def get_section_header_elements(self, extra_xpaths=[]):
section_header_xpaths = [
'/html/body//b',
'/html/body/div/p//b',
'/html/body//strong',
'/html/body/div/p/span/span[@style=\'font-family:"Courier New"\']'
]
xpaths = section_header_xpaths + extra_xpaths
return hrs_xml.get_elements_by_xpaths(self.tree, xpaths)
def get_elements(self):
return hrs_xml.get_elements_by_xpaths(self.tree, ["/html/body/div/p"])
def get_paragraphs(self):
paragraphs = hrs_xml.get_elements_by_xpaths(self.tree, ["/html/body/div/p"])
return [p.text for p in paragraphs]