20+ examples for flattening lists in Python

Flattening lists means converting a multidimensional or nested list into a one-dimensional list. For example, the process of converting this [[1,2], [3,4]] list to [1,2,3,4] is called flattening.

The process of flattening is very easy as we’ll see. You will learn how to flatten different shapes of lists with different techniques. So, let’s jump in.

 

 

Using list comprehension

List comprehension is a way to create lists in one line of code. Let’s see how we can use list comprehension for flattening the lists.

l = [[0,1], [2,3]]
flatten_list = [item for subl in l for item in subl]
print(flatten_list)

List comprehension

flatten_list = [item for subl in l for item in subl]

Let’s break this line of code.

The first loop is “for subl in l” and the second nested loop is “for item in subl”.

Check this page for more about list comprehension.

 

Deep flattening

When we try to flatten a list of varying depth like this [ [ 0, 1 ], [ [ 2 ] ][ 3, 4 ] ] list with shallow flattening, the output will be as follows:

l = [ [0,1], [ [2 ] ], [ 3, 4 ] ]
flatten_list = [item for subl in l for item in subl]
print(flatten_list)

Flatten varying depth

But our goal is to convert [ [ 0, 1 ], [ [ 2 ] ], [ 3, 4 ] ] this list to this [ 0, 1, 2, 3, 4 ] list. You can solve this problem with deep flattening. In deep flattening, the process undergoes multiple levels of depths to create a flattened list.

There is a built-in function named deepflatten in the iteration_utilities library. You need to install this library using:

pip install iteration-utilities
from iteration_utilities import deepflatten
multi_depth_list = [[0,1], [[2]], [3,4]]
flatten_list = list(deepflatten(multi_depth_list))
print(flatten_list)

Varying depth (iteration_utilities)

We have successfully achieved our target. Let’s take another example by changing the depth of the list.

from iteration_utilities import deepflatten
multi_depth_list = [[0,1], [[2, [3, [4, [5, [6 ]]]]]], [7,8]]
flatten_list = list(deepflatten(multi_depth_list))
print(flatten_list)

Varying depth (change depth)

Using recursion

To flatten a list recursively, we will call the function inside itself to run until the end:

if len(L) == 1:
        if type(L[0]) == list:
                result = flatten(L[0])
        else:
                result = L

Check whether the list length is equal to 1. If true, then check whether the type of the first index of the list is a “list” .if true, then call the function that flattens the list else, store the number in the result.

The function will be like this:

def flatten(L):
    if len(L) == 1:
            if type(L[0]) == list:
                    result = flatten(L[0])
            else:
                    result = L
    elif type(L[0]) == list:
            result = flatten(L[0]) + flatten(L[1:])
    else:
            result = [L[0]] + flatten(L[1:])
    return result

When we run this code against this [[0,1], [2], [3,4]] list, the results will be:

Flatten list recursively

Flatten without recursion

To flatten a list without recursion, we will use a while loop until we pop all the elements from it. Take a look at the code; you will have a better understanding:

def flatten_without_rec(non_flat):
    flat = []
    while non_flat: #runs until the given list is empty.
            e = non_flat.pop()
            if type(e) == list: #checks the type of the poped item.
                    non_flat.extend(e) #if list extend the item to given list.
            else:
                    flat.append(e) #if not list then add it to the flat list.
    flat.sort()
    return flat

Flatten without recursion

 

Flatten nested lists

To flatten a nested list, you can use deep flattening. For deep flattening lists within lists, use the given below code:

from iteration_utilities import deepflatten
multi_depth_list = [[0,1],[[2,[3,[4,[5,[6]]]]]],[7,8]]
flatten_list = list(deepflatten(multi_depth_list))
print(flatten_list)

Flatten nested lists

Also, you can use the recursive function as we did above.

A list of tuples

Flattening a list of tuples of a single depth is the same as flattening lists within lists. For shallow flattening of tuples, use the following code:

list_of_tuples = [(1,2),(3,4)]
flatten_list = [item for subl in list_of_tuples for item in subl]
print(flatten_list)

The following will be the output:

Flatten a list of tuples

For deep flattening, a list of tuples with varying depth (nested), you can use the code given below:

from iteration_utilities import deepflatten
multi_depth_list = [(0,1),((2,(3,(4,(5,(6)))))),(7,8)]
flatten_list = list(deepflatten(multi_depth_list))
print(flatten_list)

Flatten a list of tuples (deep flattening)

Flatten 2d array

Let’s take a 2d array of 5×5 dimensions and convert it to a flattened list. Check the following code:

from iteration_utilities import deepflatten
rows = 5
cols = 5
array = [[i for i in range(cols)] for j in range(rows)]
print(array)
print("After applying flattening")
flatten_list = list(deepflatten(array))
print(flatten_list)

The following will be the output of the above code:

Flatten 2d array

A list of NumPy arrays

There are three built-in functions defined in NumPy library that can convert the NumPy array into flattened lists.

numpy.ravel()

numpy.flatten()

numpy.reshape(-1)

 

numpy.ravel()

import numpy as np
lst = np.array([[1,2,3], [4,5,6], [7,8,9]])
out = lst.ravel()
print(out)

The output of the code will be:

Flatten NumPy ravel

numpy.flatten()

import numpy as np
lst = np.array([[1,2,3], [4,5,6], [7,8,9]])
out = lst.flatten()
print(out)

The output of the code will be:

NumPy flatten

numpy.reshape(-1)

import numpy as np
lst = np.array([[1,2,3], [4,5,6], [7,8,9]])
out = lst.reshape(-1)
print(out)

The output of the code will be:

NumPy reshape

The difference between these three functions is speed. The flatten function returns a copy every time it flattens the array. So, if you have a large data set, don’t use the flatten function; it’s the slower one.

