├── .github ├── FUNDING.yml └── workflows │ ├── doconfly.yml │ ├── release.yml │ └── tests.yml ├── .gitignore ├── LICENSE ├── README.rst ├── cssselect2 ├── __init__.py ├── compiler.py ├── parser.py └── tree.py ├── docs ├── api_reference.rst ├── changelog.rst ├── conf.py ├── contribute.rst ├── example.py ├── first_steps.rst ├── index.rst └── support.rst ├── pyproject.toml └── tests ├── LICENSE ├── __init__.py ├── content.xhtml ├── ids.html ├── make_selectors.py ├── shakespeare.html ├── test_cssselect2.py └── w3_selectors.py /.github/FUNDING.yml: -------------------------------------------------------------------------------- 1 | # These are supported funding model platforms 2 | 3 | open_collective: courtbouillon 4 | -------------------------------------------------------------------------------- /.github/workflows/doconfly.yml: -------------------------------------------------------------------------------- 1 | name: doconfly 2 | on: 3 | push: 4 | branches: 5 | - main 6 | tags: 7 | - "*" 8 | 9 | jobs: 10 | doconfly: 11 | name: doconfly job 12 | runs-on: ubuntu-latest 13 | env: 14 | PORT: ${{ secrets.PORT }} 15 | SSH_PRIVATE_KEY: ${{ secrets.SSH_PRIVATE_KEY }} 16 | TAKOYAKI: ${{ secrets.TAKOYAKI }} 17 | USER: ${{ secrets.USER }} 18 | DOCUMENTATION_PATH: ${{ secrets.DOCUMENTATION_PATH }} 19 | DOCUMENTATION_URL: ${{ secrets.DOCUMENTATION_URL }} 20 | steps: 21 | - run: | 22 | which ssh-agent || ( apt-get update -y && apt-get install openssh-client -y ) 23 | eval $(ssh-agent -s) 24 | echo "$SSH_PRIVATE_KEY" | tr -d '\r' | ssh-add - 25 | mkdir -p ~/.ssh 26 | chmod 700 ~/.ssh 27 | ssh-keyscan -p $PORT $TAKOYAKI >> ~/.ssh/known_hosts 28 | chmod 644 ~/.ssh/known_hosts 29 | ssh $USER@$TAKOYAKI -p $PORT "doconfly/doconfly.sh $GITHUB_REPOSITORY $GITHUB_REF $DOCUMENTATION_PATH $DOCUMENTATION_URL" 30 | -------------------------------------------------------------------------------- /.github/workflows/release.yml: -------------------------------------------------------------------------------- 1 | name: Release new version 2 | on: 3 | push: 4 | tags: 5 | - '*' 6 | 7 | jobs: 8 | pypi-publish: 9 | name: Upload release to PyPI 10 | runs-on: ubuntu-latest 11 | environment: 12 | name: pypi 13 | url: https://pypi.org/p/cssselect2 14 | permissions: 15 | id-token: write 16 | steps: 17 | - uses: actions/checkout@v4 18 | - uses: actions/setup-python@v5 19 | - name: Install requirements 20 | run: python -m pip install flit 21 | - name: Build packages 22 | run: flit build 23 | - name: Publish package distributions to PyPI 24 | uses: pypa/gh-action-pypi-publish@release/v1 25 | add-version: 26 | name: Add version to GitHub 27 | runs-on: ubuntu-latest 28 | permissions: 29 | contents: write 30 | steps: 31 | - name: Checkout code 32 | uses: actions/checkout@v4 33 | - name: Install requirements 34 | run: sudo apt-get install pandoc 35 | - name: Generate content 36 | run: | 37 | pandoc docs/changelog.rst -f rst -t gfm | csplit - /##/ "{1}" -f .part 38 | sed -r "s/^([A-Z].*)\:\$/## \1/" .part01 | sed -r "s/^ *//" | sed -rz "s/([^\n])\n([^\n^-])/\1 \2/g" | tail -n +5 > .body 39 | - name: Create Release 40 | uses: softprops/action-gh-release@v2 41 | with: 42 | body_path: .body 43 | -------------------------------------------------------------------------------- /.github/workflows/tests.yml: -------------------------------------------------------------------------------- 1 | name: cssselect2's tests 2 | on: [push, pull_request] 3 | 4 | jobs: 5 | tests: 6 | name: ${{ matrix.os }} - ${{ matrix.python-version }} 7 | runs-on: ${{ matrix.os }} 8 | strategy: 9 | matrix: 10 | os: [ubuntu-latest, macos-latest, windows-latest] 11 | python-version: ['3.13'] 12 | include: 13 | - os: ubuntu-latest 14 | python-version: '3.9' 15 | - os: ubuntu-latest 16 | python-version: 'pypy-3.10' 17 | steps: 18 | - uses: actions/checkout@v4 19 | with: 20 | submodules: true 21 | - uses: actions/setup-python@v5 22 | with: 23 | python-version: ${{ matrix.python-version }} 24 | - name: Upgrade pip and setuptools 25 | run: python -m pip install --upgrade pip setuptools 26 | - name: Install tests' requirements 27 | run: python -m pip install .[test] 28 | - name: Launch tests 29 | run: python -m pytest 30 | - name: Check coding style 31 | run: python -m ruff check 32 | -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | *.pyc 2 | .cache 3 | /.coverage 4 | /build 5 | /dist 6 | /docs/_build 7 | /pytest_cache 8 | /venv 9 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | BSD 3-Clause License 2 | 3 | Copyright (c) 2012-2018, Simon Sapin and contributors (see AUTHORS). 4 | All rights reserved. 5 | 6 | Redistribution and use in source and binary forms, with or without 7 | modification, are permitted provided that the following conditions are met: 8 | 9 | * Redistributions of source code must retain the above copyright notice, this 10 | list of conditions and the following disclaimer. 11 | 12 | * Redistributions in binary form must reproduce the above copyright notice, 13 | this list of conditions and the following disclaimer in the documentation 14 | and/or other materials provided with the distribution. 15 | 16 | * Neither the name of the copyright holder nor the names of its 17 | contributors may be used to endorse or promote products derived from 18 | this software without specific prior written permission. 19 | 20 | THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" 21 | AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 | IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE 23 | DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE 24 | FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 | DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR 26 | SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER 27 | CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 28 | OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE 29 | OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 30 | -------------------------------------------------------------------------------- /README.rst: -------------------------------------------------------------------------------- 1 | cssselect2 is a straightforward implementation of CSS4 Selectors for markup 2 | documents (HTML, XML, etc.) that can be read by ElementTree-like parsers 3 | (including cElementTree, lxml, html5lib, etc.) 4 | 5 | * Free software: BSD license 6 | * For Python 3.9+, tested on CPython and PyPy 7 | * Documentation: https://doc.courtbouillon.org/cssselect2 8 | * Changelog: https://github.com/Kozea/cssselect2/releases 9 | * Code, issues, tests: https://github.com/Kozea/cssselect2 10 | * Code of conduct: https://www.courtbouillon.org/code-of-conduct.html 11 | * Professional support: https://www.courtbouillon.org 12 | * Donation: https://opencollective.com/courtbouillon 13 | 14 | cssselect2 has been created and developed by Kozea (https://kozea.fr/). 15 | Professional support, maintenance and community management is provided by 16 | CourtBouillon (https://www.courtbouillon.org/). 17 | 18 | Copyrights are retained by their contributors, no copyright assignment is 19 | required to contribute to cssselect2. Unless explicitly stated otherwise, any 20 | contribution intentionally submitted for inclusion is licensed under the BSD 21 | 3-clause license, without any additional terms or conditions. For full 22 | authorship information, see the version control history. 23 | -------------------------------------------------------------------------------- /cssselect2/__init__.py: -------------------------------------------------------------------------------- 1 | """CSS4 selectors for Python. 2 | 3 | cssselect2 is a straightforward implementation of CSS4 Selectors for markup 4 | documents (HTML, XML, etc.) that can be read by ElementTree-like parsers 5 | (including cElementTree, lxml, html5lib, etc.) 6 | 7 | """ 8 | 9 | from webencodings import ascii_lower 10 | 11 | # Classes are imported here to expose them at the top level of the module 12 | from .compiler import compile_selector_list # noqa 13 | from .parser import SelectorError # noqa 14 | from .tree import ElementWrapper # noqa 15 | 16 | VERSION = __version__ = '0.8.0' 17 | 18 | 19 | class Matcher: 20 | """A CSS selectors storage that can match against HTML elements.""" 21 | def __init__(self): 22 | self.id_selectors = {} 23 | self.class_selectors = {} 24 | self.lower_local_name_selectors = {} 25 | self.namespace_selectors = {} 26 | self.lang_attr_selectors = [] 27 | self.other_selectors = [] 28 | self.order = 0 29 | 30 | def add_selector(self, selector, payload): 31 | """Add a selector and its payload to the matcher. 32 | 33 | :param selector: 34 | A :class:`compiler.CompiledSelector` object. 35 | :param payload: 36 | Some data associated to the selector, 37 | such as :class:`declarations ` 38 | parsed from the :attr:`tinycss2.ast.QualifiedRule.content` 39 | of a style rule. 40 | It can be any Python object, 41 | and will be returned as-is by :meth:`match`. 42 | 43 | """ 44 | self.order += 1 45 | 46 | if selector.never_matches: 47 | return 48 | 49 | entry = ( 50 | selector.test, selector.specificity, self.order, selector.pseudo_element, 51 | payload) 52 | if selector.id is not None: 53 | self.id_selectors.setdefault(selector.id, []).append(entry) 54 | elif selector.class_name is not None: 55 | self.class_selectors.setdefault(selector.class_name, []).append(entry) 56 | elif selector.local_name is not None: 57 | self.lower_local_name_selectors.setdefault( 58 | selector.lower_local_name, []).append(entry) 59 | elif selector.namespace is not None: 60 | self.namespace_selectors.setdefault(selector.namespace, []).append(entry) 61 | elif selector.requires_lang_attr: 62 | self.lang_attr_selectors.append(entry) 63 | else: 64 | self.other_selectors.append(entry) 65 | 66 | def match(self, element): 67 | """Match selectors against the given element. 68 | 69 | :param element: 70 | An :class:`ElementWrapper`. 71 | :returns: 72 | A list of the payload objects associated to selectors that match 73 | element, in order of lowest to highest 74 | :attr:`compiler.CompiledSelector` specificity and in order of 75 | addition with :meth:`add_selector` among selectors of equal 76 | specificity. 77 | 78 | """ 79 | relevant_selectors = [] 80 | 81 | if element.id is not None and element.id in self.id_selectors: 82 | self.add_relevant_selectors( 83 | element, self.id_selectors[element.id], relevant_selectors) 84 | 85 | for class_name in element.classes: 86 | if class_name in self.class_selectors: 87 | self.add_relevant_selectors( 88 | element, self.class_selectors[class_name], relevant_selectors) 89 | 90 | lower_name = ascii_lower(element.local_name) 91 | if lower_name in self.lower_local_name_selectors: 92 | self.add_relevant_selectors( 93 | element, self.lower_local_name_selectors[lower_name], 94 | relevant_selectors) 95 | if element.namespace_url in self.namespace_selectors: 96 | self.add_relevant_selectors( 97 | element, self.namespace_selectors[element.namespace_url], 98 | relevant_selectors) 99 | 100 | if 'lang' in element.etree_element.attrib: 101 | self.add_relevant_selectors( 102 | element, self.lang_attr_selectors, relevant_selectors) 103 | 104 | self.add_relevant_selectors(element, self.other_selectors, relevant_selectors) 105 | 106 | relevant_selectors.sort() 107 | return relevant_selectors 108 | 109 | @staticmethod 110 | def add_relevant_selectors(element, selectors, relevant_selectors): 111 | for test, specificity, order, pseudo, payload in selectors: 112 | if test(element): 113 | relevant_selectors.append((specificity, order, pseudo, payload)) 114 | -------------------------------------------------------------------------------- /cssselect2/compiler.py: -------------------------------------------------------------------------------- 1 | import re 2 | from urllib.parse import urlparse 3 | 4 | from tinycss2.nth import parse_nth 5 | from webencodings import ascii_lower 6 | 7 | from . import parser 8 | from .parser import SelectorError 9 | 10 | # http://dev.w3.org/csswg/selectors/#whitespace 11 | split_whitespace = re.compile('[^ \t\r\n\f]+').findall 12 | 13 | 14 | def compile_selector_list(input, namespaces=None): 15 | """Compile a (comma-separated) list of selectors. 16 | 17 | :param input: 18 | A string, or an iterable of tinycss2 component values such as 19 | the :attr:`tinycss2.ast.QualifiedRule.prelude` of a style rule. 20 | :param namespaces: 21 | A optional dictionary of all `namespace prefix declarations 22 | `_ in scope for this selector. 23 | Keys are namespace prefixes as strings, or ``None`` for the default 24 | namespace. 25 | Values are namespace URLs as strings. 26 | If omitted, assume that no prefix is declared. 27 | :returns: 28 | A list of opaque :class:`compiler.CompiledSelector` objects. 29 | 30 | """ 31 | return [CompiledSelector(selector) for selector in parser.parse(input, namespaces)] 32 | 33 | 34 | class CompiledSelector: 35 | """Abstract representation of a selector.""" 36 | def __init__(self, parsed_selector): 37 | source = _compile_node(parsed_selector.parsed_tree) 38 | self.never_matches = source == '0' 39 | eval_globals = { 40 | 'split_whitespace': split_whitespace, 41 | 'ascii_lower': ascii_lower, 42 | 'urlparse': urlparse, 43 | } 44 | self.test = eval('lambda el: ' + source, eval_globals, {}) 45 | self.specificity = parsed_selector.specificity 46 | self.pseudo_element = parsed_selector.pseudo_element 47 | self.id = None 48 | self.class_name = None 49 | self.local_name = None 50 | self.lower_local_name = None 51 | self.namespace = None 52 | self.requires_lang_attr = False 53 | 54 | node = parsed_selector.parsed_tree 55 | if isinstance(node, parser.CombinedSelector): 56 | node = node.right 57 | for simple_selector in node.simple_selectors: 58 | if isinstance(simple_selector, parser.IDSelector): 59 | self.id = simple_selector.ident 60 | elif isinstance(simple_selector, parser.ClassSelector): 61 | self.class_name = simple_selector.class_name 62 | elif isinstance(simple_selector, parser.LocalNameSelector): 63 | self.local_name = simple_selector.local_name 64 | self.lower_local_name = simple_selector.lower_local_name 65 | elif isinstance(simple_selector, parser.NamespaceSelector): 66 | self.namespace = simple_selector.namespace 67 | elif isinstance(simple_selector, parser.AttributeSelector): 68 | if simple_selector.name == 'lang': 69 | self.requires_lang_attr = True 70 | 71 | 72 | def _compile_node(selector): 73 | """Return a boolean expression, as a Python source string. 74 | 75 | When evaluated in a context where the `el` variable is an 76 | :class:`cssselect2.tree.Element` object, tells whether the element is a 77 | subject of `selector`. 78 | 79 | """ 80 | # To avoid precedence-related bugs, any sub-expression that is passed 81 | # around must be "atomic": add parentheses when the top-level would be 82 | # an operator. Bare literals and function calls are fine. 83 | 84 | # 1 and 0 are used for True and False to avoid global lookups. 85 | 86 | if isinstance(selector, parser.CombinedSelector): 87 | left_inside = _compile_node(selector.left) 88 | if left_inside == '0': 89 | return '0' # 0 and x == 0 90 | elif left_inside == '1': 91 | # 1 and x == x, but the element matching 1 still needs to exist. 92 | if selector.combinator in (' ', '>'): 93 | left = 'el.parent is not None' 94 | elif selector.combinator in ('~', '+'): 95 | left = 'el.previous is not None' 96 | else: 97 | raise SelectorError('Unknown combinator', selector.combinator) 98 | # Rebind the `el` name inside a generator-expressions (in a new scope) 99 | # so that 'left_inside' applies to different elements. 100 | elif selector.combinator == ' ': 101 | left = f'any(({left_inside}) for el in el.ancestors)' 102 | elif selector.combinator == '>': 103 | left = ( 104 | f'next(el is not None and ({left_inside}) ' 105 | 'for el in [el.parent])') 106 | elif selector.combinator == '+': 107 | left = ( 108 | f'next(el is not None and ({left_inside}) ' 109 | 'for el in [el.previous])') 110 | elif selector.combinator == '~': 111 | left = f'any(({left_inside}) for el in el.previous_siblings)' 112 | else: 113 | raise SelectorError('Unknown combinator', selector.combinator) 114 | 115 | right = _compile_node(selector.right) 116 | if right == '0': 117 | return '0' # 0 and x == 0 118 | elif right == '1': 119 | return left # 1 and x == x 120 | else: 121 | # Evaluate combinators right to left 122 | return f'({right}) and ({left})' 123 | 124 | elif isinstance(selector, parser.CompoundSelector): 125 | sub_expressions = [ 126 | expr for expr in [ 127 | _compile_node(selector) 128 | for selector in selector.simple_selectors] 129 | if expr != '1'] 130 | if len(sub_expressions) == 1: 131 | return sub_expressions[0] 132 | elif '0' in sub_expressions: 133 | return '0' 134 | elif sub_expressions: 135 | return ' and '.join(f'({el})' for el in sub_expressions) 136 | else: 137 | return '1' # all([]) == True 138 | 139 | elif isinstance(selector, parser.NegationSelector): 140 | sub_expressions = [ 141 | expr for expr in [ 142 | _compile_node(selector.parsed_tree) 143 | for selector in selector.selector_list] 144 | if expr != '1'] 145 | if not sub_expressions: 146 | return '0' 147 | return f'not ({" or ".join(f"({expr})" for expr in sub_expressions)})' 148 | 149 | elif isinstance(selector, parser.RelationalSelector): 150 | sub_expressions = [] 151 | for relative_selector in selector.selector_list: 152 | expression = _compile_node(relative_selector.selector.parsed_tree) 153 | if expression == '0': 154 | continue 155 | if relative_selector.combinator == ' ': 156 | elements = 'list(el.iter_subtree())[1:]' 157 | elif relative_selector.combinator == '>': 158 | elements = 'el.iter_children()' 159 | elif relative_selector.combinator == '+': 160 | elements = 'list(el.iter_next_siblings())[:1]' 161 | elif relative_selector.combinator == '~': 162 | elements = 'el.iter_next_siblings()' 163 | sub_expressions.append(f'(any({expression} for el in {elements}))') 164 | return ' or '.join(sub_expressions) 165 | 166 | elif isinstance(selector, ( 167 | parser.MatchesAnySelector, parser.SpecificityAdjustmentSelector)): 168 | sub_expressions = [ 169 | expr for expr in [ 170 | _compile_node(selector.parsed_tree) 171 | for selector in selector.selector_list] 172 | if expr != '0'] 173 | if not sub_expressions: 174 | return '0' 175 | return ' or '.join(f'({expr})' for expr in sub_expressions) 176 | 177 | elif isinstance(selector, parser.LocalNameSelector): 178 | if selector.lower_local_name == selector.local_name: 179 | return f'el.local_name == {selector.local_name!r}' 180 | else: 181 | return ( 182 | f'el.local_name == ({selector.lower_local_name!r} ' 183 | f'if el.in_html_document else {selector.local_name!r})') 184 | 185 | elif isinstance(selector, parser.NamespaceSelector): 186 | return f'el.namespace_url == {selector.namespace!r}' 187 | 188 | elif isinstance(selector, parser.ClassSelector): 189 | return f'{selector.class_name!r} in el.classes' 190 | 191 | elif isinstance(selector, parser.IDSelector): 192 | return f'el.id == {selector.ident!r}' 193 | 194 | elif isinstance(selector, parser.AttributeSelector): 195 | if selector.namespace is not None: 196 | if selector.namespace: 197 | if selector.name == selector.lower_name: 198 | key = repr(f'{{{selector.namespace}}}{selector.name}') 199 | else: 200 | lower = f'{{{selector.namespace}}}{selector.lower_name}' 201 | name = f'{{{selector.namespace}}}{selector.name}' 202 | key = f'({lower!r} if el.in_html_document else {name!r})' 203 | else: 204 | if selector.name == selector.lower_name: 205 | key = repr(selector.name) 206 | else: 207 | lower, name = selector.lower_name, selector.name 208 | key = f'({lower!r} if el.in_html_document else {name!r})' 209 | value = selector.value 210 | attribute_value = f'el.etree_element.get({key}, "")' 211 | if selector.case_sensitive is False: 212 | value = value.lower() 213 | attribute_value += '.lower()' 214 | if selector.operator is None: 215 | return f'{key} in el.etree_element.attrib' 216 | elif selector.operator == '=': 217 | return ( 218 | f'{key} in el.etree_element.attrib and ' 219 | f'{attribute_value} == {value!r}') 220 | elif selector.operator == '~=': 221 | return ( 222 | '0' if len(value.split()) != 1 or value.strip() != value 223 | else f'{value!r} in split_whitespace({attribute_value})') 224 | elif selector.operator == '|=': 225 | return ( 226 | f'{key} in el.etree_element.attrib and ' 227 | f'{attribute_value} == {value!r} or ' 228 | f'{attribute_value}.startswith({(value + "-")!r})') 229 | elif selector.operator == '^=': 230 | if value: 231 | return f'{attribute_value}.startswith({value!r})' 232 | else: 233 | return '0' 234 | elif selector.operator == '$=': 235 | return ( 236 | f'{attribute_value}.endswith({value!r})' if value else '0') 237 | elif selector.operator == '*=': 238 | return f'{value!r} in {attribute_value}' if value else '0' 239 | else: 240 | raise SelectorError('Unknown attribute operator', selector.operator) 241 | else: # In any namespace 242 | raise NotImplementedError # TODO 243 | 244 | elif isinstance(selector, parser.PseudoClassSelector): 245 | if selector.name in ('link', 'any-link', 'local-link'): 246 | test = html_tag_eq('a', 'area', 'link') 247 | test += ' and el.etree_element.get("href") is not None ' 248 | if selector.name == 'local-link': 249 | test += 'and not urlparse(el.etree_element.get("href")).scheme' 250 | return test 251 | elif selector.name == 'enabled': 252 | input = html_tag_eq( 253 | 'button', 'input', 'select', 'textarea', 'option') 254 | group = html_tag_eq('optgroup', 'menuitem', 'fieldset') 255 | a = html_tag_eq('a', 'area', 'link') 256 | return ( 257 | f'({input} and el.etree_element.get("disabled") is None' 258 | ' and not el.in_disabled_fieldset) or' 259 | f'({group} and el.etree_element.get("disabled") is None) or ' 260 | f'({a} and el.etree_element.get("href") is not None)') 261 | elif selector.name == 'disabled': 262 | input = html_tag_eq( 263 | 'button', 'input', 'select', 'textarea', 'option') 264 | group = html_tag_eq('optgroup', 'menuitem', 'fieldset') 265 | return ( 266 | f'({input} and (el.etree_element.get("disabled") is not None' 267 | ' or el.in_disabled_fieldset)) or' 268 | f'({group} and el.etree_element.get("disabled") is not None)') 269 | elif selector.name == 'checked': 270 | input = html_tag_eq('input', 'menuitem') 271 | option = html_tag_eq('option') 272 | return ( 273 | f'({input} and el.etree_element.get("checked") is not None and' 274 | ' ascii_lower(el.etree_element.get("type", "")) ' 275 | ' in ("checkbox", "radio")) or (' 276 | f'{option} and el.etree_element.get("selected") is not None)') 277 | elif selector.name in ( 278 | 'visited', 'hover', 'active', 'focus', 'focus-within', 279 | 'focus-visible', 'target', 'target-within', 'current', 'past', 280 | 'future', 'playing', 'paused', 'seeking', 'buffering', 281 | 'stalled', 'muted', 'volume-locked', 'user-valid', 282 | 'user-invalid'): 283 | # Not applicable in a static context: never match. 