0.2.11.

upsert(items: Sequence[Document], /, **kwargs: Any) → UpsertResponse[source]¶

Upsert documents into the vectorstore.

Parameters

  • items (Sequence[Document]) – Sequence of documents to upsert.

  • kwargs (Any) – vectorstore specific parameters

Returns

UpsertResponse

Return type

UpsertResponse

Examples using PGVector¶