datasourcelib 0.1.3__py3-none-any.whl → 0.1.4__py3-none-any.whl

datasourcelib/indexes/azure_search_index.py

@@ -110,7 +110,7 @@ class AzureSearchIndexer:
             logger.exception(f"Failed to get embeddings for text: {text[:100]}...")
             raise
 
-    def _build_vector_search_config(self):
+    def _build_vector_search_config_old(self):
         AzureKeyCredential, SearchClient, SearchIndexClient, AzureOpenAI, SearchIndex, SearchField, SearchFieldDataType, SimpleField, SearchableField, VectorSearch, VectorSearchProfile, HnswAlgorithmConfiguration, SemanticSearch, SemanticField, SemanticConfiguration, SemanticPrioritizedFields = self._ensure_sdk()
         vector_config = self.config.get("vector_config", {})
         dimensions = vector_config.get("dimensions", 1536)
@@ -121,6 +121,107 @@ class AzureSearchIndexer:
             )
 
         return vector_search, dimensions
+    
+    def _build_vector_search_config(self):
+        AzureKeyCredential, SearchClient, SearchIndexClient, AzureOpenAI, SearchIndex, SearchField, SearchFieldDataType, SimpleField, SearchableField, VectorSearch, VectorSearchProfile, HnswAlgorithmConfiguration, SemanticSearch, SemanticField, SemanticConfiguration, SemanticPrioritizedFields = self._ensure_sdk()
+
+        vector_config = self.config.get("vector_config", {})
+        dimensions = vector_config.get("dimensions", 1536)
+        algorithm = vector_config.get("algorithm", "hnsw").lower()
+
+        # Build algorithm configuration (SDK model if available)
+        alg_cfg = HnswAlgorithmConfiguration(name="algorithms-config-1")
+
+        # Build vectorizer settings using Azure OpenAI config from vector_db_config
+        deployment = self.config.get("embedding_deployment")
+        endpoint = self.config.get("embedding_endpoint")
+        api_key = self.config.get("embedding_key")
+        # modelName required for API version 2025-09-01 — prefer explicit embedding_model, fall back to deployment
+        model_name = self.config.get("embedding_model") or deployment
+        content_field = self.config.get("content_field", "content")
+        vector_field = self.config.get("vector_field", "contentVector")
+
+        if not model_name:
+            raise RuntimeError("Vectorizer configuration requires 'embedding_model' or 'embedding_deployment' in vector_db_config")
+
+        # Define vectorizer with explicit name and required azureOpenAIParameters including modelName
+        vectorizer_name = "azure-openai-vectorizer"
+        vectorizer = {
+            "name": vectorizer_name,
+            "kind": "azureOpenAI",
+            "azureOpenAIParameters": {
+                "resourceUri": endpoint.rstrip('/') if endpoint else None,
+                # include both modelName (required) and deploymentId (if provided)
+                "modelName": model_name,
+                **({"deploymentId": deployment} if deployment else {}),
+                "apiKey": api_key
+            },
+            "options": {
+                "fieldMapping": [
+                    {
+                        "sourceContext": f"/document/{content_field}",
+                        "outputs": [
+                            {
+                                "targetContext": f"/document/{vector_field}",
+                                "targetDimensions": dimensions
+                            }
+                        ]
+                    }
+                ]
+            }
+        }
+
+        profile_name = "vector-profile-1"
+        try:
+            # Create profile with vectorizer reference (SDK may expect vectorizer_name or vectorizer depending on version)
+            try:
+                profile = VectorSearchProfile(
+                    name=profile_name,
+                    algorithm_configuration_name="algorithms-config-1",
+                    vectorizer_name=vectorizer_name
+                )
+            except TypeError:
+                # fallback if SDK constructor uses different parameter names
+                profile = VectorSearchProfile(name=profile_name, algorithm_configuration_name="algorithms-config-1")
+                try:
+                    setattr(profile, "vectorizer_name", vectorizer_name)
+                except Exception:
+                    pass
+
+            try:
+                # Construct full vector search config with both profile and vectorizer
+                vector_search = VectorSearch(
+                    profiles=[profile],
+                    algorithms=[alg_cfg],
+                    vectorizers=[vectorizer]
+                )
+            except Exception:
+                # Fallback to dict if SDK constructor differs
+                vector_search = {
+                    "profiles": [{
+                        "name": profile_name,
+                        "algorithmConfigurationName": "algorithms-config-1",
+                        "vectorizerName": vectorizer_name
+                    }],
+                    "algorithms": [{"name": "algorithms-config-1"}],
+                    "vectorizers": [vectorizer]
+                }
+        except Exception:
+            # Full dict fallback
+            vector_search = {
+                "profiles": [{
+                    "name": profile_name,
+                    "algorithmConfigurationName": "algorithms-config-1",
+                    "vectorizerName": vectorizer_name
+                }],
+                "algorithms": [{"name": "algorithms-config-1"}],
+                "vectorizers": [vectorizer]
+            }
+
+        logger.info("Built vector_search config (dimensions=%s, model=%s, vectorizer=%s)",
+                    dimensions, model_name, vectorizer_name)
+        return vector_search, dimensions
+
 
