fix(ingest/dbt): always encode tag urns (datahub-project#10799)

aviv-julienjehannet · Jul 25, 2024 · f168925 · f168925
1 parent 207c729
commit f168925
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 7 deletions.
diff --git a/metadata-ingestion/src/datahub/emitter/mce_builder.py b/metadata-ingestion/src/datahub/emitter/mce_builder.py
@@ -476,7 +476,7 @@ def get_or_add_aspect(mce: MetadataChangeEventClass, default: Aspect) -> Aspect:
 
 def make_global_tag_aspect_with_tag_list(tags: List[str]) -> GlobalTagsClass:
  return GlobalTagsClass(
- tags=[TagAssociationClass(f"urn:li:tag:{tag}") for tag in tags]
+ tags=[TagAssociationClass(make_tag_urn(tag)) for tag in tags]
  )
 
 

diff --git a/metadata-ingestion/src/datahub/ingestion/source/dbt/dbt_common.py b/metadata-ingestion/src/datahub/ingestion/source/dbt/dbt_common.py
@@ -1483,7 +1483,7 @@ def get_patched_mce(self, mce):
  transformed_tag_list = self.get_transformed_tags_by_prefix(
  tag_aspect.tags,
  mce.proposedSnapshot.urn,
- mce_builder.make_tag_urn(self.config.tag_prefix),
+ tag_prefix=self.config.tag_prefix,
  )
  tag_aspect.tags = transformed_tag_list
 
@@ -1874,16 +1874,19 @@ def get_transformed_tags_by_prefix(
  self,
  new_tags: List[TagAssociationClass],
  entity_urn: str,
- tags_prefix_filter: str,
+ tag_prefix: str,
  ) -> List[TagAssociationClass]:
  tag_set = {new_tag.tag for new_tag in new_tags}
 
  if self.ctx.graph:
  existing_tags_class = self.ctx.graph.get_tags(entity_urn)
  if existing_tags_class and existing_tags_class.tags:
- for exiting_tag in existing_tags_class.tags:
- if not exiting_tag.tag.startswith(tags_prefix_filter):
- tag_set.add(exiting_tag.tag)
+ for existing_tag in existing_tags_class.tags:
+ if tag_prefix and existing_tag.tag.startswith(
+ mce_builder.make_tag_urn(tag_prefix)
+ ):
+ continue
+ tag_set.add(existing_tag.tag)
  return [TagAssociationClass(tag) for tag in sorted(tag_set)]
 
  # This method attempts to read-modify and return the glossary terms of a dataset.

diff --git a/metadata-ingestion/tests/unit/test_dbt_source.py b/metadata-ingestion/tests/unit/test_dbt_source.py
@@ -148,7 +148,7 @@ def test_dbt_source_patching_tags():
  ["new_non_dbt", "dbt:new_dbt"]
  )
  transformed_tags = source.get_transformed_tags_by_prefix(
- new_tag_aspect.tags, "urn:li:dataset:dummy", "urn:li:tag:dbt:"
+ new_tag_aspect.tags, "urn:li:dataset:dummy", "dbt:"
  )
  expected_tags = {
  "urn:li:tag:new_non_dbt",