feat(ingest/metabase): Allow to configure a different url for links

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

@@ -6,7 +6,7 @@
 import dateutil.parser as dp
 import pydantic
 import requests
-from pydantic import Field, validator
+from pydantic import Field, root_validator, validator
 from requests.models import HTTPError
 
 import datahub.emitter.mce_builder as builder
@@ -53,6 +53,10 @@ class MetabaseConfig(DatasetLineageProviderConfigBase):
     # See the Metabase /api/session endpoint for details
     # https://www.metabase.com/docs/latest/api-documentation.html#post-apisession
     connect_uri: str = Field(default="localhost:3000", description="Metabase host URL.")
+    display_uri: Optional[str] = Field(
+        default=None,
+        description="optional URL to use in links (if `connect_uri` is only for ingestion)",
+    )
     username: Optional[str] = Field(default=None, description="Metabase username.")
     password: Optional[pydantic.SecretStr] = Field(
         default=None, description="Metabase password."
@@ -76,10 +80,17 @@ class MetabaseConfig(DatasetLineageProviderConfigBase):
         description="Default schema name to use when schema is not provided in an SQL query",
     )
 
-    @validator("connect_uri")
+    @validator("connect_uri", "display_uri")
     def remove_trailing_slash(cls, v):
         return config_clean.remove_trailing_slashes(v)
 
+    @root_validator(skip_on_failure=True)
+    def default_display_uri_to_connect_uri(cls, values):
+        base = values.get("display_uri")
+        if base is None:
+            values["display_uri"] = values.get("connect_uri")
+        return values
+
 
 @platform_name("Metabase")
 @config_class(MetabaseConfig)
@@ -239,7 +250,7 @@ def construct_dashboard_from_api_data(
         self, dashboard_info: dict
     ) -> Optional[DashboardSnapshot]:
         dashboard_id = dashboard_info.get("id", "")
-        dashboard_url = f"{self.config.connect_uri}/api/dashboard/{dashboard_id}"
+        dashboard_url = f"{self.config.display_uri}/api/dashboard/{dashboard_id}"
         try:
             dashboard_response = self.session.get(dashboard_url)
             dashboard_response.raise_for_status()
@@ -297,7 +308,7 @@ def construct_dashboard_from_api_data(
 
     @lru_cache(maxsize=None)
     def _get_ownership(self, creator_id: int) -> Optional[OwnershipClass]:
-        user_info_url = f"{self.config.connect_uri}/api/user/{creator_id}"
+        user_info_url = f"{self.config.display_uri}/api/user/{creator_id}"
         try:
             user_info_response = self.session.get(user_info_url)
             user_info_response.raise_for_status()
@@ -362,7 +373,7 @@ def get_card_details_by_id(self, card_id: Union[int, str]) -> dict:
         :param int datasource_id: Numeric datasource ID received from Metabase API
         :return: dict with info or empty dict
         """
-        card_url = f"{self.config.connect_uri}/api/card/{card_id}"
+        card_url = f"{self.config.display_uri}/api/card/{card_id}"
         try:
             card_response = self.session.get(card_url)
             card_response.raise_for_status()

metadata-ingestion/tests/unit/test_metabase_source.py

@@ -40,3 +40,12 @@ def test_get_platform_instance():
 
     # database_id_to_instance_map is missing, platform_instance_map is defined and key missing
     assert metabase.get_platform_instance("missing-platform", 999) is None
+
+
+def test_set_display_uri():
+    display_uri = "some_host:1234"
+
+    config = MetabaseConfig.parse_obj({"display_uri": display_uri})
+
+    assert config.connect_uri == "localhost:3000"
+    assert config.display_uri == display_uri