From f4bcb31f605f040d697dfe362d69616de47f2b48 Mon Sep 17 00:00:00 2001
From: ryanrib14 <147437431+ryanrib14@users.noreply.github.com>
Date: Tue, 13 Feb 2024 18:53:25 -0300
Subject: [PATCH] fix the bug on vector_search_profile default(this profile did
 not exist) and add the possibility to choose language_analyzer of searchable
 fields, now on v0.10.3 format (#10675)

---
 .../vector_stores/azureaisearch/base.py        | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/llama-index-integrations/vector_stores/llama-index-vector-stores-azureaisearch/llama_index/vector_stores/azureaisearch/base.py b/llama-index-integrations/vector_stores/llama-index-vector-stores-azureaisearch/llama_index/vector_stores/azureaisearch/base.py
index 16a51c6532452..b81e5e5512761 100644
--- a/llama-index-integrations/vector_stores/llama-index-vector-stores-azureaisearch/llama_index/vector_stores/azureaisearch/base.py
+++ b/llama-index-integrations/vector_stores/llama-index-vector-stores-azureaisearch/llama_index/vector_stores/azureaisearch/base.py
@@ -143,14 +143,14 @@ def _create_index(self, index_name: Optional[str]) -> None:
             SearchableField(
                 name=self._field_mapping["chunk"],
                 type="Edm.String",
-                analyzer_name="en.microsoft",
+                analyzer_name=self.language_analyzer,
             ),
             SearchField(
                 name=self._field_mapping["embedding"],
                 type=SearchFieldDataType.Collection(SearchFieldDataType.Single),
                 searchable=True,
                 vector_search_dimensions=self.embedding_dimensionality,
-                vector_search_profile_name="default",
+                vector_search_profile_name=self.vector_profile_name,
             ),
             SimpleField(name=self._field_mapping["metadata"], type="Edm.String"),
             SimpleField(
@@ -243,6 +243,10 @@ def __init__(
         ] = None,
         index_management: IndexManagement = IndexManagement.NO_VALIDATION,
         embedding_dimensionality: int = 1536,
+        vector_algorithm_type: str = "exhaustiveKnn",
+        # If we have content in other languages, it is better to enable the language analyzer to be adjusted in searchable fields.
+        # https://learn.microsoft.com/en-us/azure/search/index-add-language-analyzers
+        language_analyzer: str = "en.lucene",
         **kwargs: Any,
     ) -> None:
         # ruff: noqa: E501
@@ -306,6 +310,16 @@ def __init__(
         self._search_client: SearchClient = cast(SearchClient, None)
         self.embedding_dimensionality = embedding_dimensionality
 
+        if vector_algorithm_type == "exhaustiveKnn":
+            self.vector_profile_name = "myExhaustiveKnnProfile"
+        elif vector_algorithm_type == "hnsw":
+            self.vector_profile_name = "myHnswProfile"
+        else:
+            raise ValueError(
+                "Only 'exhaustiveKnn' and 'hnsw' are supported for vector_algorithm_type"
+            )
+
+        self.language_analyzer = language_analyzer
         # Validate search_or_index_client
         if search_or_index_client is not None:
             if isinstance(search_or_index_client, SearchIndexClient):