Merge pull request #87 from sortafreel/css-immediate-children

CSS immediate children (:scope selector)

Author: dangra

cssselect/parser.py

@@ -452,6 +452,13 @@ def parse_simple_selector(stream, inside_negation=False):
                 continue
             if stream.peek() != ('DELIM', '('):
                 result = Pseudo(result, ident)
+                if result.__repr__() == 'Pseudo[Element[*]:scope]':
+                    if not (len(stream.used) == 2 or
+                            (len(stream.used) == 3
+                             and stream.used[0].type == 'S')):
+                        raise SelectorSyntaxError(
+                            'Got immediate child pseudo-element ":scope" '
+                            'not at the start of a selector')
                 continue
             stream.next()
             stream.skip_whitespace()

cssselect/xpath.py

@@ -541,6 +541,14 @@ def xpath_lang_function(self, xpath, function):
     def xpath_root_pseudo(self, xpath):
         return xpath.add_condition("not(parent::*)")
 
+    # CSS immediate children (CSS ":scope > div" to XPath "child::div" or "./div")
+    # Works only at the start of a selector
+    # Needed to get immediate children of a processed selector in Scrapy
+    # for product in response.css('.product'):
+    #     description = product.css(':scope > div::text').get()
+    def xpath_scope_pseudo(self, xpath):
+        return xpath.add_condition("1")
+
     def xpath_first_child_pseudo(self, xpath):
         return xpath.add_condition('count(preceding-sibling::*) = 0')
 

docs/index.rst

@@ -108,8 +108,10 @@ in the Level 3 specification:
 * ``:not()`` accepts a *sequence of simple selectors*, not just single
   *simple selector*. For example, ``:not(a.important[rel])`` is allowed,
   even though the negation contains 3 *simple selectors*.
+* ``:scope`` allows to access immediate children of a selector: ``product.css(':scope > div::text')``, simillar to XPath ``child::div``. Must be used at the start of a selector. Simplified version of `level 4 reference`_.
 
 .. _an early draft: http://www.w3.org/TR/2001/CR-css3-selectors-20011113/#content-selectors
+.. _level 4 reference: https://developer.mozilla.org/en-US/docs/Web/CSS/:scope
 
 ..
     The following claim was copied from lxml:

tests/test_cssselect.py

@@ -146,6 +146,20 @@ def parse_many(first, *others):
             'Negation[Element[div]:not(Class[Element[div].foo])]']
         assert parse_many('td ~ th') == [
             'CombinedSelector[Element[td] ~ Element[th]]']
+        assert parse_many(':scope > foo') == [
+            'CombinedSelector[Pseudo[Element[*]:scope] > Element[foo]]'
+        ]
+        assert parse_many(' :scope > foo') == [
+            'CombinedSelector[Pseudo[Element[*]:scope] > Element[foo]]'
+        ]
+        assert parse_many(':scope > foo bar > div') == [
+            'CombinedSelector[CombinedSelector[CombinedSelector[Pseudo[Element[*]:scope] > '
+            'Element[foo]] <followed> Element[bar]] > Element[div]]'
+        ]
+        assert parse_many(':scope > #foo #bar') == [
+            'CombinedSelector[CombinedSelector[Pseudo[Element[*]:scope] > '
+            'Hash[Element[*]#foo]] <followed> Hash[Element[*]#bar]]'
+        ]
 
     def test_pseudo_elements(self):
         def parse_pseudo(css):
@@ -164,9 +178,16 @@ def parse_one(css):
             assert len(result) == 1
             return result[0]
 
+        def test_pseudo_repr(css):
+            result = parse(css)
+            assert len(result) == 1
+            selector = result[0]
+            return selector.parsed_tree.__repr__()
+
         assert parse_one('foo') == ('Element[foo]', None)
         assert parse_one('*') == ('Element[*]', None)
         assert parse_one(':empty') == ('Pseudo[Element[*]:empty]', None)
+        assert parse_one(':scope') == ('Pseudo[Element[*]:scope]', None)
 
