示例#1
0
 def start(self):
     city_map = dict()
     f = codecs.open('../miejscowosci.csv', 'r', 'utf-8')
     for line in f:
         line = unicode(line)
         line = line.replace('  ', ' ')
         basic_form = line.split(';')[0]
         ending = line.split(';')[1] #.replace('-', ' -').replace('  ', ' ').strip()
         if not basic_form.__contains__(' ') and not basic_form.__contains__('-'):
             # if self.types.__contains__(line.strip().split(';')[2]):
             if basic_form != '' and ending != '':
                 form = Util.get_form(basic_form, ending)
                 if not city_map.__contains__(basic_form) or city_map[basic_form] != form:
                     city_map[basic_form] = form
     return city_map
示例#2
0
 def get_forms(self, cities):
     city_map = dict()
     for city_tuple in cities:
         # has_hyphen = False
         city_parts = city_tuple[0].split(' ')
         if city_parts.__len__() == 1:
             city_parts = city_tuple[0].split('-')
             has_hyphen = True
         ending_parts = city_tuple[1].split(' ')
         form_parts = ''
         for i in range(city_parts.__len__()):
             print city_parts[i], ending_parts[i]
             form_parts = form_parts + Util.get_form(city_parts[i], ending_parts[i])
             if has_hyphen:
                 form_parts += '-'
             else:
                 form_parts += ' '
         city_map[city_tuple[0]] = form_parts[0:form_parts.__len__() - 1]
     return city_map