284 | return '0' 285 | elif selector.name in ('root', 'scope'): 286 | return 'el.parent is None' 287 | elif selector.name == 'first-child': 288 | return 'el.index == 0' 289 | elif selector.name == 'last-child': 290 | return 'el.index + 1 == len(el.etree_siblings)' 291 | elif selector.name == 'first-of-type': 292 | return ( 293 | 'all(s.tag != el.etree_element.tag' 294 | ' for s in el.etree_siblings[:el.index])') 295 | elif selector.name == 'last-of-type': 296 | return ( 297 | 'all(s.tag != el.etree_element.tag' 298 | ' for s in el.etree_siblings[el.index + 1:])') 299 | elif selector.name == 'only-child': 300 | return 'len(el.etree_siblings) == 1' 301 | elif selector.name == 'only-of-type': 302 | return ( 303 | 'all(s.tag != el.etree_element.tag or i == el.index' 304 | ' for i, s in enumerate(el.etree_siblings))') 305 | elif selector.name == 'empty': 306 | return 'not (el.etree_children or el.etree_element.text)' 307 | else: 308 | raise SelectorError('Unknown pseudo-class', selector.name) 309 | 310 | elif isinstance(selector, parser.FunctionalPseudoClassSelector): 311 | if selector.name == 'lang': 312 | langs = [] 313 | tokens = [ 314 | token for token in selector.arguments 315 | if token.type not in ('whitespace', 'comment')] 316 | while tokens: 317 | token = tokens.pop(0) 318 | if token.type == 'ident': 319 | langs.append(token.lower_value) 320 | elif token.type == 'string': 321 | langs.append(ascii_lower(token.value)) 322 | else: 323 | raise SelectorError('Invalid arguments for :lang()') 324 | if tokens: 325 | token = tokens.pop(0) 326 | if token.type != 'ident' and token.value != ',': 327 | raise SelectorError('Invalid arguments for :lang()') 328 | return ' or '.join( 329 | f'el.lang == {lang!r} or el.lang.startswith({(lang + "-")!r})' 330 | for lang in langs) 331 | else: 332 | nth = [] 333 | selector_list = [] 334 | current_list = nth 335 | for argument in selector.arguments: 336 | if argument.type == 'ident' and argument.value == 'of': 337 | if current_list is nth: 338 | current_list = selector_list 339 | continue 340 | current_list.append(argument) 341 | 342 | if selector_list: 343 | test = ' and '.join( 344 | _compile_node(selector.parsed_tree) 345 | for selector in parser.parse(selector_list)) 346 | if selector.name == 'nth-child': 347 | count = ( 348 | f'sum(1 for el in el.previous_siblings if ({test}))') 349 | elif selector.name == 'nth-last-child': 350 | count = ( 351 | 'sum(1 for el in' 352 | ' tuple(el.iter_siblings())[el.index + 1:]' 353 | f' if ({test}))') 354 | elif selector.name == 'nth-of-type': 355 | count = ( 356 | 'sum(1 for s in (' 357 | ' el for el in el.previous_siblings' 358 | f' if ({test}))' 359 | ' if s.etree_element.tag == el.etree_element.tag)') 360 | elif selector.name == 'nth-last-of-type': 361 | count = ( 362 | 'sum(1 for s in (' 363 | ' el for el in' 364 | ' tuple(el.iter_siblings())[el.index + 1:]' 365 | f' if ({test}))' 366 | ' if s.etree_element.tag == el.etree_element.tag)') 367 | else: 368 | raise SelectorError('Unknown pseudo-class', selector.name) 369 | count += f'if ({test}) else float("nan")' 370 | else: 371 | if current_list is selector_list: 372 | raise SelectorError( 373 | f'Invalid arguments for :{selector.name}()') 374 | if selector.name == 'nth-child': 375 | count = 'el.index' 376 | elif selector.name == 'nth-last-child': 377 | count = 'len(el.etree_siblings) - el.index - 1' 378 | elif selector.name == 'nth-of-type': 379 | count = ( 380 | 'sum(1 for s in el.etree_siblings[:el.index]' 381 | ' if s.tag == el.etree_element.tag)') 382 | elif selector.name == 'nth-last-of-type': 383 | count = ( 384 | 'sum(1 for s in el.etree_siblings[el.index + 1:]' 385 | ' if s.tag == el.etree_element.tag)') 386 | else: 387 | raise SelectorError('Unknown pseudo-class', selector.name) 388 | 389 | result = parse_nth(nth) 390 | if result is None: 391 | raise SelectorError( 392 | f'Invalid arguments for :{selector.name}()') 393 | a, b = result 394 | # x is the number of siblings before/after the element 395 | # Matches if a positive or zero integer n exists so that: 396 | # x = a*n + b-1 397 | # x = a*n + B 398 | B = b - 1 # noqa: N806 399 | if a == 0: 400 | # x = B 401 | return f'({count}) == {B}' 402 | else: 403 | # n = (x - B) / a 404 | return ( 405 | 'next(r == 0 and n >= 0' 406 | f' for n, r in [divmod(({count}) - {B}, {a})])') 407 | 408 | else: 409 | raise TypeError(type(selector), selector) 410 | 411 | 412 | def html_tag_eq(*local_names): 413 | """Generate expression testing equality with HTML local names.""" 414 | if len(local_names) == 1: 415 | tag = f'{{http://www.w3.org/1999/xhtml}}{local_names[0]}' 416 | return ( 417 | f'((el.local_name == {local_names[0]!r}) if el.in_html_document ' 418 | f'else (el.etree_element.tag == {tag!r}))') 419 | else: 420 | names = ', '.join(repr(n) for n in local_names) 421 | tags = ', '.join( 422 | repr(f'{{http://www.w3.org/1999/xhtml}}{name}') 423 | for name in local_names) 424 | return ( 425 | f'((el.local_name in ({names})) if el.in_html_document ' 426 | f'else (el.etree_element.tag in ({tags})))') 427 | -------------------------------------------------------------------------------- /cssselect2/parser.py: -------------------------------------------------------------------------------- 1 | from tinycss2 import parse_component_value_list 2 | 3 | __all__ = ['parse'] 4 | 5 | SUPPORTED_PSEUDO_ELEMENTS = { 6 | # As per CSS Pseudo-Elements Module Level 4 7 | 'first-line', 'first-letter', 'prefix', 'postfix', 'selection', 8 | 'target-text', 'spelling-error', 'grammar-error', 'before', 'after', 9 | 'marker', 'placeholder', 'file-selector-button', 10 | # As per CSS Generated Content for Paged Media Module 11 | 'footnote-call', 'footnote-marker', 12 | # As per CSS Scoping Module Level 1 13 | 'content', 'shadow', 14 | } 15 | 16 | 17 | def parse(input, namespaces=None, forgiving=False, relative=False): 18 | """Yield tinycss2 selectors found in given ``input``. 19 | 20 | :param input: 21 | A string, or an iterable of tinycss2 component values. 22 | 23 | """ 24 | if isinstance(input, str): 25 | input = parse_component_value_list(input) 26 | tokens = TokenStream(input) 27 | namespaces = namespaces or {} 28 | try: 29 | yield parse_selector(tokens, namespaces, relative) 30 | except SelectorError as exception: 31 | if forgiving: 32 | return 33 | raise exception 34 | while 1: 35 | next = tokens.next() 36 | if next is None: 37 | return 38 | elif next == ',': 39 | try: 40 | yield parse_selector(tokens, namespaces, relative) 41 | except SelectorError as exception: 42 | if not forgiving: 43 | raise exception 44 | else: 45 | if not forgiving: 46 | raise SelectorError(next, f'unexpected {next.type} token.') 47 | 48 | 49 | def parse_selector(tokens, namespaces, relative=False): 50 | tokens.skip_whitespace_and_comment() 51 | if relative: 52 | peek = tokens.peek() 53 | if peek in ('>', '+', '~'): 54 | initial_combinator = peek.value 55 | tokens.next() 56 | else: 57 | initial_combinator = ' ' 58 | tokens.skip_whitespace_and_comment() 59 | result, pseudo_element = parse_compound_selector(tokens, namespaces) 60 | while 1: 61 | has_whitespace = tokens.skip_whitespace() 62 | while tokens.skip_comment(): 63 | has_whitespace = tokens.skip_whitespace() or has_whitespace 64 | selector = Selector(result, pseudo_element) 65 | if relative: 66 | selector = RelativeSelector(initial_combinator, selector) 67 | if pseudo_element is not None: 68 | return selector 69 | peek = tokens.peek() 70 | if peek is None or peek == ',': 71 | return selector 72 | elif peek in ('>', '+', '~'): 73 | combinator = peek.value 74 | tokens.next() 75 | elif has_whitespace: 76 | combinator = ' ' 77 | else: 78 | return selector 79 | compound, pseudo_element = parse_compound_selector(tokens, namespaces) 80 | result = CombinedSelector(result, combinator, compound) 81 | 82 | 83 | def parse_compound_selector(tokens, namespaces): 84 | type_selectors = parse_type_selector(tokens, namespaces) 85 | simple_selectors = type_selectors if type_selectors is not None else [] 86 | while 1: 87 | simple_selector, pseudo_element = parse_simple_selector( 88 | tokens, namespaces) 89 | if pseudo_element is not None or simple_selector is None: 90 | break 91 | simple_selectors.append(simple_selector) 92 | 93 | if simple_selectors or (type_selectors, pseudo_element) != (None, None): 94 | return CompoundSelector(simple_selectors), pseudo_element 95 | 96 | peek = tokens.peek() 97 | peek_type = peek.type if peek else 'EOF' 98 | raise SelectorError(peek, f'expected a compound selector, got {peek_type}') 99 | 100 | 101 | def parse_type_selector(tokens, namespaces): 102 | tokens.skip_whitespace() 103 | qualified_name = parse_qualified_name(tokens, namespaces) 104 | if qualified_name is None: 105 | return None 106 | 107 | simple_selectors = [] 108 | namespace, local_name = qualified_name 109 | if local_name is not None: 110 | simple_selectors.append(LocalNameSelector(local_name)) 111 | if namespace is not None: 112 | simple_selectors.append(NamespaceSelector(namespace)) 113 | return simple_selectors 114 | 115 | 116 | def parse_simple_selector(tokens, namespaces): 117 | peek = tokens.peek() 118 | if peek is None: 119 | return None, None 120 | if peek.type == 'hash' and peek.is_identifier: 121 | tokens.next() 122 | return IDSelector(peek.value), None 123 | elif peek == '.': 124 | tokens.next() 125 | next = tokens.next() 126 | if next is None or next.type != 'ident': 127 | raise SelectorError(next, f'Expected a class name, got {next}') 128 | return ClassSelector(next.value), None 129 | elif peek.type == '[] block': 130 | tokens.next() 131 | attr = parse_attribute_selector(TokenStream(peek.content), namespaces) 132 | return attr, None 133 | elif peek == ':': 134 | tokens.next() 135 | next = tokens.next() 136 | if next == ':': 137 | next = tokens.next() 138 | if next is None or next.type != 'ident': 139 | raise SelectorError(next, f'Expected a pseudo-element name, got {next}') 140 | value = next.lower_value 141 | if value not in SUPPORTED_PSEUDO_ELEMENTS: 142 | raise SelectorError( 143 | next, f'Expected a supported pseudo-element, got {value}') 144 | return None, value 145 | elif next is not None and next.type == 'ident': 146 | name = next.lower_value 147 | if name in ('before', 'after', 'first-line', 'first-letter'): 148 | return None, name 149 | else: 150 | return PseudoClassSelector(name), None 151 | elif next is not None and next.type == 'function': 152 | name = next.lower_name 153 | if name in ('is', 'where', 'not', 'has'): 154 | return parse_logical_combination(next, namespaces, name), None 155 | else: 156 | return (FunctionalPseudoClassSelector(name, next.arguments), None) 157 | else: 158 | raise SelectorError(next, f'unexpected {next} token.') 159 | else: 160 | return None, None 161 | 162 | 163 | def parse_logical_combination(matches_any_token, namespaces, name): 164 | forgiving = True 165 | relative = False 166 | if name == 'is': 167 | selector_class = MatchesAnySelector 168 | elif name == 'where': 169 | selector_class = SpecificityAdjustmentSelector 170 | elif name == 'not': 171 | forgiving = False 172 | selector_class = NegationSelector 173 | elif name == 'has': 174 | relative = True 175 | selector_class = RelationalSelector 176 | 177 | selectors = [ 178 | selector for selector in 179 | parse(matches_any_token.arguments, namespaces, forgiving, relative) 180 | if selector.pseudo_element is None] 181 | return selector_class(selectors) 182 | 183 | 184 | def parse_attribute_selector(tokens, namespaces): 185 | tokens.skip_whitespace() 186 | qualified_name = parse_qualified_name(tokens, namespaces, is_attribute=True) 187 | if qualified_name is None: 188 | next = tokens.next() 189 | raise SelectorError(next, f'expected attribute name, got {next}') 190 | namespace, local_name = qualified_name 191 | 192 | tokens.skip_whitespace() 193 | peek = tokens.peek() 194 | if peek is None: 195 | operator = None 196 | value = None 197 | elif peek in ('=', '~=', '|=', '^=', '$=', '*='): 198 | operator = peek.value 199 | tokens.next() 200 | tokens.skip_whitespace() 201 | next = tokens.next() 202 | if next is None or next.type not in ('ident', 'string'): 203 | next_type = 'None' if next is None else next.type 204 | raise SelectorError(next, f'expected attribute value, got {next_type}') 205 | value = next.value 206 | else: 207 | raise SelectorError(peek, f'expected attribute selector operator, got {peek}') 208 | 209 | tokens.skip_whitespace() 210 | next = tokens.next() 211 | case_sensitive = None 212 | if next is not None: 213 | if next.type == 'ident' and next.value.lower() == 'i': 214 | case_sensitive = False 215 | elif next.type == 'ident' and next.value.lower() == 's': 216 | case_sensitive = True 217 | else: 218 | raise SelectorError(next, f'expected ], got {next.type}') 219 | return AttributeSelector(namespace, local_name, operator, value, case_sensitive) 220 | 221 | 222 | def parse_qualified_name(tokens, namespaces, is_attribute=False): 223 | """Return ``(namespace, local)`` for given tokens. 224 | 225 | Can also return ``None`` for a wildcard. 226 | 227 | The empty string for ``namespace`` means "no namespace". 228 | 229 | """ 230 | peek = tokens.peek() 231 | if peek is None: 232 | return None 233 | if peek.type == 'ident': 234 | first_ident = tokens.next() 235 | peek = tokens.peek() 236 | if peek != '|': 237 | namespace = '' if is_attribute else namespaces.get(None, None) 238 | return namespace, (first_ident.value, first_ident.lower_value) 239 | tokens.next() 240 | namespace = namespaces.get(first_ident.value) 241 | if namespace is None: 242 | raise SelectorError( 243 | first_ident, f'undefined namespace prefix: {first_ident.value}') 244 | elif peek == '*': 245 | next = tokens.next() 246 | peek = tokens.peek() 247 | if peek != '|': 248 | if is_attribute: 249 | raise SelectorError(next, f'expected local name, got {next.type}') 250 | return namespaces.get(None, None), None 251 | tokens.next() 252 | namespace = None 253 | elif peek == '|': 254 | tokens.next() 255 | namespace = '' 256 | else: 257 | return None 258 | 259 | # If we get here, we just consumed '|' and set ``namespace`` 260 | next = tokens.next() 261 | if next.type == 'ident': 262 | return namespace, (next.value, next.lower_value) 263 | elif next == '*' and not is_attribute: 264 | return namespace, None 265 | else: 266 | raise SelectorError(next, f'expected local name, got {next.type}') 267 | 268 | 269 | class SelectorError(ValueError): 270 | """A specialized ``ValueError`` for invalid selectors.""" 271 | 272 | 273 | class TokenStream: 274 | def __init__(self, tokens): 275 | self.tokens = iter(tokens) 276 | self.peeked = [] # In reversed order 277 | 278 | def next(self): 279 | if self.peeked: 280 | return self.peeked.pop() 281 | else: 282 | return next(self.tokens, None) 283 | 284 | def peek(self): 285 | if not self.peeked: 286 | self.peeked.append(next(self.tokens, None)) 287 | return self.peeked[-1] 288 | 289 | def skip(self, skip_types): 290 | found = False 291 | while 1: 292 | peek = self.peek() 293 | if peek is None or peek.type not in skip_types: 294 | break 295 | self.next() 296 | found = True 297 | return found 298 | 299 | def skip_whitespace(self): 300 | return self.skip(['whitespace']) 301 | 302 | def skip_comment(self): 303 | return self.skip(['comment']) 304 | 305 | def skip_whitespace_and_comment(self): 306 | return self.skip(['comment', 'whitespace']) 307 | 308 | 309 | class Selector: 310 | def __init__(self, tree, pseudo_element=None): 311 | self.parsed_tree = tree 312 | self.pseudo_element = pseudo_element 313 | if pseudo_element is None: 314 | #: Tuple of 3 integers: http://www.w3.org/TR/selectors/#specificity 315 | self.specificity = tree.specificity 316 | else: 317 | a, b, c = tree.specificity 318 | self.specificity = a, b, c + 1 319 | 320 | def __repr__(self): 321 | pseudo = f'::{self.pseudo_element}' if self.pseudo_element else '' 322 | return f'{self.parsed_tree!r}{pseudo}' 323 | 324 | 325 | class RelativeSelector: 326 | def __init__(self, combinator, selector): 327 | self.combinator = combinator 328 | self.selector = selector 329 | 330 | @property 331 | def specificity(self): 332 | return self.selector.specificity 333 | 334 | @property 335 | def pseudo_element(self): 336 | return self.selector.pseudo_element 337 | 338 | def __repr__(self): 339 | return ( 340 | f'{self.selector!r}' if self.combinator == ' ' 341 | else f'{self.combinator} {self.selector!r}') 342 | 343 | 344 | class CombinedSelector: 345 | def __init__(self, left, combinator, right): 346 | #: Combined or compound selector 347 | self.left = left 348 | # One of `` `` (a single space), ``>``, ``+`` or ``~``. 349 | self.combinator = combinator 350 | #: compound selector 351 | self.right = right 352 | 353 | @property 354 | def specificity(self): 355 | a1, b1, c1 = self.left.specificity 356 | a2, b2, c2 = self.right.specificity 357 | return a1 + a2, b1 + b2, c1 + c2 358 | 359 | def __repr__(self): 360 | return f'{self.left!r}{self.combinator}{self.right!r}' 361 | 362 | 363 | class CompoundSelector: 364 | def __init__(self, simple_selectors): 365 | self.simple_selectors = simple_selectors 366 | 367 | @property 368 | def specificity(self): 369 | if self.simple_selectors: 370 | # zip(*foo) turns [(a1, b1, c1), (a2, b2, c2), ...] 371 | # into [(a1, a2, ...), (b1, b2, ...), (c1, c2, ...)] 372 | return tuple(map(sum, zip( 373 | *(sel.specificity for sel in self.simple_selectors)))) 374 | else: 375 | return 0, 0, 0 376 | 377 | def __repr__(self): 378 | return ''.join(map(repr, self.simple_selectors)) 379 | 380 | 381 | class LocalNameSelector: 382 | specificity = 0, 0, 1 383 | 384 | def __init__(self, local_name): 385 | self.local_name, self.lower_local_name = local_name 386 | 387 | def __repr__(self): 388 | return self.local_name 389 | 390 | 391 | class NamespaceSelector: 392 | specificity = 0, 0, 0 393 | 394 | def __init__(self, namespace): 395 | #: The namespace URL as a string, 396 | #: or the empty string for elements not in any namespace. 397 | self.namespace = namespace 398 | 399 | def __repr__(self): 400 | return '|' if self.namespace == '' else f'{{{self.namespace}}}|' 401 | 402 | 403 | class IDSelector: 404 | specificity = 1, 0, 0 405 | 406 | def __init__(self, ident): 407 | self.ident = ident 408 | 409 | def __repr__(self): 410 | return f'#{self.ident}' 411 | 412 | 413 | class ClassSelector: 414 | specificity = 0, 1, 0 415 | 416 | def __init__(self, class_name): 417 | self.class_name = class_name 418 | 419 | def __repr__(self): 420 | return f'.{self.class_name}' 421 | 422 | 423 | class AttributeSelector: 424 | specificity = 0, 1, 0 425 | 426 | def __init__(self, namespace, name, operator, value, case_sensitive): 427 | self.namespace = namespace 428 | self.name, self.lower_name = name 429 | #: A string like ``=`` or ``~=``, or None for ``[attr]`` selectors 430 | self.operator = operator 431 | #: A string, or None for ``[attr]`` selectors 432 | self.value = value 433 | #: ``True`` if case-sensitive, ``False`` if case-insensitive, ``None`` 434 | #: if depends on the document language 435 | self.case_sensitive = case_sensitive 436 | 437 | def __repr__(self): 438 | namespace = '*|' if self.namespace is None else f'{{{self.namespace}}}' 439 | case_sensitive = ( 440 | '' if self.case_sensitive is None else 441 | f' {"s" if self.case_sensitive else "i"}') 442 | return ( 443 | f'[{namespace}{self.name}{self.operator}{self.value!r}' 444 | f'{case_sensitive}]') 445 | 446 | 447 | class PseudoClassSelector: 448 | specificity = 0, 1, 0 449 | 450 | def __init__(self, name): 451 | self.name = name 452 | 453 | def __repr__(self): 454 | return ':' + self.name 455 | 456 | 457 | class FunctionalPseudoClassSelector: 458 | specificity = 0, 1, 0 459 | 460 | def __init__(self, name, arguments): 461 | self.name = name 462 | self.arguments = arguments 463 | 464 | def __repr__(self): 465 | return f':{self.name}{tuple(self.arguments)!r}' 466 | 467 | 468 | class NegationSelector: 469 | def __init__(self, selector_list): 470 | self.selector_list = selector_list 471 | 472 | @property 473 | def specificity(self): 474 | if self.selector_list: 475 | return max(selector.specificity for selector in self.selector_list) 476 | else: 477 | return (0, 0, 0) 478 | 479 | def __repr__(self): 480 | return f':not({", ".join(repr(sel) for sel in self.selector_list)})' 481 | 482 | 483 | class RelationalSelector: 484 | def __init__(self, selector_list): 485 | self.selector_list = selector_list 486 | 487 | @property 488 | def specificity(self): 489 | if self.selector_list: 490 | return max(selector.specificity for selector in self.selector_list) 491 | else: 492 | return (0, 0, 0) 493 | 494 | def __repr__(self): 495 | return f':has({", ".join(repr(sel) for sel in self.selector_list)})' 496 | 497 | 498 | class MatchesAnySelector: 499 | def __init__(self, selector_list): 500 | self.selector_list = selector_list 501 | 502 | @property 503 | def specificity(self): 504 | if self.selector_list: 505 | return max(selector.specificity for selector in self.selector_list) 506 | else: 507 | return (0, 0, 0) 508 | 509 | def __repr__(self): 510 | return f':is({", ".join(repr(sel) for sel in self.selector_list)})' 511 | 512 | 513 | class SpecificityAdjustmentSelector: 514 | def __init__(self, selector_list): 515 | self.selector_list = selector_list 516 | 517 | @property 518 | def specificity(self): 519 | return (0, 0, 0) 520 | 521 | def __repr__(self): 522 | return f':where({", ".join(repr(sel) for sel in self.selector_list)})' 523 | -------------------------------------------------------------------------------- /cssselect2/tree.py: -------------------------------------------------------------------------------- 1 | from functools import cached_property 2 | from warnings import warn 3 | 4 | from webencodings import ascii_lower 5 | 6 | from .compiler import compile_selector_list, split_whitespace 7 | 8 | 9 | class ElementWrapper: 10 | """Wrapper of :class:`xml.etree.ElementTree.Element` for Selector matching. 