Merge pull request #24 from voi-oss/tableau-exposure-crawler-refactor

chore(tableau-exposer-crawler): using batch API to fetch data

src/exposurescrawler/crawlers/tableau.py

@@ -23,7 +23,7 @@ def _should_ignore_workbook(workbook, projects_to_ignore: Collection[str]) -> bo
     # by workbooks under projects without a name.
     if not workbook.project_name:
         return True
-    
+
     return workbook.project_name in projects_to_ignore
 
 
@@ -41,7 +41,6 @@ def _parse_tables_from_sql(workbooks_sqls: WorkbookModelsMapping, models) -> Wor
     logger().info('⚙️ Parsing SQL: looking for references to models')
 
     output: WorkbookModelsMapping = {}
-
     for workbook_reference, custom_sqls in workbooks_sqls.items():
         # a list of dbt model represented as their original dicts from the manifest
         all_found: List[dict] = []
@@ -66,11 +65,38 @@ def _parse_tables_from_sql(workbooks_sqls: WorkbookModelsMapping, models) -> Wor
     return output
 
 
+def retrieve_all_workbook_owner_map(tableau_client: TableauRestClient):
+    """
+
+    :param tableau_client: Tableau rest client
+    :return: the dictionary of {workbook_id, WorkbookItem}
+    """
+    logger().info('⚙️ Retrieving all workbooks (batch)')
+
+    all_workbooks = tableau_client.retrieve_all_workbooks()
+    logger().info(f'✅ Fetched {len(all_workbooks)} workbooks')
+    return dict((workbook.id, workbook) for workbook in all_workbooks)
+
+
+def retrieve_all_user_id_map(tableau_client: TableauRestClient):
+    """
+
+    :param tableau_client: Tableau rest client
+    :return: the dictionary of {user_id, UserItem}
+    """
+    logger().info('⚙️ Retrieving all users (batch)')
+
+    all_users = tableau_client.retrieve_all_users()
+    logger().info(f'⚙️ Fetched {len(all_users)} users')
+
+    return dict((user.id, user) for user in all_users)
+
+
 def tableau_crawler(
-    manifest_path: str,
-    dbt_package_name: str,
-    tableau_projects_to_ignore: Collection[str],
-    verbose: bool,
+        manifest_path: str,
+        dbt_package_name: str,
+        tableau_projects_to_ignore: Collection[str],
+        verbose: bool,
 ) -> None:
     # Enable verbose logging
     if verbose:
@@ -107,7 +133,7 @@ def tableau_crawler(
     workbooks_models: WorkbookModelsMapping = {}
 
     for workbook_reference, found in itertools.chain(
-        workbooks_custom_sql_models.items(), workbooks_native_sql_models.items()
+            workbooks_custom_sql_models.items(), workbooks_native_sql_models.items()
     ):
         workbooks_models.setdefault(workbook_reference, []).extend(found)
 
@@ -123,12 +149,15 @@ def tableau_crawler(
     logger().info('')
     logger().info('🌏 Retrieving workbooks and authors metadata from the Tableau REST API')
 
+    # Fetching all workbooks and users using Tableau batch API and keep in a dictionary.
+    workbook_owner_map = retrieve_all_workbook_owner_map(tableau_client)
+    user_userid_map = retrieve_all_user_id_map(tableau_client)
+
     # For every workbook and the models found, create exposures and add
     # to the manifest (in-memory)
     for workbook_reference, found in workbooks_models.items():
-        workbook = tableau_client.retrieve_workbook(workbook_reference.id)
-        owner = tableau_client.retrieve_user(workbook.owner_id)
-
+        workbook = workbook_owner_map[workbook_reference.id]
+        owner = user_userid_map[workbook.owner_id]
         if _should_ignore_workbook(workbook, tableau_projects_to_ignore):
             logger().debug(
                 f'⏩ Skipping workbook: {workbook.name} ({workbook.project_name} is ignored)'
@@ -137,7 +166,6 @@ def tableau_crawler(
 
