From 1ed73f1992305e9aaeb405f720ebbbcac96398ba Mon Sep 17 00:00:00 2001 From: Junyoung Park Date: Wed, 7 Feb 2024 03:50:50 +0900 Subject: [PATCH] community[minor]: Add SelfQueryRetriever support to PGVector (#16991) - **Description:** Add SelfQueryRetriever support to PGVector - **Issue:** - - **Dependencies:** - - **Twitter handle:** - --------- Co-authored-by: Bagatur --- .../self_query/pgvector_self_query.ipynb | 308 ++++++++++++++++++ .../langchain/retrievers/self_query/base.py | 3 + .../retrievers/self_query/pgvector.py | 52 +++ .../retrievers/self_query/test_pgvector.py | 87 +++++ 4 files changed, 450 insertions(+) create mode 100644 docs/docs/integrations/retrievers/self_query/pgvector_self_query.ipynb create mode 100644 libs/langchain/langchain/retrievers/self_query/pgvector.py create mode 100644 libs/langchain/tests/unit_tests/retrievers/self_query/test_pgvector.py diff --git a/docs/docs/integrations/retrievers/self_query/pgvector_self_query.ipynb b/docs/docs/integrations/retrievers/self_query/pgvector_self_query.ipynb new file mode 100644 index 00000000000..b6475109362 --- /dev/null +++ b/docs/docs/integrations/retrievers/self_query/pgvector_self_query.ipynb @@ -0,0 +1,308 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "id": "13afcae7", + "metadata": {}, + "source": [ + "# PGVector\n", + "\n", + ">[PGVector](https://github.com/pgvector/pgvector) is a vector similarity search for Postgres.\n", + "\n", + "In the notebook, we'll demo the `SelfQueryRetriever` wrapped around a `PGVector` vector store." + ] + }, + { + "cell_type": "markdown", + "id": "68e75fb9", + "metadata": {}, + "source": [ + "## Creating a PGVector vector store\n", + "First we'll want to create a PGVector vector store and seed it with some data. We've created a small demo set of documents that contain summaries of movies.\n", + "\n", + "**Note:** The self-query retriever requires you to have `lark` installed (`pip install lark`). We also need the `` package." + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "63a8af5b", + "metadata": { + "tags": [] + }, + "outputs": [], + "source": [ + "%pip install --upgrade --quiet lark pgvector psycopg2-binary" + ] + }, + { + "cell_type": "markdown", + "id": "83811610-7df3-4ede-b268-68a6a83ba9e2", + "metadata": {}, + "source": [ + "We want to use `OpenAIEmbeddings` so we have to get the OpenAI API Key." + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "dd01b61b-7d32-4a55-85d6-b2d2d4f18840", + "metadata": { + "tags": [] + }, + "outputs": [], + "source": [ + "import getpass\n", + "import os\n", + "\n", + "os.environ[\"OPENAI_API_KEY\"] = getpass.getpass(\"OpenAI API Key:\")" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "cb4a5787", + "metadata": { + "tags": [] + }, + "outputs": [], + "source": [ + "from langchain.schema import Document\n", + "from langchain_community.vectorstores import PGVector\n", + "from langchain_openai import OpenAIEmbeddings\n", + "\n", + "collection = \"Name of your collection\"\n", + "embeddings = OpenAIEmbeddings()" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "bcbe04d9", + "metadata": { + "tags": [] + }, + "outputs": [], + "source": [ + "docs = [\n", + " Document(\n", + " page_content=\"A bunch of scientists bring back dinosaurs and mayhem breaks loose\",\n", + " metadata={\"year\": 1993, \"rating\": 7.7, \"genre\": \"science fiction\"},\n", + " ),\n", + " Document(\n", + " page_content=\"Leo DiCaprio gets lost in a dream within a dream within a dream within a ...