 

Flatten JSON objects

For flattening JSON objects, you can use the built-in function flatten() from the flatten_json library.

You first need to install it using pip:

pip install flatten_json

Then you can use this function in our code:

from flatten_json import flatten
sample_object = {'FirstName':'Ali', 'Address':{'City':'Lahore','State':'Punjab'}}
flat = flatten(sample_object)
print(flat)

The following will be the output of the code:

Flatten json object

Flatten a list of objects

You can flatten a list of objects using a built-in function available in the itertools library with function name itertools.chain.from_iterable() Let’s see how to use this function:

import itertools
class numbers:
    def __init__(self):
        pass
Object1, Object2, Object3 = [numbers() for _ in range(3)]
List_of_objects = [Object1, Object2, Object3]
Object1.myList = [1, 2, 3]
Object2.myList = [4, 5, 6]
Object3.myList = [7, 8, 9]
print(list(itertools.chain.from_iterable(x.myList for x in List_of_objects)))

The following will be the output of the above code:

Flatten a list of objects

You can achieve the same operation using list comprehension too:

class numbers:
    def __init__(self):
        pass
Object1, Object2, Object3 = [numbers() for _ in range(3)]
List_of_objects = [Object1, Object2, Object3]
Object1.myList = [1, 2, 3]
Object2.myList = [4, 5, 6]
Object3.myList = [7, 8, 9]
[i for obj in List_of_objects for i in obj.myList]

The following will be the output:

Flatten a list of objects (list comprehension)

Flatten a list of DataFrames

For flattening a list of DataFrames, the pandas library has a built-in function for flattening called df.concat() Let’s take a look at the code:

import pandas as df
dataframe1 = df.DataFrame({'colum1' : [1, 2, 3, 4], 'colum2' : [4., 3., 2., 1.]})
dataframe2 = df.DataFrame({'colum1' : [5, 6, 7, 8], 'colum2' : [9., 10., 11., 12.]})
dataframe3 = df.DataFrame({'colum1' : [15, 16, 17, 18], 'colum2' : [19., 10., 11., 12.]})
list_of_dataframes = [dataframe1, dataframe2, dataframe3]
flatten_df = df.concat(list_of_dataframes)
print(flatten_df)

The following will be the output.

Flatten a list of DataFrames

Flatten & remove duplicates

First, we will flatten our list; then we will remove the duplicates.

For flattening the list, we will use our own flatten_without_rec() function, and then we will remove the duplicates.

Let us have a look at the code:

def flatten_without_rec(non_flat):
    flat = []
    while non_flat:  # runs until the given list is empty.
        e = non_flat.pop()
        if type(e) == list:  # checks the type of popped item.
            non_flat.extend(e)  # if list extend the item in it to given list.

        else:
            flat.append(e)  # if not list then add it to the flat list.
    flat.sort()
    return flat
nested_list = [[0, 1], [[2, [3, [4, [5, [2]]]]]], [1, 2]]
flat_list = flatten_without_rec(nested_list)
set_tuple = set(flat_list)  # converting the list into set to remove duplicates
flat_list = list(set_tuple)  # converting the set back to list
print(flat_list)

The following will be the output:

Flatten remove duplicates

Flatten a dictionary into a list

You can flatten a dictionary to a list using a simple for loop:

Let us have a look at the code:

dic = {
      'alex': 1,
      'sharukh': 2,
      'flex': 3
}
flat = []
for k in dic:
    flat.append(k)
    flat.append(dic[k])
print(flat)

The output of the following code will be like this:

Flatten a dictionary into a list

You can also achieve the same using list comprehension:

dic = {
'alex': 1,
'sharukh': 2,
'flex': 3
}
[item for k in dic for item in (k, dic[k])]

The following will be the output of the code:

Flatten a dictionary into a list (list comprehension)

 

Using reduce

The reduce() function is defined in the functools library. You first need to import reduce from the functools.

Let’s take a look at the code:

from functools import reduce
multi_depth_list = [[1,2,3],[3,4,5]]
reduce(list.__add__, (list(items) for items in multi_depth_list))

The output will be:

Flatten with reduce

We flattened lists with different shapes & types in different ways. I hope you find the tutorial useful. Keep coming back.

Thank you.

Leave a Reply

Your email address will not be published. Required fields are marked *