11 | 12 | This class should not be instanciated directly. :meth:`from_xml_root` or 13 | :meth:`from_html_root` should be used for the root element of a document, 14 | and other elements should be accessed (and wrappers generated) using 15 | methods such as :meth:`iter_children` and :meth:`iter_subtree`. 16 | 17 | :class:`ElementWrapper` objects compare equal if their underlying 18 | :class:`xml.etree.ElementTree.Element` do. 19 | 20 | """ 21 | @classmethod 22 | def from_xml_root(cls, root, content_language=None): 23 | """Wrap for selector matching the root of an XML or XHTML document. 24 | 25 | :param root: 26 | An ElementTree :class:`xml.etree.ElementTree.Element` 27 | for the root element of a document. 28 | If the given element is not the root, 29 | selector matching will behave is if it were. 30 | In other words, selectors will be not be `scoped`_ 31 | to the subtree rooted at that element. 32 | :returns: 33 | A new :class:`ElementWrapper` 34 | 35 | .. _scoped: https://drafts.csswg.org/selectors-4/#scoping 36 | 37 | """ 38 | return cls._from_root(root, content_language, in_html_document=False) 39 | 40 | @classmethod 41 | def from_html_root(cls, root, content_language=None): 42 | """Same as :meth:`from_xml_root` with case-insensitive attribute names. 43 | 44 | Useful for documents parsed with an HTML parser like html5lib, which 45 | should be the case of documents with the ``text/html`` MIME type. 46 | 47 | """ 48 | return cls._from_root(root, content_language, in_html_document=True) 49 | 50 | @classmethod 51 | def _from_root(cls, root, content_language, in_html_document=True): 52 | if hasattr(root, 'getroot'): 53 | root = root.getroot() 54 | return cls( 55 | root, parent=None, index=0, previous=None, 56 | in_html_document=in_html_document, content_language=content_language) 57 | 58 | def __init__(self, etree_element, parent, index, previous, 59 | in_html_document, content_language=None): 60 | #: The underlying ElementTree :class:`xml.etree.ElementTree.Element` 61 | self.etree_element = etree_element 62 | #: The parent :class:`ElementWrapper`, 63 | #: or :obj:`None` for the root element. 64 | self.parent = parent 65 | #: The previous sibling :class:`ElementWrapper`, 66 | #: or :obj:`None` for the root element. 67 | self.previous = previous 68 | if parent is not None: 69 | #: The :attr:`parent`’s children 70 | #: as a list of 71 | #: ElementTree :class:`xml.etree.ElementTree.Element`\ s. 72 | #: For the root (which has no parent) 73 | self.etree_siblings = parent.etree_children 74 | else: 75 | self.etree_siblings = [etree_element] 76 | #: The position within the :attr:`parent`’s children, counting from 0. 77 | #: ``e.etree_siblings[e.index]`` is always ``e.etree_element``. 78 | self.index = index 79 | self.in_html_document = in_html_document 80 | self.transport_content_language = content_language 81 | 82 | # Cache 83 | self._ancestors = None 84 | self._previous_siblings = None 85 | 86 | def __eq__(self, other): 87 | return ( 88 | type(self) is type(other) and 89 | self.etree_element == other.etree_element) 90 | 91 | def __ne__(self, other): 92 | return not (self == other) 93 | 94 | def __hash__(self): 95 | return hash((type(self), self.etree_element)) 96 | 97 | def __iter__(self): 98 | yield from self.iter_children() 99 | 100 | @property 101 | def ancestors(self): 102 | """Tuple of existing ancestors. 103 | 104 | Tuple of existing :class:`ElementWrapper` objects for this element’s 105 | ancestors, in reversed tree order, from :attr:`parent` to the root. 106 | 107 | """ 108 | if self._ancestors is None: 109 | self._ancestors = ( 110 | () if self.parent is None else (*self.parent.ancestors, self.parent)) 111 | return self._ancestors 112 | 113 | @property 114 | def previous_siblings(self): 115 | """Tuple of previous siblings. 116 | 117 | Tuple of existing :class:`ElementWrapper` objects for this element’s 118 | previous siblings, in reversed tree order. 119 | 120 | """ 121 | if self._previous_siblings is None: 122 | self._previous_siblings = ( 123 | () if self.previous is None else 124 | (*self.previous.previous_siblings, self.previous)) 125 | return self._previous_siblings 126 | 127 | def iter_ancestors(self): 128 | """Iterate over ancestors. 129 | 130 | Return an iterator of existing :class:`ElementWrapper` objects for this 131 | element’s ancestors, in reversed tree order (from :attr:`parent` to the 132 | root). 133 | 134 | The element itself is not included, this is an empty sequence for the 135 | root element. 136 | 137 | This method is deprecated and will be removed in version 0.7.0. Use 138 | :attr:`ancestors` instead. 139 | 140 | """ 141 | warn( 142 | 'This method is deprecated and will be removed in version 0.7.0. ' 143 | 'Use the "ancestors" attribute instead.', 144 | DeprecationWarning) 145 | yield from self.ancestors 146 | 147 | def iter_previous_siblings(self): 148 | """Iterate over previous siblings. 149 | 150 | Return an iterator of existing :class:`ElementWrapper` objects for this 151 | element’s previous siblings, in reversed tree order. 152 | 153 | The element itself is not included, this is an empty sequence for a 154 | first child or the root element. 155 | 156 | This method is deprecated and will be removed in version 0.7.0. Use 157 | :attr:`previous_siblings` instead. 158 | 159 | """ 160 | warn( 161 | 'This method is deprecated and will be removed in version 0.7.0. ' 162 | 'Use the "previous_siblings" attribute instead.', 163 | DeprecationWarning) 164 | yield from self.previous_siblings 165 | 166 | def iter_siblings(self): 167 | """Iterate over siblings. 168 | 169 | Return an iterator of newly-created :class:`ElementWrapper` objects for 170 | this element’s siblings, in tree order. 171 | 172 | """ 173 | if self.parent is None: 174 | yield self 175 | else: 176 | yield from self.parent.iter_children() 177 | 178 | def iter_next_siblings(self): 179 | """Iterate over next siblings. 180 | 181 | Return an iterator of newly-created :class:`ElementWrapper` objects for 182 | this element’s next siblings, in tree order. 183 | 184 | """ 185 | found = False 186 | for sibling in self.iter_siblings(): 187 | if found: 188 | yield sibling 189 | if sibling == self: 190 | found = True 191 | 192 | def iter_children(self): 193 | """Iterate over children. 194 | 195 | Return an iterator of newly-created :class:`ElementWrapper` objects for 196 | this element’s child elements, in tree order. 197 | 198 | """ 199 | child = None 200 | for i, etree_child in enumerate(self.etree_children): 201 | child = type(self)( 202 | etree_child, parent=self, index=i, previous=child, 203 | in_html_document=self.in_html_document) 204 | yield child 205 | 206 | def iter_subtree(self): 207 | """Iterate over subtree. 208 | 209 | Return an iterator of newly-created :class:`ElementWrapper` objects for 210 | the entire subtree rooted at this element, in tree order. 211 | 212 | Unlike in other methods, the element itself *is* included. 213 | 214 | This loops over an entire document: 215 | 216 | .. code-block:: python 217 | 218 | for element in ElementWrapper.from_root(root_etree).iter_subtree(): 219 | ... 220 | 221 | """ 222 | stack = [iter([self])] 223 | while stack: 224 | element = next(stack[-1], None) 225 | if element is None: 226 | stack.pop() 227 | else: 228 | yield element 229 | stack.append(element.iter_children()) 230 | 231 | @staticmethod 232 | def _compile(selectors): 233 | return [ 234 | compiled_selector.test 235 | for selector in selectors 236 | for compiled_selector in ( 237 | [selector] if hasattr(selector, 'test') 238 | else compile_selector_list(selector)) 239 | if compiled_selector.pseudo_element is None and 240 | not compiled_selector.never_matches] 241 | 242 | def matches(self, *selectors): 243 | """Return wether this elememt matches any of the given selectors. 244 | 245 | :param selectors: 246 | Each given selector is either a :class:`compiler.CompiledSelector`, 247 | or an argument to :func:`compile_selector_list`. 248 | 249 | """ 250 | return any(test(self) for test in self._compile(selectors)) 251 | 252 | def query_all(self, *selectors): 253 | """Return elements, in tree order, that match any of given selectors. 254 | 255 | Selectors are `scoped`_ to the subtree rooted at this element. 256 | 257 | .. _scoped: https://drafts.csswg.org/selectors-4/#scoping 258 | 259 | :param selectors: 260 | Each given selector is either a :class:`compiler.CompiledSelector`, 261 | or an argument to :func:`compile_selector_list`. 262 | :returns: 263 | An iterator of newly-created :class:`ElementWrapper` objects. 264 | 265 | """ 266 | tests = self._compile(selectors) 267 | if len(tests) == 1: 268 | return filter(tests[0], self.iter_subtree()) 269 | elif selectors: 270 | return ( 271 | element for element in self.iter_subtree() 272 | if any(test(element) for test in tests)) 273 | else: 274 | return iter(()) 275 | 276 | def query(self, *selectors): 277 | """Return first element that matches any of given selectors. 278 | 279 | :param selectors: 280 | Each given selector is either a :class:`compiler.CompiledSelector`, 281 | or an argument to :func:`compile_selector_list`. 282 | :returns: 283 | A newly-created :class:`ElementWrapper` object, 284 | or :obj:`None` if there is no match. 285 | 286 | """ 287 | return next(self.query_all(*selectors), None) 288 | 289 | @cached_property 290 | def etree_children(self): 291 | """Children as a list of :class:`xml.etree.ElementTree.Element`. 292 | 293 | Other ElementTree nodes such as 294 | :func:`comments ` and 295 | :func:`processing instructions 296 | ` 297 | are not included. 298 | 299 | """ 300 | return [ 301 | element for element in self.etree_element 302 | if isinstance(element.tag, str)] 303 | 304 | @cached_property 305 | def local_name(self): 306 | """The local name of this element, as a string.""" 307 | namespace_url, local_name = _split_etree_tag(self.etree_element.tag) 308 | self.__dict__[str('namespace_url')] = namespace_url 309 | return local_name 310 | 311 | @cached_property 312 | def namespace_url(self): 313 | """The namespace URL of this element, as a string.""" 314 | namespace_url, local_name = _split_etree_tag(self.etree_element.tag) 315 | self.__dict__[str('local_name')] = local_name 316 | return namespace_url 317 | 318 | @cached_property 319 | def id(self): 320 | """The ID of this element, as a string.""" 321 | return self.etree_element.get('id') 322 | 323 | @cached_property 324 | def classes(self): 325 | """The classes of this element, as a :class:`set` of strings.""" 326 | return set(split_whitespace(self.etree_element.get('class', ''))) 327 | 328 | @cached_property 329 | def lang(self): 330 | """The language of this element, as a string.""" 331 | # http://whatwg.org/C#language 332 | xml_lang = self.etree_element.get('{http://www.w3.org/XML/1998/namespace}lang') 333 | if xml_lang is not None: 334 | return ascii_lower(xml_lang) 335 | is_html = ( 336 | self.in_html_document or 337 | self.namespace_url == 'http://www.w3.org/1999/xhtml') 338 | if is_html: 339 | lang = self.etree_element.get('lang') 340 | if lang is not None: 341 | return ascii_lower(lang) 342 | if self.parent is not None: 343 | return self.parent.lang 344 | # Root elememnt 345 | if is_html: 346 | content_language = None 347 | iterator = self.etree_element.iter('{http://www.w3.org/1999/xhtml}meta') 348 | for meta in iterator: 349 | http_equiv = meta.get('http-equiv', '') 350 | if ascii_lower(http_equiv) == 'content-language': 351 | content_language = _parse_content_language(meta.get('content')) 352 | if content_language is not None: 353 | return ascii_lower(content_language) 354 | # Empty string means unknown 355 | return _parse_content_language(self.transport_content_language) or '' 356 | 357 | @cached_property 358 | def in_disabled_fieldset(self): 359 | if self.parent is None: 360 | return False 361 | fieldset = '{http://www.w3.org/1999/xhtml}fieldset' 362 | legend = '{http://www.w3.org/1999/xhtml}legend' 363 | disabled_fieldset = ( 364 | self.parent.etree_element.tag == fieldset and 365 | self.parent.etree_element.get('disabled') is not None and ( 366 | self.etree_element.tag != legend or any( 367 | sibling.etree_element.tag == legend 368 | for sibling in self.iter_previous_siblings()))) 369 | return disabled_fieldset or self.parent.in_disabled_fieldset 370 | 371 | 372 | def _split_etree_tag(tag): 373 | position = tag.rfind('}') 374 | if position == -1: 375 | return '', tag 376 | else: 377 | assert tag[0] == '{' 378 | return tag[1:position], tag[position+1:] 379 | 380 | 381 | def _parse_content_language(value): 382 | if value is not None and ',' not in value: 383 | parts = split_whitespace(value) 384 | if len(parts) == 1: 385 | return parts[0] 386 | -------------------------------------------------------------------------------- /docs/api_reference.rst: -------------------------------------------------------------------------------- 1 | API Reference 2 | ============= 3 | 4 | 5 | .. module:: cssselect2 6 | .. autoclass:: Matcher 7 | :members: 8 | .. autofunction:: compile_selector_list 9 | .. autoclass:: ElementWrapper 10 | :members: 11 | .. autoclass:: SelectorError 12 | 13 | .. module:: cssselect2.compiler 14 | .. autoclass:: CompiledSelector 15 | -------------------------------------------------------------------------------- /docs/changelog.rst: -------------------------------------------------------------------------------- 1 | Changelog 2 | --------- 3 | 4 | 5 | Version 0.8.0 6 | ............. 7 | 8 | Released on 2025-03-05. 9 | 10 | * Drop support of Python 3.8 and 3.9, support 3.12 and 3.13 11 | * Handle case-sensitive and case-insensitive attribute selectors 12 | 13 | 14 | Version 0.7.0 15 | ............. 16 | 17 | Released on 2022-09-19. 18 | 19 | * Support :has selector 20 | 21 | 22 | Version 0.6.0 23 | ............. 24 | 25 | Released on 2022-04-15. 26 | 27 | **This version deprecates the ``iter_ancestors`` and ``iter_previous_siblings`` 28 | methods, that will be removed in 0.7.0. Use the ``ancestors`` and 29 | ``previous_siblings`` properties instead.** 30 | 31 | * Improve speed of ancestors and previous siblings 32 | 33 | 34 | Version 0.5.0 35 | ............. 36 | 37 | Released on 2022-02-27. 38 | 39 | * Support Python 3.10 40 | * Drop support of Python 3.6 41 | * Handle many CSS4 selectors 42 | * Ignore comments at the beginning of selectors 43 | 44 | 45 | Version 0.4.1 46 | ............. 47 | 48 | Released on 2020-10-29. 49 | 50 | * Fix PyPI description and various links. 51 | 52 | 53 | Version 0.4.0 54 | ............. 55 | 56 | Released on 2020-10-29. 57 | 58 | * Drop support of Python 3.5, add support of Python 3.9. 59 | * Don’t crash on empty :not() selectors. 60 | * New code structure, new packaging, new documentation. 61 | 62 | 63 | Version 0.3.0 64 | ............. 65 | 66 | Released on 2020-03-16. 67 | 68 | * Drop Python2 support. 69 | * Improve packaging and testing. 70 | 71 | 72 | Version 0.2.2 73 | ............. 74 | 75 | Released on 2019-09-06. 76 | 77 | * Optimize lang attribute selectors. 78 | 79 | 80 | Version 0.2.1 81 | ............. 82 | 83 | Released on 2017-10-02. 84 | 85 | * Fix documentation. 86 | 87 | 88 | Version 0.2.0 89 | ............. 90 | 91 | Released on 2017-08-16. 92 | 93 | * Fix some selectors for HTML documents with no namespace. 94 | * Don't crash when the attribute comparator is unknown. 95 | * Don't crash when there are empty attribute classes. 96 | * Follow semantic versioning. 97 | 98 | 99 | Version 0.1 100 | ........... 101 | 102 | Released on 2017-07-07. 103 | 104 | * Initial release. 105 | -------------------------------------------------------------------------------- /docs/conf.py: -------------------------------------------------------------------------------- 1 | # cssselect2 documentation build configuration file. 2 | 3 | import cssselect2 4 | 5 | # Add any Sphinx extension module names here, as strings. They can be 6 | # extensions coming with Sphinx (named 'sphinx.ext.*') or your custom ones. 7 | extensions = ['sphinx.ext.autodoc', 'sphinx.ext.intersphinx'] 8 | 9 | # Add any paths that contain templates here, relative to this directory. 10 | templates_path = ['_templates'] 11 | 12 | # The suffix of source filenames. 13 | source_suffix = '.rst' 14 | 15 | # The master toctree document. 16 | master_doc = 'index' 17 | 18 | # General information about the project. 19 | project = 'cssselect2' 20 | copyright = 'Simon Sapin and contributors' 21 | 22 | # The version info for the project you're documenting, acts as replacement for 23 | # |version| and |release|, also used in various other places throughout the 24 | # built documents. 25 | # 26 | # The full version, including alpha/beta/rc tags. 27 | release = cssselect2.__version__ 28 | 29 | # The short X.Y version. 30 | version = '.'.join(release.split('.')[:2]) 31 | 32 | # List of patterns, relative to source directory, that match files and 33 | # directories to ignore when looking for source files. 34 | exclude_patterns = ['_build'] 35 | 36 | # The name of the Pygments (syntax highlighting) style to use. 37 | pygments_style = 'monokai' 38 | 39 | # The theme to use for HTML and HTML Help pages. See the documentation for 40 | # a list of builtin themes. 41 | html_theme = 'furo' 42 | 43 | html_theme_options = { 44 | 'top_of_page_buttons': ['edit'], 45 | 'source_edit_link': 'https://github.com/Kozea/cssselect2/edit/main/docs/{filename}', 46 | } 47 | 48 | # Favicon URL 49 | html_favicon = 'https://www.courtbouillon.org/static/images/favicon.png' 50 | 51 | # Add any paths that contain custom static files (such as style sheets) here, 52 | # relative to this directory. They are copied after the builtin static files, 53 | # so a file named "default.css" will overwrite the builtin "default.css". 54 | html_static_path = [] 55 | 56 | # These paths are either relative to html_static_path 57 | # or fully qualified paths (eg. https://...) 58 | html_css_files = [ 59 | 'https://www.courtbouillon.org/static/docs-furo.css', 60 | ] 61 | 62 | # Output file base name for HTML help builder. 63 | htmlhelp_basename = 'cssselect2doc' 64 | 65 | # One entry per manual page. List of tuples 66 | # (source start file, name, description, authors, manual section). 67 | man_pages = [ 68 | ('index', 'cssselect2', 'cssselect2 Documentation', 69 | ['Simon Sapin and contributors'], 1) 70 | ] 71 | 72 | # Grouping the document tree into Texinfo files. List of tuples 73 | # (source start file, target name, title, author, 74 | # dir menu entry, description, category) 75 | texinfo_documents = [ 76 | ('index', 'cssselect2', 'cssselect2 Documentation', 77 | 'Simon Sapin', 'cssselect2', 78 | 'A straightforward implementation of CSS3 Selectors.', 79 | 'Miscellaneous'), 80 | ] 81 | 82 | # Example configuration for intersphinx: refer to the Python standard library. 83 | intersphinx_mapping = { 84 | 'python': ('https://docs.python.org/3/', None), 85 | 'webencodings': ('https://pythonhosted.org/webencodings/', None), 86 | 'tinycss2': ('https://doc.courtbouillon.org/tinycss2/latest/', None), 87 | } 88 | -------------------------------------------------------------------------------- /docs/contribute.rst: -------------------------------------------------------------------------------- 1 | Contribute 2 | ========== 3 | 4 | You want to add some code to cssselect2, launch its tests or improve its 5 | documentation? Thank you very much! Here are some tips to help you play with 6 | cssselect2 in good conditions. 7 | 8 | The first step is to clone the repository, create a virtual environment and 9 | install cssselect2 dependencies. 10 | 11 | .. code-block:: shell 12 | 13 | git clone https://github.com/Kozea/cssselect2.git 14 | cd cssselect2 15 | python -m venv venv 16 | venv/bin/pip install .[doc,test] 17 | 18 | You can then let your terminal in the current directory and launch Python to 19 | test your changes. ``import cssselect2`` will then import the working directory 20 | code, so that you can modify it and test your changes. 21 | 22 | .. code-block:: shell 23 | 24 | venv/bin/python 25 | 26 | 27 | Code & Issues 28 | ------------- 29 | 30 | If you’ve found a bug in cssselect2, it’s time to report it, and to fix it if you 31 | can! 32 | 33 | You can report bugs and feature requests on `GitHub`_. If you want to add or 34 | fix some code, please fork the repository and create a pull request, we’ll be 35 | happy to review your work. 36 | 37 | .. _GitHub: https://github.com/Kozea/cssselect2 38 | 39 | 40 | Tests 41 | ----- 42 | 43 | Tests are stored in the ``tests`` folder at the top of the repository. They use 44 | the `pytest`_ library. 45 | 46 | You can launch tests using the following command:: 47 | 48 | venv/bin/python -m pytest 49 | 50 | cssselect2 also uses ruff_ to check the coding style:: 51 | 52 | venv/bin/python -m ruff check 53 | 54 | .. _pytest: https://docs.pytest.org/ 55 | .. _ruff: https://docs.astral.sh/ruff/ 56 | 57 | 58 | Documentation 59 | ------------- 60 | 61 | Documentation is stored in the ``docs`` folder at the top of the repository. It 62 | relies on the `Sphinx`_ library. 63 | 64 | You can build the documentation using the following command:: 65 | 66 | venv/bin/sphinx-build docs docs/_build 67 | 68 | The documentation home page can now be found in the ``docs/_build/index.html`` 69 | file. You can open this file in a browser to see the final rendering. 70 | 71 | .. _Sphinx: https://www.sphinx-doc.org/ 72 | -------------------------------------------------------------------------------- /docs/example.py: -------------------------------------------------------------------------------- 1 | from xml.etree import ElementTree 2 | 3 | import cssselect2 4 | import tinycss2 5 | 6 | # Parse CSS and add rules to the matcher 7 | 8 | matcher = cssselect2.Matcher() 9 | 10 | rules = tinycss2.parse_stylesheet(''' 11 | body { font-size: 2em } 12 | body p { background: red } 13 | p { color: blue } 14 | ''', skip_whitespace=True) 15 | 16 | for rule in rules: 17 | selectors = cssselect2.compile_selector_list(rule.prelude) 18 | selector_string = tinycss2.serialize(rule.prelude) 19 | content_string = tinycss2.serialize(rule.content) 20 | payload = (selector_string, content_string) 21 | for selector in selectors: 22 | matcher.add_selector(selector, payload) 23 | 24 | 25 | # Parse HTML and find CSS rules applying to each tag 26 | 27 | html_tree = ElementTree.fromstring(''' 28 | 29 | 30 |

