Fixing argument parsing and handling absolute paths

harvester/source/s3.py

@@ -114,23 +114,31 @@ class S3Source(Source):
         return item
 
 
-class S3CatalogSource(S3Base, Source):
+class S3CatalogSource(S3Base):
     type = "S3Catalog"
 
-    def __init__(self, root_href: str, **kwargs):
-        super().__init__(**kwargs)
-        self.root_href = root_href
+    def __init__(self, parameters: dict, **kwargs):
+        self.root_href = parameters.pop("root_href")
+        self.default_catalog_name = parameters.pop("default_catalog_name", None)
+        super().__init__(**parameters)
 
     def harvest(self) -> Iterator[dict]:
         logger.info("Starting S3 Catalog harvesting")
         parsed = urlparse(self.root_href)
-        yield from self.harvest_catalog(parsed.netloc, parsed.path)
+        path = parsed.path
+        if path.startswith("/"):
+            path = parsed.path[1:]
+        if path.endswith("/") and self.default_catalog_name:
+            path = join(path, self.default_catalog_name)
+        yield from self.harvest_catalog(parsed.netloc, path)
 
     def fetch_json(self, bucket: str, key: str) -> dict:
         """
         Loads the given object identifier by bucket and key and loads it as
         JSON.
         """
+        if key.startswith("/"):
+            key = key[1:]
         response = self.client.get_object(Bucket=bucket, Key=key)
         return json.load(response["Body"])