     def _build_semantic_settings(self):
         """

datasourcelib/strategies/full_load.py

@@ -1,6 +1,6 @@
 from datasourcelib.core.sync_base import SyncBase
 from datasourcelib.utils.logger import get_logger
-from datasourcelib.indexes.azure_search_index_vector import AzureSearchIndexer
+from datasourcelib.indexes.azure_search_index import AzureSearchIndexer
 logger = get_logger(__name__)
 
 class FullLoadStrategy(SyncBase):

{datasourcelib-0.1.3.dist-info → datasourcelib-0.1.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datasourcelib
-Version: 0.1.3
+Version: 0.1.4
 Summary: Data source sync strategies for vector DBs
 Home-page: https://github.com/jaiprakash0217/datasourcelib
 Author: Jai Prakash

{datasourcelib-0.1.3.dist-info → datasourcelib-0.1.4.dist-info}/RECORD

@@ -12,12 +12,10 @@ datasourcelib/datasources/sharepoint_source - Copy.py,sha256=7V1c-zyvTo4IuPN_YMr
 datasourcelib/datasources/sharepoint_source.py,sha256=Pv9735Gu2FylVeeT9e_cZlCvgGUwxn-pVRRZQe2PHU8,20196
 datasourcelib/datasources/sql_source.py,sha256=sCYHrmeD82fQVcdQjL9Y2TTTjaqlv2v8B5noAng3Bl4,5450
 datasourcelib/indexes/__init__.py,sha256=S8dz-lyxy1BTuDuLGRJNLrZD_1ku_FIUnDEm6HhMyT0,94
-datasourcelib/indexes/azure_search_index.py,sha256=o3BoSxURBk5jCC3AlNz-v9_igg-dXYS4yUxXZwSfqFg,17265
-datasourcelib/indexes/azure_search_index_only.py,sha256=SulrYPehWGaf3Wi_Dw8UvFneSY-UwEK9viVYXwIlQuI,7120
-datasourcelib/indexes/azure_search_index_vector.py,sha256=4By1vJHv1ORiWOpTqO5wR0sTrq1TaEHP6t8MoOINhok,13410
+datasourcelib/indexes/azure_search_index.py,sha256=kznAz06UXgyT1Clqj6gRhnBQ5HFw40ZQHJElRFIcbRo,22115
 datasourcelib/strategies/__init__.py,sha256=kot3u62KIAqYBg9M-KRE4mkMII_zwrDBZNf8Dj1vmX8,399
 datasourcelib/strategies/daily_load.py,sha256=Rh-veUhxKYsplwHTyko_Zp9C6NkUJV5VAGtg-p7Iy34,856
-datasourcelib/strategies/full_load.py,sha256=zqDZZcmyJKXQ4v3coq5njjadlBNI9V8f_lfXVZCoLbQ,1698
+datasourcelib/strategies/full_load.py,sha256=U1a9wO_ZLRnMInvU0IRW-ZKnhu0Cv437VcNMKIYuzMA,1691
 datasourcelib/strategies/incremental_load.py,sha256=TVqmDLu3m571nqGvzo_69i36QtYe4sBpllFwfPNL0TE,1178
 datasourcelib/strategies/ondemand_load.py,sha256=VxzAYgrW2ebTOC3xm61CerL2AFehZUJLnKrqtGRGJoE,644
 datasourcelib/strategies/timerange_load.py,sha256=c62BN2yXwVFaA_dQV54qenP4vrb4rcFqbx6m-nqhaTA,900
@@ -27,8 +25,8 @@ datasourcelib/utils/exceptions.py,sha256=mgcDaW1k3VndgpMOwSm7NqgyRTvvE2a5ehn3x4f
 datasourcelib/utils/file_reader.py,sha256=Zr0rwNTRWE6KeVJEXgTOPS1_JI74LiUSiX5-6qojmN0,7301
 datasourcelib/utils/logger.py,sha256=Sl6lNlvubxtK9ztzyq7vjGVyA8_-pZ_ixpk5jfVsh6U,424
 datasourcelib/utils/validators.py,sha256=fLgmRAb5OZSdMVlHu_n0RKJUDl-G8dI8JsRSfxIquh8,205
-datasourcelib-0.1.3.dist-info/licenses/LICENSE,sha256=9S0AcKETmp9XOcC73jEjN7WSkuSWGFGreiBat6ONClo,1087
-datasourcelib-0.1.3.dist-info/METADATA,sha256=cPVrPEkPN22sTYOoO20byXcpu5hvKVQIPu3elgyyEko,1185
-datasourcelib-0.1.3.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-datasourcelib-0.1.3.dist-info/top_level.txt,sha256=wIwiwdIj8T9pAvE2TkGLUvT2oIi43C2vkkTKibUlv3U,14
-datasourcelib-0.1.3.dist-info/RECORD,,
+datasourcelib-0.1.4.dist-info/licenses/LICENSE,sha256=9S0AcKETmp9XOcC73jEjN7WSkuSWGFGreiBat6ONClo,1087
+datasourcelib-0.1.4.dist-info/METADATA,sha256=LR3db7O_rnbTmF_owLl-lH06xAfP-iZu4aXPtmjVtRo,1185
+datasourcelib-0.1.4.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+datasourcelib-0.1.4.dist-info/top_level.txt,sha256=wIwiwdIj8T9pAvE2TkGLUvT2oIi43C2vkkTKibUlv3U,14
+datasourcelib-0.1.4.dist-info/RECORD,,