\",\n", + " metadata={\"year\": 2010, \"director\": \"Christopher Nolan\", \"rating\": 8.2},\n", + " ),\n", + " Document(\n", + " page_content=\"A psychologist / detective gets lost in a series of dreams within dreams within dreams and Inception reused the idea\",\n", + " metadata={\"year\": 2006, \"director\": \"Satoshi Kon\", \"rating\": 8.6},\n", + " ),\n", + " Document(\n", + " page_content=\"A bunch of normal-sized women are supremely wholesome and some men pine after them\",\n", + " metadata={\"year\": 2019, \"director\": \"Greta Gerwig\", \"rating\": 8.3},\n", + " ),\n", + " Document(\n", + " page_content=\"Toys come alive and have a blast doing so\",\n", + " metadata={\"year\": 1995, \"genre\": \"animated\"},\n", + " ),\n", + " Document(\n", + " page_content=\"Three men walk into the Zone, three men walk out of the Zone\",\n", + " metadata={\n", + " \"year\": 1979,\n", + " \"director\": \"Andrei Tarkovsky\",\n", + " \"genre\": \"science fiction\",\n", + " \"rating\": 9.9,\n", + " },\n", + " ),\n", + "]\n", + "vectorstore = PGVector.from_documents(\n", + " docs,\n", + " embeddings,\n", + " collection_name=collection,\n", + ")" + ] + }, + { + "cell_type": "markdown", + "id": "5ecaab6d", + "metadata": {}, + "source": [ + "## Creating our self-querying retriever\n", + "Now we can instantiate our retriever. To do this we'll need to provide some information upfront about the metadata fields that our documents support and a short description of the document contents." + ] + }, + { + "cell_type": "code", + "execution_count": 6, + "id": "86e34dbf", + "metadata": { + "tags": [] + }, + "outputs": [], + "source": [ + "from langchain.chains.query_constructor.base import AttributeInfo\n", + "from langchain.retrievers.self_query.base import SelfQueryRetriever\n", + "from langchain_openai import OpenAI\n", + "\n", + "metadata_field_info = [\n", + " AttributeInfo(\n", + " name=\"genre\",\n", + " description=\"The genre of the movie\",\n", + " type=\"string or list[string]\",\n", + " ),\n", + " AttributeInfo(\n", + " name=\"year\",\n", + " description=\"The year the movie was released\",\n", + " type=\"integer\",\n", + " ),\n", + " AttributeInfo(\n", + " name=\"director\",\n", + " description=\"The name of the movie director\",\n", + " type=\"string\",\n", + " ),\n", + " AttributeInfo(\n", + " name=\"rating\", description=\"A 1-10 rating for the movie\", type=\"float\"\n", + " ),\n", + "]\n", + "document_content_description = \"Brief summary of a movie\"\n", + "llm = OpenAI(temperature=0)\n", + "retriever = SelfQueryRetriever.from_llm(\n", + " llm, vectorstore, document_content_description, metadata_field_info, verbose=True\n", + ")" + ] + }, + { + "cell_type": "markdown", + "id": "ea9df8d4", + "metadata": {}, + "source": [ + "## Testing it out\n", + "And now we can try actually using our retriever!" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "38a126e9", + "metadata": {}, + "outputs": [], + "source": [ + "# This example only specifies a relevant query\n", + "retriever.get_relevant_documents(\"What are some movies about dinosaurs\")" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "fc3f1e6e", + "metadata": {}, + "outputs": [], + "source": [ + "# This example only specifies a filter\n", + "retriever.get_relevant_documents(\"I want to watch a movie rated higher than 8.5\")" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "b19d4da0", + "metadata": {}, + "outputs": [], + "source": [ + "# This example specifies a query and a filter\n", + "retriever.get_relevant_documents(\"Has Greta Gerwig directed any movies about women\")" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "f900e40e", + "metadata": {}, + "outputs": [], + "source": [ + "# This example specifies a composite filter\n", + "retriever.get_relevant_documents(\n", + " \"What's a highly rated (above 8.5) science fiction film?