fix(tableau): fixes wrong argument when reauthenticating (#12216)

Author: sgomezvillamor

metadata-ingestion/src/datahub/ingestion/source/tableau/tableau.py

@@ -621,6 +621,12 @@ def update_table(
                 self.parsed_columns = parsed_columns
 
 
+@dataclass
+class SiteIdContentUrl:
+    site_id: str
+    site_content_url: str
+
+
 class TableauSourceReport(StaleEntityRemovalSourceReport):
     get_all_datasources_query_failed: bool = False
     num_get_datasource_query_failures: int = 0
@@ -773,7 +779,6 @@ def get_workunits_internal(self) -> Iterable[MetadataWorkUnit]:
                         config=self.config,
                         ctx=self.ctx,
                         site=site,
-                        site_id=site.id,
                         report=self.report,
                         server=self.server,
                         platform=self.platform,
@@ -792,8 +797,11 @@ def get_workunits_internal(self) -> Iterable[MetadataWorkUnit]:
                 site_source = TableauSiteSource(
                     config=self.config,
                     ctx=self.ctx,
-                    site=site,
-                    site_id=self.server.site_id,
+                    site=site
+                    if site
+                    else SiteIdContentUrl(
+                        site_id=self.server.site_id, site_content_url=self.config.site
+                    ),
                     report=self.report,
                     server=self.server,
                     platform=self.platform,
@@ -826,8 +834,7 @@ def __init__(
         self,
         config: TableauConfig,
         ctx: PipelineContext,
-        site: Optional[SiteItem],
-        site_id: Optional[str],
+        site: Union[SiteItem, SiteIdContentUrl],
         report: TableauSourceReport,
         server: Server,
         platform: str,
@@ -838,13 +845,18 @@ def __init__(
         self.ctx: PipelineContext = ctx
         self.platform = platform
 
-        self.site: Optional[SiteItem] = site
-        if site_id is not None:
-            self.site_id: str = site_id
+        self.site: Optional[SiteItem] = None
+        if isinstance(site, SiteItem):
+            self.site = site
+            assert site.id is not None, "Site ID is required"
+            self.site_id = site.id
+            self.site_content_url = site.content_url
+        elif isinstance(site, SiteIdContentUrl):
+            self.site = None
+            self.site_id = site.site_id
+            self.site_content_url = site.site_content_url
         else:
-            assert self.site is not None, "site or site_id is required"
-            assert self.site.id is not None, "site_id is required when site is provided"
-            self.site_id = self.site.id
+            raise AssertionError("site or site id+content_url pair is required")
 
         self.database_tables: Dict[str, DatabaseTable] = {}
         self.tableau_stat_registry: Dict[str, UsageStat] = {}
@@ -898,16 +910,14 @@ def dataset_browse_prefix(self) -> str:
         # datasets also have the env in the browse path
         return f"/{self.config.env.lower()}{self.no_env_browse_prefix}"
 
-    def _re_authenticate(self):
+    def _re_authenticate(self) -> None:
+        self.report.info(
+            message="Re-authenticating to Tableau",
+            context=f"site='{self.site_content_url}'",
+        )
         # Sign-in again may not be enough because Tableau sometimes caches invalid sessions
         # so we need to recreate the Tableau Server object
-        self.server = self.config.make_tableau_client(self.site_id)
-
-    @property
-    def site_content_url(self) -> Optional[str]:
-        if self.site and self.site.content_url:
-            return self.site.content_url
-        return None
+        self.server = self.config.make_tableau_client(self.site_content_url)
 
     def _populate_usage_stat_registry(self) -> None:
         if self.server is None:

metadata-ingestion/tests/integration/tableau/test_tableau_ingest.py

@@ -26,6 +26,7 @@
 from datahub.ingestion.run.pipeline import Pipeline, PipelineContext
 from datahub.ingestion.source.tableau import tableau_constant as c
 from datahub.ingestion.source.tableau.tableau import (
+    SiteIdContentUrl,
     TableauConfig,
     TableauProject,
     TableauSiteSource,
@@ -1008,8 +1009,7 @@ def check_lineage_metadata(
         config=config,
         ctx=context,
         platform="tableau",
-        site=SiteItem(name="Site 1", content_url="site1"),
-        site_id="site1",
+        site=SiteIdContentUrl(site_id="id1", site_content_url="site1"),
         report=TableauSourceReport(),
         server=Server("https://test-tableau-server.com"),
     )
@@ -1313,8 +1313,7 @@ def test_permission_warning(pytestconfig, tmp_path, mock_datahub_graph):
                 platform="tableau",
                 config=mock.MagicMock(),
                 ctx=mock.MagicMock(),
-                site=mock.MagicMock(),
-                site_id=None,
+                site=mock.MagicMock(spec=SiteItem, id="Site1", content_url="site1"),
                 server=mock_sdk.return_value,
                 report=reporter,
             )
@@ -1371,8 +1370,7 @@ def test_extract_project_hierarchy(extract_project_hierarchy, allowed_projects):
         config=config,
         ctx=context,
         platform="tableau",
-        site=SiteItem(name="Site 1", content_url="site1"),
-        site_id="site1",
+        site=mock.MagicMock(spec=SiteItem, id="Site1", content_url="site1"),
         report=TableauSourceReport(),
         server=Server("https://test-tableau-server.com"),
     )