handle path that end with * better

pipeline/validation.py

@@ -34,15 +34,15 @@ def assert_valid_glob_pattern(pattern: str, privacy_level: str) -> None:
 
     path = Path(pattern)
 
-    if path.suffix == "":
+    if path.suffix == "" or path.suffix.endswith("*"):
         raise InvalidPatternError(
             "output paths must have a file type extension at the end"
         )
 
     if privacy_level == "moderately_sensitive":
         if path.suffix not in LEVEL4_FILE_TYPES:
             raise InvalidPatternError(
-                f"{path.suffix} is not an allowed file type for moderately_sensitive outputs"
+                f"{path} is not an allowed file type for moderately_sensitive outputs"
             )
 
     # Check that the path is in normal form

tests/fixtures/valid_yaml/project.yaml

@@ -23,21 +23,21 @@ actions:
     needs: [generate_cohort]
     outputs:
       highly_sensitive:
-        male_cohort: male.*
+        male_cohort: male*.csv
 
   prepare_data_f:
     run: python:latest python analysis/filter_by_sex.py F output/input.csv female.csv
     needs: [generate_cohort]
     outputs:
       highly_sensitive:
-        female_cohort: female.*
+        female_cohort: female*.csv
 
   prepare_data_with_quote_in_filename:
     run: python:latest python analysis/filter_by_sex.py F output/input.csv "qu'ote.csv"
     needs: [generate_cohort]
     outputs:
       highly_sensitive:
-        quote_cohort: "qu'ote.*"
+        quote_cohort: "qu'ote*.csv"
 
   analyse_data:
     run: python:latest python analysis/count_lines.py counts.txt

tests/test_validation.py

@@ -19,6 +19,7 @@ def test_assert_valid_glob_pattern():
         ("metadata", "highly_sensitive"),
         ("metadata/test.txt", "highly_sensitive"),
         ("outputs/*", "highly_sensitive"),
+        ("outputs/foo.*", "highly_sensitive"),
         ("outputs/output.rds", "moderately_sensitive"),
     ]
     for pattern, sensitivity in bad_patterns: