Converting Nested For Loops To Flattened List Comprehension When Outer Variable Is Used In Inner Scope

I have a pair of nested loops where the operation inside the inner loop depend on both the loop elements. def ngram(inp='', mn=2, mx=60): ''' EG inp='the' => ['th', 'the

Solution 1:

You can do this:

def ngram(inp='', mn=2, mx=60):
    return [inp[i:j] 
            for i in range(0, len(inp) + 1 - mn)
            for j in range(i + mn, min(i + mx + 1, len(inp) + 1))]

In general, whenever you have a for loop of this form:

result = []
for sub_1 in collection:
    for sub_2 in sub_1:
        …
            for sub_n in sub_(n - 1):
                result.append(element)

You can construct an equivalent list comprehension of the following form:

[element for sub_1 in collection for sub_2 in sub_1 … for sub_n in sub_(n - 1)]

Solution 2:

What's wrong with:

def ngram(inp='', mn=2, mx=60): 
    return [
        inp[i:j]
        for i in range(0, len(inp) + 1 - mn)
        for j in range(i+mn, min(i + mx + 1, len(inp) + 1))
    ]