datasourcelib/indexes/azure_search_index_only.py

@@ -1,162 +0,0 @@
-from typing import List, Dict, Any, Optional
-from datasourcelib.utils.logger import get_logger
-
-logger = get_logger(__name__)
-
-class AzureSearchIndexer:
-    """
-    Minimal Azure Cognitive Search indexer wrapper.
-    Expects vector_db_config with:
-      - service_endpoint: str
-      - index_name: str
-      - api_key: str
-    Optional:
-      - key_field: name of unique key in documents (default 'id')
-    """
-
-    def __init__(self, vector_db_config: Dict[str, Any]):
-        self.config = vector_db_config or {}
-        self._client = None
-        self._index_client = None
-
-    def validate_config(self) -> bool:
-        required = ("aisearch_endpoint", "aisearch_index_name", "aisearch_api_key")
-        missing = [k for k in required if k not in self.config]
-        if missing:
-            logger.error("AzureSearchIndexer.validate_config missing: %s", missing)
-            return False
-        return True
-    
-    def _ensure_sdk(self):
-        try:
-            from azure.core.credentials import AzureKeyCredential  # type: ignore
-            from azure.search.documents import SearchClient  # type: ignore
-            from azure.search.documents.indexes import SearchIndexClient  # type: ignore
-            from azure.search.documents.indexes.models import (
-                SearchIndex,
-                SimpleField,
-                SearchableField,
-                SearchFieldDataType,
-            )  # type: ignore
-        except Exception as e:
-            raise RuntimeError("azure-search-documents package is required: install azure-search-documents") from e
-
-        return AzureKeyCredential, SearchClient, SearchIndexClient, SearchIndex, SimpleField, SearchableField, SearchFieldDataType
-    
-    def _infer_field_type(self, value) -> Any:
-        """
-        Map Python types to SearchFieldDataType
-        """
-        *_, SearchFieldDataType = self._ensure_sdk()
-        if value is None:
-            return SearchFieldDataType.String
-        t = type(value)
-        if t is str:
-            return SearchFieldDataType.String
-        if t is bool:
-            return SearchFieldDataType.Boolean
-        if t is int:
-            return SearchFieldDataType.Int32
-        if t is float:
-            return SearchFieldDataType.Double
-        # fallback to string
-        return SearchFieldDataType.String
-
-    def _build_fields(self, sample: Dict[str, Any], key_field: str):
-        AzureKeyCredential, SearchClient, SearchIndexClient, SearchIndex, SimpleField, SearchableField, SearchFieldDataType = self._ensure_sdk()
-        
-        fields = []
-        # ensure key field present
-        if key_field not in sample:
-            # we'll create a string key, uploader will populate unique ids
-            fields.append(SimpleField(name=key_field, type=SearchFieldDataType.String, key=True))
-        else:
-            typ = self._infer_field_type(sample[key_field])
-            fields.append(SimpleField(name=key_field, type=SearchFieldDataType.String, key=True))
-
-        for k, v in sample.items():
-            logger.info(f"================={k}============")
-            if k == key_field:
-                continue
-            typ = self._infer_field_type(v)
-            # for strings use SearchableField so full text queries work
-            if typ == SearchFieldDataType.String:
-                fields.append(SearchableField(name=k, type=SearchFieldDataType.String))
-            else:
-                fields.append(SimpleField(name=k, type=typ))
