label_encode.py

class LabelEncoder:
    
    def __init__(self):
        self._values2index = {}
        self._next_index = 0
        self._index2value = {}
        
    def encode(self, value) -> int:
        if value in self._values2index:
            return self._values2index[value]
        # new
        self._values2index[value] = self._next_index
        
        self._index2value[self._next_index] = value
        
        self._next_index += 1
        return self._next_index - 1
    
    def decode(self, index: int):
        if index in self._index2value:
            return self._index2value[index]
        return None
    
    def encode_batch(self, values: list) -> list:
        return [self.encode(val) for val in values]
    
    def decode_batch(self, indices: list) -> list:
        return [self.decode(i) for i in indices]
    
x = ['a', 'b', 'c', 'd', 'a']
l = LabelEncoder()
l.encode_batch(x)
l.decode_batch([0, 1, 2, 3, 0])