IceCat.py 文件源码

def _parse(self, xml_file):
        if xml_file.endswith('.gz'):
            with gzip.open(xml_file, 'rb') as f:
                data = ET.parse(f).getroot()
        else:
             data = ET.parse(xml_file).getroot()

        '''
        Data is an XML ElementTree Object
        '''
        self.id_map = {}
        self.catid = ''
        self.catname = ''
        self.findpath = 'Name[@langid="' + langid + '"]'
        for elem in data.iter('Category'):
            self.catid = elem.attrib['ID']
            for name in elem.iterfind(self.findpath):
                self.catname = name.attrib['Value']
                # only need one match
                break
            if not self.catname:
                self.catname = "Unknown"
            self.id_map[self.catid] = self.catname

        self.log.info("Parsed {} Categories from IceCat CategoriesList".format(str(len(self.id_map.keys()))))