示例#1
0
    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)
示例#2
0
 def get_elements(self):
     return hrs_xml.get_elements_by_xpaths(self.tree, ["/html/body/div/p"])
示例#3
0
 def get_paragraphs(self):
     paragraphs = hrs_xml.get_elements_by_xpaths(self.tree, ["/html/body/div/p"])
     return [p.text for p in paragraphs]