perf: Attempt to parse XML strings without cleaning (which is expensive) before trying again with cleaning

Author: eliasbenb

plexapi/client.py

@@ -197,8 +197,7 @@ def query(self, path, method=None, headers=None, timeout=None, **kwargs):
                 raise NotFound(message)
             else:
                 raise BadRequest(message)
-        data = utils.cleanXMLString(response.text).encode('utf8')
-        return ElementTree.fromstring(data) if data.strip() else None
+        return utils.parseXMLString(response.text)
 
     def sendCommand(self, command, proxy=None, **params):
         """ Convenience wrapper around :func:`~plexapi.client.PlexClient.query` to more easily

plexapi/myplex.py

@@ -250,8 +250,7 @@ def query(self, url, method=None, headers=None, timeout=None, **kwargs):
             return response.json()
         elif 'text/plain' in response.headers.get('Content-Type', ''):
             return response.text.strip()
-        data = utils.cleanXMLString(response.text).encode('utf8')
-        return ElementTree.fromstring(data) if data.strip() else None
+        return utils.parseXMLString(response.text)
 
     def ping(self):
         """ Ping the Plex.tv API.

plexapi/server.py

@@ -768,8 +768,7 @@ def query(self, key, method=None, headers=None, params=None, timeout=None, **kwa
                 raise NotFound(message)
             else:
                 raise BadRequest(message)
-        data = utils.cleanXMLString(response.text).encode('utf8')
-        return ElementTree.fromstring(data) if data.strip() else None
+        return utils.parseXMLString(response.text)
 
     def search(self, query, mediatype=None, limit=None, sectionId=None):
         """ Returns a list of media items or filter categories from the resulting

plexapi/utils.py

@@ -23,6 +23,8 @@
 
 from plexapi.exceptions import BadRequest, NotFound, Unauthorized
 
+from xml.etree import ElementTree
+
 try:
     from tqdm import tqdm
 except ImportError:
@@ -718,3 +720,13 @@ def sha1hash(guid):
 
 def cleanXMLString(s):
     return _illegal_XML_re.sub('', s)
+
+
+def parseXMLString(s: str):
+    """ Parse an XML string and return an ElementTree object. """
+    if not s.strip():
+        return None
+    try:  # Attempt to parse the string as-is without cleaning (which is expensive)
+        return ElementTree.fromstring(s.encode('utf-8'))
+    except ElementTree.ParseError:  # If it fails, clean the string and try again
+        return ElementTree.fromstring(cleanXMLString(s).encode('utf-8'))