pyobvector 0.2.19__tar.gz → 0.2.20__tar.gz

{pyobvector-0.2.19 → pyobvector-0.2.20}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pyobvector
-Version: 0.2.19
+Version: 0.2.20
 Summary: A python SDK for OceanBase Vector Store, based on SQLAlchemy, compatible with Milvus API.
 License-File: LICENSE
 Author: shanhaikang.shk
@@ -38,7 +38,7 @@ poetry install
 - install with pip:
 
 ```shell
-pip install pyobvector==0.2.19
+pip install pyobvector==0.2.20
 ```
 
 ## Build Doc
@@ -234,17 +234,17 @@ res = self.client.ann_search(
 The `ann_search` method supports flexible output column selection through the `output_columns` parameter:
 
 - **`output_columns`** (recommended): Accepts SQLAlchemy Column objects, expressions, or a mix of both
-
+  
   - Column objects: `table.c.id`, `table.c.name`
   - Expressions: `(table.c.age + 10).label('age_plus_10')`
   - JSON queries: `text("JSON_EXTRACT(meta, '$.key') as extracted_key")`
   - String functions: `func.concat(table.c.name, ' (', table.c.age, ')').label('name_age')`
 - **`output_column_names`** (legacy): Accepts list of column name strings
-
+  
   - Example: `['id', 'name', 'meta']`
 - **Parameter Priority**: `output_columns` takes precedence over `output_column_names` when both are provided
 - **`distance_threshold`** (optional): Filter results by distance threshold
-
+  
   - Type: `Optional[float]`
   - Only returns results where `distance <= threshold`
   - Example: `distance_threshold=0.5` returns only results with distance <= 0.5
@@ -454,3 +454,4 @@ sql = client.get_sql(index=test_table_name, body=body)
 print(sql)  # prints the SQL query
 ```
 
+

{pyobvector-0.2.19 → pyobvector-0.2.20}/README.md

@@ -15,7 +15,7 @@ poetry install
 - install with pip:
 
 ```shell
-pip install pyobvector==0.2.19
+pip install pyobvector==0.2.20
 ```
 
 ## Build Doc
@@ -211,17 +211,17 @@ res = self.client.ann_search(
 The `ann_search` method supports flexible output column selection through the `output_columns` parameter:
 
 - **`output_columns`** (recommended): Accepts SQLAlchemy Column objects, expressions, or a mix of both
-
+  
   - Column objects: `table.c.id`, `table.c.name`
   - Expressions: `(table.c.age + 10).label('age_plus_10')`
   - JSON queries: `text("JSON_EXTRACT(meta, '$.key') as extracted_key")`
   - String functions: `func.concat(table.c.name, ' (', table.c.age, ')').label('name_age')`
 - **`output_column_names`** (legacy): Accepts list of column name strings
-
+  
   - Example: `['id', 'name', 'meta']`
 - **Parameter Priority**: `output_columns` takes precedence over `output_column_names` when both are provided
 - **`distance_threshold`** (optional): Filter results by distance threshold
-
+  
   - Type: `Optional[float]`
   - Only returns results where `distance <= threshold`
   - Example: `distance_threshold=0.5` returns only results with distance <= 0.5
@@ -430,3 +430,4 @@ You can also get the actual SQL that will be executed:
 sql = client.get_sql(index=test_table_name, body=body)
 print(sql)  # prints the SQL query
 ```
+

{pyobvector-0.2.19 → pyobvector-0.2.20}/pyobvector/client/hybrid_search.py

@@ -30,13 +30,9 @@ class HybridSearch(Client):
         
         if self.ob_version < min_required_version:
             # For versions < 4.4.1.0, check if it's SeekDB
-            with self.engine.connect() as conn:
-                with conn.begin():
-                    res = conn.execute(text("SELECT version()"))
-                    version_str = [r[0] for r in res][0]
-                    if "SeekDB" in version_str:
-                        logger.info(f"SeekDB detected in version string: {version_str}, allowing hybrid search")
-                        return
+            if self._is_seekdb():
+                logger.info("SeekDB detected, allowing hybrid search")
+                return
             raise ClusterVersionException(
                 code=ErrorCode.NOT_SUPPORTED,
                 message=ExceptionsMessage.ClusterVersionIsLow % ("Hybrid Search", "4.4.1.0"),

