Use https link
wRAR
| 1 |
try: |
|
| 2 |
from functools import lru_cache |
|
| 3 |
except ImportError: |
|
| 4 |
from functools32 import lru_cache |
|
| 5 |
|
|
| 6 |
from cssselect import GenericTranslator as OriginalGenericTranslator |
|
| 7 |
from cssselect import HTMLTranslator as OriginalHTMLTranslator |
|
| 8 |
from cssselect.xpath import XPathExpr as OriginalXPathExpr |
|
| 9 |
from cssselect.xpath import _unicode_safe_getattr, ExpressionError |
|
| 10 |
from cssselect.parser import FunctionalPseudoElement |
|
| 11 |
|
|
| 12 |
|
|
| 13 |
class XPathExpr(OriginalXPathExpr): |
|
| 14 |
|
|
| 15 |
textnode = False |
|
| 16 |
attribute = None |
|
| 17 |
|
|
| 18 |
@classmethod
|
|
| 19 |
def from_xpath(cls, xpath, textnode=False, attribute=None): |
|
| 20 |
x = cls(path=xpath.path, element=xpath.element, condition=xpath.condition) |
|
| 21 |
x.textnode = textnode |
|
| 22 |
x.attribute = attribute |
|
| 23 |
return x |
|
| 24 |
|
|
| 25 |
def __str__(self): |
|
| 26 |
path = super(XPathExpr, self).__str__() |
|
| 27 |
if self.textnode: |
|
| 28 |
if path == '*': |
|
| 29 |
path = 'text()' |
|
| 30 |
elif path.endswith('::*/*'): |
|
| 31 |
path = path[:-3] + 'text()' |
|
| 32 |
else: |
|
| 33 |
path += '/text()' |
|
| 34 |
|
|
| 35 |
if self.attribute is not None: |
|
| 36 |
if path.endswith('::*/*'): |
|
| 37 |
path = path[:-2] |
|
| 38 |
path += '/@%s' % self.attribute |
|
| 39 |
|
|
| 40 |
return path |
|
| 41 |
|
|
| 42 |
def join(self, combiner, other): |
|
| 43 |
super(XPathExpr, self).join(combiner, other) |
|
| 44 |
self.textnode = other.textnode |
|
| 45 |
self.attribute = other.attribute |
|
| 46 |
return self |
|
| 47 |
|
|
| 48 |
|
|
| 49 |
class TranslatorMixin(object): |
|
| 50 |
"""This mixin adds support to CSS pseudo elements via dynamic dispatch.
|
|
| 51 |
|
|
| 52 |
Currently supported pseudo-elements are ``::text`` and ``::attr(ATTR_NAME)``.
|
|
| 53 |
"""
|
|
| 54 |
|
|
| 55 |
def xpath_element(self, selector): |
|
| 56 |
xpath = super(TranslatorMixin, self).xpath_element(selector) |
|
| 57 |
return XPathExpr.from_xpath(xpath) |
|
| 58 |
|
|
| 59 |
def xpath_pseudo_element(self, xpath, pseudo_element): |
|
| 60 |
"""
|
|
| 61 |
Dispatch method that transforms XPath to support pseudo-element
|
|
| 62 |
"""
|
|
| 63 |
if isinstance(pseudo_element, FunctionalPseudoElement): |
|
| 64 |
method = 'xpath_%s_functional_pseudo_element' % ( |
|
| 65 |
pseudo_element.name.replace('-', '_')) |
|
| 66 |
method = _unicode_safe_getattr(self, method, None) |
|
| 67 |
if not method: |
|
| 68 |
raise ExpressionError( |
|
| 69 |
"The functional pseudo-element ::%s() is unknown" |
|
| 70 |
% pseudo_element.name) |
|
| 71 |
xpath = method(xpath, pseudo_element) |
|
| 72 |
else: |
|
| 73 |
method = 'xpath_%s_simple_pseudo_element' % ( |
|
| 74 |
pseudo_element.replace('-', '_')) |
|
| 75 |
method = _unicode_safe_getattr(self, method, None) |
|
| 76 |
if not method: |
|
| 77 |
raise ExpressionError( |
|
| 78 |
"The pseudo-element ::%s is unknown" |
|
| 79 |
% pseudo_element) |
|
| 80 |
xpath = method(xpath) |
|
| 81 |
return xpath |
|
| 82 |
|
|
| 83 |
def xpath_attr_functional_pseudo_element(self, xpath, function): |
|
| 84 |
"""Support selecting attribute values using ::attr() pseudo-element
|
|
| 85 |
"""
|
|
| 86 |
if function.argument_types() not in (['STRING'], ['IDENT']): |
|
| 87 |
raise ExpressionError( |
|
| 88 |
"Expected a single string or ident for ::attr(), got %r" |
|
| 89 |
% function.arguments) |
|
| 90 |
return XPathExpr.from_xpath(xpath, |
|
| 91 |
attribute=function.arguments[0].value) |
|
| 92 |
|
|
| 93 |
def xpath_text_simple_pseudo_element(self, xpath): |
|
| 94 |
"""Support selecting text nodes using ::text pseudo-element"""
|
|
| 95 |
return XPathExpr.from_xpath(xpath, textnode=True) |
|
| 96 |
|
|
| 97 |
|
|
| 98 |
class GenericTranslator(TranslatorMixin, OriginalGenericTranslator): |
|
| 99 |
@lru_cache(maxsize=256) |
|
| 100 |
def css_to_xpath(self, css, prefix='descendant-or-self::'): |
|
| 101 |
return super(GenericTranslator, self).css_to_xpath(css, prefix) |
|
| 102 |
|
|
| 103 |
|
|
| 104 |
class HTMLTranslator(TranslatorMixin, OriginalHTMLTranslator): |
|
| 105 |
@lru_cache(maxsize=256) |
|
| 106 |
def css_to_xpath(self, css, prefix='descendant-or-self::'): |
|
| 107 |
return super(HTMLTranslator, self).css_to_xpath(css, prefix) |
|
| 108 |
|
|
| 109 |
|
|
| 110 |
_translator = HTMLTranslator() |
|
| 111 |
|
|
| 112 |
|
|
| 113 |
def css2xpath(query): |
|
| 114 |
"Return translated XPath version of a given CSS query"
|
|
| 115 |
return _translator.css_to_xpath(query) |
Read our documentation on viewing source code .