         # Special cases for CSS 2.1 pseudo-elements
         assert parse_one(':BEfore') == ('Element[*]', 'before')
@@ -190,11 +211,14 @@ def parse_one(css):
             'CombinedSelector[Hash[Element[lorem]#ipsum] ~ '
                 'Pseudo[Attrib[Class[Hash[Element[a]#b].c][href]]:empty]]',
             'selection')
-
-        parse_pseudo('foo:before, bar, baz:after') == [
-            ('Element[foo]', 'before'),
-            ('Element[bar]', None),
-            ('Element[baz]', 'after')]
+        assert parse_pseudo(':scope > div, foo bar') == [
+            ('CombinedSelector[Pseudo[Element[*]:scope] > Element[div]]', None),
+            ('CombinedSelector[Element[foo] <followed> Element[bar]]', None)
+        ]
+        assert parse_pseudo('foo:before, bar, baz:after') == [
+            ('Element[foo]', 'before'), ('Element[bar]', None),
+            ('Element[baz]', 'after')
+        ]
 
         # Special cases for CSS 2.1 pseudo-elements are ignored by default
         for pseudo in ('after', 'before', 'first-line', 'first-letter'):
@@ -211,6 +235,11 @@ def parse_one(css):
         self.assertRaises(ExpressionError, tr.selector_to_xpath, selector,
                           translate_pseudo_elements=True)
 
+        # Special test for the unicode symbols and ':scope' element if check
+        # Errors if use repr() instead of __repr__()
+        assert test_pseudo_repr(u':fİrst-child') == u'Pseudo[Element[*]:fİrst-child]'
+        assert test_pseudo_repr(':scope') == 'Pseudo[Element[*]:scope]'
+
     def test_specificity(self):
         def specificity(css):
             selectors = parse(css)
@@ -310,6 +339,13 @@ def get_error(css):
             "Got pseudo-element ::before inside :not() at 12")
         assert get_error(':not(:not(a))') == (
             "Got nested :not()")
+        assert get_error(':scope > div :scope header') == (
+            'Got immediate child pseudo-element ":scope" not at the start of a selector'
+        )
+        assert get_error('div :scope header') == (
+            'Got immediate child pseudo-element ":scope" not at the start of a selector'
+        )
+        assert get_error('> div p') == ("Expected selector, got <DELIM '>' at 0>")
 
     def test_translation(self):
         def xpath(css):
@@ -483,6 +519,8 @@ def test_quoting(self):
             '''descendant-or-self::*[@aval = '"']''')
         assert css_to_xpath('*[aval=\'"""\']') == (
             '''descendant-or-self::*[@aval = '"""']''')
+        assert css_to_xpath(':scope > div[dataimg="<testmessage>"]') == (
+            "descendant-or-self::*[1]/div[@dataimg = '<testmessage>']")
 
     def test_unicode_escapes(self):
         # \22 == '"'  \20 == ' '
@@ -560,6 +598,7 @@ def xpath(css):
         assert xpath('::attr-href') == "descendant-or-self::*/@href"
         assert xpath('p img::attr(src)') == (
             "descendant-or-self::p/descendant-or-self::*/img/@src")
+        assert xpath(':scope') == "descendant-or-self::*[1]"
 
     def test_series(self):
         def series(css):
@@ -672,6 +711,11 @@ def pcss(main, *selectors, **kwargs):
         assert pcss(':lang("EN")', '*:lang(en-US)', html_only=True) == [
             'second-li', 'li-div']
         assert pcss(':lang("e")', html_only=True) == []
+        assert pcss(':scope > div') == []
+        assert pcss(':scope body') == ['nil']
+        assert pcss(':scope body > div') == ['outer-div', 'foobar-div']
+        assert pcss(':scope head') == ['nil']
+        assert pcss(':scope html') == []
 
         # --- nth-* and nth-last-* -------------------------------------
 
@@ -853,6 +897,9 @@ def count(selector):
         assert count('div[class|=dialog]') == 50 # ? Seems right
         assert count('div[class!=madeup]') == 243 # ? Seems right
         assert count('div[class~=dialog]') == 51 # ? Seems right
+        assert count(':scope > div') == 1
+        assert count(':scope > div > div[class=dialog]') == 1
+        assert count(':scope > div div') == 242
 
 XMLLANG_IDS = '''
 <test>