Merge pull request #354 from ckan/python3.13_support

Python 3.13 support

Author: amercader

ckanext/spatial/harvesters/base.py

@@ -2,8 +2,7 @@
 from urllib.request import urlopen
 
 import re
-import cgitb
-import warnings
+import traceback
 
 import sys
 import logging
@@ -41,12 +40,7 @@
 
 
 def text_traceback():
-    with warnings.catch_warnings():
-        warnings.simplefilter("ignore")
-        res = 'the original traceback:'.join(
-            cgitb.text(sys.exc_info()).split('the original traceback:')[1:]
-        ).strip()
-    return res
+    return "".join(traceback.format_exception(*sys.exc_info()))
 
 
 def guess_standard(content):
@@ -855,7 +849,7 @@ def _get_content_as_unicode(self, url):
         content = response.text
 
         # Remove original XML declaration
-        content = re.sub('<\?xml(.*)\?>', '', content)
+        content = re.sub('<xml(.*)>', '', content)
 
         # Get rid of the BOM and other rubbish at the beginning of the file
         content = re.sub('.*?<', '<', content, 1)
@@ -878,7 +872,7 @@ def _validate_document(self, document_string, harvest_object, validator=None):
         if not validator:
             validator = self._get_validator()
 
-        document_string = re.sub('<\?xml(.*)\?>', '', document_string)
+        document_string = re.sub('<xml(.*)>', '', document_string)
 
         try:
             xml = etree.fromstring(document_string)

ckanext/spatial/harvesters/csw.py

@@ -175,7 +175,7 @@ def fetch_stage(self,harvest_object):
             # Save the fetch contents in the HarvestObject
             # Contents come from csw_client already declared and encoded as utf-8
             # Remove original XML declaration
-            content = re.sub('<\?xml(.*)\?>', '', record['xml'])
+            content = re.sub('<xml(.*)>', '', record['xml'])
 
             harvest_object.content = content.strip()
             harvest_object.save()