scandata: fix scandata path when source file is page_numbers.json

derivermodule/scandata.py

@@ -25,7 +25,11 @@ def get_scandata_xml(identifier, source_file):
     zip_path = item_dir/'scandata.zip'
     raw_xml_path = item_dir/'scandata.xml'
     identifier_xml_path = item_dir/f'{identifier}_scandata.xml'
-    source_xml_path = Path(f'{source_file.rsplit("_", 1)[0]}_scandata.xml')
+    if source_file:
+        if source_file.endswith('page_numbers.json'):
+            source_xml_path = Path(f'{source_file[:-18]}_scandata.xml')
+        else:
+            source_xml_path = Path(f'{source_file.rsplit("_", 1)[0]}_scandata.xml')
 
     if source_xml_path.exists():
         return source_xml_path