fix: serialize text nodes in PyQuery str/html

pyquery/pyquery.py

@@ -344,6 +344,15 @@ def remove_namespaces(self):
                     el.tag = el.tag.split('}', 1)[1]
         return self
 
+    def _serialize_nodes(self, dumps, method):
+        # Text nodes from contents() cannot go through tostring.
+        has_text_nodes = any(isinstance(e, str) for e in self)
+        return ''.join(
+            e if isinstance(e, str) else dumps(
+                e, encoding=str, method=method, with_tail=not has_text_nodes)
+            for e in self
+        )
+
     def __str__(self):
         """xml representation of current nodes::
 
@@ -353,12 +362,11 @@ def __str__(self):
             <script>&lt;![[CDATA[ ]&gt;</script>
 
         """
-        return ''.join([etree.tostring(e, encoding=str) for e in self])
+        return self._serialize_nodes(etree.tostring, method='xml')
 
     def __unicode__(self):
         """xml representation of current nodes"""
-        return u''.join([etree.tostring(e, encoding=str)
-                         for e in self])
+        return self._serialize_nodes(etree.tostring, method='xml')
 
     def __html__(self):
         """html representation of current nodes::
@@ -369,8 +377,7 @@ def __html__(self):
             <script><![[CDATA[ ]></script>
 
         """
-        return u''.join([lxml.html.tostring(e, encoding=str)
-                         for e in self])
+        return self._serialize_nodes(lxml.html.tostring, method='html')
 
     def __repr__(self):
         r = []

tests/test_pyquery.py

@@ -359,6 +359,19 @@ def test_comment(self):
         self.assertEqual(doc.text(), 'bar')
 
 
+class TestContentsStr(TestCase):
+
+    def test_str_contents_with_text_nodes(self):
+        doc = pq('hello <b>bold</b> world')
+        contents = doc.contents()
+        self.assertEqual(str(contents), 'hello <b>bold</b> world')
+        self.assertEqual(contents.__html__(), 'hello <b>bold</b> world')
+
+    def test_str_contents_text_only(self):
+        doc = pq('<div>only text</div>')
+        self.assertEqual(str(doc.contents()), 'only text')
+
+
 class TestCallback(TestCase):
     html = """
         <ol>