Test

31 | 32 | 33 | ''') 34 | wrapper = cssselect2.ElementWrapper.from_html_root(html_tree) 35 | for element in wrapper.iter_subtree(): 36 | tag = element.etree_element.tag.split('}')[-1] 37 | print('Found tag "{}" in HTML'.format(tag)) 38 | 39 | matches = matcher.match(element) 40 | if matches: 41 | for match in matches: 42 | specificity, order, pseudo, payload = match 43 | selector_string, content_string = payload 44 | print('Matching selector "{}" ({})'.format( 45 | selector_string, content_string)) 46 | else: 47 | print('No rule matching this tag') 48 | print() 49 | -------------------------------------------------------------------------------- /docs/first_steps.rst: -------------------------------------------------------------------------------- 1 | First Steps 2 | =========== 3 | 4 | .. currentmodule:: cssselect2 5 | 6 | 7 | Installation 8 | ------------ 9 | 10 | The easiest way to use tinycss2 is to install it in a Python `virtual 11 | environment`_. When your virtual environment is activated, you can then install 12 | cssselect2 with pip_:: 13 | 14 | pip install cssselect2 15 | 16 | This will also automatically install tinycss2’s only dependencies, tinycss2_ 17 | and webencodings_. cssselect2, tinycss2 and webencodings only contain Python 18 | code and should work on any Python implementation. 19 | 20 | cssselect2 also is packaged for many Linux distributions (Debian, Ubuntu, 21 | Fedora, Archlinux, Gentoo…). 22 | 23 | .. _virtual environment: https://packaging.python.org/guides/installing-using-pip-and-virtual-environments/ 24 | .. _pip: https://pip.pypa.io/ 25 | .. _webencodings: https://pythonhosted.org/webencodings/ 26 | .. _tinycss2: https://doc.courtbouillon.org/tinycss2/ 27 | 28 | 29 | Basic Example 30 | ------------- 31 | 32 | Here is a classical cssselect2 workflow: 33 | 34 | - parse a CSS stylesheet using tinycss2_, 35 | - store the CSS rules in a :class:`Matcher` object, 36 | - parse an HTML document using an ElementTree-like parser, 37 | - wrap the HTML tree in a :class:`ElementWrapper` object, 38 | - find the CSS rules matching each HTML tag, using the matcher and the wrapper. 39 | 40 | .. literalinclude:: example.py 41 | -------------------------------------------------------------------------------- /docs/index.rst: -------------------------------------------------------------------------------- 1 | cssselect2 2 | ========== 3 | 4 | .. currentmodule:: cssselect2 5 | 6 | .. include:: ../README.rst 7 | 8 | .. toctree:: 9 | :caption: Documentation 10 | :maxdepth: 2 11 | 12 | first_steps 13 | api_reference 14 | 15 | .. toctree:: 16 | :caption: Extra Information 17 | :maxdepth: 2 18 | 19 | changelog 20 | contribute 21 | support 22 | -------------------------------------------------------------------------------- /docs/support.rst: -------------------------------------------------------------------------------- 1 | Support 2 | ======= 3 | 4 | 5 | Sponsorship 6 | ----------- 7 | 8 | With `donations and sponsorship`_, you help the projects to be 9 | better. Donations allow the CourtBouillon team to have more time dedicated to 10 | add new features, fix bugs, and improve documentation. 11 | 12 | .. _donations and sponsorship: https://opencollective.com/courtbouillon 13 | 14 | 15 | Professionnal Support 16 | --------------------- 17 | 18 | You can improve your experience with CourtBouillon’s tools thanks to our 19 | professional support. You want bugs fixed as soon as possible? You projects 20 | would highly benefit from some new features? You or your team would like to get 21 | new skills with one of the technologies we master? 22 | 23 | Please contact us by mail_, by chat_ or by tweet_ to get in touch and find the 24 | best way we can help you. 25 | 26 | .. _mail: mailto:contact@courtbouillon.org 27 | .. _chat: https://gitter.im/CourtBouillon/cssselect2 28 | .. _tweet: https://twitter.com/BouillonCourt 29 | -------------------------------------------------------------------------------- /pyproject.toml: -------------------------------------------------------------------------------- 1 | [build-system] 2 | requires = ['flit_core >=3.2,<4'] 3 | build-backend = 'flit_core.buildapi' 4 | 5 | [project] 6 | name = 'cssselect2' 7 | description = 'CSS selectors for Python ElementTree' 8 | keywords = ['css', 'elementtree'] 9 | authors = [{name = 'Simon Sapin', email = 'simon.sapin@exyr.org'}] 10 | maintainers = [{name = 'CourtBouillon', email = 'contact@courtbouillon.org'}] 11 | requires-python = '>=3.9' 12 | readme = {file = 'README.rst', content-type = 'text/x-rst'} 13 | license = {file = 'LICENSE'} 14 | dependencies = ['tinycss2', 'webencodings'] 15 | classifiers = [ 16 | 'Development Status :: 5 - Production/Stable', 17 | 'Intended Audience :: Developers', 18 | 'License :: OSI Approved :: BSD License', 19 | 'Operating System :: OS Independent', 20 | 'Programming Language :: Python', 21 | 'Programming Language :: Python :: 3', 22 | 'Programming Language :: Python :: 3 :: Only', 23 | 'Programming Language :: Python :: 3.9', 24 | 'Programming Language :: Python :: 3.10', 25 | 'Programming Language :: Python :: 3.11', 26 | 'Programming Language :: Python :: 3.12', 27 | 'Programming Language :: Python :: 3.13', 28 | 'Programming Language :: Python :: Implementation :: CPython', 29 | 'Programming Language :: Python :: Implementation :: PyPy', 30 | 'Topic :: Internet :: WWW/HTTP', 31 | ] 32 | dynamic = ['version'] 33 | 34 | [project.urls] 35 | Homepage = 'https://doc.courtbouillon.org/cssselect2/' 36 | Documentation = 'https://doc.courtbouillon.org/cssselect2/' 37 | Code = 'https://github.com/Kozea/cssselect2/' 38 | Issues = 'https://github.com/Kozea/cssselect2/issues' 39 | Changelog = 'https://github.com/Kozea/cssselect2/releases' 40 | Donation = 'https://opencollective.com/courtbouillon' 41 | 42 | [project.optional-dependencies] 43 | doc = ['sphinx', 'furo'] 44 | test = ['pytest', 'ruff'] 45 | 46 | [tool.flit.sdist] 47 | exclude = ['.*'] 48 | 49 | [tool.coverage.run] 50 | branch = true 51 | include = ['tests/*', 'cssselect2/*'] 52 | 53 | [tool.coverage.report] 54 | exclude_lines = ['pragma: no cover', 'def __repr__', 'raise NotImplementedError'] 55 | omit = ['.*'] 56 | 57 | [tool.ruff.lint] 58 | select = ['E', 'W', 'F', 'I', 'N', 'RUF'] 59 | ignore = ['RUF001', 'RUF002', 'RUF003'] 60 | 61 | [tool.ruff.lint.extend-per-file-ignores] 62 | 'docs/example.py' = ['I001'] 63 | -------------------------------------------------------------------------------- /tests/LICENSE: -------------------------------------------------------------------------------- 1 | These files are taken form the web-platform-test repository 2 | and used under a 3-clause BSD License. 3 | 4 | https://github.com/w3c/web-platform-tests/tree/master/selectors-api 5 | -------------------------------------------------------------------------------- /tests/__init__.py: -------------------------------------------------------------------------------- 1 | """ 2 | 3 | Test suite for cssselect2. 4 | 5 | """ 6 | -------------------------------------------------------------------------------- /tests/content.xhtml: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | Selectors-API Test Suite: HTML with Selectors Level 2 using TestHarness: Test Document 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 |
13 |
14 | 15 |
16 |

Universal selector tests inside element with id="universal".

17 |
18 | 
Some preformatted text with some embedded code
19 |

This is a normal link: W3C

20 |
Some more nested elements code hyperlink
21 |
22 | 23 |
24 |
25 |
26 |
27 |
28 |

29 | 

 30 | 		

 31 | 		
    
 32 | 
 33 | 		
 39 | 		
 45 | 		
 51 | 	
    
 52 | 
 53 | 	
    
 54 | 		
    
 55 | 	    
    
 56 | 	    
    
 57 | 	    
    
 58 | 
 59 | 		
    
 60 | 			
 61 | 			
 62 | 			
 63 | 			
 64 | 			
 65 | 			
 66 | 			
 67 | 			
 68 | 			
 69 | 		
    
 70 | 
 71 | 		
    
 72 | 	
    
 73 | 
 74 | 	
    
 75 | 		
    
 76 | 	    
    
 77 | 	    
    
 78 | 
 79 | 	    
    
 80 | 	    
    
 81 | 
 82 | 		
 83 | 		
 84 | 		
 85 | 		 
 86 | 		
 87 | 		
 88 | 		
 89 | 		
 90 | 		
    
 91 | 	
    
 92 | 
 93 | 	
    
 94 | 		
    
 95 | 	    
    
 96 | 	    
    
 97 | 	    
    
 98 | 	
    
 99 | 
100 | 	
    
101 | 		
102 | 		
103 | 		
104 | 
105 | 	    
    
106 | 	    
    
107 | 	    
    
108 | 	    
    
109 | 	    
    
110 | 
111 | 		
     
112 | 	
    
113 | 
114 | 	
    
115 | 		
116 | 		
117 | 		
118 | 
119 | 	    
    
120 | 	    
    
121 | 	    
    
122 | 	    
    
123 | 
124 | 		
     
125 | 	
    
126 | 
127 | 	
    
128 | 		
129 | 		
130 | 		
131 | 
132 | 	    
    
133 | 	    
    
134 | 	    
    
135 | 	    
    
136 | 	    
    
137 | 	    
    
138 | 
139 | 		
    
140 | 	
    
141 | 
142 | 	
    
143 | 		
144 | 			
145 | 			
146 | 			
147 | 		
    
148 | 		
149 | 		
      
150 | 			
    1. 
151 | 			
    2. 
152 | 			
    3. 
153 | 			
    4. 
154 | 			
    5. 
155 | 			
    6. 
156 | 			
    7. 
157 | 			
    8. 
158 | 			
    9. 
159 | 			
    10. 
160 | 			
    11. 
161 | 			
    12. 
162 | 		
    
163 | 
164 | 		
    
165 | 			span1
166 | 			em1
167 | 			
168 | 			em2
169 | 			span2
170 | 			strong1
171 | 			em3
172 | 			span3
173 | 			span4
174 | 			strong2
175 | 			em4
176 | 		
    
177 | 	
    
178 | 
179 | 	
    
180 | 		
    
181 | 		
    
182 | 		
    
183 | 
184 | 		
    
185 | 		
    
186 | 		
    
187 | 	
    
188 | 
189 | 	
    
190 | 		
    
191 | 		
    
192 | 		
    
193 | 
194 | 		
    
195 | 		
    
196 | 		
    
197 | 	
    
198 | 
199 | 	
    
200 | 		
    
201 | 			
202 | 		
    
203 | 		
    
204 | 			
205 | 			
206 | 		
    
207 | 		
    
208 | 			
209 | 			
210 | 			
211 | 		
    
212 | 	
    >
213 | 
214 | 	
    
215 | 		
    
216 | 		
    
217 | 		
     
    
218 | 		
    Text node
    
219 | 		
    
220 | 	
    
221 | 
222 | 	
231 | 
232 | 	
    
233 | 		
    
234 | 	    
    
235 | 	    
    
236 | 	    
    
237 | 	
    
238 | 
239 | 	
    
240 | 		
241 | 		
242 | 		
243 | 		
244 | 		
245 | 		
246 | 		
247 | 		
248 | 		
249 | 		
250 | 		
251 | 
252 | 		
253 | 		
254 | 		
255 | 		
256 | 		
257 | 		
258 | 		
259 | 		
260 | 		
261 | 		
262 | 		
263 | 	
    
264 | 
265 | 	
    
266 | 		
    
267 | 		
    
268 | 		
    
269 | 
270 | 		
    
271 | 		
    
272 | 		
    
273 | 	
    
274 | 
275 | 	
    All pseudo-element tests
    
276 | 
277 | 	
    
278 | 		
    
279 | 		
    
280 | 		
    
281 | 
282 | 		
283 | 		
    
284 | 		
    
285 | 		
    
286 | 		
    
287 | 		
    
288 | 		
    
289 | 		
    
290 | 		
    
291 | 		
    
292 | 
293 | 		
294 | 		
295 | 
296 | 		
297 | 		
298 | 	
    
299 | 
300 | 	
    
301 | 		
    
302 | 		
    
303 | 
304 | 		
      
305 | 			
    • 
306 | 			
    • 
307 | 			
    • 
308 | 			
    • 
309 | 		
    
310 | 
311 | 		
312 | 		
313 | 
314 | 		
315 | 		
316 | 	
    
317 | 
318 | 	
    
319 | 		
    
320 | 			
    
321 | 				
    
322 | 				
    				
323 | 			
    
324 | 		
    
325 | 		
    
326 | 	
    
327 | 
328 | 	
    
329 | 		
    
330 | 			
    
331 | 				
    
332 | 				
    				
333 | 			
    
334 | 		
    
335 | 		
    
336 | 	
    
337 | 
338 | 	
    
339 | 		
    
340 | 		
    
341 | 			
    
342 | 		
    
343 | 		
    
344 | 			
    
345 | 			
    
346 | 		
    
347 | 		
    
348 | 		
    
349 | 		
    
350 | 	
    
351 | 
352 | 	
    
353 | 		
    
354 | 		
    
355 | 			
    
356 | 		
    
357 | 		
    
358 | 			
    
359 | 			
    
360 | 		
    
361 | 		
    
362 | 		
    
363 | 		
    
364 | 	
    
365 | 
366 | 	
    
367 | 		
368 | 		
369 | 	
    
370 | 
    
371 | 
372 | 
373 | 


--------------------------------------------------------------------------------
/tests/ids.html:
--------------------------------------------------------------------------------
 1 | 
 2 |     
 3 |     
 4 |   
 5 |     
    
 6 |       
 7 |       
 8 |       
 9 |         link
10 |       
      
11 |         
    1. content
      2. 
12 |         
    2. 
13 |           
      
14 |           
      
15 |         
      3. 
16 |         
    3. 
17 |         
    4. 
19 |         
    5. 
20 |         
    6. 
21 |         
    7.   
      8. 
22 |       
    
23 |       
    
24 |         hi there
25 |         guy
26 |         
27 |         
28 |         
29 |         
30 |         
31 |         
32 |         
34 |         
    
35 |           
36 |           
37 |         
    
38 |       
    
39 |       
      
40 |       
    
41 |       
42 |         
43 |         
44 |       
45 |     
    
46 |     
    
48 | 
49 | 


--------------------------------------------------------------------------------
/tests/make_selectors.py:
--------------------------------------------------------------------------------
 1 | #!/usr/bin/env python
 2 | 
 3 | import json
 4 | import re
 5 | from pathlib import Path
 6 | from pprint import pformat
 7 | from urllib.request import urlopen
 8 | 
 9 | with urlopen('http://wpt.live/dom/nodes/selectors.js') as fd:
10 |     js = fd.read().decode()
11 | 
12 | js = js.replace(r'\_', '_')  # Unescape underscores
13 | js = re.sub(r'/\*.*?\*/', '', js, flags=re.S)  # Remove /* comments */
14 | js = re.sub(r'( +|^)//.*$', '', js, flags=re.M)  # Remove // comments
15 | js = re.sub(r',\s+testType:[0-fx\s\|]+\}', '}', js)  # Remove testType
16 | js = re.sub(r'(\{|,\s+)(\w+):', r'\1"\2":', js)  # Use strings for keys
17 | js = re.sub(r',\s+(\]|\})', r'\1', js, flags=re.M)  # Remove trailing commas
18 | 
19 | invalid_selectors = json.loads(re.search(
20 |     r'var invalidSelectors = (\[.*?\]);', js, flags=re.S).group(1))
21 | valid_selectors = json.loads(re.search(
22 |     r'var validSelectors = (\[.*?\]);', js, flags=re.S).group(1))
23 | 
24 | python = f'''# File generated by make_selectors.py, do not edit
25 | 
26 | invalid_selectors = {pformat(invalid_selectors, indent=4, width=79)}
27 | 
28 | valid_selectors = {pformat(valid_selectors, indent=4, width=79)}
29 | '''
30 | python = python.replace('= [   ', '= [\n    ')
31 | python = python.replace('    {   ', '    {\n        ')
32 | python = python.replace(': [   ', ': [\n            ')
33 | python = python.replace('                      ', '            ')
34 | 
35 | Path(__file__).parent.joinpath('w3_selectors.py').write_text(python)
36 | 


--------------------------------------------------------------------------------
/tests/shakespeare.html:
--------------------------------------------------------------------------------
  1 | 
  2 | 
  3 |   
  4 |     
  5 |   
  6 |   
  7 |     
    
  8 |       
    
  9 |         
    As You Like It
    
 10 |         
    
 11 |           by William Shakespeare
 12 |         
    
 13 |         
    
 14 |           
    ACT I, SCENE III. A room in the palace.
    
 15 |           
    
 16 |             
    Enter CELIA and ROSALIND
    
 17 |           
    
 18 |           
    CELIA
    
 19 |           
    
 20 |             
    Why, cousin! why, Rosalind! Cupid have mercy! not a word?
    
 21 |           
    
 22 |           
    ROSALIND
    
 23 |           
    
 24 |             
    Not one to throw at a dog.
    
 25 |           
    
 26 |           
    CELIA
    
 27 |           
    
 28 |             
    No, thy words are too precious to be cast away upon
    
 29 |             
    curs; throw some of them at me; come, lame me with reasons.
    
 30 |           
    
 31 |           
    ROSALIND
    
 32 |           
    CELIA
    
 33 |           
    
 34 |             
    But is all this for your father?
    
 35 |           
    
 36 |           
    
 37 |             
    Then there were two cousins laid up; when the one
    
 38 |             
    should be lamed with reasons and the other mad
    
 39 |             
    without any.
    
 40 |           
    
 41 |           
    ROSALIND
    
 42 |           
    
 43 |             
    No, some of it is for my child's father. O, how
    
 44 |             
    full of briers is this working-day world!
    
 45 |           
    
 46 |           
    CELIA
    
 47 |           
    
 48 |             
    They are but burs, cousin, thrown upon thee in
    
 49 |             
    holiday foolery: if we walk not in the trodden
    
 50 |             
    paths our very petticoats will catch them.
    
 51 |           
    
 52 |           
    ROSALIND
    
 53 |           
    
 54 |             
    I could shake them off my coat: these burs are in my heart.
    
 55 |           
    
 56 |           
    CELIA
    
 57 |           
    
 58 |             
    Hem them away.
    
 59 |           
    
 60 |           
    ROSALIND
    
 61 |           
    
 62 |             
    I would try, if I could cry 'hem' and have him.
    
 63 |           
    
 64 |           
    CELIA
    
 65 |           
    
 66 |             
    Come, come, wrestle with thy affections.
    
 67 |           
    
 68 |           
    ROSALIND
    
 69 |           
    
 70 |             
    O, they take the part of a better wrestler than myself!
    
 71 |           
    
 72 |           
    CELIA
    
 73 |           
    
 74 |             
    O, a good wish upon you! you will try in time, in
    
 75 |             
    despite of a fall. But, turning these jests out of
    
 76 |             
    service, let us talk in good earnest: is it
    
 77 |             
    possible, on such a sudden, you should fall into so
    
 78 |             
    strong a liking with old Sir Rowland's youngest son?
    
 79 |           
    
 80 |           
    ROSALIND
    
 81 |           
    
 82 |             
    The duke my father loved his father dearly.
    
 83 |           
    
 84 |           
    CELIA
    
 85 |           
    
 86 |             
    Doth it therefore ensue that you should love his son
    
 87 |             
    dearly? By this kind of chase, I should hate him,
    
 88 |             
    for my father hated his father dearly; yet I hate
    
 89 |             
    not Orlando.
    
 90 |           
    
 91 |           
    ROSALIND
    
 92 |           
    
 93 |             
    No, faith, hate him not, for my sake.
    
 94 |           
    
 95 |           
    CELIA
    
 96 |           
    
 97 |             
    Why should I not? doth he not deserve well?
    
 98 |           
    
 99 |           
    ROSALIND
    
100 |           
    
101 |             
    Let me love him for that, and do you love him
    
102 |             
    because I do. Look, here comes the duke.
    
103 |           
    
104 |           
    CELIA
    
105 |           
    
106 |             
    With his eyes full of anger.
    
107 |             
    Enter DUKE FREDERICK, with Lords
    
108 |           
    
109 |           
    DUKE FREDERICK
    
110 |           
    
111 |             
    Mistress, dispatch you with your safest haste
    
112 |             
    And get you from our court.
    
113 |           
    
114 |           
    ROSALIND
    
115 |           
    
116 |             
    Me, uncle?
    
117 |           
    
118 |           
    DUKE FREDERICK
    
119 |           
    
120 |             
    You, cousin
    
121 |             
    Within these ten days if that thou be'st found
    
122 |             
    So near our public court as twenty miles,
    
123 |             
    Thou diest for it.
    
124 |           
    
125 |           
    ROSALIND
    
126 |           
    
127 |             
                      I do beseech your grace,
    
128 |             
    Let me the knowledge of my fault bear with me:
    
129 |             
    If with myself I hold intelligence
    
130 |             
    Or have acquaintance with mine own desires,
    
131 |             
    If that I do not dream or be not frantic,--
    
132 |             
    As I do trust I am not--then, dear uncle,
    
133 |             
    Never so much as in a thought unborn
    
134 |             
    Did I offend your highness.
    
