fix(importer): tolerate blobs disappearing

I noticed some errors getting logged by the GCS deletion code because
the blob disappeared in between the bucket getting listed and the blobs
being retrieved (this can happen, but led me to also find the fix in

Also give validation failure the same exception-raising treatment as the
refactor in google#2644 did for import-time validation failure.

docker/importer/importer.py

@@ -35,6 +35,7 @@
 from google.cloud import pubsub_v1
 from google.cloud import storage
 from google.cloud.storage import retry
+from google.cloud.exceptions import NotFound
 import pygit2.enums
 
 import osv
@@ -292,31 +293,34 @@ def _vuln_ids_from_gcs_blob(self, client: storage.Client,
       source_repo: the osv.SourceRepository the blob relates to
       blob: the storage.Blob object to operate on
 
+    Raises:
+      jsonschema.exceptions.ValidationError when self._strict_validation is True
+      input fails OSV JSON Schema validation
+
     Returns:
       a list of one or more vulnerability IDs (from the Vulnerability proto) or
-      None when the blob has an unexpected name or fails to parse
+      None when the blob has an unexpected name or fails to retrieve
     """
     if not _is_vulnerability_file(source_repo, blob.name):
       return None
 
     # Download in a blob generation agnostic way to cope with the blob
     # changing between when it was listed and now (if the generation doesn't
     # match, retrieval fails otherwise).
-    blob_bytes = storage.Blob(
-        blob.name, blob.bucket, generation=None).download_as_bytes(client)
-
-    vuln_ids = []
     try:
-      vulns = osv.parse_vulnerabilities_from_data(
-          blob_bytes,
-          os.path.splitext(blob.name)[1],
-          strict=self._strict_validation)
-    except Exception as e:
-      logging.error('Failed to parse vulnerability %s: %s', blob.name, e)
-      # TODO(andrewpollock): I think this needs to be reraised here...
-      # a jsonschema.exceptions.ValidationError only gets raised in strict
-      # validation mode.
+      blob_bytes = storage.Blob(
+          blob.name, blob.bucket, generation=None).download_as_bytes(client)
+    except NotFound:
+      # The file can disappear between bucket listing and blob retrieval.
       return None
+
+    vuln_ids = []
+    # When self._strict_validation is True,
+    # this *may* raise a jsonschema.exceptions.ValidationError
+    vulns = osv.parse_vulnerabilities_from_data(
+        blob_bytes,
+        os.path.splitext(blob.name)[1],
+        strict=self._strict_validation)
     for vuln in vulns:
       vuln_ids.append(vuln.id)
     return vuln_ids