Merge pull request #227 from hkad98/data_sources

PSDK-110 PSDK-124 MS SQL and Databricks support

Reviewed-by: https://github.com/pcerny

Author: gdgate

docs/content/en/docs/data/data-source/_index.md

@@ -143,3 +143,40 @@ CatalogDataSourceGreenplum(
     ),
 )
 ```
+
+### Microsoft SQL Server
+
+```python
+CatalogDataSourceMsSql(
+    id=data_source_id,
+    name=data_source_name,
+    db_specific_attributes=MsSqlAttributes(
+        host=os.environ["MSSQL_HOST"],
+        db_name=os.environ["MSSQL_DBNAME"]
+    ),
+    schema=os.environ["MSSQL_SCHEMA"],
+    credentials=BasicCredentials(
+        username=os.environ["MSSQL_USER"],
+        password=os.environ["MSSQL_PASSWORD"],
+    ),
+)
+```
+
+### Databricks
+
+```python
+CatalogDataSourceDatabricks(
+    id=data_source_id,
+    name=data_source_name,
+    db_specific_attributes=DatabricksAttributes(
+        host=os.environ["DATABRICKS_HOST"],
+        http_path=os.environ["DATABRICKS_HTTP_PATH"]
+    ),
+    schema=xyz,
+    parameters=[{"name":"catalog", "value": os.environ["DATABRICKS_CATALOG"]}],
+    credentials=BasicCredentials(
+        username=os.environ["DATABRICKS_USER"],
+        password=os.environ["DATABRICKS_PASSWORD"],
+    ),
+)
+```

gooddata-sdk/gooddata_sdk/__init__.py

@@ -29,12 +29,16 @@
 from gooddata_sdk.catalog.data_source.entity_model.data_source import (
     CatalogDataSource,
     CatalogDataSourceBigQuery,
+    CatalogDataSourceDatabricks,
     CatalogDataSourceGreenplum,
+    CatalogDataSourceMsSql,
     CatalogDataSourcePostgres,
     CatalogDataSourceRedshift,
     CatalogDataSourceSnowflake,
     CatalogDataSourceVertica,
+    DatabricksAttributes,
     GreenplumAttributes,
+    MsSqlAttributes,
     PostgresAttributes,
     RedshiftAttributes,
     SnowflakeAttributes,

gooddata-sdk/gooddata_sdk/catalog/data_source/entity_model/data_source.py

@@ -206,3 +206,41 @@ class CatalogDataSourceBigQuery(CatalogDataSource):
 class CatalogDataSourceGreenplum(CatalogDataSourcePostgres):
     type: str = "GREENPLUM"
     db_vendor: str = "postgresql"
+
+
+@attr.s(auto_attribs=True, kw_only=True)
+class MsSqlAttributes(DatabaseAttributes):
+    host: str
+    db_name: str
+    port: str = "1433"
+
+
+@attr.s(auto_attribs=True, kw_only=True)
+class CatalogDataSourceMsSql(CatalogDataSource):
+    _URL_TMPL: ClassVar[str] = "jdbc:{db_vendor}://{host}:{port};databaseName={db_name}"
+    type: str = "MSSQL"
+    db_vendor: str = "sqlserver"
+    db_specific_attributes: MsSqlAttributes
+
+
+@attr.s(auto_attribs=True, kw_only=True)
+class DatabricksAttributes(DatabaseAttributes):
+    host: str
+    http_path: str
+    port: str = "443"
+
+
+@attr.s(auto_attribs=True, kw_only=True)
+class CatalogDataSourceDatabricks(CatalogDataSource):
+    _URL_TMPL: ClassVar[str] = "jdbc:{db_vendor}://{host}:{port}/default;httpPath={http_path}"
+    type: str = "DATABRICKS"
+    parameters: List[Dict[str, str]]
+    db_specific_attributes: DatabricksAttributes
+
+    def __attrs_post_init__(self) -> None:
+        mandatory_parameter = [parameter.get("name") == "catalog" for parameter in self.parameters]
+        if not any(mandatory_parameter):
+            raise ValueError(f"'catalog' is mandatory parameter for data source type {self.type}")
+
+        self.db_vendor = self.type.lower()
+        self.url = self._make_url()

gooddata-sdk/tests/catalog/test_catalog_data_source.py

@@ -15,6 +15,8 @@
     BasicCredentials,
     CatalogDataSource,
     CatalogDataSourceBigQuery,
+    CatalogDataSourceDatabricks,
+    CatalogDataSourceMsSql,
     CatalogDataSourcePostgres,
     CatalogDataSourceRedshift,
     CatalogDataSourceSnowflake,
@@ -25,9 +27,11 @@
     CatalogPdmLdmRequest,
     CatalogPdmSql,
     CatalogScanModelRequest,
+    DatabricksAttributes,
     ExecutionDefinition,
     GoodDataApiClient,
     GoodDataSdk,
+    MsSqlAttributes,
     PostgresAttributes,
     RedshiftAttributes,
     ScanSqlRequest,
@@ -746,3 +750,32 @@ def test_allowed_data_source_type(test_config):
         pass
     else:
         assert False, "ValueError was not raised for nonsense database type"
+
+
+def test_catalog_data_source_mssql(test_config):
+    data_source = CatalogDataSourceMsSql(
+        id="test",
+        name="Test",
+        db_specific_attributes=MsSqlAttributes(host="Host", db_name="DbName"),
+        schema="Schema",
+        credentials=BasicCredentials(
+            username="demouser",
+            password="demopass",
+        ),
+    )
+    assert data_source.url == "jdbc:sqlserver://Host:1433;databaseName=DbName"
+
+
+def test_catalog_data_source_databricks(test_config):
+    data_source = CatalogDataSourceDatabricks(
+        id="test",
+        name="Test",
+        db_specific_attributes=DatabricksAttributes(host="Host", http_path="xyz123abc"),
+        schema="SCHEMA",
+        parameters=[{"name": "catalog", "value": "super_catalog"}],
+        credentials=BasicCredentials(
+            username="demouser",
+            password="demospass",
+        ),
+    )
+    assert data_source.url == "jdbc:databricks://Host:443/default;httpPath=xyz123abc"