fix(backend): Metadata Writer - Fixed regression with artifact type r…

…etrieval. Fixes #3971 (#4231)

* Metadata Writer - Fixed regression with artifact type retrieval

The DSL compiler has changed the output name sanitization rules, so we should change them here accordingly.

* Added the code link

backend/metadata_writer/src/metadata_writer.py

@@ -80,7 +80,9 @@ def patch_pod_metadata(
 
 def output_name_to_argo(name: str) -> str:
     import re
-    return re.sub('-+', '-', re.sub('[^-0-9a-z]+', '-', name.lower())).strip('-')
+    # This sanitization code should be kept in sync with the code in the DSL compiler.
+    # See https://github.com/kubeflow/pipelines/blob/39975e3cde7ba4dcea2bca835b92d0fe40b1ae3c/sdk/python/kfp/compiler/_k8s_helper.py#L33
+    return re.sub('-+', '-', re.sub('[^-_0-9A-Za-z]+', '-', name)).strip('-')
 
 def is_s3_endpoint(endpoint: str) -> bool:
     return re.search('^.*s3.*amazonaws.com.*$', endpoint)