\"\n", + ")" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "12a51522", + "metadata": {}, + "outputs": [], + "source": [ + "# This example specifies a query and composite filter\n", + "retriever.get_relevant_documents(\n", + " \"What's a movie after 1990 but before 2005 that's all about toys, and preferably is animated\"\n", + ")" + ] + }, + { + "cell_type": "markdown", + "id": "39bd1de1-b9fe-4a98-89da-58d8a7a6ae51", + "metadata": {}, + "source": [ + "## Filter k\n", + "\n", + "We can also use the self query retriever to specify `k`: the number of documents to fetch.\n", + "\n", + "We can do this by passing `enable_limit=True` to the constructor." + ] + }, + { + "cell_type": "code", + "execution_count": 7, + "id": "bff36b88-b506-4877-9c63-e5a1a8d78e64", + "metadata": { + "tags": [] + }, + "outputs": [], + "source": [ + "retriever = SelfQueryRetriever.from_llm(\n", + " llm,\n", + " vectorstore,\n", + " document_content_description,\n", + " metadata_field_info,\n", + " enable_limit=True,\n", + " verbose=True,\n", + ")" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "2758d229-4f97-499c-819f-888acaf8ee10", + "metadata": { + "tags": [] + }, + "outputs": [], + "source": [ + "# This example only specifies a relevant query\n", + "retriever.get_relevant_documents(\"what are two movies about dinosaurs\")" + ] + } + ], + "metadata": { + "kernelspec": { + "display_name": "Python 3 (ipykernel)", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.9.16" + } + }, + "nbformat": 4, + "nbformat_minor": 5 +} diff --git a/libs/langchain/langchain/retrievers/self_query/base.py b/libs/langchain/langchain/retrievers/self_query/base.py index 9c6a584a6e1..d54120ccfc3 100644 --- a/libs/langchain/langchain/retrievers/self_query/base.py +++ b/libs/langchain/langchain/retrievers/self_query/base.py @@ -12,6 +12,7 @@ from langchain_community.vectorstores import ( MongoDBAtlasVectorSearch, MyScale, OpenSearchVectorSearch, + PGVector, Pinecone, Qdrant, Redis, @@ -43,6 +44,7 @@ from langchain.retrievers.self_query.milvus import MilvusTranslator from langchain.retrievers.self_query.mongodb_atlas import MongoDBAtlasTranslator from langchain.retrievers.self_query.myscale import MyScaleTranslator from langchain.retrievers.self_query.opensearch import OpenSearchTranslator +from langchain.retrievers.self_query.pgvector import PGVectorTranslator from langchain.retrievers.self_query.pinecone import PineconeTranslator from langchain.retrievers.self_query.qdrant import QdrantTranslator from langchain.retrievers.self_query.redis import RedisTranslator @@ -58,6 +60,7 @@ def _get_builtin_translator(vectorstore: VectorStore) -> Visitor: """Get the translator class corresponding to the vector store class.""" BUILTIN_TRANSLATORS: Dict[Type[VectorStore], Type[Visitor]] = { AstraDB: AstraDBTranslator, + PGVector: PGVectorTranslator, Pinecone: PineconeTranslator, Chroma: ChromaTranslator, DashVector: DashvectorTranslator, diff --git a/libs/langchain/langchain/retrievers/self_query/pgvector.py b/libs/langchain/langchain/retrievers/self_query/pgvector.py new file mode 100644 index 00000000000..ebe5bf42ffc --- /dev/null +++ b/libs/langchain/langchain/retrievers/self_query/pgvector.py @@ -0,0 +1,52 @@ +from typing import Dict, Tuple, Union + +from langchain.chains.query_constructor.ir import ( + Comparator, + Comparison, + Operation, + Operator, + StructuredQuery, + Visitor, +) + + +class PGVectorTranslator(Visitor): + """Translate `PGVector` internal query language elements