Implement rudimentary support for the relational pseudo-class :has()

Author: Gallaecio

cssselect/parser.py

@@ -250,6 +250,30 @@ def specificity(self):
         return a1 + a2, b1 + b2, c1 + c2
 
 
+class Relation(object):
+    """
+    Represents selector:has(subselector)
+    """
+    def __init__(self, selector, subselector):
+        self.selector = selector
+        self.subselector = subselector
+
+    def __repr__(self):
+        return '%s[%r:has(%r)]' % (
+            self.__class__.__name__, self.selector, self.subselector)
+
+    def canonical(self):
+        subsel = self.subselector.canonical()
+        if len(subsel) > 1:
+            subsel = subsel.lstrip('*')
+        return '%s:has(%s)' % (self.selector.canonical(), subsel)
+
+    def specificity(self):
+        a1, b1, c1 = self.selector.specificity()
+        a2, b2, c2 = self.subselector.specificity()
+        return a1 + a2, b1 + b2, c1 + c2
+
+
 class Attrib(object):
     """
     Represents selector[namespace|attrib operator value]
@@ -456,7 +480,7 @@ def parse_selector(stream):
     return result, pseudo_element
 
 
-def parse_simple_selector(stream, inside_negation=False):
+def parse_simple_selector(stream, nestable=True):
     stream.skip_whitespace()
     selector_start = len(stream.used)
     peek = stream.peek()
@@ -479,7 +503,7 @@ def parse_simple_selector(stream, inside_negation=False):
     while 1:
         peek = stream.peek()
         if peek.type in ('S', 'EOF') or peek.is_delim(',', '+', '>', '~') or (
-                inside_negation and peek == ('DELIM', ')')):
+                not nestable and peek == ('DELIM', ')')):
             break
         if pseudo_element:
             raise SelectorSyntaxError(
@@ -507,7 +531,8 @@ def parse_simple_selector(stream, inside_negation=False):
                         pseudo_element, parse_arguments(stream))
                 continue
             ident = stream.next_ident()
-            if ident.lower() in ('first-line', 'first-letter',
+            lowercase_indent = ident.lower()
+            if lowercase_indent in ('first-line', 'first-letter',
                                  'before', 'after'):
                 # Special case: CSS 2.1 pseudo-elements can have a single ':'
                 # Any new pseudo-element must have two.
@@ -523,13 +548,16 @@ def parse_simple_selector(stream, inside_negation=False):
                             'Got immediate child pseudo-element ":scope" '
                             'not at the start of a selector')
                 continue
+
             stream.next()
             stream.skip_whitespace()
-            if ident.lower() == 'not':
-                if inside_negation:
-                    raise SelectorSyntaxError('Got nested :not()')
+
+            if lowercase_indent == 'not':
+                if not nestable:
+                    raise SelectorSyntaxError(
+                        'Got :not() within :has() or another :not()')
                 argument, argument_pseudo_element = parse_simple_selector(
-                    stream, inside_negation=True)
+                    stream, nestable=False)
                 next = stream.next()
                 if argument_pseudo_element:
                     raise SelectorSyntaxError(
@@ -538,8 +566,25 @@ def parse_simple_selector(stream, inside_negation=False):
                 if next != ('DELIM', ')'):
                     raise SelectorSyntaxError("Expected ')', got %s" % (next,))
                 result = Negation(result, argument)
-            else:
-                result = Function(result, ident, parse_arguments(stream))
+                continue
+
+            if lowercase_indent == 'has':
+                if not nestable:
+                    raise SelectorSyntaxError(
+                        'Got :has() within :not() or another :has()')
+                argument, argument_pseudo_element = parse_simple_selector(
+                    stream, nestable=False)
+                next = stream.next()
+                if argument_pseudo_element:
+                    raise SelectorSyntaxError(
+                        'Got pseudo-element ::%s inside :has() at %s'
+                        % (argument_pseudo_element, next.pos))
+                if next != ('DELIM', ')'):
+                    raise SelectorSyntaxError("Expected ')', got %s" % (next,))
+                result = Relation(result, argument)
+                continue
+
+            result = Function(result, ident, parse_arguments(stream))
         else:
             raise SelectorSyntaxError(
                 "Expected selector, got %s" % (peek,))

cssselect/xpath.py

@@ -272,6 +272,14 @@ def xpath_negation(self, negation):
         else:
             return xpath.add_condition('0')
 
