remove selector support for LxmlDocument DOM cache and add deprecation warning for _root argument

scrapy/selector/unified.py

@@ -2,13 +2,13 @@
 XPath selectors based on lxml
 """
 
+import warnings
+from parsel import Selector as ParselSelector, SelectorList
 from scrapy.utils.trackref import object_ref
 from scrapy.utils.python import to_bytes
 from scrapy.http import HtmlResponse, XmlResponse
 from scrapy.utils.decorators import deprecated
-from parsel import Selector as ParselSelector, SelectorList
-from parsel.unified import _ctgroup
-from .lxmldocument import LxmlDocument
+from scrapy.exceptions import ScrapyDeprecationWarning
 
 
 __all__ = ['Selector', 'SelectorList']
@@ -33,22 +33,21 @@ class Selector(ParselSelector, object_ref):
 
     __slots__ = ['response']
 
-    def __init__(self, response=None, text=None, type=None, root=None, **kwargs):
+    def __init__(self, response=None, text=None, type=None, root=None, _root=None, **kwargs):
         st = _st(response, type or self._default_type)
 
-        # supporting legacy _root argument
-        root = kwargs.get('_root', root)
-
-        self._parser = _ctgroup[st]['_parser']
+        if root is None and _root is not None:
+            warnings.warn("Argument `_root` is deprecated, use `root` instead",
+                          ScrapyDeprecationWarning, stacklevel=2)
+            root = _root
 
         if text is not None:
             response = _response_from_text(text, st)
 
         if response is not None:
-            root = LxmlDocument(response, self._parser)
+            text = response.body_as_unicode()
 
         self.response = response
-        text = response.body_as_unicode() if response else None
         super(Selector, self).__init__(text=text, type=st, root=root, **kwargs)
 
     # Deprecated api
@@ -59,4 +58,3 @@ class Selector(ParselSelector, object_ref):
     @deprecated(use_instead='.extract()')
     def extract_unquoted(self):
         return self.extract()
-

tests/test_selector.py

@@ -7,6 +7,8 @@ from scrapy.exceptions import ScrapyDeprecationWarning
 from scrapy.http import TextResponse, HtmlResponse, XmlResponse
 from scrapy.selector import Selector
 from scrapy.selector.lxmlsel import XmlXPathSelector, HtmlXPathSelector, XPathSelector
+from lxml import etree
+from tests import mock
 
 
 class SelectorTestCase(unittest.TestCase):
@@ -37,6 +39,15 @@ class SelectorTestCase(unittest.TestCase):
         self.assertEqual([x.extract() for x in sel.xpath("concat(//input[@name='a']/@value, //input[@name='b']/@value)")],
                          [u'12'])
 
+    @mock.patch('scrapy.selector.unified.warnings')
+    def test_deprecated_root_argument(self, warnings):
+        root = etree.fromstring(u'<html/>')
+        sel = self.sscls(_root=root)
+        self.assertEqual(root, sel._root)
+        warnings.warn.assert_called_once_with(
+            'Argument `_root` is deprecated, use `root` instead',
+            mock.ANY, stacklevel=2)
+
     def test_representation_slice(self):
         body = u"<p><input name='{}' value='\xa9'/></p>".format(50 * 'b')
         response = TextResponse(url="http://example.com", body=body, encoding='utf8')