-        return fields
-
-    def create_index(self, sample: Dict[str, Any]) -> bool:
-        try:
-            AzureKeyCredential, SearchClient, SearchIndexClient, SearchIndex, SimpleField, SearchableField, SearchFieldDataType = self._ensure_sdk()
-            endpoint = self.config["aisearch_endpoint"]
-            api_key = self.config["aisearch_api_key"]
-            index_name = self.config["aisearch_index_name"]
-            key_field = self.config.get("key_field", "id")
-
-            index_client = SearchIndexClient(endpoint, AzureKeyCredential(api_key))
-            fields = self._build_fields(sample, key_field)
-            logger.info("=================Creating Index============")
-            index = SearchIndex(name=index_name, fields=fields)
-            # create or update index
-            index_client.create_or_update_index(index)
-            logger.info("Azure Search index '%s' created/updated", index_name)
-            return True
-        except Exception as ex:
-            logger.exception("AzureSearchIndexer.create_index failed")
-            return False
-
-    def upload_documents(self, docs: List[Dict[str, Any]]) -> bool:
-        try:
-            AzureKeyCredential, SearchClient, SearchIndexClient, SearchIndex, SimpleField, SearchableField, SearchFieldDataType = self._ensure_sdk()
-            endpoint = self.config["aisearch_endpoint"]
-            api_key = self.config["aisearch_api_key"]
-            index_name = self.config["aisearch_index_name"]
-            key_field = self.config.get("key_field", "id")
-
-            # ensure each doc has key_field
-            from uuid import uuid4
-            for d in docs:
-                if key_field not in d:
-                    d[key_field] = str(uuid4())
-            # ensure each doc has key_field is of string type
-            for d in docs:
-                if key_field in d:                    
-                    typ = self._infer_field_type(d[key_field])
-                    if typ != SearchFieldDataType.String:
-                        d[key_field] = str(d[key_field])
-
-            client = SearchClient(endpoint=endpoint, index_name=index_name, credential=AzureKeyCredential(api_key))
-            logger.info("Uploading %d documents to index %s", len(docs), index_name)
-            result = client.upload_documents(documents=docs)
-            # Check results for failures
-            failed = [r for r in result if not r.succeeded]
-            if failed:
-                logger.error("Some documents failed to upload: %s", failed)
-                return False
-            logger.info("Uploaded documents successfully")
-            return True
-        except Exception:
-            logger.exception("AzureSearchIndexer.upload_documents failed")
-            return False
-
-    def index(self, rows: List[Dict[str, Any]]) -> bool:
-        """
-        High level: create index (based on first row) and upload all rows.
-        """
-        if not rows:
-            logger.error("AzureSearchIndexer.index called with empty rows")
-            return False
-        try:
-            if not self.validate_config():
-                return False
-            sample = rows[0]
-            logger.info(f"================={sample}============")
-            ok = self.create_index(sample)
-            if not ok:
-                return False
-            ok2 = self.upload_documents(rows)
-            return ok2
-        except Exception:
-            logger.exception("AzureSearchIndexer.index failed")
-            return False