{pyobvector-0.2.19 → pyobvector-0.2.20}/pyobvector/client/index_param.py

@@ -134,8 +134,11 @@ class IndexParam:
                 if 'efSearch' in params:
                     ob_params['ef_search'] = params['efSearch']
         
-        if self.is_index_type_sparse_vector() and ob_params['distance'] != 'inner_product':
-            raise ValueError("Metric type should be 'inner_product' for sparse vector index.")
+        if self.is_index_type_sparse_vector():
+            if ob_params['distance'] != 'inner_product':
+                raise ValueError("Metric type should be 'inner_product' for sparse vector index.")
+            if 'sparse_index_type' in self.kwargs:
+                ob_params['type'] = self.kwargs['sparse_index_type']
         return ob_params
 
     def param_str(self):

{pyobvector-0.2.19 → pyobvector-0.2.20}/pyobvector/client/ob_client.py

@@ -93,6 +93,26 @@ class ObClient:
             self.metadata_obj.clear()
             self.metadata_obj.reflect(bind=self.engine, extend_existing=True)
 
+    def _is_seekdb(self) -> bool:
+        """Check if the database is SeekDB by querying version.
+        
+        Returns:
+            bool: True if database is SeekDB, False otherwise
+        """
+        is_seekdb = False
+        try:
+            if hasattr(self, '_is_seekdb_cached'):
+                return self._is_seekdb_cached
+            with self.engine.connect() as conn:
+                result = conn.execute(text("SELECT VERSION()"))
+                version_str = [r[0] for r in result][0]
+                is_seekdb = "SeekDB" in version_str
+                self._is_seekdb_cached = is_seekdb
+                logger.debug(f"Version query result: {version_str}, is_seekdb: {is_seekdb}")
+        except Exception as e:
+            logger.warning(f"Failed to query version: {e}")
+        return is_seekdb
+
     def _insert_partition_hint_for_query_sql(self, sql: str, partition_hint: str):
         from_index = sql.find("FROM")
         assert from_index != -1

{pyobvector-0.2.19 → pyobvector-0.2.20}/pyobvector/client/ob_vec_client.py

@@ -99,7 +99,11 @@ class ObVecClient(ObClient):
                     create_table_sql = str(CreateTable(table).compile(self.engine))
                     new_sql = create_table_sql[:create_table_sql.rfind(')')]
                     for sparse_vidx in sparse_vidxs:
-                        new_sql += f",\n\tVECTOR INDEX {sparse_vidx.index_name}({sparse_vidx.field_name}) with (distance=inner_product)"
+                        sparse_params = sparse_vidx._parse_kwargs()
+                        if 'type' in sparse_params:
+                            new_sql += f",\n\tVECTOR INDEX {sparse_vidx.index_name}({sparse_vidx.field_name}) with (type={sparse_params['type']}, distance=inner_product)"
+                        else:
+                            new_sql += f",\n\tVECTOR INDEX {sparse_vidx.index_name}({sparse_vidx.field_name}) with (distance=inner_product)"
                     new_sql += "\n)"
                     conn.execute(text(new_sql))
                 else:

{pyobvector-0.2.19 → pyobvector-0.2.20}/pyobvector/client/ob_vec_json_table_client.py

@@ -817,11 +817,12 @@ class ObVecJsonTableClient(ObVecClient):
     ):
         real_user_id = opt_user_id or self.user_id
 
-        table_name = ast.args['from'].this.this.this
+        from_key = 'from_' if 'from_' in ast.args else 'from'
+        table_name = ast.args[from_key].this.this.this
         if not self._check_table_exists(table_name):
             raise ValueError(f"Table {table_name} does not exists")
         
-        ast.args['from'].args['this'].args['this'] = to_identifier(name=JSON_TABLE_DATA_TABLE_NAME, quoted=False)
+        ast.args[from_key].args['this'].args['this'] = to_identifier(name=JSON_TABLE_DATA_TABLE_NAME, quoted=False)
 
         col_meta = self.jmetadata.meta_cache[table_name]
         json_table_meta_str = []

{pyobvector-0.2.19 → pyobvector-0.2.20}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "pyobvector"
-version = "0.2.19"
+version = "0.2.20"
 description = "A python SDK for OceanBase Vector Store, based on SQLAlchemy, compatible with Milvus API."
 authors = ["shanhaikang.shk <shanhaikang.shk@oceanbase.com>"]
 readme = "README.md"