Revert "reverted retries on the command line - due to issues with passing to solids in the ingest pipeline. Hard coded at 12 for now."

This reverts commit 78c2dab.

Author: bahill

hca/staging_area_validator.py

@@ -42,13 +42,12 @@ def main(self):
 
     date_format = "%Y-%m-%dT%H:%M:%S.%fZ"
 
-    # removing total_retries as a config for now
     def __init__(
         self,
         staging_area: str,
         ignore_dangling_inputs: bool,
         validate_json: bool,
-        # total_retries,
+        total_retries,
     ) -> None:
         super().__init__()
         self.staging_area = staging_area
@@ -58,7 +57,7 @@ def __init__(
         self.gcs = gcs.Client()
 
         # Number of retries for validation
-        # self.total_retries = total_retries
+        self.total_retries = total_retries
         # A boolean to tell us if this is a delta or non-delta staging area
         self.is_delta = None
         # A mapping of data file name to metadata id
@@ -299,8 +298,7 @@ def validate_file_json(
         if self.validate_json:
             print(f"Validating JSON of {file_name}")
             try:
-                # self.validator.validate_json(file_json, self.total_retries, schema)
-                self.validator.validate_json(file_json, schema)
+                self.validator.validate_json(file_json, self.total_retries, schema)
             except Exception as e:
                 log.error("File %s failed json validation.", file_name)
                 self.file_errors[file_name] = e
@@ -371,31 +369,27 @@ class SchemaValidator:
     def validate_json(
         cls,
         file_json: JSON,
-        # total_retries: int,
+        total_retries: int,
         schema: Optional[JSON] = None,
     ) -> None:
         if schema is None:
             try:
-                # schema = cls._download_schema(file_json["describedBy"], total_retries)
-                schema = cls._download_schema(file_json["describedBy"])
+                schema = cls._download_schema(file_json["describedBy"], total_retries)
             except json.decoder.JSONDecodeError as e:
                 schema_url = file_json["describedBy"]
                 raise Exception("Failed to parse schema JSON", schema_url) from e
         validate(file_json, schema, format_checker=FormatChecker())
 
     @classmethod
     # setting to maxsize=None so as not to evict old values, and maybe help avoid connectivity issues (DI-22)
-    # Could also have used the dagster.RetryPolicy for this
     @lru_cache(maxsize=None)
-    # def _download_schema(cls, schema_url: str, total_retries: int) -> JSON:
-    def _download_schema(cls, schema_url: str) -> JSON:
+    def _download_schema(cls, schema_url: str, total_retries: int) -> JSON:
         log.debug("Downloading schema %s", schema_url)
 
         s = requests.Session()
-        # log.debug(f"total_retries = {total_retries}")
+        log.debug(f"total_retries = {total_retries}")
         retries = Retry(
-            # total=total_retries,
-            total=12,
+            total=total_retries,
             backoff_factor=0.2,
             status_forcelist=[500, 502, 503, 504],
         )

pyproject.toml

@@ -5,7 +5,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "hca-import-validation"
-version = "0.0.12"
+version = "0.0.11"
 description = "HCA Staging Import Validation"
 urls = {"Source" = "https://github.com/dataBiosphere/hca-import-validation"}
 classifiers = [

validate_staging_area.py

@@ -33,15 +33,14 @@ def _parse_args(argv):
         dest="validate_json",
         help="Do not validate JSON documents against their schema.",
     )
-    # May add this back later if we really need to set this in the config
-    # parser.add_argument(
-    #     "--total-retries",
-    #     "-t",
-    #     type=int,
-    #     default=10,
-    #     dest="total_retries",
-    #     help="The number of times to retry the validation process.",
-    # )
+    parser.add_argument(
+        "--total-retries",
+        "-t",
+        type=int,
+        default=10,
+        dest="total_retries",
+        help="The number of times to retry the validation process.",
+    )
     return parser.parse_args(argv)
 
 
@@ -51,6 +50,6 @@ def _parse_args(argv):
         staging_area=args.staging_area,
         ignore_dangling_inputs=args.ignore_dangling_inputs,
         validate_json=args.validate_json,
-        # total_retries=args.total_retries,
+        total_retries=args.total_retries,
     )
     sys.exit(adapter.main())