@@ -370,25 +370,25 @@ def extract_metadata_from_xml(record_xml):
370370 root = ET .fromstring (record_xml )
371371
372372 # Extract category (primary category from categories field)
373- categories_element = root .find (
373+ categories_elem = root .find (
374374 ".//{http://arxiv.org/OAI/arXiv/}categories"
375375 )
376- category = "Uncategorized "
377- if categories_element is not None and categories_element .text :
376+ category = "Unknown "
377+ if categories_elem is not None and categories_elem .text :
378378 # Take first category as primary
379- category = categories_element .text .strip ().split ()[0 ]
379+ category = categories_elem .text .strip ().split ()[0 ]
380380
381381 # Extract year from created date
382- created_element = root .find (".//{http://arxiv.org/OAI/arXiv/}created" )
383- year = "Undated "
384- if created_element is not None and created_element .text :
382+ created_elem = root .find (".//{http://arxiv.org/OAI/arXiv/}created" )
383+ year = "Unknown "
384+ if created_elem is not None and created_elem .text :
385385 try :
386- year = created_element .text .strip ()[:4 ] # Extract year
386+ year = created_elem .text .strip ()[:4 ] # Extract year
387387 except (AttributeError , IndexError ) as e :
388388 LOGGER .warning (
389- f"Failed to extract year from '{ created_element .text } ': { e } "
389+ f"Failed to extract year from '{ created_elem .text } ': { e } "
390390 )
391- year = "Undated "
391+ year = "Unknown "
392392
393393 # Extract author count
394394 authors = root .findall (".//{http://arxiv.org/OAI/arXiv/}author" )
@@ -407,10 +407,10 @@ def extract_metadata_from_xml(record_xml):
407407 except Exception as e :
408408 LOGGER .error (f"Metadata extraction error: { e } " )
409409 return {
410- "category" : "Uncategorized " ,
411- "year" : "Undated " ,
410+ "category" : "Unknown " ,
411+ "year" : "Unknown " ,
412412 "author_count" : 0 ,
413- "license" : "Unspecified " ,
413+ "license" : "Unknown " ,
414414 }
415415
416416
0 commit comments