feat(ingest): add urn -> url helpers (#13410)

metadata-ingestion/setup.py

@@ -147,14 +147,14 @@ great_expectations_lib = {
     "acryl-great-expectations==0.15.50.1",
 }
 
-sql_common_slim = {
+sqlalchemy_lib = {
     # Required for all SQL sources.
     # This is temporary lower bound that we're open to loosening/tightening as requirements show up
     "sqlalchemy>=1.4.39, <2",
 }
 sql_common = (
     {
-        *sql_common_slim,
+        *sqlalchemy_lib,
         # Required for SQL profiling.
         *great_expectations_lib,
         "pydantic<2",  # keeping this for now, but can be removed eventually
@@ -551,7 +551,7 @@ plugins: Dict[str, Set[str]] = {
     "unity-catalog": databricks | sql_common,
     # databricks is alias for unity-catalog and needs to be kept in sync
     "databricks": databricks | sql_common,
-    "fivetran": snowflake_common | bigquery_common | sqlglot_lib,
+    "fivetran": snowflake_common | bigquery_common | sqlalchemy_lib | sqlglot_lib,
     "qlik-sense": sqlglot_lib | {"requests", "websocket-client"},
     "sigma": sqlglot_lib | {"requests"},
     "sac": sac,

metadata-ingestion/src/datahub/ingestion/graph/client.py

@@ -54,6 +54,7 @@ from datahub.ingestion.graph.filters import (
     RemovedStatusFilter,
     generate_filter,
 )
+from datahub.ingestion.graph.links import make_url_for_urn
 from datahub.ingestion.source.state.checkpoint import Checkpoint
 from datahub.metadata.com.linkedin.pegasus2avro.mxe import (
     MetadataChangeEvent,
@@ -187,6 +188,7 @@ class DataHubGraph(DatahubRestEmitter, EntityVersioningAPI):
         """Get the public-facing base url of the frontend
 
         This url can be used to construct links to the frontend. The url will not include a trailing slash.
+
         Note: Only supported with DataHub Cloud.
         """
 
@@ -198,6 +200,20 @@ class DataHubGraph(DatahubRestEmitter, EntityVersioningAPI):
             raise ValueError("baseUrl not found in server config")
         return base_url
 
+    def url_for(self, entity_urn: Union[str, Urn]) -> str:
+        """Get the UI url for an entity.
+
+        Note: Only supported with DataHub Cloud.
+
+        Args:
+            entity_urn: The urn of the entity to get the url for.
+
+        Returns:
+            The public-facing url for the entity.
+        """
+
+        return make_url_for_urn(self.frontend_base_url, str(entity_urn))
+
     @classmethod
     def from_emitter(cls, emitter: DatahubRestEmitter) -> "DataHubGraph":
         session_config = emitter._session_config

metadata-ingestion/src/datahub/ingestion/graph/links.py

@@ -0,0 +1,53 @@
+from typing import Optional
+
+import datahub.metadata.urns as urns
+from datahub.utilities.urns.urn import guess_entity_type
+
+_url_prefixes = {
+    # Atypical mappings.
+    urns.DataJobUrn.ENTITY_TYPE: "tasks",
+    urns.DataFlowUrn.ENTITY_TYPE: "pipelines",
+    urns.CorpUserUrn.ENTITY_TYPE: "user",
+    urns.CorpGroupUrn.ENTITY_TYPE: "group",
+    # Normal mappings - matches the entity type.
+    urns.ChartUrn.ENTITY_TYPE: "chart",
+    urns.ContainerUrn.ENTITY_TYPE: "container",
+    urns.DataProductUrn.ENTITY_TYPE: "dataProduct",
+    urns.DatasetUrn.ENTITY_TYPE: "dataset",
+    urns.DashboardUrn.ENTITY_TYPE: "dashboard",
+    urns.DomainUrn.ENTITY_TYPE: "domain",
+    urns.GlossaryNodeUrn.ENTITY_TYPE: "glossaryNode",
+    urns.GlossaryTermUrn.ENTITY_TYPE: "glossaryTerm",
+    urns.TagUrn.ENTITY_TYPE: "tag",
+}
+
+
+def make_url_for_urn(
+    frontend_base_url: str,
+    entity_urn: str,
+    *,
+    tab: Optional[str] = None,
+) -> str:
+    """Build the public-facing URL for an entity urn.
+
+    Args:
+        frontend_url: The public-facing base url of the frontend.
+        entity_urn: The urn of the entity to get the url for.
+        tab: The tab to deep link into. If not provided, the default tab for the entity will be shown.
+
+    Returns:
+        The public-facing url for the entity.
+
+    Examples:
+        >>> make_url_for_urn("https://demo.datahub.com", "urn:li:container:b41c14bc5cb3ccfbb0433c8cbdef2992", tab="Contents")
+        'https://demo.datahub.com/container/urn:li:container:b41c14bc5cb3ccfbb0433c8cbdef2992/Contents'
+        >>> make_url_for_urn("https://demo.datahub.com", "urn:li:dataset:(urn:li:dataPlatform:snowflake,long_tail_companions.adoption.actuating,PROD)")
+        'https://demo.datahub.com/dataset/urn:li:dataset:(urn:li:dataPlatform:snowflake,long_tail_companions.adoption.actuating,PROD)/'
+    """
+    entity_type = guess_entity_type(entity_urn)
+
+    url_prefix = _url_prefixes.get(entity_type, entity_type)
+    url = f"{frontend_base_url}/{url_prefix}/{entity_urn}/"
+    if tab:
+        url += f"{tab}"
+    return url

metadata-ingestion/tests/unit/sdk/test_links.py

@@ -0,0 +1,6 @@
+import datahub.ingestion.graph.links as links
+from datahub.testing.doctest import assert_doctest
+
+
+def test_links() -> None:
+    assert_doctest(links)