Merge lp:~ceronman/txsolr/xml-input-encoding-867786 into lp:txsolr

[Manuel Cerón]

This branch introduces the following changes:

 - The XML Input encoder now forces UTF-8 encoding when generating XML documents. This prevents unicode errors in Python 2.6.

[Esteve Fernandez (esteve)]

+1 LGTM

review: Approve

[Nicholas Tollervey (ntoll)]

+1 LGTM

review: Approve

=== modified file 'txsolr/input.py'
--- txsolr/input.py	2010-10-14 10:58:34 +0000
+++ txsolr/input.py	2011-10-04 19:05:24 +0000
@@ -127,7 +127,7 @@
                     docElement.append(fieldElement)
             addElement.append(docElement)
 
-        result = ElementTree.tostring(addElement)
+        result = ElementTree.tostring(addElement, encoding='utf-8')
         return StringProducer(result)
 
     def createDelete(self, id):
@@ -143,7 +143,7 @@
             idElement.text = self._encodeValue(id)
             deleteElement.append(idElement)
 
-        result = ElementTree.tostring(deleteElement)
+        result = ElementTree.tostring(deleteElement, encoding='utf-8')
         return StringProducer(result)
 
     def createDeleteByQuery(self, query):

=== modified file 'txsolr/test/test_input.py'
--- txsolr/test/test_input.py	2011-05-24 02:26:43 +0000
+++ txsolr/test/test_input.py	2011-10-04 19:05:24 +0000
@@ -56,6 +56,16 @@
         input = self.input.createAdd(document).body
         self.assertEqual(input, expected, 'Wrong input')
 
+    def testCreateAddEncoding(self):
+        """
+        L{SimpleXMLInputFactory.createAdd} encodes the C{add} request as UTF-8.
+        """
+        document = {'id': 1, 'text': u'\U0001d1b6'}
+        expected = ('<add><doc><field name="text">\xf0\x9d\x86\xb6</field>'
+                    '<field name="id">1</field></doc></add>')
+        input = self.input.createAdd(document).body
+        self.assertEqual(input, expected, 'Wrong input')
+
     def testCreateAddWithCollection(self):
         """
         L{SimpleXMLInputFactory.createAdd} creates a correct body for an C{add}
@@ -111,6 +121,15 @@
         expected = '<delete><id>123</id></delete>'
         self.assertEqual(self.input.createDelete(id).body, expected)
 
+    def testCreateDeleteEncoding(self):
+        """
+        L{SimpleXMLInputFactory.createDelete} encodes a C{delete} request as
+        UTF-8.
+        """
+        id = u'\U0001d1b6'
+        expected = '<delete><id>\xf0\x9d\x86\xb6</id></delete>'
+        self.assertEqual(self.input.createDelete(id).body, expected)
+
     def testCreateDeleteWithEncoding(self):
         """
         L{SimpleXMLInputFactory.createDelete} correctly escapes XML special