135 |           
    
136 |           
    DUKE FREDERICK
    
137 |           
    
138 |             
    Thus do all traitors:
    
139 |             
    If their purgation did consist in words,
    
140 |             
    They are as innocent as grace itself:
    
141 |             
    Let it suffice thee that I trust thee not.
    
142 |           
    
143 |           
    ROSALIND
    
144 |           
    
145 |             
    Yet your mistrust cannot make me a traitor:
    
146 |             
    Tell me whereon the likelihood depends.
    
147 |           
    
148 |           
    DUKE FREDERICK
    
149 |           
    
150 |             
    Thou art thy father's daughter; there's enough.
    
151 |           
    
152 |           
    ROSALIND
    
153 |           
    
154 |             
    So was I when your highness took his dukedom;
    
155 |             
    So was I when your highness banish'd him:
    
156 |             
    Treason is not inherited, my lord;
    
157 |             
    Or, if we did derive it from our friends,
    
158 |             
    What's that to me? my father was no traitor:
    
159 |             
    Then, good my liege, mistake me not so much
    
160 |             
    To think my poverty is treacherous.
    
161 |           
    
162 |           
    CELIA
    
163 |           
    
164 |             
    Dear sovereign, hear me speak.
    
165 |           
    
166 |           
    DUKE FREDERICK
    
167 |           
    
168 |             
    Ay, Celia; we stay'd her for your sake,
    
169 |             
    Else had she with her father ranged along.
    
170 |           
    
171 |           
    CELIA
    
172 |           
    
173 |             
    I did not then entreat to have her stay;
    
174 |             
    It was your pleasure and your own remorse:
    
175 |             
    I was too young that time to value her;
    
176 |             
    But now I know her: if she be a traitor,
    
177 |             
    Why so am I; we still have slept together,
    
178 |             
    Rose at an instant, learn'd, play'd, eat together,
    
179 |             
    And wheresoever we went, like Juno's swans,
    
180 |             
    Still we went coupled and inseparable.
    
181 |           
    
182 |           
    DUKE FREDERICK
    
183 |           
    
184 |             
    She is too subtle for thee; and her smoothness,
    
185 |             
    Her very silence and her patience
    
186 |             
    Speak to the people, and they pity her.
    
187 |             
    Thou art a fool: she robs thee of thy name;
    
188 |             
    And thou wilt show more bright and seem more virtuous
    
189 |             
    When she is gone. Then open not thy lips:
    
190 |             
    Firm and irrevocable is my doom
    
191 |             
    Which I have pass'd upon her; she is banish'd.
    
192 |           
    
193 |           
    CELIA
    
194 |           
    
195 |             
    Pronounce that sentence then on me, my liege:
    
196 |             
    I cannot live out of her company.
    
197 |           
    
198 |           
    DUKE FREDERICK
    
199 |           
    
200 |             
    You are a fool. You, niece, provide yourself:
    
201 |             
    If you outstay the time, upon mine honour,
    
202 |             
    And in the greatness of my word, you die.
    
203 |             
    Exeunt DUKE FREDERICK and Lords
    
204 |           
    
205 |           
    CELIA
    
206 |           
    
207 |             
    O my poor Rosalind, whither wilt thou go?
    
208 |             
    Wilt thou change fathers? I will give thee mine.
    
209 |             
    I charge thee, be not thou more grieved than I am.
    
210 |           
    
211 |           
    ROSALIND
    
212 |           
    
213 |             
    I have more cause.
    
214 |           
    
215 |           
    CELIA
    
216 |           
    
217 |             
                      Thou hast not, cousin;
    
218 |             
    Prithee be cheerful: know'st thou not, the duke
    
219 |             
    Hath banish'd me, his daughter?
    
220 |           
    
221 |           
    ROSALIND
    
222 |           
    
223 |             
    That he hath not.
    
224 |           
    
225 |           
    CELIA
    
226 |           
    
227 |             
    No, hath not? Rosalind lacks then the love
    
228 |             
    Which teacheth thee that thou and I am one:
    
229 |             
    Shall we be sunder'd? shall we part, sweet girl?
    
230 |             
    No: let my father seek another heir.
    
231 |             
    Therefore devise with me how we may fly,
    
232 |             
    Whither to go and what to bear with us;
    
233 |             
    And do not seek to take your change upon you,
    
234 |             
    To bear your griefs yourself and leave me out;
    
235 |             
    For, by this heaven, now at our sorrows pale,
    
236 |             
    Say what thou canst, I'll go along with thee.
    
237 |           
    
238 |           
    ROSALIND
    
239 |           
    
240 |             
    Why, whither shall we go?
    
241 |           
    
242 |           
    CELIA
    
243 |           
    
244 |             
    To seek my uncle in the forest of Arden.
    
245 |           
    
246 |           
    ROSALIND
    
247 |           
    
248 |             
    Alas, what danger will it be to us,
    
249 |             
    Maids as we are, to travel forth so far!
    
250 |             
    Beauty provoketh thieves sooner than gold.
    
251 |           
    
252 |           
    CELIA
    
253 |           
    
254 |             
    I'll put myself in poor and mean attire
    
255 |             
    And with a kind of umber smirch my face;
    
256 |             
    The like do you: so shall we pass along
    
257 |             
    And never stir assailants.
    
258 |           
    
259 |           
    ROSALIND
    
260 |           
    
261 |             
    Were it not better,
    
262 |             
    Because that I am more than common tall,
    
263 |             
    That I did suit me all points like a man?
    
264 |             
    A gallant curtle-axe upon my thigh,
    
265 |             
    A boar-spear in my hand; and--in my heart
    
266 |             
    Lie there what hidden woman's fear there will--
    
267 |             
    We'll have a swashing and a martial outside,
    
268 |             
    As many other mannish cowards have
    
269 |             
    That do outface it with their semblances.
    
270 |           
    
271 |           
    CELIA
    
272 |           
    
273 |             
    What shall I call thee when thou art a man?
    
274 |           
    
275 |           
    ROSALIND
    
276 |           
    
277 |             
    I'll have no worse a name than Jove's own page;
    
278 |             
    And therefore look you call me Ganymede.
    
279 |             
    But what will you be call'd?
    
280 |           
    
281 |           
    CELIA
    
282 |           
    
283 |             
    Something that hath a reference to my state
    
284 |             
    No longer Celia, but Aliena.
    
285 |           
    
286 |           
    ROSALIND
    
287 |           
    
288 |             
    But, cousin, what if we assay'd to steal
    
289 |             
    The clownish fool out of your father's court?
    
290 |             
    Would he not be a comfort to our travel?
    
291 |           
    
292 |           
    CELIA
    
293 |           
    
294 |             
    He'll go along o'er the wide world with me;
    
295 |             
    Leave me alone to woo him. Let's away,
    
296 |             
    And get our jewels and our wealth together,
    
297 |             
    Devise the fittest time and safest way
    
298 |             
    To hide us from pursuit that will be made
    
299 |             
    After my flight. Now go we in content
    
300 |             
    To liberty and not to banishment.
    
301 |             
    Exeunt
    
302 |           
    
303 |         
    
304 |       
    
305 |     
    
306 |   
307 | 
308 | 


--------------------------------------------------------------------------------
/tests/test_cssselect2.py:
--------------------------------------------------------------------------------
  1 | """
  2 | 
  3 | Test suite for cssselect2.
  4 | 
  5 | """
  6 | 
  7 | import xml.etree.ElementTree as etree  # noqa: N813
  8 | from pathlib import Path
  9 | 
 10 | import pytest
 11 | 
 12 | from cssselect2 import ElementWrapper, SelectorError, compile_selector_list
 13 | 
 14 | from .w3_selectors import invalid_selectors, valid_selectors
 15 | 
 16 | CURRENT_FOLDER = Path(__file__).parent
 17 | IDS_ROOT = etree.parse(CURRENT_FOLDER / 'ids.html')
 18 | ALL_IDS = [
 19 |     element.etree_element.get('id', 'nil') for element in
 20 |     ElementWrapper.from_html_root(IDS_ROOT).query_all('*')]
 21 | SHAKESPEARE_BODY = (
 22 |     ElementWrapper.from_xml_root(
 23 |         etree.parse(CURRENT_FOLDER / 'shakespeare.html').find(
 24 |             './/{http://www.w3.org/1999/xhtml}body')))
 25 | 
 26 | 
 27 | def get_test_document():
 28 |     document = etree.parse(CURRENT_FOLDER / 'content.xhtml')
 29 |     parent = document.find(".//*[@id='root']")
 30 | 
 31 |     # Setup namespace tests
 32 |     for id in ('any-namespace', 'no-namespace'):
 33 |         div = etree.SubElement(parent, '{http://www.w3.org/1999/xhtml}div')
 34 |         div.set('id', id)
 35 |         div1 = etree.SubElement(div, '{http://www.w3.org/1999/xhtml}div')
 36 |         div1.set('id', id + '-div1')
 37 |         div2 = etree.SubElement(div, '{http://www.w3.org/1999/xhtml}div')
 38 |         div2.set('id', id + '-div2')
 39 |         div3 = etree.SubElement(div, 'div')
 40 |         div3.set('id', id + '-div3')
 41 |         div4 = etree.SubElement(div, '{http://www.example.org/ns}div')
 42 |         div4.set('id', id + '-div4')
 43 | 
 44 |     return document
 45 | 
 46 | 
 47 | TEST_DOCUMENT = get_test_document()
 48 | 
 49 | 
 50 | # Remove unsuitable tests
 51 | valid_selectors = [
 52 |     test for test in valid_selectors
 53 |     if not set(test.get('exclude', ())) & {'document', 'xhtml'}]
 54 | 
 55 | # Mark failing tests
 56 | for failing in (2, 9, 104, 105, 111, 197, 198):
 57 |     valid_selectors[failing] = pytest.param(
 58 |         valid_selectors[failing], marks=pytest.mark.xfail)
 59 | 
 60 | 
 61 | @pytest.mark.parametrize('test', invalid_selectors)
 62 | def test_invalid_selectors(test):
 63 |     try:
 64 |         compile_selector_list(test['selector'])
 65 |     except SelectorError:
 66 |         pass
 67 |     else:  # pragma: no cover
 68 |         raise AssertionError(
 69 |             f'Should be invalid: {test["selector"]!r} ({test["name"]})')
 70 | 
 71 | 
 72 | @pytest.mark.parametrize('test', valid_selectors)
 73 | def test_valid_selectors(test):
 74 |     root = ElementWrapper.from_xml_root(TEST_DOCUMENT)
 75 |     result = [element.id for element in root.query_all(test['selector'])]
 76 |     if result != test['expect']:  # pragma: no cover
 77 |         raise AssertionError(
 78 |             f'{test["selector"]!r}: {result} != {test["expect"]} ({test["name"]})')
 79 | 
 80 | 
 81 | def test_lang():
 82 |     doc = etree.fromstring('''
 83 |         
 84 |     ''')
 85 |     assert not ElementWrapper.from_xml_root(doc).matches(':lang(fr)')
 86 | 
 87 |     doc = etree.fromstring('''
 88 |         
 89 |             
 90 |         
 91 |     ''')
 92 |     root = ElementWrapper.from_xml_root(doc, content_language='en')
 93 |     assert root.matches(':lang(fr)')
 94 | 
 95 |     doc = etree.fromstring('''
 96 |         
 97 |             
 98 |         
 99 |     ''')
100 |     root = ElementWrapper.from_xml_root(doc, content_language='en')
101 |     assert root.matches(':lang(en)')
102 | 
103 |     doc = etree.fromstring('')
104 |     root = ElementWrapper.from_xml_root(doc, content_language='en')
105 |     assert root.matches(':lang(en)')
106 | 
107 |     root = ElementWrapper.from_xml_root(doc, content_language='en, es')
108 |     assert not root.matches(':lang(en)')
109 | 
110 |     root = ElementWrapper.from_xml_root(doc)
111 |     assert not root.matches(':lang(en)')
112 | 
113 |     doc = etree.fromstring('')
114 |     root = ElementWrapper.from_html_root(doc)
115 |     assert root.matches(':lang(en)')
116 | 
117 |     doc = etree.fromstring('')
118 |     root = ElementWrapper.from_xml_root(doc)
119 |     assert not root.matches(':lang(en)')
120 | 
121 | 
122 | @pytest.mark.parametrize('selector, result', (
123 |     ('*', ALL_IDS),
124 |     ('div', ['outer-div', 'li-div', 'foobar-div']),
125 |     ('div div', ['li-div']),
126 |     ('div, div div', ['outer-div', 'li-div', 'foobar-div']),
127 |     ('div , div div', ['outer-div', 'li-div', 'foobar-div']),
128 |     ('a[name]', ['name-anchor']),
129 |     ('a[rel]', ['tag-anchor', 'nofollow-anchor']),
130 |     ('a[rel="tag"]', ['tag-anchor']),
131 |     ('a[href*="localhost"]', ['tag-anchor']),
132 |     ('a[href*=""]', []),
133 |     ('a[href^="http"]', ['tag-anchor', 'nofollow-anchor']),
134 |     ('a[href^="http:"]', ['tag-anchor']),
135 |     ('a[href^=""]', []),
136 |     ('a[href$="org"]', ['nofollow-anchor']),
137 |     ('a[href$=""]', []),
138 |     ('div[foobar~="bc"]', ['foobar-div']),
139 |     ('div[foobar~="cde"]', ['foobar-div']),
140 |     ('[foobar~="ab bc"]', []),
141 |     ('[foobar~=""]', []),
142 |     ('[foobar~=" \t"]', []),
143 |     ('div[foobar~="cd"]', []),
144 | 
145 |     ('a[rel="tAg"]', []),
146 |     ('a[rel="tAg" s]', []),
147 |     ('a[rel="tAg" i]', ['tag-anchor']),
148 |     ('a[href*="localHOST"]', []),
149 |     ('a[href*="localHOST" s]', []),
150 |     ('a[href*="localHOST" i]', ['tag-anchor']),
151 |     ('a[href^="hTtp"]', []),
152 |     ('a[href^="hTtp" s]', []),
153 |     ('a[href^="hTtp" i]', ['tag-anchor', 'nofollow-anchor']),
154 |     ('a[href$="Org"]', []),
155 |     ('a[href$="Org" S]', []),
156 |     ('a[href$="Org" I]', ['nofollow-anchor']),
157 |     ('div[foobar~="BC"]', []),
158 |     ('div[foobar~="BC" s]', []),
159 |     ('div[foobar~="BC" i]', ['foobar-div']),
160 | 
161 |     # Attribute values are case sensitive…
162 |     ('*[lang|="En"]', ['second-li']),
163 |     ('[lang|="En-us"]', ['second-li']),
164 |     ('*[lang|="en"]', []),
165 |     ('[lang|="en-US"]', []),
166 |     ('*[lang|="e"]', []),
167 |     # … but :lang() is not.
168 |     (':lang(EN)', ['second-li', 'li-div']),
169 |     ('*:lang(en-US)', ['second-li', 'li-div']),
170 |     (':lang(En)', ['second-li', 'li-div']),
171 |     (':lang(e)', []),
172 |     (':lang("en-US")', ['second-li', 'li-div']),
173 |     pytest.param(
174 |         ':lang("*-US")', ['second-li', 'li-div'], marks=pytest.mark.xfail),
175 |     pytest.param(
176 |         ':lang(\\*-US)', ['second-li', 'li-div'], marks=pytest.mark.xfail),
177 |     (':lang(en /* English */, fr /* French */)', ['second-li', 'li-div']),
178 | 
179 |     ('li:nth-child(3)', ['third-li']),
180 |     ('li:nth-child(10)', []),
181 |     ('li:nth-child(2n)', ['second-li', 'fourth-li', 'sixth-li']),
182 |     ('li:nth-child(even)', ['second-li', 'fourth-li', 'sixth-li']),
183 |     ('li:nth-child(+2n+0)', ['second-li', 'fourth-li', 'sixth-li']),
184 |     ('li:nth-child(2n+1)', ['first-li', 'third-li', 'fifth-li', 'seventh-li']),
185 |     ('li:nth-child(odd)', ['first-li', 'third-li', 'fifth-li', 'seventh-li']),
186 |     ('li:nth-child(2n+4)', ['fourth-li', 'sixth-li']),
187 |     ('li:nth-child(3n+1)', ['first-li', 'fourth-li', 'seventh-li']),
188 |     ('p > input:nth-child(2n of p input[type=checkbox])', [
189 |         'checkbox-disabled', 'checkbox-disabled-checked']),
190 |     ('li:nth-last-child(1)', ['seventh-li']),
191 |     ('li:nth-last-child(0)', []),
192 |     ('li:nth-last-child(2n+2)', ['second-li', 'fourth-li', 'sixth-li']),
193 |     ('li:nth-last-child(even)', ['second-li', 'fourth-li', 'sixth-li']),
194 |     ('li:nth-last-child(2n+4)', ['second-li', 'fourth-li']),
195 |     (':nth-last-child(1 of [type=checkbox])', [
196 |         'checkbox-disabled-checked', 'checkbox-fieldset-disabled']),
197 |     ('ol:first-of-type', ['first-ol']),
198 |     ('ol:nth-child(1)', []),
199 |     ('ol:nth-of-type(2)', ['second-ol']),
200 |     (':nth-of-type(1 of .e)', ['tag-anchor', 'first-ol']),
201 |     ('ol:nth-last-of-type(2)', ['first-ol']),
202 |     (':nth-last-of-type(1 of .e)', ['tag-anchor', 'second-ol']),
203 |     ('span:only-child', ['foobar-span']),
204 |     ('div:only-child', ['li-div']),
205 |     ('div *:only-child', ['li-div', 'foobar-span']),
206 |     ('p *:only-of-type', ['p-em', 'fieldset']),
207 |     ('p:only-of-type', ['paragraph']),
208 | 
209 |     ('a:empty', ['name-anchor']),
210 |     ('a:EMpty', ['name-anchor']),
211 |     ('li:empty', ['third-li', 'fourth-li', 'fifth-li', 'sixth-li']),
212 |     (':root', ['html']),
213 |     ('html:root', ['html']),
214 |     ('li:root', []),
215 |     ('* :root', []),
216 |     ('.a', ['first-ol']),
217 |     ('.b', ['first-ol']),
218 |     ('*.a', ['first-ol']),
219 |     ('ol.a', ['first-ol']),
220 |     ('.c', ['first-ol', 'third-li', 'fourth-li']),
221 |     ('*.c', ['first-ol', 'third-li', 'fourth-li']),
222 |     ('ol *.c', ['third-li', 'fourth-li']),
223 |     ('ol li.c', ['third-li', 'fourth-li']),
224 |     ('li ~ li.c', ['third-li', 'fourth-li']),
225 |     ('ol > li.c', ['third-li', 'fourth-li']),
226 |     ('#first-li', ['first-li']),
227 |     ('li#first-li', ['first-li']),
228 |     ('*#first-li', ['first-li']),
229 |     ('li div', ['li-div']),
230 |     ('li > div', ['li-div']),
231 |     ('div div', ['li-div']),
232 |     ('div > div', []),
233 |     ('div>.c', ['first-ol']),
234 |     ('div > .c', ['first-ol']),
235 |     ('div + div', ['foobar-div']),
236 |     ('a ~ a', ['tag-anchor', 'nofollow-anchor']),
237 |     ('a[rel="tag"] ~ a', ['nofollow-anchor']),
238 |     ('ol#first-ol li:last-child', ['seventh-li']),
239 |     ('ol#first-ol *:last-child', ['li-div', 'seventh-li']),
240 |     ('#outer-div:first-child', ['outer-div']),
241 |     ('#outer-div :first-child', [
242 |         'name-anchor', 'first-li', 'li-div', 'p-b',
243 |         'checkbox-fieldset-disabled', 'area-href']),
244 |     ('a[href]', ['tag-anchor', 'nofollow-anchor']),
245 |     (':not(*)', []),
246 |     ('a:not([href])', ['name-anchor']),
247 |     ('ol :Not([class])', [
248 |         'first-li', 'second-li', 'li-div',
249 |         'fifth-li', 'sixth-li', 'seventh-li']),
250 |     ('li:not(:nth-child(odd), #second-li)', ['fourth-li', 'sixth-li']),
251 |     ('li:not(li)', []),
252 |     (':is(*)', ALL_IDS),
253 |     (':is(div)', ['outer-div', 'li-div', 'foobar-div']),
254 |     (':is(div, fieldset)', ['outer-div', 'li-div', 'fieldset', 'foobar-div']),
255 |     (':is(:::wrong)', []),
256 |     (':is(div, :::wrong, fieldset)', [
257 |         'outer-div', 'li-div', 'fieldset', 'foobar-div']),
258 |     ('div :is(div, div)', ['li-div']),
259 |     ('li:is(.c)', ['third-li', 'fourth-li']),
260 |     ('input:is([type="text"])', ['text-checked']),
261 |     ('div:is(:not(#outer-div))', ['li-div', 'foobar-div']),
262 |     ('div:is(div::before)', []),
263 |     (':where(*)', ALL_IDS),
264 |     (':where(div)', ['outer-div', 'li-div', 'foobar-div']),
265 |     (':where(div, fieldset)', [
266 |         'outer-div', 'li-div', 'fieldset', 'foobar-div']),
267 |     (':where(:::wrong)', []),
268 |     (':where(div, :::wrong, fieldset)', [
269 |         'outer-div', 'li-div', 'fieldset', 'foobar-div']),
270 |     ('div :where(div, div)', ['li-div']),
271 |     ('li:where(.c)', ['third-li', 'fourth-li']),
272 |     ('input:where([type="text"])', ['text-checked']),
273 |     ('div:where(:not(#outer-div))', ['li-div', 'foobar-div']),
274 |     ('div:where(div::before)', []),
275 |     ('p:has(input)', ['paragraph']),
276 |     ('p:has(fieldset input)', ['paragraph']),
277 |     ('p:has(> fieldset)', ['paragraph']),
278 |     ('ol:has(> div)', []),
279 |     ('ol:has(input, li)', ['first-ol']),
280 |     ('ol:has(input, fieldset)', []),
281 |     ('ol:has(+ p)', ['first-ol']),
282 |     ('ol:has(~ ol)', ['first-ol']),
283 |     ('ol:has(>a, ~ ol)', ['first-ol']),
284 |     ('ol:has(a,ol,  li  )', ['first-ol']),
285 |     ('ol:has(*)', ['first-ol']),
286 |     ('ol:has(:not(li))', ['first-ol']),
287 |     ('ol:has( > :not( li ))', []),
288 |     ('ol:has(:not(li, div))', []),
289 | 
290 |     # Invalid characters in XPath element names, should not crash
291 |     (r'di\a0 v', []),
292 |     (r'div\[', []),
293 |     (r'[h\a0 ref]', []),
294 |     (r'[h\]ref]', []),
295 | 
296 |     (':link', ['link-href', 'tag-anchor', 'nofollow-anchor', 'area-href']),
297 |     (':any-link', ['link-href', 'tag-anchor', 'nofollow-anchor', 'area-href']),
298 |     (':local-link', ['link-href', 'area-href']),
299 |     (':visited', []),
300 |     (':hover', []),
301 |     (':active', []),
302 |     (':focus', []),
303 |     (':target', []),
304 |     (':enabled', [
305 |         'link-href', 'tag-anchor', 'nofollow-anchor', 'checkbox-unchecked',
306 |         'text-checked', 'input-hidden', 'checkbox-checked', 'area-href']),
307 |     (':disabled', [
308 |         'checkbox-disabled', 'input-hidden-disabled',
309 |         'checkbox-disabled-checked', 'fieldset', 'checkbox-fieldset-disabled',
310 |         'hidden-fieldset-disabled']),
311 |     (':checked', ['checkbox-checked', 'checkbox-disabled-checked']),
312 | 
313 |     ('a:not([href]), div div', ['name-anchor', 'li-div']),
314 |     ('a:not([href]) /* test */, div div', ['name-anchor', 'li-div']),
315 |     ('a:not([href]), /* test */ div div', ['name-anchor', 'li-div']),
316 |     ('/* test */a:not([href]),div div', ['name-anchor', 'li-div']),
317 |     ('a:not([href]) , div div/* test */', ['name-anchor', 'li-div']),
318 |     ('/* test */a:not([href]), /* test */ div div', ['name-anchor', 'li-div']),
319 |     ('/* test */a:not([href])/* test */,div  div', ['name-anchor', 'li-div']),
320 |     ('/* test */ a:not([href]), div/* test */ div', ['name-anchor', 'li-div']),
321 |     ('a:not([href]) /* test */,/* test */div  div', ['name-anchor', 'li-div']),
322 | ))
323 | def test_select(selector, result):
324 |     xml_ids = [
325 |         element.etree_element.get('id', 'nil') for element in
326 |         ElementWrapper.from_xml_root(IDS_ROOT).query_all(selector)]
327 |     html_ids = [
328 |         element.etree_element.get('id', 'nil') for element in
329 |         ElementWrapper.from_html_root(IDS_ROOT).query_all(selector)]
330 |     assert xml_ids == html_ids == result
331 | 
332 | 
333 | @pytest.mark.parametrize('selector, result', (
334 |     ('DIV', ['outer-div', 'li-div', 'foobar-div']),
335 |     ('a[NAme]', ['name-anchor']),
336 |     ('HTML :link', [
337 |         'link-href', 'tag-anchor', 'nofollow-anchor', 'area-href']),
338 | ))
339 | def test_html_select(selector, result):
340 |     assert not [
341 |         element.etree_element.get('id', 'nil') for element in
342 |         ElementWrapper.from_xml_root(IDS_ROOT).query_all(selector)]
343 |     assert result == [
344 |         element.etree_element.get('id', 'nil') for element in
345 |         ElementWrapper.from_html_root(IDS_ROOT).query_all(selector)]
346 | 
347 | 
348 | # Data borrowed from http://mootools.net/slickspeed/
349 | @pytest.mark.parametrize('selector, result', (
350 |     # Changed from original because we’re only searching the body.
351 |     # ('*', 252),
352 |     ('*', 246),
353 |     # ('div:contains(CELIA)', 26),
354 |     ('div:only-child', 22),  # ?
355 |     ('div:nth-child(even)', 106),
356 |     ('div:nth-child(2n)', 106),
357 |     ('div:nth-child(odd)', 137),
358 |     ('div:nth-child(2n+1)', 137),
359 |     ('div:nth-child(n)', 243),
360 |     ('div:last-child', 53),
361 |     ('div:first-child', 51),
362 |     ('div > div', 242),
363 |     ('div + div', 190),
364 |     ('div ~ div', 190),
365 |     ('body', 1),
366 |     ('body div', 243),
367 |     ('div', 243),
368 |     ('div div', 242),
369 |     ('div div div', 241),
370 |     ('div, div, div', 243),
371 |     ('div, a, span', 243),
372 |     ('.dialog', 51),
373 |     ('div.dialog', 51),
374 |     ('div .dialog', 51),
375 |     ('div.character, div.dialog', 99),
376 |     ('div.direction.dialog', 0),
377 |     ('div.dialog.direction', 0),
378 |     ('div.dialog.scene', 1),
379 |     ('div.scene.scene', 1),
380 |     ('div.scene .scene', 0),
381 |     ('div.direction .dialog ', 0),
382 |     ('div .dialog .direction', 4),
383 |     ('div.dialog .dialog .direction', 4),
384 |     ('#speech5', 1),
385 |     ('div#speech5', 1),
386 |     ('div #speech5', 1),
387 |     ('div.scene div.dialog', 49),
388 |     ('div#scene1 div.dialog div', 142),
389 |     ('#scene1 #speech1', 1),
390 |     ('div[class]', 103),
391 |     ('div[class=dialog]', 50),
392 |     ('div[class^=dia]', 51),
393 |     ('div[class$=log]', 50),
394 |     ('div[class*=sce]', 1),
395 |     ('div[class|=dialog]', 50),  # ? Seems right
396 |     # assert count('div[class!=madeup]', 243),  # ? Seems right
397 |     ('div[class~=dialog]', 51),  # ? Seems right
398 | ))
399 | def test_select_shakespeare(selector, result):
400 |     assert sum(1 for _ in SHAKESPEARE_BODY.query_all(selector)) == result
401 | 


