def xpath(self, selector: str, *, clean: bool = False, first: bool = False, _encoding: str = None) -> _XPath:
"""Given an XPath selector, returns a list of
:class:`Element <Element>` objects or a single one.
:param selector: XPath Selector to use.
:param clean: Whether or not to sanitize the found HTML of ``<script>`` and ``<style>`` tags.
:param first: Whether or not to return just the first result.
:param _encoding: The encoding format.
If a sub-selector is specified (e.g. ``//a/@href``), a simple
list of results is returned.
See W3School's `XPath Examples
<https://www.w3schools.com/xml/xpath_examples.asp>`_
for more details.
If ``first`` is ``True``, only returns the first
:class:`Element <Element>` found.
"""
selected = self.lxml.xpath(selector)
elements = [
Element(element=selection, url=self.url, default_encoding=_encoding or self.encoding)
if not isinstance(selection, etree._ElementUnicodeResult) else str(selection)
for selection in selected
]
# Sanitize the found HTML.
if clean:
elements_copy = elements.copy()
elements = []
for element in elements_copy:
element.raw_html = lxml_html_tostring(cleaner.clean_html(element.lxml))
elements.append(element)
return _get_first_or_list(elements, first)
def html_tostring(tree):
"""
html to string
"""
return lxml_html_tostring(
tree,
pretty_print=True,
# method="html",
encoding='utf-8',
).strip()
def find(self, selector: str = "*", *, containing: _Containing = None, clean: bool = False, first: bool = False, _encoding: str = None) -> _Find:
"""Given a CSS Selector, returns a list of
:class:`Element <Element>` objects or a single one.
:param selector: CSS Selector to use.
:param clean: Whether or not to sanitize the found HTML of ``<script>`` and ``<style>`` tags.
:param containing: If specified, only return elements that contain the provided text.
:param first: Whether or not to return just the first result.
:param _encoding: The encoding format.
Example CSS Selectors:
- ``a``
- ``a.someClass``
- ``a#someID``
- ``a[target=_blank]``
See W3School's `CSS Selectors Reference
<https://www.w3schools.com/cssref/css_selectors.asp>`_
for more details.
If ``first`` is ``True``, only returns the first
:class:`Element <Element>` found.
"""
# Convert a single containing into a list.
if isinstance(containing, str):
containing = [containing]
encoding = _encoding or self.encoding
elements = [
Element(element=found, url=self.url, default_encoding=encoding)
for found in self.pq(selector)
]
if containing:
elements_copy = elements.copy()
elements = []
for element in elements_copy:
if any([c.lower() in element.full_text.lower() for c in containing]):
elements.append(element)
elements.reverse()
# Sanitize the found HTML.
if clean:
elements_copy = elements.copy()
elements = []
for element in elements_copy:
element.raw_html = lxml_html_tostring(cleaner.clean_html(element.lxml))
elements.append(element)
return _get_first_or_list(elements, first)