fix license-check transform

Signed-off-by: Shivdeep Singh <[email protected]>

transforms/code/license_check/Makefile

@@ -21,6 +21,10 @@ image::
 	@# Help: Recursively make $@ in all subdirs 
 	@$(MAKE) RULE=$@ .recurse
 
+set-versions:  
+	@# Help: Recursively $@ in all subdirs 
+	@$(MAKE) RULE=$@ .recurse
+
 publish:: 
 	@# Help: Recursively make $@ in all subdirs 
 	@$(MAKE) RULE=$@ .recurse
@@ -43,21 +47,17 @@ load-image::
 
 .PHONY: workflow-venv
 workflow-venv:
-	$(MAKE) -C kfp_ray/v1 workflow-venv
+	$(MAKE) -C  kfp_ray workflow-venv
 
 .PHONY: workflow-build
 workflow-build:
-	$(MAKE) -C kfp_ray/v1 workflow-build
+	$(MAKE) -C  kfp_ray workflow-build
 
 .PHONY: workflow-test
 workflow-test:
-	$(MAKE) -C kfp_ray/v1 workflow-test
+	$(MAKE) -C  kfp_ray workflow-test
 
 .PHONY: workflow-upload
 workflow-upload:
-	$(MAKE) -C kfp_ray/v1 workflow-upload
-
-.PHONY: workflow-reconcile-requirements
-workflow-reconcile-requirements:
-	$(MAKE) -C kfp_ray/v1 workflow-reconcile-requirements
+	$(MAKE) -C  kfp_ray workflow-upload
 

transforms/code/license_check/python/src/license_check_local.py

@@ -36,7 +36,7 @@
     transform = LicenseCheckTransform(license_check_params)
 
     data_access = DataAccessLocal()
-    table = data_access.get_table(os.path.join(input_folder, "sample_1.parquet"))
+    table, _ = data_access.get_table(os.path.join(input_folder, "sample_1.parquet"))
     print(f"input table: {table}")
     # Transform the table
     table_list, metadata = transform.transform(table)

transforms/code/license_check/python/src/license_check_transform.py

@@ -34,12 +34,14 @@ def _get_supported_licenses(license_file: str, data_access: DataAccess) -> list[
         licenses_list = json.loads(licenses_list_json.decode("utf-8"))
         logger.info(f"Read a list of {len(licenses_list)} licenses.")
     except Exception as e:
-        logger.error(f"got and exception {e}")
+        logger.error(f"Failed to read file: {license_file} due to {e}")
     return licenses_list
 
 
 class LicenseCheckTransform(AbstractTableTransform):
-    """ """
+    """It can be used to select the rows/records of data with licenses
+    matching those in the approved/deny list.
+    """
 
     def __init__(self, config: dict):
         super().__init__(config)
@@ -66,13 +68,10 @@ def __init__(self, config: dict):
             )
 
     def transform(self, table: pa.Table, file_name: str = None) -> tuple[list[pa.Table], dict]:
-        """ """
-        if type(table) is tuple:
-            logger.info(f"We got a tuple. Extract table from tuple, if possible.")
-            extracted_table, _ = table
-            if type(extracted_table) is pa.Table:
-                table = extracted_table
-
+        """
+        Transforms input tables by adding a boolean `license_status` column
+        indicating whether the license is approved/denied.
+        """
         if not TransformUtils.validate_columns(table, [self.license_check["license_column_name"]]):
             return [], {}
         new_table = self.transformer.transform(table)

transforms/code/license_check/python/test-data/expected/metadata.json

@@ -208,11 +208,11 @@
         "result_doc_count": 4
     },
     "source": {
-        "name": "/Users/shivdeep/workspace/projects/current/oss-data-prep/data-prep-lab/transforms/code/license_check/python/test-data/input",
+        "name": "data-prep-kit/transforms/code/license_check/python/test-data/input",
         "type": "path"
     },
     "target": {
-        "name": "/Users/shivdeep/workspace/projects/current/oss-data-prep/data-prep-lab/transforms/code/license_check/python/test-data/output",
+        "name": "data-prep-kit/transforms/code/license_check/python/test-data/output",
         "type": "path"
     }
 }