Change "<>" selector to ":scope"

Author: sortafreel

cssselect/parser.py

@@ -400,22 +400,9 @@ def parse_simple_selector(stream, inside_negation=False):
     stream.skip_whitespace()
     selector_start = len(stream.used)
     peek = stream.peek()
-    if peek.type == 'IDENT' or peek == ('DELIM', '*') or peek == ('DELIM', '<'):
+    if peek.type == 'IDENT' or peek == ('DELIM', '*'):
         if peek.type == 'IDENT':
             namespace = stream.next().value
-        elif peek == ('DELIM', '<'):
-            if not (len(stream.used) == 0 or
-                    (len(stream.used) == 1 and stream.used[0].type == 'S')):
-                raise SelectorSyntaxError(
-                    'Got immediate child pseudo-element "<>" not at the start of a selector'
-                )
-            namespace = stream.next().value
-            stream.skip_whitespace()
-            peek = stream.peek()
-            if not peek == ('DELIM', '>'):
-                raise SelectorSyntaxError(
-                    'Got incomplete immediate child pseudo-element "<>" (no ">")'
-                )
         else:
             stream.next()
             namespace = None
@@ -465,6 +452,13 @@ def parse_simple_selector(stream, inside_negation=False):
                 continue
             if stream.peek() != ('DELIM', '('):
                 result = Pseudo(result, ident)
+                if result.ident == '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

@@ -220,18 +220,7 @@ def selector_to_xpath(self, selector, prefix='descendant-or-self::',
         assert isinstance(xpath, self.xpathexpr_cls)  # help debug a missing 'return'
         if translate_pseudo_elements and selector.pseudo_element:
             xpath = self.xpath_pseudo_element(xpath, selector.pseudo_element)
-
-        unicode_xpath = _unicode(xpath)
-        # CSS immediate children (CSS "<> 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
-        # product = response.css('.product')
-        # name = product.css('<> div')
-        child_re = r'^[ \t\r\n\f]*\<[ \t\r\n\f]*\/'
-        if re.match(child_re, unicode_xpath):
-            prefix = 'child::'
-            unicode_xpath = re.sub(child_re, '', unicode_xpath)
-        return (prefix or '') + unicode_xpath
+        return (prefix or '') + _unicode(xpath)
 
     def xpath_pseudo_element(self, xpath, pseudo_element):
         """Translate a pseudo-element.
@@ -343,8 +332,6 @@ def xpath_element(self, selector):
         if not element:
             element = '*'
             safe = True
-        if element == '<':
-            safe = True
         else:
             safe = is_safe_name(element)
             if self.lower_case_element_names:
@@ -554,6 +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')
 

tests/test_cssselect.py

@@ -147,18 +147,19 @@ 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('<> foo') == [
-            'CombinedSelector[Element[<] > Element[foo]]'
+        assert parse_many(':scope > foo') == [
+            'CombinedSelector[Pseudo[Element[*]:scope] > Element[foo]]'
         ]
-        assert parse_many('<> foo bar > div') == [
-            'CombinedSelector[CombinedSelector[CombinedSelector[Element[<] > Element[foo]] '
-            '<followed> Element[bar]] > Element[div]]'
+        assert parse_many(':scope > foo bar > div') == [
+            'CombinedSelector[CombinedSelector[CombinedSelector[Pseudo[Element[*]:scope] > '
+            'Element[foo]] <followed> Element[bar]] > Element[div]]'
         ]
-        assert parse_many('<> #foo #bar') == [
-            'CombinedSelector[CombinedSelector[Element[<] > Hash[Element[*]#foo]] '
-            '<followed> Hash[Element[*]#bar]]'
+        assert parse_many(':scope > #foo #bar') == [
+            'CombinedSelector[CombinedSelector[Pseudo[Element[*]:scope] > '
+            'Hash[Element[*]#foo]] <followed> Hash[Element[*]#bar]]'
         ]
 
+    # TODO ADD TESTS
     def test_pseudo_elements(self):
         def parse_pseudo(css):
             result = []
@@ -179,6 +180,7 @@ def parse_one(css):
         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')
@@ -322,11 +324,9 @@ def get_error(css):
             "Got pseudo-element ::before inside :not() at 12")
         assert get_error(':not(:not(a))') == (
             "Got nested :not()")
-        assert get_error('<> div <> header') == (
-            'Got immediate child pseudo-element "<>" not at the start of a selector'
+        assert get_error(':scope > div :scope header') == (
+            'Got immediate child pseudo-element ":scope" not at the start of a selector'
         )
-        assert get_error('< div p') == (
-            'Got incomplete immediate child pseudo-element "<>" (no ">")')
         assert get_error('> div p') == ("Expected selector, got <DELIM '>' at 0>")
 
     def test_translation(self):
@@ -501,8 +501,8 @@ def test_quoting(self):
             '''descendant-or-self::*[@aval = '"']''')
         assert css_to_xpath('*[aval=\'"""\']') == (
             '''descendant-or-self::*[@aval = '"""']''')
-        assert css_to_xpath('<> div[dataimg="<testmessage>"]') == (
-            "child::div[@dataimg = '<testmessage>']")
+        assert css_to_xpath(':scope > div[dataimg="<testmessage>"]') == (
+            "descendant-or-self::*[1]/div[@dataimg = '<testmessage>']")
 
     def test_unicode_escapes(self):
         # \22 == '"'  \20 == ' '
@@ -580,6 +580,7 @@ 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):
@@ -692,11 +693,11 @@ 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('<> div') == []
-        assert pcss('<> body') == ['nil']
-        assert pcss('<> body > div') == ['outer-div', 'foobar-div']
-        assert pcss('<> head') == ['nil']
-        assert pcss('<> html') == []
+        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-* -------------------------------------
 
@@ -878,7 +879,9 @@ 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('<> div') == 1
+        assert count(':scope > div') == 1
+        assert count(':scope > div > div[class=dialog]') == 1
+        assert count(':scope > div div') == 242
 
 XMLLANG_IDS = '''
 <test>