to valid filters.""" + + allowed_operators = [Operator.AND, Operator.OR] + """Subset of allowed logical operators.""" + allowed_comparators = [ + Comparator.EQ, + Comparator.NE, + Comparator.GT, + Comparator.LT, + Comparator.IN, + Comparator.NIN, + Comparator.CONTAIN, + Comparator.LIKE, + ] + """Subset of allowed logical comparators.""" + + def _format_func(self, func: Union[Operator, Comparator]) -> str: + self._validate_func(func) + return f"{func.value}" + + def visit_operation(self, operation: Operation) -> Dict: + args = [arg.accept(self) for arg in operation.arguments] + return {self._format_func(operation.operator): args} + + def visit_comparison(self, comparison: Comparison) -> Dict: + return { + comparison.attribute: { + self._format_func(comparison.comparator): comparison.value + } + } + + def visit_structured_query( + self, structured_query: StructuredQuery + ) -> Tuple[str, dict]: + if structured_query.filter is None: + kwargs = {} + else: + kwargs = {"filter": structured_query.filter.accept(self)} + return structured_query.query, kwargs diff --git a/libs/langchain/tests/unit_tests/retrievers/self_query/test_pgvector.py b/libs/langchain/tests/unit_tests/retrievers/self_query/test_pgvector.py new file mode 100644 index 00000000000..451508bdab3 --- /dev/null +++ b/libs/langchain/tests/unit_tests/retrievers/self_query/test_pgvector.py @@ -0,0 +1,87 @@ +from typing import Dict, Tuple + +import pytest as pytest + +from langchain.chains.query_constructor.ir import ( + Comparator, + Comparison, + Operation, + Operator, + StructuredQuery, +) +from langchain.retrievers.self_query.pgvector import PGVectorTranslator + +DEFAULT_TRANSLATOR = PGVectorTranslator() + + +def test_visit_comparison() -> None: + comp = Comparison(comparator=Comparator.LT, attribute="foo", value=1) + expected = {"foo": {"lt": 1}} + actual = DEFAULT_TRANSLATOR.visit_comparison(comp) + assert expected == actual + + +@pytest.mark.skip("Not implemented") +def test_visit_operation() -> None: + op = Operation( + operator=Operator.AND, + arguments=[ + Comparison(comparator=Comparator.LT, attribute="foo", value=2), + Comparison(comparator=Comparator.EQ, attribute="bar", value="baz"), + Comparison(comparator=Comparator.GT, attribute="abc", value=2.0), + ], + ) + expected = { + "foo": {"lt": 2}, + "bar": {"eq": "baz"}, + "abc": {"gt": 2.0}, + } + actual = DEFAULT_TRANSLATOR.visit_operation(op) + assert expected == actual + + +def test_visit_structured_query() -> None: + query = "What is the capital of France?" + structured_query = StructuredQuery( + query=query, + filter=None, + ) + expected: Tuple[str, Dict] = (query, {}) + actual = DEFAULT_TRANSLATOR.visit_structured_query(structured_query) + assert expected == actual + + comp = Comparison(comparator=Comparator.LT, attribute="foo", value=1) + structured_query = StructuredQuery( + query=query, + filter=comp, + ) + expected = (query, {"filter": {"foo": {"lt": 1}}}) + actual = DEFAULT_TRANSLATOR.visit_structured_query(structured_query) + assert expected == actual + + op = Operation( + operator=Operator.AND, + arguments=[ + Comparison(comparator=Comparator.LT, attribute="foo", value=2), + Comparison(comparator=Comparator.EQ, attribute="bar", value="baz"), + Comparison(comparator=Comparator.GT, attribute="abc", value=2.0), + ], + ) + structured_query = StructuredQuery( + query=query, + filter=op, + ) + expected = ( + query, + { + "filter": { + "and": [ + {"foo": {"lt": 2}}, + {"bar": {"eq": "baz"}}, + {"abc": {"gt": 2.0}}, + ] + } + }, + ) + actual = DEFAULT_TRANSLATOR.visit_structured_query(structured_query) + assert expected == actual