--------------------------------------------------------------------------------
/tests/w3_selectors.py:
--------------------------------------------------------------------------------
   1 | # File generated by make_selectors.py, do not edit
   2 | 
   3 | invalid_selectors = [
   4 |     {'name': 'Empty String', 'selector': ''},
   5 |     {'name': 'Invalid character', 'selector': '['},
   6 |     {'name': 'Invalid character', 'selector': ']'},
   7 |     {'name': 'Invalid character', 'selector': '('},
   8 |     {'name': 'Invalid character', 'selector': ')'},
   9 |     {'name': 'Invalid character', 'selector': '{'},
  10 |     {'name': 'Invalid character', 'selector': '}'},
  11 |     {'name': 'Invalid character', 'selector': '<'},
  12 |     {'name': 'Invalid character', 'selector': '>'},
  13 |     {'name': 'Invalid ID', 'selector': '#'},
  14 |     {'name': 'Invalid group of selectors', 'selector': 'div,'},
  15 |     {'name': 'Invalid class', 'selector': '.'},
  16 |     {'name': 'Invalid class', 'selector': '.5cm'},
  17 |     {'name': 'Invalid class', 'selector': '..test'},
  18 |     {'name': 'Invalid class', 'selector': '.foo..quux'},
  19 |     {'name': 'Invalid class', 'selector': '.bar.'},
  20 |     {'name': 'Invalid combinator', 'selector': 'div & address, p'},
  21 |     {'name': 'Invalid combinator', 'selector': 'div ++ address, p'},
  22 |     {'name': 'Invalid combinator', 'selector': 'div ~~ address, p'},
  23 |     {'name': 'Invalid [att=value] selector', 'selector': '[*=test]'},
  24 |     {'name': 'Invalid [att=value] selector', 'selector': '[*|*=test]'},
  25 |     {
  26 |         'name': 'Invalid [att=value] selector',
  27 |         'selector': '[class= space unquoted ]'},
  28 |     {'name': 'Unknown pseudo-class', 'selector': 'div:example'},
  29 |     {'name': 'Unknown pseudo-class', 'selector': ':example'},
  30 |     {'name': 'Unknown pseudo-class', 'selector': 'div:linkexample'},
  31 |     {'name': 'Unknown pseudo-element', 'selector': 'div::example'},
  32 |     {'name': 'Unknown pseudo-element', 'selector': '::example'},
  33 |     {'name': 'Invalid pseudo-element', 'selector': ':::before'},
  34 |     {'name': 'Invalid pseudo-element', 'selector': ':: before'},
  35 |     {'name': 'Undeclared namespace', 'selector': 'ns|div'},
  36 |     {'name': 'Undeclared namespace', 'selector': ':not(ns|div)'},
  37 |     {'name': 'Invalid arguments', 'selector': ':nth-child(3 of)'},
  38 |     {'name': 'Invalid namespace', 'selector': '^|div'},
  39 |     {'name': 'Invalid namespace', 'selector': '$|div'},
  40 |     {'name': 'Relative selector', 'selector': '>*'}]
  41 | 
  42 | valid_selectors = [
  43 |     {
  44 |         'exclude': ['element', 'fragment', 'detached'],
  45 |         'expect': ['html'],
  46 |         'level': 1,
  47 |         'name': 'Type selector, matching html element',
  48 |         'selector': 'html'},
  49 |     {
  50 |         'exclude': ['document'],
  51 |         'expect': [],
  52 |         'level': 1,
  53 |         'name': 'Type selector, matching html element',
  54 |         'selector': 'html'},
  55 |     {
  56 |         'exclude': ['element', 'fragment', 'detached'],
  57 |         'expect': ['body'],
  58 |         'level': 1,
  59 |         'name': 'Type selector, matching body element',
  60 |         'selector': 'body'},
  61 |     {
  62 |         'exclude': ['document'],
  63 |         'expect': [],
  64 |         'level': 1,
  65 |         'name': 'Type selector, matching body element',
  66 |         'selector': 'body'},
  67 |     {
  68 |         'expect': [
  69 |             'universal',
  70 |             'universal-p1',
  71 |             'universal-code1',
  72 |             'universal-hr1',
  73 |             'universal-pre1',
  74 |             'universal-span1',
  75 |             'universal-p2',
  76 |             'universal-a1',
  77 |             'universal-address1',
  78 |             'universal-code2',
  79 |             'universal-a2'],
  80 |         'level': 2,
  81 |         'name': 'Universal selector, matching all elements',
  82 |         'selector': '*'},
  83 |     {
  84 |         'expect': [
  85 |             'universal-p1',
  86 |             'universal-hr1',
  87 |             'universal-pre1',
  88 |             'universal-p2',
  89 |             'universal-address1'],
  90 |         'level': 2,
  91 |         'name': 'Universal selector, matching all children of element with '
  92 |                 'specified ID',
  93 |         'selector': '#universal>*'},
  94 |     {
  95 |         'expect': [
  96 |             'universal-code1',
  97 |             'universal-span1',
  98 |             'universal-a1',
  99 |             'universal-code2'],
 100 |         'level': 2,
 101 |         'name': 'Universal selector, matching all grandchildren of element '
 102 |                 'with specified ID',
 103 |         'selector': '#universal>*>*'},
 104 |     {
 105 |         'expect': [],
 106 |         'level': 2,
 107 |         'name': 'Universal selector, matching all children of empty element '
 108 |                 'with specified ID',
 109 |         'selector': '#empty>*'},
 110 |     {
 111 |         'expect': [
 112 |             'universal-p1',
 113 |             'universal-code1',
 114 |             'universal-hr1',
 115 |             'universal-pre1',
 116 |             'universal-span1',
 117 |             'universal-p2',
 118 |             'universal-a1',
 119 |             'universal-address1',
 120 |             'universal-code2',
 121 |             'universal-a2'],
 122 |         'level': 2,
 123 |         'name': 'Universal selector, matching all descendants of element with '
 124 |                 'specified ID',
 125 |         'selector': '#universal *'},
 126 |     {
 127 |         'expect': ['attr-presence-div1'],
 128 |         'level': 2,
 129 |         'name': 'Attribute presence selector, matching align attribute with '
 130 |                 'value',
 131 |         'selector': '.attr-presence-div1[align]'},
 132 |     {
 133 |         'expect': ['attr-presence-div2'],
 134 |         'level': 2,
 135 |         'name': 'Attribute presence selector, matching align attribute with '
 136 |                 'empty value',
 137 |         'selector': '.attr-presence-div2[align]'},
 138 |     {
 139 |         'exclude': ['xhtml'],
 140 |         'expect': [
 141 |             'attr-presence-a1',
 142 |             'attr-presence-span1',
 143 |             'attr-presence-i1'],
 144 |         'level': 2,
 145 |         'name': 'Attribute presence selector, matching title attribute, case '
 146 |                 'insensitivity',
 147 |         'selector': '#attr-presence [*|TiTlE]'},
 148 |     {
 149 |         'exclude': ['html'],
 150 |         'expect': [],
 151 |         'level': 2,
 152 |         'name': 'Attribute presence selector, not matching title attribute, '
 153 |                 'case sensitivity',
 154 |         'selector': '#attr-presence [*|TiTlE]'},
 155 |     {
 156 |         'expect': ['attr-presence-pre1', 'attr-presence-blockquote1'],
 157 |         'level': 2,
 158 |         'name': 'Attribute presence selector, matching custom data-* '
 159 |                 'attribute',
 160 |         'selector': '[data-attr-presence]'},
 161 |     {
 162 |         'expect': [],
 163 |         'level': 2,
 164 |         'name': 'Attribute presence selector, not matching attribute with '
 165 |                 'similar name',
 166 |         'selector': '.attr-presence-div3[align], .attr-presence-div4[align]'},
 167 |     {
 168 |         'expect': ['attr-presence-ul1'],
 169 |         'level': 2,
 170 |         'name': 'Attribute presence selector, matching attribute with '
 171 |                 'non-ASCII characters',
 172 |         'selector': 'ul[data-中文]'},
 173 |     {
 174 |         'expect': [],
 175 |         'level': 2,
 176 |         'name': 'Attribute presence selector, not matching default option '
 177 |                 'without selected attribute',
 178 |         'selector': '#attr-presence-select1 option[selected]'},
 179 |     {
 180 |         'expect': ['attr-presence-select2-option4'],
 181 |         'level': 2,
 182 |         'name': 'Attribute presence selector, matching option with selected '
 183 |                 'attribute',
 184 |         'selector': '#attr-presence-select2 option[selected]'},
 185 |     {
 186 |         'expect': [
 187 |             'attr-presence-select3-option2',
 188 |             'attr-presence-select3-option3'],
 189 |         'level': 2,
 190 |         'name': 'Attribute presence selector, matching multiple options with '
 191 |                 'selected attributes',
 192 |         'selector': '#attr-presence-select3 option[selected]'},
 193 |     {
 194 |         'expect': ['attr-value-div1'],
 195 |         'level': 2,
 196 |         'name': 'Attribute value selector, matching align attribute with '
 197 |                 'value',
 198 |         'selector': '#attr-value [align="center"]'},
 199 |     {
 200 |         'expect': ['attr-value-div1'],
 201 |         'level': 2,
 202 |         'name': 'Attribute value selector, matching align attribute with '
 203 |                 'value, unclosed bracket',
 204 |         'selector': '#attr-value [align="center"'},
 205 |     {
 206 |         'expect': ['attr-value-div2'],
 207 |         'level': 2,
 208 |         'name': 'Attribute value selector, matching align attribute with '
 209 |                 'empty value',
 210 |         'selector': '#attr-value [align=""]'},
 211 |     {
 212 |         'expect': [],
 213 |         'level': 2,
 214 |         'name': 'Attribute value selector, not matching align attribute with '
 215 |                 'partial value',
 216 |         'selector': '#attr-value [align="c"]'},
 217 |     {
 218 |         'expect': [],
 219 |         'level': 2,
 220 |         'name': 'Attribute value selector, not matching align attribute with '
 221 |                 'incorrect value',
 222 |         'selector': '#attr-value [align="centera"]'},
 223 |     {
 224 |         'expect': ['attr-value-div3'],
 225 |         'level': 2,
 226 |         'name': 'Attribute value selector, matching custom data-* attribute '
 227 |                 'with unicode escaped value',
 228 |         'selector': '[data-attr-value="\\e9"]'},
 229 |     {
 230 |         'expect': ['attr-value-div4'],
 231 |         'level': 2,
 232 |         'name': 'Attribute value selector, matching custom data-* attribute '
 233 |                 'with escaped character',
 234 |         'selector': '[data-attr-value_foo="\\e9"]'},
 235 |     {
 236 |         'expect': [
 237 |             'attr-value-input3',
 238 |             'attr-value-input4',
 239 |             'attr-value-input6',
 240 |             'attr-value-input8',
 241 |             'attr-value-input9'],
 242 |         'level': 2,
 243 |         'name': 'Attribute value selector with single-quoted value, matching '
 244 |                 'multiple inputs with type attributes',
 245 |         'selector': "#attr-value input[type='hidden'],#attr-value "
 246 |                     "input[type='radio']"},
 247 |     {
 248 |         'expect': [
 249 |             'attr-value-input3',
 250 |             'attr-value-input4',
 251 |             'attr-value-input6',
 252 |             'attr-value-input8',
 253 |             'attr-value-input9'],
 254 |         'level': 2,
 255 |         'name': 'Attribute value selector with double-quoted value, matching '
 256 |                 'multiple inputs with type attributes',
 257 |         'selector': '#attr-value input[type="hidden"],#attr-value '
 258 |                     "input[type='radio']"},
 259 |     {
 260 |         'expect': [
 261 |             'attr-value-input3',
 262 |             'attr-value-input4',
 263 |             'attr-value-input6',
 264 |             'attr-value-input8',
 265 |             'attr-value-input9'],
 266 |         'level': 2,
 267 |         'name': 'Attribute value selector with unquoted value, matching '
 268 |                 'multiple inputs with type attributes',
 269 |         'selector': '#attr-value input[type=hidden],#attr-value '
 270 |                     'input[type=radio]'},
 271 |     {
 272 |         'expect': ['attr-value-div5'],
 273 |         'level': 2,
 274 |         'name': 'Attribute value selector, matching attribute with value '
 275 |                 'using non-ASCII characters',
 276 |         'selector': '[data-attr-value=中文]'},
 277 |     {
 278 |         'expect': ['attr-whitespace-div1'],
 279 |         'level': 2,
 280 |         'name': 'Attribute whitespace-separated list selector, matching class '
 281 |                 'attribute with value',
 282 |         'selector': '#attr-whitespace [class~="div1"]'},
 283 |     {
 284 |         'expect': [],
 285 |         'level': 2,
 286 |         'name': 'Attribute whitespace-separated list selector, not matching '
 287 |                 'class attribute with empty value',
 288 |         'selector': '#attr-whitespace [class~=""]'},
 289 |     {
 290 |         'expect': [],
 291 |         'level': 2,
 292 |         'name': 'Attribute whitespace-separated list selector, not matching '
 293 |                 'class attribute with partial value',
 294 |         'selector': '[data-attr-whitespace~="div"]'},
 295 |     {
 296 |         'expect': ['attr-whitespace-div4'],
 297 |         'level': 2,
 298 |         'name': 'Attribute whitespace-separated list selector, matching '
 299 |                 'custom data-* attribute with unicode escaped value',
 300 |         'selector': '[data-attr-whitespace~="\\0000e9"]'},
 301 |     {
 302 |         'expect': ['attr-whitespace-div5'],
 303 |         'level': 2,
 304 |         'name': 'Attribute whitespace-separated list selector, matching '
 305 |                 'custom data-* attribute with escaped character',
 306 |         'selector': '[data-attr-whitespace_foo~="\\e9"]'},
 307 |     {
 308 |         'expect': [
 309 |             'attr-whitespace-a1',
 310 |             'attr-whitespace-a2',
 311 |             'attr-whitespace-a3',
 312 |             'attr-whitespace-a5',
 313 |             'attr-whitespace-a7'],
 314 |         'level': 2,
 315 |         'name': 'Attribute whitespace-separated list selector with '
 316 |                 'single-quoted value, matching multiple links with rel '
 317 |                 'attributes',
 318 |         'selector': "#attr-whitespace a[rel~='bookmark'],  #attr-whitespace "
 319 |                     "a[rel~='nofollow']"},
 320 |     {
 321 |         'expect': [
 322 |             'attr-whitespace-a1',
 323 |             'attr-whitespace-a2',
 324 |             'attr-whitespace-a3',
 325 |             'attr-whitespace-a5',
 326 |             'attr-whitespace-a7'],
 327 |         'level': 2,
 328 |         'name': 'Attribute whitespace-separated list selector with '
 329 |                 'double-quoted value, matching multiple links with rel '
 330 |                 'attributes',
 331 |         'selector': '#attr-whitespace a[rel~="bookmark"],#attr-whitespace '
 332 |                     "a[rel~='nofollow']"},
 333 |     {
 334 |         'expect': [
 335 |             'attr-whitespace-a1',
 336 |             'attr-whitespace-a2',
 337 |             'attr-whitespace-a3',
 338 |             'attr-whitespace-a5',
 339 |             'attr-whitespace-a7'],
 340 |         'level': 2,
 341 |         'name': 'Attribute whitespace-separated list selector with unquoted '
 342 |                 'value, matching multiple links with rel attributes',
 343 |         'selector': '#attr-whitespace a[rel~=bookmark],    #attr-whitespace '
 344 |                     'a[rel~=nofollow]'},
 345 |     {
 346 |         'expect': [],
 347 |         'level': 2,
 348 |         'name': 'Attribute whitespace-separated list selector with '
 349 |                 'double-quoted value, not matching value with space',
 350 |         'selector': '#attr-whitespace a[rel~="book mark"]'},
 351 |     {
 352 |         'expect': ['attr-whitespace-p1'],
 353 |         'level': 2,
 354 |         'name': 'Attribute whitespace-separated list selector, matching title '
 355 |                 'attribute with value using non-ASCII characters',
 356 |         'selector': '#attr-whitespace [title~=中文]'},
 357 |     {
 358 |         'expect': [],
 359 |         'level': 2,
 360 |         'name': 'Attribute hyphen-separated list selector, not matching '
 361 |                 'unspecified lang attribute',
 362 |         'selector': '#attr-hyphen-div1[lang|="en"]'},
 363 |     {
 364 |         'expect': ['attr-hyphen-div2'],
 365 |         'level': 2,
 366 |         'name': 'Attribute hyphen-separated list selector, matching lang '
 367 |                 'attribute with exact value',
 368 |         'selector': '#attr-hyphen-div2[lang|="fr"]'},
 369 |     {
 370 |         'expect': ['attr-hyphen-div3'],
 371 |         'level': 2,
 372 |         'name': 'Attribute hyphen-separated list selector, matching lang '
 373 |                 'attribute with partial value',
 374 |         'selector': '#attr-hyphen-div3[lang|="en"]'},
 375 |     {
 376 |         'expect': [],
 377 |         'level': 2,
 378 |         'name': 'Attribute hyphen-separated list selector, not matching '
 379 |                 'incorrect value',
 380 |         'selector': '#attr-hyphen-div4[lang|="es-AR"]'},
 381 |     {
 382 |         'expect': ['attr-begins-a1', 'attr-begins-a3'],
 383 |         'level': 3,
 384 |         'name': 'Attribute begins with selector, matching href attributes '
 385 |                 'beginning with specified substring',
 386 |         'selector': '#attr-begins a[href^="http://www"]'},
 387 |     {
 388 |         'expect': ['attr-begins-div2', 'attr-begins-div4'],
 389 |         'level': 3,
 390 |         'name': 'Attribute begins with selector, matching lang attributes '
 391 |                 'beginning with specified substring, ',
 392 |         'selector': '#attr-begins [lang^="en-"]'},
 393 |     {
 394 |         'expect': [],
 395 |         'level': 3,
 396 |         'name': 'Attribute begins with selector, not matching class attribute '
 397 |                 'with empty value',
 398 |         'selector': '#attr-begins [class^=""]'},
 399 |     {
 400 |         'expect': [],
 401 |         'level': 3,
 402 |         'name': 'Attribute begins with selector, not matching class attribute '
 403 |                 'not beginning with specified substring',
 404 |         'selector': '#attr-begins [class^=apple]'},
 405 |     {
 406 |         'expect': ['attr-begins-p1'],
 407 |         'level': 3,
 408 |         'name': 'Attribute begins with selector with single-quoted value, '
 409 |                 'matching class attribute beginning with specified substring',
 410 |         'selector': "#attr-begins [class^=' apple']"},
 411 |     {
 412 |         'expect': ['attr-begins-p1'],
 413 |         'level': 3,
 414 |         'name': 'Attribute begins with selector with double-quoted value, '
 415 |                 'matching class attribute beginning with specified substring',
 416 |         'selector': '#attr-begins [class^=" apple"]'},
 417 |     {
 418 |         'expect': [],
 419 |         'level': 3,
 420 |         'name': 'Attribute begins with selector with unquoted value, not '
 421 |                 'matching class attribute not beginning with specified '
 422 |                 'substring',
 423 |         'selector': '#attr-begins [class^= apple]'},
 424 |     {
 425 |         'expect': ['attr-ends-a1', 'attr-ends-a3'],
 426 |         'level': 3,
 427 |         'name': 'Attribute ends with selector, matching href attributes '
 428 |                 'ending with specified substring',
 429 |         'selector': '#attr-ends a[href$=".org"]'},
 430 |     {
 431 |         'expect': ['attr-ends-div2', 'attr-ends-div4'],
 432 |         'level': 3,
 433 |         'name': 'Attribute ends with selector, matching lang attributes '
 434 |                 'ending with specified substring, ',
 435 |         'selector': '#attr-ends [lang$="-CH"]'},
 436 |     {
 437 |         'expect': [],
 438 |         'level': 3,
 439 |         'name': 'Attribute ends with selector, not matching class attribute '
 440 |                 'with empty value',
 441 |         'selector': '#attr-ends [class$=""]'},
 442 |     {
 443 |         'expect': [],
 444 |         'level': 3,
 445 |         'name': 'Attribute ends with selector, not matching class attribute '
 446 |                 'not ending with specified substring',
 447 |         'selector': '#attr-ends [class$=apple]'},
 448 |     {
 449 |         'expect': ['attr-ends-p1'],
 450 |         'level': 3,
 451 |         'name': 'Attribute ends with selector with single-quoted value, '
 452 |                 'matching class attribute ending with specified substring',
 453 |         'selector': "#attr-ends [class$='apple ']"},
 454 |     {
 455 |         'expect': ['attr-ends-p1'],
 456 |         'level': 3,
 457 |         'name': 'Attribute ends with selector with double-quoted value, '
 458 |                 'matching class attribute ending with specified substring',
 459 |         'selector': '#attr-ends [class$="apple "]'},
 460 |     {
 461 |         'expect': [],
 462 |         'level': 3,
 463 |         'name': 'Attribute ends with selector with unquoted value, not '
 464 |                 'matching class attribute not ending with specified substring',
 465 |         'selector': '#attr-ends [class$=apple ]'},
