Check if file exists on S3

ckanext/qa/tasks.py

@@ -14,6 +14,7 @@
 import ckan.lib.helpers as ckan_helpers
 from ckanext.qa.sniff_format import sniff_file_format
 from ckanext.archiver.model import Archival, Status
+from google.cloud import storage
 
 import logging
 
@@ -295,15 +296,26 @@ def score_by_sniffing_data(archival, resource, score_reasons):
         score_reasons.append(_('This file had not been downloaded at the time of scoring it.'))
         return (None, None)
     # Analyse the cached file
-    cloudstorage_enabled = config.get('ckanext.qa.cloudstorage_enabled', True)
+    cloudstorage_enabled = config.get('ckanext.qa.cloudstorage_enabled', False)
 
     if cloudstorage_enabled:
 
-        # check if blob exists
-        pass
+        # check if blob exists in cloudstorage
+        path_to_json = config.get(
+            'ckanext.cloudstorage.google_service_account_json')
+        bucket_name = config.get('ckanext.cloudstorage.container_name')
+        storage_client = storage.Client.from_service_account_json(path_to_json)
+        bucket = storage_client.bucket(bucket_name)
+        s3_filepath = "resources/" + resource.id + "/" + resource.url
+        blob = bucket.blob(s3_filepath)
+        if blob.exists():
+            return (None, None)
+        else:
+            score_reasons.append(_('File does not exists on S3: "%s".') % s3_filepath)
+            return (None, None)
 
     else:
-
+        # check if file exists in filestore
         filepath = archival.cache_filepath
         if not os.path.exists(filepath):
             score_reasons.append(_('Cache filepath does not exist: "%s".') % filepath)