get_marc_record_from_ia: use existing metadata if present (#8332)

* get_marc_record_from_ia: use existing metadata if present

openlibrary/catalog/get_ia.py

@@ -27,14 +27,20 @@ def urlopen_keep_trying(url: str, headers=None, **kwargs):
         sleep(2)
 
 
-def get_marc_record_from_ia(identifier: str) -> MarcBinary | MarcXml | None:
+def get_marc_record_from_ia(
+    identifier: str, ia_metadata: dict | None = None
+) -> MarcBinary | MarcXml | None:
     """
-    Takes IA identifiers and returns MARC record instance.
+    Takes IA identifiers and optional IA metadata and returns MARC record instance.
     08/2018: currently called by openlibrary/plugins/importapi/code.py
     when the /api/import/ia endpoint is POSTed to.
+
+    :param ia_metadata: The full ia metadata; e.g. https://archive.org/metadata/goody,
+                        not https://archive.org/metadata/goody/metadata
     """
-    metadata = ia.get_metadata(identifier)
-    filenames = metadata['_filenames']
+    if ia_metadata is None:
+        ia_metadata = ia.get_metadata(identifier)
+    filenames = ia_metadata['_filenames']  # type: ignore[index]
 
     marc_xml_filename = identifier + '_marc.xml'
     marc_bin_filename = identifier + '_meta.mrc'

openlibrary/plugins/importapi/code.py

@@ -230,7 +230,9 @@ def ia_import(
             raise BookImportError(status, 'Prohibited Item %s' % identifier)
 
         # Case 4 - Does this item have a marc record?
-        marc_record = get_marc_record_from_ia(identifier)
+        marc_record = get_marc_record_from_ia(
+            identifier=identifier, ia_metadata=metadata
+        )
         if require_marc and not marc_record:
             raise BookImportError('no-marc-record')
         if marc_record: