Update publication_parser.py

for arrow errors, fallback to regex year extraction
scholarly-python-package · Feb 3, 2025 · a16d8b2 · a16d8b2
1 parent 568d4ad
commit a16d8b2
Showing 1 changed file with 5 additions and 1 deletion.
diff --git a/scholarly/publication_parser.py b/scholarly/publication_parser.py
@@ -312,7 +312,11 @@ def fill(self, publication: Publication)->Publication:
                                 'YYYY/M/DD',
                                 'YYYY/M/D',
                                 'YYYY/MM/D']
-                    publication['bib']['pub_year'] = arrow.get(val.text, patterns).year
+                    try:
+                        publication['bib']['pub_year'] = arrow.get(val.text, patterns).year
+                    except ValueError:
+                        # fallback to regex year extraction if arrow fails
+                         publication['bib']['pub_year'] = re.search(r'\d{4}', val.text).group()
                     publication['bib']['pub_date'] = val.text
                 elif key == 'description':
                     # try to find all the gsh_csp if they exist