Ambiguity for the pyspark reduce method

def reduce(self, f):                                             #1 parameter
    def func(iterator): 
        iterator = iter(iterator) 
        try: 
            initial = next(iterator) 
        except StopIteration: 
            return 
        yield reduce(f, iterator, initial)                             #3 parameter 

    vals = self.mapPartitions(func).collect() 
    if vals: 
        return reduce(f, vals)                                         #2 parameter 
    raise ValueError("Can not reduce() empty RDD") 

The reduce function only takes 1 parameter, but it gives 2 or 3 parameters.

I search the whole python code without any other definition about reduce.