datasourcelib/indexes/azure_search_index_vector.py

@@ -1,286 +0,0 @@
-from typing import List, Dict, Any, Optional
-from datasourcelib.utils.logger import get_logger
-
-logger = get_logger(__name__)
-
-class AzureSearchIndexer:
-    """
-    Azure Cognitive Search indexer with vector search support.
-    Required vector_db_config:
-      - aisearch_endpoint: str
-      - aisearch_index_name: str
-      - aisearch_api_key
-    
-    Optional vector search config:
-      - vectorization: bool (enable vector search)
-      - vector_config: dict
-        - dimensions: int (default 1024)
-        - algorithm: str ('hnsw' or 'flat', default 'hnsw')
-        - metric: str ('cosine', 'euclidean', 'dotProduct', default 'cosine')
-      - key_field: str (default 'id')
-      - vector_field: str (default 'contentVector')
-      - embedding_endpoint: str (Azure OpenAI endpoint for embeddings)
-      - embedding_key: str (Azure OpenAI API key)
-      - embedding_deployment: str (Azure OpenAI model deployment name)
-    """
-
-    def __init__(self, vector_db_config: Dict[str, Any]):
-        self.config = vector_db_config or {}
-        self._client = None
-        self._index_client = None
-        self._embedding_client = None
-
-    def validate_config(self) -> bool:
-        required = ("aisearch_endpoint", "aisearch_index_name", "aisearch_api_key")
-        missing = [k for k in required if k not in self.config]
-        
-        # Check vector search requirements if enabled
-        if self.config.get("vectorization", False):
-            vector_required = ("embedding_endpoint", "embedding_key", "embedding_deployment")
-            missing.extend([k for k in vector_required if k not in self.config])
-        
-        if missing:
-            logger.error("AzureSearchIndexer.validate_config missing: %s", missing)
-            return False
-        return True
-
-    def _ensure_sdk(self):
-        try:
-            from azure.core.credentials import AzureKeyCredential  # type: ignore
-            from azure.search.documents import SearchClient  # type: ignore
-            from azure.search.documents.indexes import SearchIndexClient  # type: ignore
-            from openai import AzureOpenAI # type: ignore
-            from azure.search.documents.indexes.models import (
-                SearchIndex,
-                SearchField,
-                SearchFieldDataType,
-                SimpleField,
-                SearchableField,
-                VectorSearch,
-                VectorSearchProfile,
-                HnswAlgorithmConfiguration
-            ) # type: ignore
-            
-        except Exception as e:
-            raise RuntimeError("Required packages missing. Install: azure-search-documents openai") from e
-
-        return (
-            AzureKeyCredential, SearchClient, SearchIndexClient, AzureOpenAI, SearchIndex, SearchField, SearchFieldDataType, SimpleField, SearchableField, VectorSearch, VectorSearchProfile, HnswAlgorithmConfiguration
-        )
-
-    def _setup_embedding_client(self):
-        if not self._embedding_client and self.config.get("vectorization"):
-            try:
-                AzureKeyCredential, SearchClient, SearchIndexClient, AzureOpenAI, SearchIndex, SearchField, SearchFieldDataType, SimpleField, SearchableField, VectorSearch, VectorSearchProfile, HnswAlgorithmConfiguration = self._ensure_sdk()
-                self._embedding_client = AzureOpenAI(
-                    api_version=self.config["embedding_api_version"],
-                    azure_endpoint=self.config["embedding_endpoint"],
-                    api_key=self.config["embedding_key"],
-                )
-                logger.info("Azure OpenAI embedding client initialized")
-            except Exception as ex:
-                logger.exception("Failed to initialize embedding client")
-                raise
-
-    def _get_embeddings(self, text: str) -> List[float]:
-        try:
-            self._setup_embedding_client()
-            response = self._embedding_client.embeddings.create(
-                model=self.config["embedding_deployment"],
-                input=text
-            )
-            return response.data[0].embedding
-        except Exception as ex:
-            logger.exception(f"Failed to get embeddings for text: {text[:100]}...")
-            raise
-
-    def _build_vector_search_config(self):
