From 104adb2d7d214c8bce3b1a465cb30c8e0f70495c Mon Sep 17 00:00:00 2001
From: Steve Majewski <sdm7g@virginia.edu>
Date: Fri, 8 Feb 2019 13:09:02 -0500
Subject: [PATCH 1/2] add recover option to be more lenient with bad metadata
 payloads

---
 src/oaipmh/client.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/oaipmh/client.py b/src/oaipmh/client.py
index 2b0c611..daffa79 100644
--- a/src/oaipmh/client.py
+++ b/src/oaipmh/client.py
@@ -107,7 +107,7 @@ def ignoreBadCharacters(self, true_or_false):
         """
         self._ignore_bad_character_hack = true_or_false
 
-    def parse(self, xml):
+    def parse(self, xml, recover=False):
         """Parse the XML to a lxml tree.
         """
         # XXX this is only safe for UTF-8 encoded content,
@@ -122,7 +122,7 @@ def parse(self, xml):
             if hasattr(xml, "encode"):
                 xml = xml.encode("utf-8")
             # xml = xml.encode("utf-8")
-        return etree.XML(xml)
+        return etree.XML(xml, etree.XMLParser(recover=recover))
 
     # implementation of the various methods, delegated here by
     # handleVerb method

From bf1b7212bef2b510088905e89d09eceef7bd6453 Mon Sep 17 00:00:00 2001
From: Steve Majewski <sdm7g@virginia.edu>
Date: Fri, 5 Apr 2019 17:22:36 -0400
Subject: [PATCH 2/2] on SyntaxError, log error and retry with parser option
 recover=True

---
 src/oaipmh/client.py | 19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/src/oaipmh/client.py b/src/oaipmh/client.py
index daffa79..b342b76 100644
--- a/src/oaipmh/client.py
+++ b/src/oaipmh/client.py
@@ -17,6 +17,8 @@
 from lxml import etree
 import time
 import codecs
+import logging
+logger = logging.getLogger( __name__ )
 
 from oaipmh import common, metadata, validation, error
 from oaipmh.datestamp import datestamp_to_datetime, datetime_to_datestamp
@@ -300,8 +302,20 @@ def makeRequestErrorHandling(self, **kw):
         xml = self.makeRequest(**kw)
         try:
             tree = self.parse(xml)
-        except SyntaxError:
-            raise error.XMLSyntaxError(kw)
+        except SyntaxError as e:
+            try: # try again
+                logger.warning( e )
+                logger.warning( 'attempting to recover...')
+                tree = self.parse(xml, recover=True )
+                ids = tree.xpath( '/oai:OAI-PMH/oai:ListRecords/oai:record/oai:header/oai:identifier', namespaces=self.getNamespaces() )
+                if len(ids) == 1:
+                    logger.warning("Recoverable parse error on: {0}".format( ids[0].text ))
+                else:
+                    logger.warning(
+                    "Recoverable parse error on one or more of:\n {0}".format(
+                       "\n".join([ id.text for id in ids ])))
+            except SyntaxError:  # can't recover
+                raise error.XMLSyntaxError(kw)
         # check whether there are errors first
         e_errors = tree.xpath('/oai:OAI-PMH/oai:error',
                               namespaces=self.getNamespaces())
@@ -320,6 +334,7 @@ def makeRequestErrorHandling(self, **kw):
                         code, msg))
                 # find exception in error module and raise with msg
                 raise getattr(error, code[0].upper() + code[1:] + 'Error')(msg)
+
         return tree
 
     def makeRequest(self, **kw):