         exposure = DbtExposure.from_tableau_workbook(dbt_package_name, workbook, owner, found)
         manifest.add_exposure(exposure, found)
-
     # Persist the modified manifest
     logger().info('')
     logger().info(f'💾 Writing results to file: {manifest_path}')
@@ -156,7 +184,7 @@ def tableau_crawler(
     required=True,
     metavar='PROJECT_NAME',
     help='The name of the dbt pacakge where the exposures should be added. If in doubt, check the '
-    'name of your dbt project on dbt_project.yml',
+         'name of your dbt project on dbt_project.yml',
 )
 @click.option(
     '--tableau-ignore-projects',
@@ -166,10 +194,10 @@ def tableau_crawler(
 )
 @click.option('-v', '--verbose', is_flag=True, default=False, help='Enable verbose logging')
 def tableau_crawler_command(
-    manifest_path: str,
-    dbt_package_name: str,
-    tableau_projects_to_ignore: Collection[str],
-    verbose: bool,
+        manifest_path: str,
+        dbt_package_name: str,
+        tableau_projects_to_ignore: Collection[str],
+        verbose: bool,
 ):
     tableau_crawler(manifest_path, dbt_package_name, tableau_projects_to_ignore, verbose)
 

src/exposurescrawler/tableau/rest_client.py

@@ -30,3 +30,15 @@ def run_metadata_api(self, query: str):
             response = self.server.metadata.query(query)
 
         return response['data']
+
+    def retrieve_all_workbooks(self):
+        with self.server.auth.sign_in(self.tableau_auth):
+            all_workbooks = list(TSC.Pager(self.server.workbooks))
+
+        return all_workbooks
+
+    def retrieve_all_users(self):
+        with self.server.auth.sign_in(self.tableau_auth):
+            all_users = list(TSC.Pager(self.server.users))
+
+        return all_users

tests/_integration/test_tableau_crawler.py

@@ -45,7 +45,7 @@ def mock_graphql_custom_sql_result():
                 },
                 {
                     'query': "select * from sample_db.public.customers left join "
-                    "sample_db.public.orders on customers.id = orders.customer_id",
+                             "sample_db.public.orders on customers.id = orders.customer_id",
                     'name': 'Custom SQL Query',
                     'isEmbedded': None,
                     'database': {'name': 'SAMPLE_DB', 'connectionType': 'snowflake'},
@@ -150,22 +150,31 @@ def mock_tableau_rest_api():
     UserDetailsMock = namedtuple('UserDetailsMock', ['id', 'fullname', 'name'])
 
     workbook_details = {
-        'customers-workbook-luid': WorkbookDetailsMock(id='aaa', name='Customers workbook'),
-        'company-kpis-workbook-luid': WorkbookDetailsMock(id='bbb', name='Company KPIs workbook'),
+        'customers-workbook-luid': WorkbookDetailsMock(id='customers-workbook-luid', name='Customers workbook',
+                                                       owner_id='user-id'),
+        'company-kpis-workbook-luid': WorkbookDetailsMock(id='company-kpis-workbook-luid', name='Company KPIs workbook',
+                                                          owner_id='user-id'),
         'orders-workbook-luid': WorkbookDetailsMock(
-            id='ccc', name='Orders workbook', tags=['certified']
+            id='orders-workbook-luid', name='Orders workbook', tags=['certified'], owner_id='user-id'
         ),
     }
+    user_details = {
+        'user-id': UserDetailsMock('user-id', 'John Doe', 'john.doe@example.com')
+    }
 
     def _get_workbook_details(workbook_id):
         return workbook_details[workbook_id]
 
     with patch('exposurescrawler.crawlers.tableau.TableauRestClient', autospec=True) as mock:
         instance = mock.return_value
         instance.retrieve_workbook.side_effect = _get_workbook_details
-        instance.retrieve_user.return_value = UserDetailsMock(
-            'user-id', 'John Doe', 'john.doe@example.com'
-        )
+        instance.retrieve_user.return_value = user_details['user-id']
+        instance.retrieve_all_users.return_value = [user_details['user-id']]
+        instance.retrieve_all_workbooks.return_value = [
+            workbook_details['customers-workbook-luid'],
+            workbook_details['company-kpis-workbook-luid'],
+            workbook_details['orders-workbook-luid']
+        ]
         yield
 
 
@@ -183,10 +192,10 @@ def test_tableau_crawler(manifest_path):
         tableau_crawler(manifest_path, 'jaffle_shop', [], True)
 
         final_manifest = mock.call_args.args[0].data
-        exposure = final_manifest['exposures']['exposure.jaffle_shop.tableau_orders_workbook_ccc']
+        exposure = final_manifest['exposures']['exposure.jaffle_shop.tableau_orders_workbook_ord']
 
         assert len(final_manifest['exposures']) == 3
-        assert exposure['name'] == 'tableau_orders_workbook_ccc'
+        assert exposure['name'] == 'tableau_orders_workbook_ord'
         assert 'Workbook description' in exposure['description']
         assert 'https://my-tableau-server.com/path/to/workbook' in exposure['description']
         assert exposure['type'] == 'Dashboard'