 466 |     {
 467 |         'expect': ['attr-contains-a1', 'attr-contains-a3'],
 468 |         'level': 3,
 469 |         'name': 'Attribute contains selector, matching href attributes '
 470 |                 'beginning with specified substring',
 471 |         'selector': '#attr-contains a[href*="http://www"]'},
 472 |     {
 473 |         'expect': ['attr-contains-a1', 'attr-contains-a2'],
 474 |         'level': 3,
 475 |         'name': 'Attribute contains selector, matching href attributes ending '
 476 |                 'with specified substring',
 477 |         'selector': '#attr-contains a[href*=".org"]'},
 478 |     {
 479 |         'expect': ['attr-contains-a1', 'attr-contains-a3'],
 480 |         'level': 3,
 481 |         'name': 'Attribute contains selector, matching href attributes '
 482 |                 'containing specified substring',
 483 |         'selector': '#attr-contains a[href*=".example."]'},
 484 |     {
 485 |         'expect': ['attr-contains-div2', 'attr-contains-div6'],
 486 |         'level': 3,
 487 |         'name': 'Attribute contains selector, matching lang attributes '
 488 |                 'beginning with specified substring, ',
 489 |         'selector': '#attr-contains [lang*="en-"]'},
 490 |     {
 491 |         'expect': ['attr-contains-div3', 'attr-contains-div5'],
 492 |         'level': 3,
 493 |         'name': 'Attribute contains selector, matching lang attributes ending '
 494 |                 'with specified substring, ',
 495 |         'selector': '#attr-contains [lang*="-CH"]'},
 496 |     {
 497 |         'expect': [],
 498 |         'level': 3,
 499 |         'name': 'Attribute contains selector, not matching class attribute '
 500 |                 'with empty value',
 501 |         'selector': '#attr-contains [class*=""]'},
 502 |     {
 503 |         'expect': ['attr-contains-p1'],
 504 |         'level': 3,
 505 |         'name': 'Attribute contains selector with single-quoted value, '
 506 |                 'matching class attribute beginning with specified substring',
 507 |         'selector': "#attr-contains [class*=' apple']"},
 508 |     {
 509 |         'expect': ['attr-contains-p1'],
 510 |         'level': 3,
 511 |         'name': 'Attribute contains selector with single-quoted value, '
 512 |                 'matching class attribute ending with specified substring',
 513 |         'selector': "#attr-contains [class*='orange ']"},
 514 |     {
 515 |         'expect': ['attr-contains-p1'],
 516 |         'level': 3,
 517 |         'name': 'Attribute contains selector with single-quoted value, '
 518 |                 'matching class attribute containing specified substring',
 519 |         'selector': "#attr-contains [class*='ple banana ora']"},
 520 |     {
 521 |         'expect': ['attr-contains-p1'],
 522 |         'level': 3,
 523 |         'name': 'Attribute contains selector with double-quoted value, '
 524 |                 'matching class attribute beginning with specified substring',
 525 |         'selector': '#attr-contains [class*=" apple"]'},
 526 |     {
 527 |         'expect': ['attr-contains-p1'],
 528 |         'level': 3,
 529 |         'name': 'Attribute contains selector with double-quoted value, '
 530 |                 'matching class attribute ending with specified substring',
 531 |         'selector': '#attr-contains [class*="orange "]'},
 532 |     {
 533 |         'expect': ['attr-contains-p1'],
 534 |         'level': 3,
 535 |         'name': 'Attribute contains selector with double-quoted value, '
 536 |                 'matching class attribute containing specified substring',
 537 |         'selector': '#attr-contains [class*="ple banana ora"]'},
 538 |     {
 539 |         'expect': ['attr-contains-p1'],
 540 |         'level': 3,
 541 |         'name': 'Attribute contains selector with unquoted value, matching '
 542 |                 'class attribute beginning with specified substring',
 543 |         'selector': '#attr-contains [class*= apple]'},
 544 |     {
 545 |         'expect': ['attr-contains-p1'],
 546 |         'level': 3,
 547 |         'name': 'Attribute contains selector with unquoted value, matching '
 548 |                 'class attribute ending with specified substring',
 549 |         'selector': '#attr-contains [class*=orange ]'},
 550 |     {
 551 |         'expect': ['attr-contains-p1'],
 552 |         'level': 3,
 553 |         'name': 'Attribute contains selector with unquoted value, matching '
 554 |                 'class attribute containing specified substring',
 555 |         'selector': '#attr-contains [class*= banana ]'},
 556 |     {
 557 |         'exclude': ['element', 'fragment', 'detached'],
 558 |         'expect': ['html'],
 559 |         'level': 3,
 560 |         'name': ':root pseudo-class selector, matching document root element',
 561 |         'selector': ':root'},
 562 |     {
 563 |         'exclude': ['document'],
 564 |         'expect': [],
 565 |         'level': 3,
 566 |         'name': ':root pseudo-class selector, not matching document root '
 567 |                 'element',
 568 |         'selector': ':root'},
 569 |     {
 570 |         'expect': [
 571 |             'pseudo-nth-td3',
 572 |             'pseudo-nth-td9',
 573 |             'pseudo-nth-tr3',
 574 |             'pseudo-nth-td15'],
 575 |         'level': 3,
 576 |         'name': ':nth-child selector, matching the third child element',
 577 |         'selector': '#pseudo-nth-table1 :nth-child(3)'},
 578 |     {
 579 |         'expect': [
 580 |             'pseudo-nth-li3',
 581 |             'pseudo-nth-li6',
 582 |             'pseudo-nth-li9',
 583 |             'pseudo-nth-li12'],
 584 |         'level': 3,
 585 |         'name': ':nth-child selector, matching every third child element',
 586 |         'selector': '#pseudo-nth li:nth-child(3n)'},
 587 |     {
 588 |         'expect': [
 589 |             'pseudo-nth-li4',
 590 |             'pseudo-nth-li6',
 591 |             'pseudo-nth-li8',
 592 |             'pseudo-nth-li10',
 593 |             'pseudo-nth-li12'],
 594 |         'level': 3,
 595 |         'name': ':nth-child selector, matching every second child element, '
 596 |                 'starting from the fourth',
 597 |         'selector': '#pseudo-nth li:nth-child(2n+4)'},
 598 |     {
 599 |         'expect': ['pseudo-nth-em2', 'pseudo-nth-span3'],
 600 |         'level': 3,
 601 |         'name': ':nth-child selector, matching every fourth child element, '
 602 |                 'starting from the third',
 603 |         'selector': '#pseudo-nth-p1 :nth-child(4n-1)'},
 604 |     {
 605 |         'expect': [
 606 |             'pseudo-nth-tr1',
 607 |             'pseudo-nth-td4',
 608 |             'pseudo-nth-td10',
 609 |             'pseudo-nth-td16'],
 610 |         'level': 3,
 611 |         'name': ':nth-last-child selector, matching the third last child '
 612 |                 'element',
 613 |         'selector': '#pseudo-nth-table1 :nth-last-child(3)'},
 614 |     {
 615 |         'expect': [
 616 |             'pseudo-nth-li1',
 617 |             'pseudo-nth-li4',
 618 |             'pseudo-nth-li7',
 619 |             'pseudo-nth-li10'],
 620 |         'level': 3,
 621 |         'name': ':nth-last-child selector, matching every third child element '
 622 |                 'from the end',
 623 |         'selector': '#pseudo-nth li:nth-last-child(3n)'},
 624 |     {
 625 |         'expect': [
 626 |             'pseudo-nth-li1',
 627 |             'pseudo-nth-li3',
 628 |             'pseudo-nth-li5',
 629 |             'pseudo-nth-li7',
 630 |             'pseudo-nth-li9'],
 631 |         'level': 3,
 632 |         'name': ':nth-last-child selector, matching every second child '
 633 |                 'element from the end, starting from the fourth last',
 634 |         'selector': '#pseudo-nth li:nth-last-child(2n+4)'},
 635 |     {
 636 |         'expect': ['pseudo-nth-span2', 'pseudo-nth-span4'],
 637 |         'level': 3,
 638 |         'name': ':nth-last-child selector, matching every fourth element from '
 639 |                 'the end, starting from the third last',
 640 |         'selector': '#pseudo-nth-p1 :nth-last-child(4n-1)'},
 641 |     {
 642 |         'expect': ['pseudo-nth-em3'],
 643 |         'level': 3,
 644 |         'name': ':nth-of-type selector, matching the third em element',
 645 |         'selector': '#pseudo-nth-p1 em:nth-of-type(3)'},
 646 |     {
 647 |         'expect': [
 648 |             'pseudo-nth-em2',
 649 |             'pseudo-nth-span2',
 650 |             'pseudo-nth-span4',
 651 |             'pseudo-nth-strong2',
 652 |             'pseudo-nth-em4'],
 653 |         'level': 3,
 654 |         'name': ':nth-of-type selector, matching every second element of '
 655 |                 'their type',
 656 |         'selector': '#pseudo-nth-p1 :nth-of-type(2n)'},
 657 |     {
 658 |         'expect': ['pseudo-nth-span1', 'pseudo-nth-span3'],
 659 |         'level': 3,
 660 |         'name': ':nth-of-type selector, matching every second elemetn of '
 661 |                 'their type, starting from the first',
 662 |         'selector': '#pseudo-nth-p1 span:nth-of-type(2n-1)'},
 663 |     {
 664 |         'expect': ['pseudo-nth-em2'],
 665 |         'level': 3,
 666 |         'name': ':nth-last-of-type selector, matching the third last em '
 667 |                 'element',
 668 |         'selector': '#pseudo-nth-p1 em:nth-last-of-type(3)'},
 669 |     {
 670 |         'expect': [
 671 |             'pseudo-nth-span1',
 672 |             'pseudo-nth-em1',
 673 |             'pseudo-nth-strong1',
 674 |             'pseudo-nth-em3',
 675 |             'pseudo-nth-span3'],
 676 |         'level': 3,
 677 |         'name': ':nth-last-of-type selector, matching every second last '
 678 |                 'element of their type',
 679 |         'selector': '#pseudo-nth-p1 :nth-last-of-type(2n)'},
 680 |     {
 681 |         'expect': ['pseudo-nth-span2', 'pseudo-nth-span4'],
 682 |         'level': 3,
 683 |         'name': ':nth-last-of-type selector, matching every second last '
 684 |                 'element of their type, starting from the last',
 685 |         'selector': '#pseudo-nth-p1 span:nth-last-of-type(2n-1)'},
 686 |     {
 687 |         'expect': ['pseudo-nth-em1'],
 688 |         'level': 3,
 689 |         'name': ':first-of-type selector, matching the first em element',
 690 |         'selector': '#pseudo-nth-p1 em:first-of-type'},
 691 |     {
 692 |         'expect': ['pseudo-nth-span1', 'pseudo-nth-em1', 'pseudo-nth-strong1'],
 693 |         'level': 3,
 694 |         'name': ':first-of-type selector, matching the first of every type of '
 695 |                 'element',
 696 |         'selector': '#pseudo-nth-p1 :first-of-type'},
 697 |     {
 698 |         'expect': ['pseudo-nth-td1', 'pseudo-nth-td7', 'pseudo-nth-td13'],
 699 |         'level': 3,
 700 |         'name': ':first-of-type selector, matching the first td element in '
 701 |                 'each table row',
 702 |         'selector': '#pseudo-nth-table1 tr :first-of-type'},
 703 |     {
 704 |         'expect': ['pseudo-nth-em4'],
 705 |         'level': 3,
 706 |         'name': ':last-of-type selector, matching the last em elemnet',
 707 |         'selector': '#pseudo-nth-p1 em:last-of-type'},
 708 |     {
 709 |         'expect': ['pseudo-nth-span4', 'pseudo-nth-strong2', 'pseudo-nth-em4'],
 710 |         'level': 3,
 711 |         'name': ':last-of-type selector, matching the last of every type of '
 712 |                 'element',
 713 |         'selector': '#pseudo-nth-p1 :last-of-type'},
 714 |     {
 715 |         'expect': ['pseudo-nth-td6', 'pseudo-nth-td12', 'pseudo-nth-td18'],
 716 |         'level': 3,
 717 |         'name': ':last-of-type selector, matching the last td element in each '
 718 |                 'table row',
 719 |         'selector': '#pseudo-nth-table1 tr :last-of-type'},
 720 |     {
 721 |         'expect': ['pseudo-first-child-div1'],
 722 |         'level': 2,
 723 |         'name': ':first-child pseudo-class selector, matching first child div '
 724 |                 'element',
 725 |         'selector': '#pseudo-first-child div:first-child'},
 726 |     {
 727 |         'expect': [],
 728 |         'level': 2,
 729 |         'name': ":first-child pseudo-class selector, doesn't match "
 730 |                 'non-first-child elements',
 731 |         'selector': '.pseudo-first-child-div2:first-child, '
 732 |                     '.pseudo-first-child-div3:first-child'},
 733 |     {
 734 |         'expect': [
 735 |             'pseudo-first-child-span1',
 736 |             'pseudo-first-child-span3',
 737 |             'pseudo-first-child-span5'],
 738 |         'level': 2,
 739 |         'name': ':first-child pseudo-class selector, matching first-child of '
 740 |                 'multiple elements',
 741 |         'selector': '#pseudo-first-child span:first-child'},
 742 |     {
 743 |         'expect': ['pseudo-last-child-div3'],
 744 |         'level': 3,
 745 |         'name': ':last-child pseudo-class selector, matching last child div '
 746 |                 'element',
 747 |         'selector': '#pseudo-last-child div:last-child'},
 748 |     {
 749 |         'expect': [],
 750 |         'level': 3,
 751 |         'name': ":last-child pseudo-class selector, doesn't match "
 752 |                 'non-last-child elements',
 753 |         'selector': '.pseudo-last-child-div1:last-child, '
 754 |                     '.pseudo-last-child-div2:first-child'},
 755 |     {
 756 |         'expect': [
 757 |             'pseudo-last-child-span2',
 758 |             'pseudo-last-child-span4',
 759 |             'pseudo-last-child-span6'],
 760 |         'level': 3,
 761 |         'name': ':last-child pseudo-class selector, matching first-child of '
 762 |                 'multiple elements',
 763 |         'selector': '#pseudo-last-child span:last-child'},
 764 |     {
 765 |         'expect': ['pseudo-only-span1'],
 766 |         'level': 3,
 767 |         'name': ':pseudo-only-child pseudo-class selector, matching all '
 768 |                 'only-child elements',
 769 |         'selector': '#pseudo-only :only-child'},
 770 |     {
 771 |         'expect': [],
 772 |         'level': 3,
 773 |         'name': ':pseudo-only-child pseudo-class selector, matching '
 774 |                 'only-child em elements',
 775 |         'selector': '#pseudo-only em:only-child'},
 776 |     {
 777 |         'expect': ['pseudo-only-span1', 'pseudo-only-em1'],
 778 |         'level': 3,
 779 |         'name': ':pseudo-only-of-type pseudo-class selector, matching all '
 780 |                 'elements with no siblings of the same type',
 781 |         'selector': '#pseudo-only :only-of-type'},
 782 |     {
 783 |         'expect': ['pseudo-only-em1'],
 784 |         'level': 3,
 785 |         'name': ':pseudo-only-of-type pseudo-class selector, matching em '
 786 |                 'elements with no siblings of the same type',
 787 |         'selector': '#pseudo-only em:only-of-type'},
 788 |     {
 789 |         'expect': ['pseudo-empty-p1', 'pseudo-empty-p2'],
 790 |         'level': 3,
 791 |         'name': ':empty pseudo-class selector, matching empty p elements',
 792 |         'selector': '#pseudo-empty p:empty'},
 793 |     {
 794 |         'expect': ['pseudo-empty-p1', 'pseudo-empty-p2', 'pseudo-empty-span1'],
 795 |         'level': 3,
 796 |         'name': ':empty pseudo-class selector, matching all empty elements',
 797 |         'selector': '#pseudo-empty :empty'},
 798 |     {
 799 |         'expect': ['pseudo-link-a1', 'pseudo-link-a2', 'pseudo-link-area1'],
 800 |         'level': 1,
 801 |         'name': ':link and :visited pseudo-class selectors, matching a and '
 802 |                 'area elements with href attributes',
 803 |         'selector': '#pseudo-link :link, #pseudo-link :visited'},
 804 |     {
 805 |         'exclude': ['element', 'fragment', 'detached'],
 806 |         'expect': [],
 807 |         'level': 1,
 808 |         'name': ':link and :visited pseudo-class selectors, matching no '
 809 |                 'elements',
 810 |         'selector': '#head :link, #head :visited'},
 811 |     {
 812 |         'exclude': ['document'],
 813 |         'expect': [],
 814 |         'level': 1,
 815 |         'name': ':link and :visited pseudo-class selectors, not matching link '
 816 |                 'elements with href attributes',
 817 |         'selector': '#head :link, #head :visited'},
 818 |     {
 819 |         'exclude': ['document'],
 820 |         'expect': [],
 821 |         'level': 1,
 822 |         'name': ':link and :visited pseudo-class selectors, chained, mutually '
 823 |                 'exclusive pseudo-classes match nothing',
 824 |         'selector': ':link:visited'},
 825 |     {
 826 |         'exclude': ['document', 'element'],
 827 |         'expect': [],
 828 |         'level': 3,
 829 |         'name': ':target pseudo-class selector, matching the element '
 830 |                 'referenced by the URL fragment identifier',
 831 |         'selector': ':target'},
 832 |     {
 833 |         'exclude': ['fragment', 'detached'],
 834 |         'expect': ['target'],
 835 |         'level': 3,
 836 |         'name': ':target pseudo-class selector, matching the element '
 837 |                 'referenced by the URL fragment identifier',
 838 |         'selector': ':target'},
 839 |     {
 840 |         'exclude': ['detached', 'fragment'],
 841 |         'expect': ['pseudo-lang-div1'],
 842 |         'level': 2,
 843 |         'name': ':lang pseudo-class selector, matching inherited language',
 844 |         'selector': '#pseudo-lang-div1:lang(en)'},
 845 |     {
 846 |         'exclude': ['document', 'element'],
 847 |         'expect': [],
 848 |         'level': 2,
 849 |         'name': ':lang pseudo-class selector, not matching element with no '
 850 |                 'inherited language',
 851 |         'selector': '#pseudo-lang-div1:lang(en)'},
 852 |     {
 853 |         'expect': ['pseudo-lang-div2'],
 854 |         'level': 2,
 855 |         'name': ':lang pseudo-class selector, matching specified language '
 856 |                 'with exact value',
 857 |         'selector': '#pseudo-lang-div2:lang(fr)'},
 858 |     {
 859 |         'expect': ['pseudo-lang-div3'],
 860 |         'level': 2,
 861 |         'name': ':lang pseudo-class selector, matching specified language '
 862 |                 'with partial value',
 863 |         'selector': '#pseudo-lang-div3:lang(en)'},
 864 |     {
 865 |         'expect': [],
 866 |         'level': 2,
 867 |         'name': ':lang pseudo-class selector, not matching incorrect language',
 868 |         'selector': '#pseudo-lang-div4:lang(es-AR)'},
 869 |     {
 870 |         'expect': [
 871 |             'pseudo-ui-input1',
 872 |             'pseudo-ui-input2',
 873 |             'pseudo-ui-input3',
 874 |             'pseudo-ui-input4',
 875 |             'pseudo-ui-input5',
 876 |             'pseudo-ui-input6',
 877 |             'pseudo-ui-input7',
 878 |             'pseudo-ui-input8',
 879 |             'pseudo-ui-input9',
 880 |             'pseudo-ui-textarea1',
 881 |             'pseudo-ui-button1'],
 882 |         'level': 3,
 883 |         'name': ':enabled pseudo-class selector, matching all enabled form '
 884 |                 'controls',
 885 |         'selector': '#pseudo-ui :enabled'},
 886 |     {
 887 |         'expect': [],
 888 |         'level': 3,
 889 |         'name': ':enabled pseudo-class selector, not matching link elements',
 890 |         'selector': '#pseudo-link :enabled',
 891 |         'unexpected': [
 892 |             'pseudo-link-a1',
 893 |                 'pseudo-link-a2',
 894 |                 'pseudo-link-a3',
 895 |                 'pseudo-link-map1',
 896 |                 'pseudo-link-area1',
 897 |                 'pseudo-link-area2']},
 898 |     {
 899 |         'expect': [
 900 |             'pseudo-ui-input10',
 901 |             'pseudo-ui-input11',
 902 |             'pseudo-ui-input12',
 903 |             'pseudo-ui-input13',
 904 |             'pseudo-ui-input14',
 905 |             'pseudo-ui-input15',
 906 |             'pseudo-ui-input16',
 907 |             'pseudo-ui-input17',
 908 |             'pseudo-ui-input18',
 909 |             'pseudo-ui-textarea2',
 910 |             'pseudo-ui-button2'],
 911 |         'level': 3,
 912 |         'name': ':disabled pseudo-class selector, matching all disabled form '
 913 |                 'controls',
 914 |         'selector': '#pseudo-ui :disabled'},
 915 |     {
 916 |         'expect': [],
 917 |         'level': 3,
 918 |         'name': ':disabled pseudo-class selector, not matching link elements',
 919 |         'selector': '#pseudo-link :disabled',
 920 |         'unexpected': [
 921 |             'pseudo-link-a1',
 922 |                 'pseudo-link-a2',
 923 |                 'pseudo-link-a3',
 924 |                 'pseudo-link-map1',
 925 |                 'pseudo-link-area1',
 926 |                 'pseudo-link-area2']},
 927 |     {
 928 |         'expect': [
 929 |             'pseudo-ui-input4',
 930 |             'pseudo-ui-input6',
 931 |             'pseudo-ui-input13',
 932 |             'pseudo-ui-input15'],
 933 |         'level': 3,
 934 |         'name': ':checked pseudo-class selector, matching checked radio '
 935 |                 'buttons and checkboxes',
