[Backend/Prepdocs] Update to latest version of search SDK (#1010)

* Update to latest version of search

* Prefilter is default for new index

* Prefilter is default for new index

* Fix typing issue

Author: pamelafox

app/backend/approaches/chatreadretrieveread.py

@@ -6,7 +6,7 @@
 import aiohttp
 import openai
 from azure.search.documents.aio import SearchClient
-from azure.search.documents.models import QueryType
+from azure.search.documents.models import QueryType, RawVectorQuery, VectorQuery
 
 from approaches.approach import Approach
 from core.messagebuilder import MessageBuilder
@@ -143,12 +143,12 @@ async def run_until_final_call(
         # STEP 2: Retrieve relevant documents from the search index with the GPT optimized query
 
         # If retrieval mode includes vectors, compute an embedding for the query
+        vectors: list[VectorQuery] = []
         if has_vector:
             embedding_args = {"deployment_id": self.embedding_deployment} if self.openai_host == "azure" else {}
             embedding = await openai.Embedding.acreate(**embedding_args, model=self.embedding_model, input=query_text)
             query_vector = embedding["data"][0]["embedding"]
-        else:
-            query_vector = None
+            vectors.append(RawVectorQuery(vector=query_vector, k=50, fields="embedding"))
 
         # Only keep the text query if the retrieval mode uses text, otherwise drop it
         if not has_text:
@@ -165,19 +165,10 @@ async def run_until_final_call(
                 semantic_configuration_name="default",
                 top=top,
                 query_caption="extractive|highlight-false" if use_semantic_captions else None,
-                vector=query_vector,
-                top_k=50 if query_vector else None,
-                vector_fields="embedding" if query_vector else None,
+                vector_queries=vectors,
             )
         else:
-            r = await self.search_client.search(
-                query_text,
-                filter=filter,
-                top=top,
-                vector=query_vector,
-                top_k=50 if query_vector else None,
-                vector_fields="embedding" if query_vector else None,
-            )
+            r = await self.search_client.search(query_text, filter=filter, top=top, vector_queries=vectors)
         if use_semantic_captions:
             results = [
                 doc[self.sourcepage_field] + ": " + nonewlines(" . ".join([c.text for c in doc["@search.captions"]]))

app/backend/approaches/retrievethenread.py

@@ -2,7 +2,7 @@
 
 import openai
 from azure.search.documents.aio import SearchClient
-from azure.search.documents.models import QueryType
+from azure.search.documents.models import QueryType, RawVectorQuery, VectorQuery
 
 from approaches.approach import Approach
 from core.messagebuilder import MessageBuilder
@@ -78,12 +78,12 @@ async def run(
         filter = self.build_filter(overrides, auth_claims)
 
         # If retrieval mode includes vectors, compute an embedding for the query
+        vectors: list[VectorQuery] = []
         if has_vector:
             embedding_args = {"deployment_id": self.embedding_deployment} if self.openai_host == "azure" else {}
             embedding = await openai.Embedding.acreate(**embedding_args, model=self.embedding_model, input=q)
             query_vector = embedding["data"][0]["embedding"]
-        else:
-            query_vector = None
+            vectors.append(RawVectorQuery(vector=query_vector, k=50, fields="embedding"))
 
         # Only keep the text query if the retrieval mode uses text, otherwise drop it
         query_text = q if has_text else ""
@@ -99,18 +99,14 @@ async def run(
                 semantic_configuration_name="default",
                 top=top,
                 query_caption="extractive|highlight-false" if use_semantic_captions else None,
-                vector=query_vector,
-                top_k=50 if query_vector else None,
-                vector_fields="embedding" if query_vector else None,
+                vector_queries=vectors,
             )
         else:
             r = await self.search_client.search(
                 query_text,
                 filter=filter,
                 top=top,
-                vector=query_vector,
-                top_k=50 if query_vector else None,
-                vector_fields="embedding" if query_vector else None,
+                vector_queries=vectors,
             )
         if use_semantic_captions:
             results = [

app/backend/requirements.in

@@ -3,7 +3,7 @@ quart
 quart-cors
 openai[datalib]==0.28.1
 tiktoken
-azure-search-documents==11.4.0b6
+azure-search-documents==11.4.0b11
 azure-storage-blob
 uvicorn
 aiohttp

app/backend/requirements.txt

@@ -37,7 +37,7 @@ azure-monitor-opentelemetry==1.1.0
     # via -r requirements.in
 azure-monitor-opentelemetry-exporter==1.0.0b18
     # via azure-monitor-opentelemetry
-azure-search-documents==11.4.0b6
+azure-search-documents==11.4.0b11
     # via -r requirements.in
 azure-storage-blob==12.19.0
     # via -r requirements.in

scripts/prepdocslib/searchmanager.py

@@ -4,6 +4,7 @@
 
 from azure.search.documents.indexes.models import (
     HnswParameters,
+    HnswVectorSearchAlgorithmConfiguration,
     PrioritizedFields,
     SearchableField,
     SearchField,
@@ -14,7 +15,8 @@
     SemanticSettings,
     SimpleField,
     VectorSearch,
-    VectorSearchAlgorithmConfiguration,
+    VectorSearchAlgorithmKind,
+    VectorSearchProfile,
 )
 
 from .blobmanager import BlobManager
@@ -70,7 +72,7 @@ async def create_index(self):
                     sortable=False,
                     facetable=False,
                     vector_search_dimensions=1536,
-                    vector_search_configuration="default",
+                    vector_search_profile="embedding_config",
                 ),
                 SimpleField(name="category", type="Edm.String", filterable=True, facetable=True),
                 SimpleField(name="sourcepage", type="Edm.String", filterable=True, facetable=True),
@@ -102,11 +104,19 @@ async def create_index(self):
                     ]
                 ),
                 vector_search=VectorSearch(
-                    algorithm_configurations=[
-                        VectorSearchAlgorithmConfiguration(
-                            name="default", kind="hnsw", hnsw_parameters=HnswParameters(metric="cosine")
+                    algorithms=[
+                        HnswVectorSearchAlgorithmConfiguration(
+                            name="hnsw_config",
+                            kind=VectorSearchAlgorithmKind.HNSW,
+                            parameters=HnswParameters(metric="cosine"),
                         )
-                    ]
+                    ],
+                    profiles=[
+                        VectorSearchProfile(
+                            name="embedding_config",
+                            algorithm="hnsw_config",
+                        ),
+                    ],
                 ),
             )
             if self.search_info.index_name not in [name async for name in search_index_client.list_index_names()]:

scripts/requirements.in

@@ -1,6 +1,6 @@
 pypdf
 azure-identity
-azure-search-documents==11.4.0b6
+azure-search-documents==11.4.0b11
 azure-ai-formrecognizer
 azure-storage-blob
 azure-storage-file-datalake

scripts/requirements.txt

@@ -28,7 +28,7 @@ azure-core==1.29.5
     #   msrest
 azure-identity==1.15.0
     # via -r requirements.in
-azure-search-documents==11.4.0b6
+azure-search-documents==11.4.0b11
     # via -r requirements.in
 azure-storage-blob==12.19.0
     # via