From cacaf487c38785520b40c9efc53a556ddb6dc3d0 Mon Sep 17 00:00:00 2001 From: Fabrizio Ruocco Date: Fri, 25 Aug 2023 11:34:09 +0200 Subject: [PATCH] Azure Cognitive Search - update sdk b8, mod user agent, search with scores (#9191) Description: Update Azure Cognitive Search SDK to version b8 (breaking change) Customizable User Agent. Implemented Similarity search with scores @baskaryan --------- Co-authored-by: Bagatur --- .../vectorstores/azuresearch.ipynb | 71 ++++++++++++++----- .../langchain/vectorstores/azuresearch.py | 66 +++++++++++++---- libs/langchain/poetry.lock | 8 +-- libs/langchain/pyproject.toml | 2 +- 4 files changed, 111 insertions(+), 36 deletions(-) diff --git a/docs/extras/integrations/vectorstores/azuresearch.ipynb b/docs/extras/integrations/vectorstores/azuresearch.ipynb index fe64621365..fc9bb75b5b 100644 --- a/docs/extras/integrations/vectorstores/azuresearch.ipynb +++ b/docs/extras/integrations/vectorstores/azuresearch.ipynb @@ -6,7 +6,9 @@ "source": [ "# Azure Cognitive Search\n", "\n", - "[Azure Cognitive Search](https://learn.microsoft.com/azure/search/search-what-is-azure-search) (formerly known as `Azure Search`) is a cloud search service that gives developers infrastructure, APIs, and tools for building a rich search experience over private, heterogeneous content in web, mobile, and enterprise applications.\n" + "[Azure Cognitive Search](https://learn.microsoft.com/azure/search/search-what-is-azure-search) (formerly known as `Azure Search`) is a cloud search service that gives developers infrastructure, APIs, and tools for building a rich search experience over private, heterogeneous content in web, mobile, and enterprise applications.\n", + "\n", + "Vector search is currently in public preview. It's available through the Azure portal, preview REST API and beta client libraries. [More info](https://learn.microsoft.com/en-us/azure/search/vector-search-overview) Beta client libraries are subject to potential breaking changes, please be sure to use the SDK package version identified below. azure-search-documents==11.4.0b8" ] }, { @@ -22,7 +24,7 @@ "metadata": {}, "outputs": [], "source": [ - "!pip install azure-search-documents==11.4.0b6\n", + "!pip install azure-search-documents==11.4.0b8\n", "!pip install azure-identity" ] }, @@ -36,13 +38,13 @@ }, { "cell_type": "code", - "execution_count": 7, + "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "import openai\n", "import os\n", - "from langchain.embeddings.openai import OpenAIEmbeddings\n", + "from langchain.embeddings import OpenAIEmbeddings\n", "from langchain.vectorstores.azuresearch import AzureSearch" ] }, @@ -57,7 +59,7 @@ }, { "cell_type": "code", - "execution_count": 8, + "execution_count": 2, "metadata": {}, "outputs": [], "source": [ @@ -79,7 +81,7 @@ }, { "cell_type": "code", - "execution_count": 9, + "execution_count": 3, "metadata": {}, "outputs": [], "source": [ @@ -98,7 +100,7 @@ }, { "cell_type": "code", - "execution_count": 10, + "execution_count": 4, "metadata": {}, "outputs": [], "source": [ @@ -151,7 +153,7 @@ }, { "cell_type": "code", - "execution_count": 15, + "execution_count": 5, "metadata": {}, "outputs": [ { @@ -178,6 +180,41 @@ "print(docs[0].page_content)" ] }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Perform a vector similarity search with relevance scores\n", + " \n", + "Execute a pure vector similarity search using the similarity_search_with_relevance_scores() method:" + ] + }, + { + "cell_type": "code", + "execution_count": 12, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "[(Document(page_content='Tonight. I call on the Senate to: Pass the Freedom to Vote Act. Pass the John Lewis Voting Rights Act. And while you’re at it, pass the Disclose Act so Americans can know who is funding our elections. \\n\\nTonight, I’d like to honor someone who has dedicated his life to serve this country: Justice Stephen Breyer—an Army veteran, Constitutional scholar, and retiring Justice of the United States Supreme Court. Justice Breyer, thank you for your service. \\n\\nOne