 936 |         'selector': '#pseudo-ui :checked'},
 937 |     {
 938 |         'expect': ['not-p1', 'not-p2', 'not-p3'],
 939 |         'level': 3,
 940 |         'name': ':not pseudo-class selector, matching ',
 941 |         'selector': '#not>:not(div)'},
 942 |     {
 943 |         'expect': ['not-em1', 'not-em2', 'not-em3'],
 944 |         'level': 3,
 945 |         'name': ':not pseudo-class selector, matching ',
 946 |         'selector': '#not * :not(:first-child)'},
 947 |     {
 948 |         'expect': [],
 949 |         'level': 3,
 950 |         'name': ':not pseudo-class selector, matching nothing',
 951 |         'selector': ':not(*)'},
 952 |     {
 953 |         'expect': [],
 954 |         'level': 3,
 955 |         'name': ':not pseudo-class selector, matching nothing',
 956 |         'selector': ':not(*|*)'},
 957 |     {
 958 |         'expect': ['not-p1', 'not-p2', 'not-p3'],
 959 |         'level': 3,
 960 |         'name': ':not pseudo-class selector argument surrounded by spaces, '
 961 |                 'matching ',
 962 |         'selector': '#not>:not( div )'},
 963 |     {
 964 |         'expect': [],
 965 |         'level': 2,
 966 |         'name': ':first-line pseudo-element (one-colon syntax) selector, not '
 967 |                 'matching any elements',
 968 |         'selector': '#pseudo-element:first-line'},
 969 |     {
 970 |         'expect': [],
 971 |         'level': 3,
 972 |         'name': '::first-line pseudo-element (two-colon syntax) selector, not '
 973 |                 'matching any elements',
 974 |         'selector': '#pseudo-element::first-line'},
 975 |     {
 976 |         'expect': [],
 977 |         'level': 2,
 978 |         'name': ':first-letter pseudo-element (one-colon syntax) selector, '
 979 |                 'not matching any elements',
 980 |         'selector': '#pseudo-element:first-letter'},
 981 |     {
 982 |         'expect': [],
 983 |         'level': 3,
 984 |         'name': '::first-letter pseudo-element (two-colon syntax) selector, '
 985 |                 'not matching any elements',
 986 |         'selector': '#pseudo-element::first-letter'},
 987 |     {
 988 |         'expect': [],
 989 |         'level': 2,
 990 |         'name': ':before pseudo-element (one-colon syntax) selector, not '
 991 |                 'matching any elements',
 992 |         'selector': '#pseudo-element:before'},
 993 |     {
 994 |         'expect': [],
 995 |         'level': 3,
 996 |         'name': '::before pseudo-element (two-colon syntax) selector, not '
 997 |                 'matching any elements',
 998 |         'selector': '#pseudo-element::before'},
 999 |     {
1000 |         'expect': [],
1001 |         'level': 2,
1002 |         'name': ':after pseudo-element (one-colon syntax) selector, not '
1003 |                 'matching any elements',
1004 |         'selector': '#pseudo-element:after'},
1005 |     {
1006 |         'expect': [],
1007 |         'level': 3,
1008 |         'name': '::after pseudo-element (two-colon syntax) selector, not '
1009 |                 'matching any elements',
1010 |         'selector': '#pseudo-element::after'},
1011 |     {
1012 |         'expect': ['class-p1', 'class-p2', 'class-p3'],
1013 |         'level': 1,
1014 |         'name': 'Class selector, matching element with specified class',
1015 |         'selector': '.class-p'},
1016 |     {
1017 |         'expect': [
1018 |             'class-div1',
1019 |             'class-div2',
1020 |             'class-p4',
1021 |             'class-div3',
1022 |             'class-p6',
1023 |             'class-div4'],
1024 |         'level': 1,
1025 |         'name': 'Class selector, chained, matching only elements with all '
1026 |                 'specified classes',
1027 |         'selector': '#class .apple.orange.banana'},
1028 |     {
1029 |         'expect': ['class-div1', 'class-div2', 'class-div3', 'class-div4'],
1030 |         'level': 1,
1031 |         'name': 'Class Selector, chained, with type selector',
1032 |         'selector': 'div.apple.banana.orange'},
1033 |     {
1034 |         'expect': ['class-span1'],
1035 |         'level': 1,
1036 |         'name': 'Class selector, matching element with class value using '
1037 |                 'non-ASCII characters (1)',
1038 |         'selector': '.台北Táiběi'},
1039 |     {
1040 |         'expect': ['class-span1', 'class-span2'],
1041 |         'level': 1,
1042 |         'name': 'Class selector, matching multiple elements with class value '
1043 |                 'using non-ASCII characters',
1044 |         'selector': '.台北'},
1045 |     {
1046 |         'expect': ['class-span1'],
1047 |         'level': 1,
1048 |         'name': 'Class selector, chained, matching element with multiple '
1049 |                 'class values using non-ASCII characters (1)',
1050 |         'selector': '.台北Táiběi.台北'},
1051 |     {
1052 |         'expect': ['class-span3'],
1053 |         'level': 1,
1054 |         'name': 'Class selector, matching element with class with escaped '
1055 |                 'character',
1056 |         'selector': '.foo\\:bar'},
1057 |     {
1058 |         'expect': ['class-span4'],
1059 |         'level': 1,
1060 |         'name': 'Class selector, matching element with class with escaped '
1061 |                 'character',
1062 |         'selector': '.test\\.foo\\[5\\]bar'},
1063 |     {
1064 |         'expect': ['id-div1'],
1065 |         'level': 1,
1066 |         'name': 'ID selector, matching element with specified id',
1067 |         'selector': '#id #id-div1'},
1068 |     {
1069 |         'expect': ['id-div1'],
1070 |         'level': 1,
1071 |         'name': 'ID selector, chained, matching element with specified id',
1072 |         'selector': '#id-div1, #id-div1'},
1073 |     {
1074 |         'expect': ['id-div1', 'id-div2'],
1075 |         'level': 1,
1076 |         'name': 'ID selector, chained, matching element with specified id',
1077 |         'selector': '#id-div1, #id-div2'},
1078 |     {
1079 |         'expect': ['id-div1', 'id-div2'],
1080 |         'level': 1,
1081 |         'name': 'ID Selector, chained, with type selector',
1082 |         'selector': 'div#id-div1, div#id-div2'},
1083 |     {
1084 |         'expect': [],
1085 |         'level': 1,
1086 |         'name': 'ID selector, not matching non-existent descendant',
1087 |         'selector': '#id #none'},
1088 |     {
1089 |         'expect': [],
1090 |         'level': 1,
1091 |         'name': 'ID selector, not matching non-existent ancestor',
1092 |         'selector': '#none #id-div1'},
1093 |     {
1094 |         'expect': [
1095 |             'id-li-duplicate',
1096 |             'id-li-duplicate',
1097 |             'id-li-duplicate',
1098 |             'id-li-duplicate'],
1099 |         'level': 1,
1100 |         'name': 'ID selector, matching multiple elements with duplicate id',
1101 |         'selector': '#id-li-duplicate'},
1102 |     {
1103 |         'expect': ['台北Táiběi'],
1104 |         'level': 1,
1105 |         'name': 'ID selector, matching id value using non-ASCII characters '
1106 |                 '(1)',
1107 |         'selector': '#台北Táiběi'},
1108 |     {
1109 |         'expect': ['台北'],
1110 |         'level': 1,
1111 |         'name': 'ID selector, matching id value using non-ASCII characters '
1112 |                 '(2)',
1113 |         'selector': '#台北'},
1114 |     {
1115 |         'expect': ['台北Táiběi', '台北'],
1116 |         'level': 1,
1117 |         'name': 'ID selector, matching id values using non-ASCII characters '
1118 |                 '(1)',
1119 |         'selector': '#台北Táiběi, #台北'},
1120 |     {
1121 |         'expect': ['#foo:bar'],
1122 |         'level': 1,
1123 |         'name': 'ID selector, matching element with id with escaped character',
1124 |         'selector': '#\\#foo\\:bar'},
1125 |     {
1126 |         'expect': ['test.foo[5]bar'],
1127 |         'level': 1,
1128 |         'name': 'ID selector, matching element with id with escaped character',
1129 |         'selector': '#test\\.foo\\[5\\]bar'},
1130 |     {
1131 |         'expect': [
1132 |             'any-namespace-div1',
1133 |             'any-namespace-div2',
1134 |             'any-namespace-div3',
1135 |             'any-namespace-div4'],
1136 |         'level': 3,
1137 |         'name': 'Namespace selector, matching element with any namespace',
1138 |         'selector': '#any-namespace *|div'},
1139 |     {
1140 |         'expect': ['no-namespace-div3'],
1141 |         'level': 3,
1142 |         'name': 'Namespace selector, matching div elements in no namespace '
1143 |                 'only',
1144 |         'selector': '#no-namespace |div'},
1145 |     {
1146 |         'expect': ['no-namespace-div3'],
1147 |         'level': 3,
1148 |         'name': 'Namespace selector, matching any elements in no namespace '
1149 |                 'only',
1150 |         'selector': '#no-namespace |*'},
1151 |     {
1152 |         'expect': [
1153 |             'descendant-div1',
1154 |             'descendant-div2',
1155 |             'descendant-div3',
1156 |             'descendant-div4'],
1157 |         'level': 1,
1158 |         'name': 'Descendant combinator, matching element that is a descendant '
1159 |                 'of an element with id',
1160 |         'selector': '#descendant div'},
1161 |     {
1162 |         'exclude': ['detached', 'fragment'],
1163 |         'expect': ['descendant-div1'],
1164 |         'level': 1,
1165 |         'name': 'Descendant combinator, matching element with id that is a '
1166 |                 'descendant of an element',
1167 |         'selector': 'body #descendant-div1'},
1168 |     {
1169 |         'expect': ['descendant-div1'],
1170 |         'level': 1,
1171 |         'name': 'Descendant combinator, matching element with id that is a '
1172 |                 'descendant of an element',
1173 |         'selector': 'div #descendant-div1'},
1174 |     {
1175 |         'expect': ['descendant-div2'],
1176 |         'level': 1,
1177 |         'name': 'Descendant combinator, matching element with id that is a '
1178 |                 'descendant of an element with id',
1179 |         'selector': '#descendant #descendant-div2'},
1180 |     {
1181 |         'expect': ['descendant-div2'],
1182 |         'level': 1,
1183 |         'name': 'Descendant combinator, matching element with class that is a '
1184 |                 'descendant of an element with id',
1185 |         'selector': '#descendant .descendant-div2'},
1186 |     {
1187 |         'expect': ['descendant-div3'],
1188 |         'level': 1,
1189 |         'name': 'Descendant combinator, matching element with class that is a '
1190 |                 'descendant of an element with class',
1191 |         'selector': '.descendant-div1 .descendant-div3'},
1192 |     {
1193 |         'expect': [],
1194 |         'level': 1,
1195 |         'name': 'Descendant combinator, not matching element with id that is '
1196 |                 'not a descendant of an element with id',
1197 |         'selector': '#descendant-div1 #descendant-div4'},
1198 |     {
1199 |         'expect': ['descendant-div2'],
1200 |         'level': 1,
1201 |         'name': 'Descendant combinator, whitespace characters',
1202 |         'selector': '#descendant\t\r\n#descendant-div2'},
1203 |     {
1204 |         'expect': ['child-div1', 'child-div4'],
1205 |         'level': 2,
1206 |         'name': 'Child combinator, matching element that is a child of an '
1207 |                 'element with id',
1208 |         'selector': '#child>div'},
1209 |     {
1210 |         'expect': ['child-div1'],
1211 |         'level': 2,
1212 |         'name': 'Child combinator, matching element with id that is a child '
1213 |                 'of an element',
1214 |         'selector': 'div>#child-div1'},
1215 |     {
1216 |         'expect': ['child-div1'],
1217 |         'level': 2,
1218 |         'name': 'Child combinator, matching element with id that is a child '
1219 |                 'of an element with id',
1220 |         'selector': '#child>#child-div1'},
1221 |     {
1222 |         'expect': ['child-div2'],
1223 |         'level': 2,
1224 |         'name': 'Child combinator, matching element with id that is a child '
1225 |                 'of an element with class',
1226 |         'selector': '#child-div1>.child-div2'},
1227 |     {
1228 |         'expect': ['child-div2'],
1229 |         'level': 2,
1230 |         'name': 'Child combinator, matching element with class that is a '
1231 |                 'child of an element with class',
1232 |         'selector': '.child-div1>.child-div2'},
1233 |     {
1234 |         'expect': [],
1235 |         'level': 2,
1236 |         'name': 'Child combinator, not matching element with id that is not a '
1237 |                 'child of an element with id',
1238 |         'selector': '#child>#child-div3'},
1239 |     {
1240 |         'expect': [],
1241 |         'level': 2,
1242 |         'name': 'Child combinator, not matching element with id that is not a '
1243 |                 'child of an element with class',
1244 |         'selector': '#child-div1>.child-div3'},
1245 |     {
1246 |         'expect': [],
1247 |         'level': 2,
1248 |         'name': 'Child combinator, not matching element with class that is '
1249 |                 'not a child of an element with class',
1250 |         'selector': '.child-div1>.child-div3'},
1251 |     {
1252 |         'expect': ['child-div2'],
1253 |         'level': 2,
1254 |         'name': 'Child combinator, surrounded by whitespace',
1255 |         'selector': '#child-div1\t\r\n>\t\r\n#child-div2'},
1256 |     {
1257 |         'expect': ['child-div2'],
1258 |         'level': 2,
1259 |         'name': 'Child combinator, whitespace after',
1260 |         'selector': '#child-div1>\t\r\n#child-div2'},
1261 |     {
1262 |         'expect': ['child-div2'],
1263 |         'level': 2,
1264 |         'name': 'Child combinator, whitespace before',
1265 |         'selector': '#child-div1\t\r\n>#child-div2'},
1266 |     {
1267 |         'expect': ['child-div2'],
1268 |         'level': 2,
1269 |         'name': 'Child combinator, no whitespace',
1270 |         'selector': '#child-div1>#child-div2'},
1271 |     {
1272 |         'expect': ['adjacent-div4'],
1273 |         'level': 2,
1274 |         'name': 'Adjacent sibling combinator, matching element that is an '
1275 |                 'adjacent sibling of an element with id',
1276 |         'selector': '#adjacent-div2+div'},
1277 |     {
1278 |         'expect': ['adjacent-div4'],
1279 |         'level': 2,
1280 |         'name': 'Adjacent sibling combinator, matching element with id that '
1281 |                 'is an adjacent sibling of an element',
1282 |         'selector': 'div+#adjacent-div4'},
1283 |     {
1284 |         'expect': ['adjacent-div4'],
1285 |         'level': 2,
1286 |         'name': 'Adjacent sibling combinator, matching element with id that '
1287 |                 'is an adjacent sibling of an element with id',
1288 |         'selector': '#adjacent-div2+#adjacent-div4'},
1289 |     {
1290 |         'expect': ['adjacent-div4'],
1291 |         'level': 2,
1292 |         'name': 'Adjacent sibling combinator, matching element with class '
1293 |                 'that is an adjacent sibling of an element with id',
1294 |         'selector': '#adjacent-div2+.adjacent-div4'},
1295 |     {
1296 |         'expect': ['adjacent-div4'],
1297 |         'level': 2,
1298 |         'name': 'Adjacent sibling combinator, matching element with class '
1299 |                 'that is an adjacent sibling of an element with class',
1300 |         'selector': '.adjacent-div2+.adjacent-div4'},
1301 |     {
1302 |         'expect': ['adjacent-p2'],
1303 |         'level': 2,
1304 |         'name': 'Adjacent sibling combinator, matching p element that is an '
1305 |                 'adjacent sibling of a div element',
1306 |         'selector': '#adjacent div+p'},
1307 |     {
1308 |         'expect': [],
1309 |         'level': 2,
1310 |         'name': 'Adjacent sibling combinator, not matching element with id '
1311 |                 'that is not an adjacent sibling of an element with id',
1312 |         'selector': '#adjacent-div2+#adjacent-p2, '
1313 |                     '#adjacent-div2+#adjacent-div1'},
1314 |     {
1315 |         'expect': ['adjacent-p3'],
1316 |         'level': 2,
1317 |         'name': 'Adjacent sibling combinator, surrounded by whitespace',
1318 |         'selector': '#adjacent-p2\t\r\n+\t\r\n#adjacent-p3'},
1319 |     {
1320 |         'expect': ['adjacent-p3'],
1321 |         'level': 2,
1322 |         'name': 'Adjacent sibling combinator, whitespace after',
1323 |         'selector': '#adjacent-p2+\t\r\n#adjacent-p3'},
1324 |     {
1325 |         'expect': ['adjacent-p3'],
1326 |         'level': 2,
1327 |         'name': 'Adjacent sibling combinator, whitespace before',
1328 |         'selector': '#adjacent-p2\t\r\n+#adjacent-p3'},
1329 |     {
1330 |         'expect': ['adjacent-p3'],
1331 |         'level': 2,
1332 |         'name': 'Adjacent sibling combinator, no whitespace',
1333 |         'selector': '#adjacent-p2+#adjacent-p3'},
1334 |     {
1335 |         'expect': ['sibling-div4', 'sibling-div6'],
1336 |         'level': 3,
1337 |         'name': 'General sibling combinator, matching element that is a '
1338 |                 'sibling of an element with id',
1339 |         'selector': '#sibling-div2~div'},
1340 |     {
1341 |         'expect': ['sibling-div4'],
1342 |         'level': 3,
1343 |         'name': 'General sibling combinator, matching element with id that is '
1344 |                 'a sibling of an element',
1345 |         'selector': 'div~#sibling-div4'},
1346 |     {
1347 |         'expect': ['sibling-div4'],
1348 |         'level': 3,
1349 |         'name': 'General sibling combinator, matching element with id that is '
1350 |                 'a sibling of an element with id',
1351 |         'selector': '#sibling-div2~#sibling-div4'},
1352 |     {
1353 |         'expect': ['sibling-div4', 'sibling-div6'],
1354 |         'level': 3,
1355 |         'name': 'General sibling combinator, matching element with class that '
1356 |                 'is a sibling of an element with id',
1357 |         'selector': '#sibling-div2~.sibling-div'},
1358 |     {
1359 |         'expect': ['sibling-p2', 'sibling-p3'],
1360 |         'level': 3,
1361 |         'name': 'General sibling combinator, matching p element that is a '
1362 |                 'sibling of a div element',
1363 |         'selector': '#sibling div~p'},
1364 |     {
1365 |         'expect': [],
1366 |         'level': 3,
1367 |         'name': 'General sibling combinator, not matching element with id '
1368 |                 'that is not a sibling after a p element',
1369 |         'selector': '#sibling>p~div'},
1370 |     {
1371 |         'expect': [],
1372 |         'level': 3,
1373 |         'name': 'General sibling combinator, not matching element with id '
1374 |                 'that is not a sibling after an element with id',
1375 |         'selector': '#sibling-div2~#sibling-div3, '
1376 |                     '#sibling-div2~#sibling-div1'},
1377 |     {
1378 |         'expect': ['sibling-p3'],
1379 |         'level': 3,
1380 |         'name': 'General sibling combinator, surrounded by whitespace',
1381 |         'selector': '#sibling-p2\t\r\n~\t\r\n#sibling-p3'},
1382 |     {
1383 |         'expect': ['sibling-p3'],
1384 |         'level': 3,
1385 |         'name': 'General sibling combinator, whitespace after',
1386 |         'selector': '#sibling-p2~\t\r\n#sibling-p3'},
1387 |     {
1388 |         'expect': ['sibling-p3'],
1389 |         'level': 3,
1390 |         'name': 'General sibling combinator, whitespace before',
1391 |         'selector': '#sibling-p2\t\r\n~#sibling-p3'},
1392 |     {
1393 |         'expect': ['sibling-p3'],
1394 |         'level': 3,
1395 |         'name': 'General sibling combinator, no whitespace',
1396 |         'selector': '#sibling-p2~#sibling-p3'},
1397 |     {
1398 |         'expect': ['group-em1', 'group-strong1'],
1399 |         'level': 1,
1400 |         'name': 'Syntax, group of selectors separator, surrounded by '
1401 |                 'whitespace',
1402 |         'selector': '#group em\t\r \n,\t\r \n#group strong'},
1403 |     {
1404 |         'expect': ['group-em1', 'group-strong1'],
1405 |         'level': 1,
1406 |         'name': 'Syntax, group of selectors separator, whitespace after',
1407 |         'selector': '#group em,\t\r\n#group strong'},
1408 |     {
1409 |         'expect': ['group-em1', 'group-strong1'],
1410 |         'level': 1,
1411 |         'name': 'Syntax, group of selectors separator, whitespace before',
1412 |         'selector': '#group em\t\r\n,#group strong'},
1413 |     {
1414 |         'expect': ['group-em1', 'group-strong1'],
1415 |         'level': 1,
1416 |         'name': 'Syntax, group of selectors separator, no whitespace',
1417 |         'selector': '#group em,#group strong'},
1418 |     {
1419 |         'expect': [],
1420 |         'level': 3,
1421 |         'name': 'Slotted selector',
1422 |         'selector': '::slotted(foo)'},
1423 |     {
1424 |         'expect': [],
1425 |         'level': 3,
1426 |         'name': 'Slotted selector (no matching closing paren)',
1427 |         'selector': '::slotted(foo'}]
1428 | 


--------------------------------------------------------------------------------