Fix inheritance order of mixin classes in lxml.html (GH-340)

As the old FIXME comment from
https://github.com/lxml/lxml/commit/8132c755adad4a75ba855d985dd257493bccc7fd
notes, the mixin should come first for the inheritance to be correct (the left-most class is the
first in the MRO, at least if no diamond inheritance is involved).

Also fix the odd `super` call in `HtmlMixin`, likely stemming from the incorrect MRO.

Fixes the inheritance order of all `HTML*` base classes though it
probably doesn't matter for other than `HtmlElement`.

3 files changed, 49 insertions, 10 deletions

diff --git a/src/lxml/html/__init__.py b/src/lxml/html/__init__.py
index 2139c75a..ef06a40b 100644
--- a/src/lxml/html/__init__.py
+++ b/src/lxml/html/__init__.py
@@ -245,7 +245,7 @@ class HtmlMixin(object):
         creates a 'boolean' attribute without value, e.g. "<form novalidate></form>"
         for ``form.set('novalidate')``.
         """
-        super(HtmlElement, self).set(key, value)
+        super(HtmlMixin, self).set(key, value)
 
     @property
     def classes(self):
@@ -685,21 +685,19 @@ iterlinks = _MethodFunc('iterlinks', copy=False)
 rewrite_links = _MethodFunc('rewrite_links', copy=True)
 
 
-class HtmlComment(etree.CommentBase, HtmlMixin):
+class HtmlComment(HtmlMixin, etree.CommentBase):
     pass
 
 
-class HtmlElement(etree.ElementBase, HtmlMixin):
-    # Override etree.ElementBase.cssselect() and set(), despite the MRO (FIXME: change base order?)
-    cssselect = HtmlMixin.cssselect
-    set = HtmlMixin.set
+class HtmlElement(HtmlMixin, etree.ElementBase):
+    pass
 
 
-class HtmlProcessingInstruction(etree.PIBase, HtmlMixin):
+class HtmlProcessingInstruction(HtmlMixin, etree.PIBase):
     pass
 
 
-class HtmlEntity(etree.EntityBase, HtmlMixin):
+class HtmlEntity(HtmlMixin, etree.EntityBase):
     pass
 
 
diff --git a/src/lxml/html/tests/test_basic.py b/src/lxml/html/tests/test_basic.py
index 6e35c274..464d4747 100644
--- a/src/lxml/html/tests/test_basic.py
+++ b/src/lxml/html/tests/test_basic.py
@@ -1,11 +1,51 @@
+import sys
 import unittest
 from lxml.tests.common_imports import make_doctest, doctest
-import lxml.html
+from lxml import html
+
+class TestBasicFeatures(unittest.TestCase):
+    def test_various_mixins(self):
+        base_url = "http://example.org"
+        doc = html.fromstring("""
+        <root>
+            <!-- comment -->
+            <?pi contents ?>
+            &entity;
+            <el/>
+        </root>
+        """, base_url=base_url)
+        self.assertEqual(doc.getroottree().docinfo.URL, base_url)
+        self.assertEqual(len(doc), 3)
+        self.assertIsInstance(doc[0], html.HtmlComment)
+        self.assertIsInstance(doc[1], html.HtmlProcessingInstruction)
+        self.assertIsInstance(doc[2], html.HtmlElement)
+        for child in doc:
+            # base_url makes sense on all nodes (kinda) whereas `classes` or
+            # `get_rel_links` not really
+            self.assertEqual(child.base_url, base_url)
+
+    def test_set_empty_attribute(self):
+        e = html.Element('e')
+        e.set('a')
+        e.set('b', None)
+        e.set('c', '')
+        self.assertEqual(
+            html.tostring(e),
+            b'<e a b c=""></e>',
+            "Attributes set to `None` should yield empty attributes"
+        )
+        self.assertEqual(e.get('a'), '', "getting the empty attribute results in an empty string")
+        self.assertEqual(e.attrib, {
+            'a': '',
+            'b': '',
+            'c': '',
+        })
 
 def test_suite():
     suite = unittest.TestSuite()
     suite.addTests([make_doctest('test_basic.txt')])
-    suite.addTests([doctest.DocTestSuite(lxml.html)])
+    suite.addTests([doctest.DocTestSuite(html)])
+    suite.addTest(unittest.TestLoader().loadTestsFromModule(sys.modules[__name__]))
     return suite
 
 if __name__ == '__main__':
diff --git a/tox.ini b/tox.ini
index 3906b1de..063a6804 100644
--- a/tox.ini
+++ b/tox.ini
@@ -7,6 +7,7 @@
 envlist = py27, py35, py36, py37, py38, py39, py310
 
 [testenv]
+allowlist_externals = make
 setenv =
     CFLAGS = -g -O0
 commands =