of the most serious constitutional responsibilities a President has is nominating someone to serve on the United States Supreme Court. \\n\\nAnd I did that 4 days ago, when I nominated Circuit Court of Appeals Judge Ketanji Brown Jackson. One of our nation’s top legal minds, who will continue Justice Breyer’s legacy of excellence.', metadata={'source': 'C:\\\\repos\\\\langchain-fruocco-acs\\\\langchain\\\\docs\\\\extras\\\\modules\\\\state_of_the_union.txt'}),\n", + " 0.8441472),\n", + " (Document(page_content='Tonight. I call on the Senate to: Pass the Freedom to Vote Act. Pass the John Lewis Voting Rights Act. And while you’re at it, pass the Disclose Act so Americans can know who is funding our elections. \\n\\nTonight, I’d like to honor someone who has dedicated his life to serve this country: Justice Stephen Breyer—an Army veteran, Constitutional scholar, and retiring Justice of the United States Supreme Court. Justice Breyer, thank you for your service. \\n\\nOne of the most serious constitutional responsibilities a President has is nominating someone to serve on the United States Supreme Court. \\n\\nAnd I did that 4 days ago, when I nominated Circuit Court of Appeals Judge Ketanji Brown Jackson. One of our nation’s top legal minds, who will continue Justice Breyer’s legacy of excellence.', metadata={'source': 'C:\\\\repos\\\\langchain-fruocco-acs\\\\langchain\\\\docs\\\\extras\\\\modules\\\\state_of_the_union.txt'}),\n", + " 0.8441472),\n", + " (Document(page_content='A former top litigator in private practice. A former federal public defender. And from a family of public school educators and police officers. A consensus builder. Since she’s been nominated, she’s received a broad range of support—from the Fraternal Order of Police to former judges appointed by Democrats and Republicans. \\n\\nAnd if we are to advance liberty and justice, we need to secure the Border and fix the immigration system. \\n\\nWe can do both. At our border, we’ve installed new technology like cutting-edge scanners to better detect drug smuggling. \\n\\nWe’ve set up joint patrols with Mexico and Guatemala to catch more human traffickers. \\n\\nWe’re putting in place dedicated immigration judges so families fleeing persecution and violence can have their cases heard faster. \\n\\nWe’re securing commitments and supporting partners in South and Central America to host more refugees and secure their own borders.', metadata={'source': 'C:\\\\repos\\\\langchain-fruocco-acs\\\\langchain\\\\docs\\\\extras\\\\modules\\\\state_of_the_union.txt'}),\n", + " 0.82153815),\n", + " (Document(page_content='A former top litigator in private practice. A former federal public defender. And from a family of public school educators and police officers. A consensus builder. Since she’s been nominated, she’s received a broad range of support—from the Fraternal Order of Police to former judges appointed by Democrats and Republicans. \\n\\nAnd if we are to advance liberty and justice, we need to secure the Border and fix the immigration system. \\n\\nWe can do both. At our border, we’ve installed new technology like cutting-edge scanners to better detect drug smuggling. \\n\\nWe’ve set up joint patrols with Mexico and Guatemala to catch more human traffickers. \\n\\nWe’re putting in place dedicated immigration judges so families fleeing persecution and violence can have their cases heard faster. \\n\\nWe’re securing commitments and supporting partners in South and Central America to host more refugees and secure their own borders.', metadata={'source': 'C:\\\\repos\\\\langchain-fruocco-acs\\\\langchain\\\\docs\\\\extras\\\\modules\\\\state_of_the_union.txt'}),\n", + " 0.82153815)]\n" + ] + } + ], + "source": [ + "docs_and_scores = vector_store.similarity_search_with_relevance_scores(query=\"What did the president say about Ketanji Brown Jackson\", k=4, score_threshold=0.80)\n", + "from pprint import pprint\n", + "pprint(docs_and_scores)" + ] + }, { "attachments": {}, "cell_type": "markdown", @@ -190,7 +227,7 @@ }, { "cell_type": "code", - "execution_count": null, + "execution_count": 13, "metadata": {}, "outputs": [ { @@ -219,7 +256,7 @@ }, { "cell_type": "code", - "execution_count": null, + "execution_count": 14, "metadata": {}, "outputs": [ { @@ -254,7 +291,7 @@ }, { "cell_type": "code", - "execution_count": 16, + "execution_count": 18, "metadata": {}, "outputs": [], "source": [ @@ -328,7 +365,7 @@ }, { "cell_type": "code", - "execution_count": 17, + "execution_count": 19, "metadata": {}, "outputs": [], "source": [ @@ -348,7 +385,7 @@ }, { "cell_type": "code", - "execution_count": 18, + "execution_count": 20, "metadata": {}, "outputs": [ { @@ -371,7 +408,7 @@ }, { "cell_type": "code", - "execution_count": 19, + "execution_count": 21, "metadata": {}, "outputs": [ { @@ -400,7 +437,7 @@ }, { "cell_type": "code", - "execution_count": 20, + "execution_count": 22, "metadata": {}, "outputs": [], "source": [ @@ -494,7 +531,7 @@ }, { "cell_type": "code", - "execution_count": 21, + "execution_count": 23, "metadata": {}, "outputs": [ { @@ -530,7 +567,7 @@ }, { "cell_type": "code", - "execution_count": 23, + "execution_count": 24, "metadata": {}, "outputs": [ { diff --git a/libs/langchain/langchain/vectorstores/azuresearch.py b/libs/langchain/langchain/vectorstores/azuresearch.py index 32b6d03f18..2fad466ada 100644 --- a/libs/langchain/langchain/vectorstores/azuresearch.py +++ b/libs/langchain/langchain/vectorstores/azuresearch.py @@ -73,6 +73,7 @@ def _get_search_client( scoring_profiles: Optional[List[ScoringProfile]] = None, default_scoring_profile: Optional[str] = None, default_fields: Optional[List[SearchField]] = None, + user_agent: Optional[str] = "langchain", ) -> SearchClient: from azure.core.credentials import AzureKeyCredential from azure.core.exceptions import ResourceNotFoundError @@ -80,13 +81,13 @@ def _get_search_client( from azure.search.documents import SearchClient from azure.search.documents.indexes import SearchIndexClient from azure.search.documents.indexes.models import ( + HnswVectorSearchAlgorithmConfiguration, PrioritizedFields, SearchIndex, SemanticConfiguration, SemanticField, SemanticSettings, VectorSearch, - VectorSearchAlgorithmConfiguration, ) default_fields = default_fields or [] @@ -95,7 +96,7 @@ def _get_search_client( else: credential = AzureKeyCredential(key) index_client: SearchIndexClient = SearchIndexClient( - endpoint=endpoint, credential=credential, user_agent="langchain" + endpoint=endpoint, credential=credential, user_agent=user_agent ) try: index_client.get_index(name=index_name) @@ -130,10 +131,10 @@ def _get_search_client( if vector_search is None: vector_search = VectorSearch( algorithm_configurations=[ - VectorSearchAlgorithmConfiguration( + HnswVectorSearchAlgorithmConfiguration( name="default", kind="hnsw", - hnsw_parameters={ # type: ignore + parameters={ # type: ignore "m": 4, "efConstruction": 400, "efSearch": 500, @@ -171,7 +172,7 @@ def _get_search_client( endpoint=endpoint, index_name=index_name, credential=credential, - user_agent="langchain", + user_agent=user_agent, ) @@ -227,6 +228,9 @@ class AzureSearch(VectorStore): type=SearchFieldDataType.String, ), ] + user_agent = "langchain" + if "user_agent" in kwargs and kwargs["user_agent"]: + user_agent += " " + kwargs["user_agent"] self.client = _get_search_client( azure_search_endpoint, azure_search_key, @@ -238,6 +242,7 @@ class AzureSearch(VectorStore): scoring_profiles=scoring_profiles, default_scoring_profile=default_scoring_profile, default_fields=default_fields, + user_agent=user_agent, ) self.search_type = search_type self.semantic_configuration_name = semantic_configuration_name @@ -321,6 +326,17 @@ class AzureSearch(VectorStore): raise ValueError(f"search_type of {search_type} not allowed.") return docs + def similarity_search_with_relevance_scores( + self, query: str, k: int = 4, **kwargs: Any + ) -> List[Tuple[Document, float]]: + score_threshold = kwargs.pop("score_threshold", None) + result = self.vector_search_with_score(query, k=k, **kwargs) + return ( + result + if score_threshold is None + else [r for r in result if r[1] >= score_threshold] + ) + def vector_search(self, query: str, k: int = 4, **kwargs: Any) -> List[Document]: """ Returns the most similar indexed documents to the query text. @@ -349,12 +365,19 @@ class AzureSearch(VectorStore): Returns: List of Documents most similar to the query and score for each """ + from azure.search.documents.models import Vector results = self.client.search( search_text="", - vector=np.array(self.embedding_function(query), dtype=np.float32).tolist(), - top_k=k, - vector_fields=FIELDS_CONTENT_VECTOR, + vectors=[ + Vector( + value=np.array( + self.embedding_function(query), dtype=np.float32 + ).tolist(), + k=k, + fields=FIELDS_CONTENT_VECTOR, + ) + ], select=[FIELDS_ID, FIELDS_CONTENT, FIELDS_METADATA], filter=filters, ) @@ -399,12 +422,19 @@ class AzureSearch(VectorStore): Returns: List of Documents most similar to the query and score for each """ + from azure.search.documents.models import Vector results = self.client.search( search_text=query, - vector=np.array(self.embedding_function(query), dtype=np.float32).tolist(), - top_k=k, - vector_fields=FIELDS_CONTENT_VECTOR, + vectors=[ + Vector( + value=np.array( + self.embedding_function(query), dtype=np.float32 + ).tolist(), + k=k, + fields=FIELDS_CONTENT_VECTOR, + ) + ], select=[FIELDS_ID, FIELDS_CONTENT, FIELDS_METADATA], filter=filters, top=k, @@ -452,11 +482,19 @@ class AzureSearch(VectorStore): Returns: List of Documents most similar to the query and score for each """ + from azure.search.documents.models import Vector + results = self.client.search( search_text=query, - vector=np.array(self.embedding_function(query), dtype=np.float32).tolist(), - top_k=50, # Hardcoded value to maximize L2 retrieval - vector_fields=FIELDS_CONTENT_VECTOR, + vectors=[ + Vector( + value=np.array( + self.embedding_function(query), dtype=np.float32 + ).tolist(), + k=50, + fields=FIELDS_CONTENT_VECTOR, + ) + ], select=[FIELDS_ID, FIELDS_CONTENT, FIELDS_METADATA], filter=filters, query_type="semantic", diff --git a/libs/langchain/poetry.lock b/libs/langchain/poetry.lock index badf9ec891..e21acf426b 100644 --- a/libs/langchain/poetry.lock +++ b/libs/langchain/poetry.lock @@ -719,13 +719,13 @@ msal-extensions = ">=0.3.0,<2.0.0" [[package]] name = "azure-search-documents" -version = "11.4.0b6" +version = "11.4.0b8" description = "Microsoft Azure Cognitive Search Client Library for Python" optional = true python-versions = ">=3.7" files = [ - {file = "azure-search-documents-11.4.0b6.zip", hash = "sha256:c9ebd7d99d3c7b879f48acad66141e1f50eae4468cfb8389a4b25d4c620e8df1"}, - {file = "azure_search_documents-11.4.0b6-py3-none-any.whl", hash = "sha256:24ff85bf2680c36b38d8092bcbbe2d90699aac7c4a228b0839c0ce595a41628c"}, + {file = "azure-search-documents-11.4.0b8.zip", hash = "sha256:b178ff52918590191a9cb7f411a9ab3cb517663666a501a3e84b715d19b0d93b"}, + {file = "azure_search_documents-11.4.0b8-py3-none-any.whl", hash = "sha256:4137daa2db75bff9484d394c16c0604822a51281cad2f50e11d7c48dd8d4b4cf"}, ] [package.dependencies] @@ -10447,4 +10447,4 @@ text-helpers = ["chardet"] [metadata] lock-version = "2.0" python-versions = ">=3.8.1,<4.0" -content-hash = "88e479307b19d991105360780f67ed3258ef1a0151f70b9e91c86c8153751e83" +content-hash = "43a6bd42efc0baf917418087f788aaf3b1bc793cb4aa81de99c52ed6a7d54d26" diff --git a/libs/langchain/pyproject.toml b/libs/langchain/pyproject.toml index bc626b4155..bab5c36c17 100644 --- a/libs/langchain/pyproject.toml +++ b/libs/langchain/pyproject.toml @@ -105,7 +105,7 @@ nebula3-python = {version = "^3.4.0", optional = true} mwparserfromhell = {version = "^0.6.4", optional = true} mwxml = {version = "^0.3.3", optional = true} awadb = {version = "^0.3.9", optional = true} -azure-search-documents = {version = "11.4.0b6", optional = true} +azure-search-documents = {version = "11.4.0b8", optional = true} esprima = {version = "^4.0.1", optional = true} streamlit = {version = "^1.18.0", optional = true, python = ">=3.8.1,<3.9.7 || >3.9.7,<4.0"} psychicapi = {version = "^0.8.0", optional = true}