-        AzureKeyCredential, SearchClient, SearchIndexClient, AzureOpenAI, SearchIndex, SearchField, SearchFieldDataType, SimpleField, SearchableField, VectorSearch, VectorSearchProfile, HnswAlgorithmConfiguration = self._ensure_sdk()
-        vector_config = self.config.get("vector_config", {})
-        dimensions = vector_config.get("dimensions", 1536)
-             
-        vector_search = VectorSearch(
-                profiles=[VectorSearchProfile(name="vector-profile-1", algorithm_configuration_name="algorithms-config-1")],
-                algorithms=[HnswAlgorithmConfiguration(name="algorithms-config-1")]
-            )
-
-        return vector_search, dimensions
-    
-    def _infer_field_type(self, value) -> Any:
-        #Map Python types to SearchFieldDataType, including collections
-
-        AzureKeyCredential, SearchClient, SearchIndexClient, AzureOpenAI, SearchIndex, SearchField, SearchFieldDataType, SimpleField, SearchableField, VectorSearch, VectorSearchProfile, HnswAlgorithmConfiguration = self._ensure_sdk()
-        
-        if value is None:
-            return SearchFieldDataType.String
-            
-        t = type(value)
-        
-        # Handle list/array types as Collections
-        if t in (list, tuple):
-            # If empty list, default to Collection of Double
-            if not value:
-                return SearchFieldDataType.Collection(SearchFieldDataType.Double)
-            # Get type of first element for non-empty lists
-            element_type = self._infer_field_type(value[0])
-            return SearchFieldDataType.Collection(element_type)
-        # Handle vector embeddings (list or tuple of floats)
-        if type(value) in (list, tuple) and all(isinstance(x, (int, float)) for x in value):
-            return SearchFieldDataType.Collection(SearchFieldDataType.Single)
-            
-        # Handle basic types
-        logger.info(f"######## Infer field type for value:[ {value} ] of type [ {t} ]")
-        if t is bool:
-            return SearchFieldDataType.Boolean
-        if t is int:
-            return SearchFieldDataType.Int32
-        if t is float:
-            return SearchFieldDataType.Double
-        print(f"############## Infer field type for value: {value} of type {t}")
-        print(t is str)
-        if t is str:
-            return SearchFieldDataType.String
-        # fallback to string
-        logger.warning(f"Falling back to string type for value: {value} of type {t}")
-        return SearchFieldDataType.String
-        
-    def _build_fields(self, sample: Dict[str, Any], key_field: str):
-        AzureKeyCredential, SearchClient, SearchIndexClient, AzureOpenAI, SearchIndex, SearchField, SearchFieldDataType, SimpleField, SearchableField, VectorSearch, VectorSearchProfile, HnswAlgorithmConfiguration = self._ensure_sdk()
-
-        fields = []
-        # Add key field
-        if key_field not in sample:
-            fields.append(SimpleField(name=key_field, type=SearchFieldDataType.String, key=True))
-        else:
-            fields.append(SimpleField(name=key_field, type=SearchFieldDataType.String, key=True))
-
-        # Add regular fields
-        for k, v in sample.items():
-            logger.info(f"================={k}============")
-            if k == key_field:
-                continue
-            logger.info(f"#### Infer field type for field: {k}")
-            typ = self._infer_field_type(v)
-            logger.info(f"#### Inferred type for field {k}: {typ}")
-            if typ == SearchFieldDataType.String:
-                fields.append(SearchableField(name=k, type=SearchFieldDataType.String))
-            else:
-                fields.append(SimpleField(name=k, type=typ))
-
-        # Add vector field if vectorization is enabled
-        if self.config.get("vectorization"):
-            vector_field = self.config.get("vector_field", "contentVector")
-            _, dimensions = self._build_vector_search_config()
-            fields.append(
-                SearchField(
-                    name=vector_field,
-                    type=SearchFieldDataType.Collection(SearchFieldDataType.Single),
-                    searchable=True,
-                    vector_search_dimensions=dimensions,
-                    vector_search_profile_name="vector-profile-1"
-                )
-            )
-
-        return fields
-
-    def create_index(self, sample: Dict[str, Any]) -> bool:
-        try:
-            AzureKeyCredential, SearchClient, SearchIndexClient, AzureOpenAI, SearchIndex, SearchField, SearchFieldDataType, SimpleField, SearchableField, VectorSearch, VectorSearchProfile, HnswAlgorithmConfiguration = self._ensure_sdk()
