view env/lib/python3.9/site-packages/lxml/ @ 0:4f3585e2f14b draft default tip

"planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"
author shellac
date Mon, 22 Mar 2021 18:12:50 +0000
line wrap: on
line source

"""CSS Selectors based on XPath.

This module supports selecting XML/HTML tags based on CSS selectors.
See the `CSSSelector` class for details.

This is a thin wrapper around cssselect 0.7 or later.

from __future__ import absolute_import

from . import etree
    import cssselect as external_cssselect
except ImportError:
    raise ImportError(
        'cssselect does not seem to be installed. '

SelectorSyntaxError = external_cssselect.SelectorSyntaxError
ExpressionError = external_cssselect.ExpressionError
SelectorError = external_cssselect.SelectorError

__all__ = ['SelectorSyntaxError', 'ExpressionError', 'SelectorError',

class LxmlTranslator(external_cssselect.GenericTranslator):
    A custom CSS selector to XPath translator with lxml-specific extensions.
    def xpath_contains_function(self, xpath, function):
        # Defined there, removed in later drafts:
        if function.argument_types() not in (['STRING'], ['IDENT']):
            raise ExpressionError(
                "Expected a single string or ident for :contains(), got %r"
                % function.arguments)
        value = function.arguments[0].value
        return xpath.add_condition(
            'contains(__lxml_internal_css:lower-case(string(.)), %s)'
            % self.xpath_literal(value.lower()))

class LxmlHTMLTranslator(LxmlTranslator, external_cssselect.HTMLTranslator):
    lxml extensions + HTML support.

def _make_lower_case(context, s):
    return s.lower()

ns = etree.FunctionNamespace('')
ns.prefix = '__lxml_internal_css'
ns['lower-case'] = _make_lower_case

class CSSSelector(etree.XPath):
    """A CSS selector.


        >>> from lxml import etree, cssselect
        >>> select = cssselect.CSSSelector("a tag > child")

        >>> root = etree.XML("<a><b><c/><tag><child>TEXT</child></tag></b></a>")
        >>> [ el.tag for el in select(root) ]

    To use CSS namespaces, you need to pass a prefix-to-namespace
    mapping as ``namespaces`` keyword argument::

        >>> rdfns = ''
        >>> select_ns = cssselect.CSSSelector('root > rdf|Description',
        ...                                   namespaces={'rdf': rdfns})

        >>> rdf = etree.XML((
        ...     '<root xmlns:rdf="%s">'
        ...       '<rdf:Description>blah</rdf:Description>'
        ...     '</root>') % rdfns)
        >>> [(el.tag, el.text) for el in select_ns(rdf)]
        [('{}Description', 'blah')]

    def __init__(self, css, namespaces=None, translator='xml'):
        if translator == 'xml':
            translator = LxmlTranslator()
        elif translator == 'html':
            translator = LxmlHTMLTranslator()
        elif translator == 'xhtml':
            translator = LxmlHTMLTranslator(xhtml=True)
        path = translator.css_to_xpath(css)
        etree.XPath.__init__(self, path, namespaces=namespaces)
        self.css = css

    def __repr__(self):
        return '<%s %s for %r>' % (