Fix & test query string parsing

dosagelib/util.py

@@ -5,13 +5,12 @@
 
 from __future__ import absolute_import, division, print_function
 
-from six.moves.urllib.parse import (
+from six.moves.urllib.parse import (parse_qs,
     quote as url_quote, unquote as url_unquote, urlparse, urlunparse, urlsplit)
 from six.moves.urllib_robotparser import RobotFileParser
 import requests
 import sys
 import os
-import cgi
 import re
 import traceback
 import time
@@ -321,9 +320,9 @@ def getRelativePath(basepath, path):
 
 def getQueryParams(url):
     """Get URL query parameters."""
-    query = urlsplit(url)[3]
+    query = urlsplit(url).query
     out.debug(u'Extracting query parameters from %r (%r)...' % (url, query))
-    return cgi.parse_qs(query)
+    return parse_qs(query)
 
 
 def internal_error(out=sys.stderr, etype=None, evalue=None, tb=None):

tests/test_helpers.py

@@ -3,7 +3,7 @@
 
 from __future__ import absolute_import, division, print_function
 
-from dosagelib.helpers import joinPathPartsNamer
+from dosagelib.helpers import joinPathPartsNamer, queryNamer
 
 
 class TestNamer(object):
@@ -11,6 +11,11 @@ class TestNamer(object):
     Tests for comic namer.
     """
 
+    def test_queryNamer(self):
+        testurl = 'http://FOO?page=result&page2=result2'
+        assert queryNamer('page')(self, testurl, "") == 'result'
+        assert queryNamer('page2', True)(self, "", testurl) == 'result2'
+
     def test_joinPathPartsNamer(self):
         imgurl = 'https://HOST/wp-content/uploads/2019/02/tennis5wp-1.png'
         pageurl = 'https://HOST/2019/03/11/12450/'