+    def xpath_relation(self, relation):
+        xpath = self.xpath(relation.selector)
+        sub_xpath = self.xpath(relation.subselector)
+        if sub_xpath.condition:
+            return xpath.add_condition('%s' % sub_xpath.condition)
+        else:
+            return xpath.add_condition('*')
+
     def xpath_function(self, function):
         """Translate a functional pseudo-class."""
         method = 'xpath_%s_function' % function.name.replace('-', '_')

tests/test_cssselect.py

@@ -145,6 +145,8 @@ def parse_many(first, *others):
             'Hash[Element[div]#foobar]']
         assert parse_many('div:not(div.foo)') == [
             'Negation[Element[div]:not(Class[Element[div].foo])]']
+        assert parse_many('div:has(div.foo)') == [
+            'Relation[Element[div]:has(Class[Element[div].foo])]']
         assert parse_many('td ~ th') == [
             'CombinedSelector[Element[td] ~ Element[th]]']
         assert parse_many(':scope > foo') == [
@@ -266,6 +268,13 @@ def specificity(css):
         assert specificity(':not(:empty)') == (0, 1, 0)
         assert specificity(':not(#foo)') == (1, 0, 0)
 
+        assert specificity(':has(*)') == (0, 0, 0)
+        assert specificity(':has(foo)') == (0, 0, 1)
+        assert specificity(':has(.foo)') == (0, 1, 0)
+        assert specificity(':has([foo])') == (0, 1, 0)
+        assert specificity(':has(:empty)') == (0, 1, 0)
+        assert specificity(':has(#foo)') == (1, 0, 0)
+
         assert specificity('foo:empty') == (0, 1, 1)
         assert specificity('foo:before') == (0, 0, 2)
         assert specificity('foo::before') == (0, 0, 2)
@@ -300,6 +309,12 @@ def css2css(css, res=None):
         css2css(':not(*[foo])', ':not([foo])')
         css2css(':not(:empty)')
         css2css(':not(#foo)')
+        css2css(':has(*)')
+        css2css(':has(foo)')
+        css2css(':has(*.foo)', ':has(.foo)')
+        css2css(':has(*[foo])', ':has([foo])')
+        css2css(':has(:empty)')
+        css2css(':has(#foo)')
         css2css('foo:empty')
         css2css('foo::before')
         css2css('foo:empty::before')
@@ -371,8 +386,8 @@ def get_error(css):
             "Got pseudo-element ::before not at the end of a selector")
         assert get_error(':not(:before)') == (
             "Got pseudo-element ::before inside :not() at 12")
-        assert get_error(':not(:not(a))') == (
-            "Got nested :not()")
+        assert get_error(':has(:before)') == (
+            "Got pseudo-element ::before inside :has() at 12")
         assert get_error(':scope > div :scope header') == (
             'Got immediate child pseudo-element ":scope" not at the start of a selector'
         )
@@ -381,6 +396,16 @@ def get_error(css):
         )
         assert get_error('> div p') == ("Expected selector, got <DELIM '>' at 0>")
 
+        # Unsupported nesting
+        assert get_error(':has(:has(a))') == (
+            'Got :has() within :not() or another :has()')
+        assert get_error(':has(:not(a))') == (
+            'Got :not() within :has() or another :not()')
+        assert get_error(':not(:has(a))') == (
+            'Got :has() within :not() or another :has()')
+        assert get_error(':not(:not(a))') == (
+            'Got :not() within :has() or another :not()')
+
     def test_translation(self):
         def xpath(css):
             return _unicode(GenericTranslator().css_to_xpath(css, prefix=''))
@@ -490,6 +515,8 @@ def xpath(css):
             "e[@id = 'myid']")
         assert xpath('e:not(:nth-child(odd))') == (
             "e[not(count(preceding-sibling::*) mod 2 = 0)]")
+        assert xpath('e:has(:nth-child(odd))') == (
+            "e[count(preceding-sibling::*) mod 2 = 0]")
         assert xpath('e:nOT(*)') == (
             "e[0]")  # never matches
         assert xpath('e f') == (
@@ -839,6 +866,9 @@ def pcss(main, *selectors, **kwargs):
         assert pcss('ol :Not(li[class])') == [
             'first-li', 'second-li', 'li-div',
             'fifth-li', 'sixth-li', 'seventh-li']
+        assert pcss('link:has(*)') == []
+        assert pcss('link:has([href])') == ['link-href']
+        assert pcss('ol:has(div)') == ['first-ol']
         assert pcss('ol.a.b.c > li.c:nth-child(3)') == ['third-li']
 
         # Invalid characters in XPath element names, should not crash
@@ -935,6 +965,7 @@ def count(selector):
         assert count(':scope > div > div[class=dialog]') == 1
         assert count(':scope > div div') == 242
 
+
 XMLLANG_IDS = '''
 <test>
   <a id="first" xml:lang="en">a</a>

tox.ini

@@ -6,4 +6,4 @@ deps=
     -r tests/requirements.txt
 
 commands =
-    py.test --cov-report term --cov=cssselect
+    py.test --cov-report term --cov=cssselect {posargs:tests}