-            
-            endpoint = self.config["aisearch_endpoint"]
-            api_key = self.config["aisearch_api_key"]
-            index_name = self.config["aisearch_index_name"]
-            key_field = self.config.get("key_field", "id")
-
-            index_client = SearchIndexClient(endpoint, AzureKeyCredential(api_key))
-            fields = self._build_fields(sample, key_field)
-
-            # Create index with vector search if enabled
-            if self.config.get("vectorization"):
-                vector_search, _ = self._build_vector_search_config()
-                index = SearchIndex(
-                    name=index_name,
-                    fields=fields,
-                    vector_search=vector_search
-                )
-            else:
-                index = SearchIndex(name=index_name, fields=fields)
-
-            index_client.create_or_update_index(index)
-            logger.info(f"Azure Search index '{index_name}' created/updated with vectorization={self.config.get('vectorization', False)}")
-            return True
-        except Exception as ex:
-            logger.exception("AzureSearchIndexer.create_index failed")
-            return False
-
-    def upload_documents(self, docs: List[Dict[str, Any]]) -> bool:
-        try:
-            AzureKeyCredential, SearchClient, SearchIndexClient, AzureOpenAI, SearchIndex, SearchField, SearchFieldDataType, SimpleField, SearchableField, VectorSearch, VectorSearchProfile, HnswAlgorithmConfiguration = self._ensure_sdk()
-            endpoint = self.config["aisearch_endpoint"]
-            api_key = self.config["aisearch_api_key"]
-            index_name = self.config["aisearch_index_name"]
-            key_field = self.config.get("key_field", "id")
-            
-            # Add IDs if missing
-            from uuid import uuid4
-            for d in docs:
-                if key_field not in d:
-                    d[key_field] = str(uuid4())
-                elif not isinstance(d[key_field], str):
-                    d[key_field] = str(d[key_field])
-
-            # Add vector embeddings if enabled
-            if self.config.get("vectorization"):
-                vector_field = self.config.get("vector_field", "contentVector")
-                content_field = self.config.get("content_field", "content")
-                
-                for doc in docs:
-                    if content_field in doc:
-                        try:
-                            embedding = self._get_embeddings(str(doc[content_field]))
-                            doc[vector_field] = embedding
-                        except Exception as e:
-                            logger.error(f"Failed to get embedding for document {doc.get(key_field)}: {str(e)}")
-                            continue
-
-            client = SearchClient(endpoint=endpoint, index_name=index_name, 
-                               credential=AzureKeyCredential(api_key))
-            
-            logger.info(f"Uploading {len(docs)} documents to index {index_name}")
-            result = client.upload_documents(documents=docs)
-            
-            failed = [r for r in result if not r.succeeded]
-            if failed:
-                logger.error(f"Some documents failed to upload: {failed}")
-                return False
-                
-            logger.info("Documents uploaded successfully")
-            return True
-            
-        except Exception:
-            logger.exception("AzureSearchIndexer.upload_documents failed")
-            return False
-
-    def index(self, rows: List[Dict[str, Any]]) -> bool:
-        """High level: create index (based on first row) and upload all rows."""
-        if not rows:
-            logger.error("AzureSearchIndexer.index called with empty rows")
-            return False
-            
-        try:
-            if not self.validate_config():
-                return False
-                
-            sample = rows[0]
-            logger.info(f"Creating/updating index with sample: {sample}")
-            
-            ok = self.create_index(sample)
-            if not ok:
-                return False
-                
-            ok2 = self.upload_documents(rows)
-            return ok2
-            
-        except Exception:
-            logger.exception("AzureSearchIndexer.index failed")
-            return False