Beispiel #1
0
 def __init__(self, params):
     super(LocalFilelistLoader, self).__init__()
     check_required_params(['src'], params)
     self._source = params['src']
     self._encoding = 'utf8' if 'encoding' not in params else params[
         'encoding']
     self._fd = None
Beispiel #2
0
 def __init__(self, params):
     super(TextLoader, self).__init__()
     check_required_params(['txt'], params)
     self._text = params['txt']
     self._encoding = 'utf8' if 'encoding' not in params else params[
         'encoding']
     self._content = None
Beispiel #3
0
 def __init__(self, params):
     super(WebLoader, self).__init__()
     check_required_params(['url'], params)
     self._url = params['url']
     self._encoding = 'utf8' if 'encoding' not in params else params[
         'encoding']
     self._doc = None
Beispiel #4
0
 def __init__(self, params):
     super(NGramTokenizer, self).__init__(params)
     check_required_params(['n'], params)
     try:
         self.n = int(params['n'])
         assert(self.n > 0)
     except ValueError:
         self.n = 1
     self._imp_tokenizer = ChunkTokenizer()
Beispiel #5
0
 def __init__(self, params):
     super(LocalKVFileLoader, self).__init__()
     check_required_params(['src'], params)
     self._source = params['src']
     self._encoding = 'utf8' if 'encoding' not in params else params['encoding']
     self._fd = None
Beispiel #6
0
 def __init__(self, params):
     super(TextLoader, self).__init__()
     check_required_params(['txt'], params)
     self._text = params['txt']
     self._encoding = 'utf8' if 'encoding' not in params else params['encoding']
     self._content = None
Beispiel #7
0
 def __init__(self, params):
     super(WebLoader, self).__init__()
     check_required_params(['url'], params)
     self._url = params['url']
     self._encoding = 'utf8' if 'encoding' not in params else params['encoding']
     self._doc = None