Jats email problem iop.20250117

adsingestp/parsers/base.py

@@ -1,8 +1,10 @@
 import html
 import re
+import warnings
 from datetime import datetime
 
 import bs4
+from bs4 import MarkupResemblesLocatorWarning
 
 from adsingestp.ingest_exceptions import WrongFormatException
 
@@ -21,6 +23,7 @@ class IngestBase(object):
     ]
 
     def __init__(self, xml_ref=True):
+        warnings.filterwarnings("ignore", category=MarkupResemblesLocatorWarning, module="bs4")
         self.xml_ref = xml_ref
 
     def _clean_empty(self, input_to_clean, keys_to_keep=required_keys):

adsingestp/parsers/jats.py

@@ -129,13 +129,25 @@ def _fix_email(self, email):
         for em in email:
             if " " in em:
                 for e in em.strip().split():
-                    if email_format.search(e):
-                        email_new.add(email_format.search(e).group(0))
-                        email_parsed = True
+                    try:
+                        if email_format.search(e):
+                            email_new.add(email_format.search(e).group(0))
+                            email_parsed = True
+                    except Exception as err:
+                        logger.warning("Bad format in _fix_email: %s" % err)
             else:
-                if email_format.search(em):
-                    email_new.add(email_format.search(em).group(0))
-                    email_parsed = True
+                try:
+                    if type(em) == str:
+                        if email_format.search(em):
+                            email_new.add(email_format.search(em).group(0))
+                            email_parsed = True
+                    elif type(em) == list:
+                        for e in em:
+                            if email_format.search(e):
+                                email_new.add(email_format.search(e).group(0))
+                                email_parsed = True
+                except Exception as err:
+                    logger.warning("Bad format in _fix_email: %s" % err)
 
         if not email_parsed:
             logger.warning("Email not verified as valid. Input email list: %s", str(email))