Question

I want to write an iterator for my ‘toy’ Trie implementation.

Adding already works like this:

class Trie:
    def __init__(self):
        self.root = dict()
        pass
    def add(self, string, value):
        global nops
        current_dict = self.root
        for letter in string:
           nops += 1
           current_dict = current_dict.setdefault(letter, {})
        current_dict = current_dict.setdefault('value', value)              
        pass

The output of the adding looks like that:

trie = Trie()
trie.add("hello",1)
trie.add("world",2)
trie.add("worlds",12)
print trie.root
{'h': {'e': {'l': {'l': {'o': {'value': 1}}}}}, 'w': {'o': {'r': {'l': {'d': {'s': {'value': 12}, 'value': 2}}}}}}

I know, that I need a __iter__ and next method.

def __iter__(self):
    self.root.__iter__()
    pass

 def next(self):
    print self.root.next()

But AttributeError: 'dict' object has no attribute 'next'. How should I do it?

[Update] In the perfect world I would like the output to be one dict with all the words/entries with their corresponding values.

Answer 1

Your __iter__ special method should return an iterator; that is, an object of a class that you can call next on. A toy iterator class would be something like:

class MyIterator:
    def __init__(self):
        self.i = 10
    def next(self):
        self.i -= 1
        if self.i == 0:
            raise StopIteration
        else:
            return self.i

Unless you have a natural object to call iter on, it’s usually easier to make __iter__ a generator:

def __iter__(self):
    for i in range(10)
        yield i

Here’s a stack-based generator iterator for your Trie:

    def __iter__(self):
        stack = [('', self.root)]
        while stack:
            prefix, d = stack.pop()
            for k, v in d.items():
                if k == 'value':
                    yield prefix, v
                else:
                    stack.append((prefix + k, v))

You could also try writing it recursively, although you’d need to use itertools.chain or yield from (only since Python 3.3).