def number_of_a_descendants(element: Element):
"""
get number of a tags in this element
:param element:
:return:
"""
if element is None:
return 0
return len(element.xpath('.//a'))
def number_of_p_children(element: Element):
"""
get number of p tags in children
:param element:
:return:
"""
if element is None:
return 0
return len(element.xpath('./p'))
def number_of_p_descendants(element: Element):
"""
get number of p tags
:param element:
:return:
"""
if element is None:
return 0
return len(element.xpath('.//p'))
def number_of_a_char(element: Element):
"""
get number of linked char, for example, result of `<a href="#">hello</a>world` = 5
:param element:
:return: length
"""
if element is None:
return 0
text = ''.join(element.xpath('.//a//text()'))
text = re.sub(r'\s*', '', text, flags=re.S)
return len(text)
def text(element: Element):
"""
get text of element
:param element:
:return:
"""
if element is None:
return 0
text = ''.join(element.xpath('.//text()'))
text = re.sub(r'\s*', '', text, flags=re.S)
return text
def _has_datetime_mata(self, element: Element):
"""
has datetime meta
:param element:
:return:
"""
for xpath in DATETIME_METAS:
datetime = element.xpath(xpath)
if datetime:
return True
return False
def number_of_punctuation(element: Element):
"""
get number of punctuation of text in this element
:param element:
:return:
"""
if element is None:
return 0
text = ''.join(element.xpath('.//text()'))
text = re.sub(r'\s*', '', text, flags=re.S)
punctuations = [c for c in text if c in PUNCTUATION]
return len(punctuations)
def a_descendants(element: Element):
"""
get
:param element:
:return:
"""
if element is None:
return []
descendants = []
for descendant in element.xpath('.//a'):
descendant.__class__ = Element
descendants.append(descendant)
return descendants
def descendants_of_body(element: Element):
"""
get descendants element of body element
:param element:
:return:
"""
if element is None:
return []
body_xpath = '//body'
elements = element.xpath(body_xpath)
if elements:
elements[0].__class__ = Element
return list(descendants(elements[0], True))
return []
def children_of_head(element: Element):
"""
get children element of body element
:param element:
:return:
"""
if element is None:
return []
body_xpath = '//head'
body_element = element.xpath(body_xpath)
if body_element:
body_element.__class__ = Element
return descendants(body_element, True)
return []
def remove_children(element: Element, xpaths):
"""
remove children from element
:param element:
:param xpaths:
:return:
"""
if element is None:
return
if not xpaths:
return
for xpath in xpaths:
nodes = element.xpath(xpath)
for node in nodes:
remove_element(node)
return element