How to decode base64 url in python?

I have shared a code snippet for parsing signed_request parameter in a python based facebook canvas application at http://sunilarora.org/parsing-signedrequest-parameter-in-python-bas:

import base64
import hashlib
import hmac
import simplejson as json

def base64_url_decode(inp):
    padding_factor = (4 - len(inp) % 4) % 4
    inp += "="*padding_factor 
    return base64.b64decode(unicode(inp).translate(dict(zip(map(ord, u'-_'), u'+/'))))

def parse_signed_request(signed_request, secret):

    l = signed_request.split('.', 2)
    encoded_sig = l[0]
    payload = l[1]

    sig = base64_url_decode(encoded_sig)
    data = json.loads(base64_url_decode(payload))

    if data.get('algorithm').upper() != 'HMAC-SHA256':
        log.error('Unknown algorithm')
        return None
    else:
        expected_sig = hmac.new(secret, msg=payload, digestmod=hashlib.sha256).digest()

    if sig != expected_sig:
        return None
    else:
        log.debug('valid signed request received..')
return data

import base64
import simplejson as json

def parse_signed_request( signed_request ):
    encoded_sig, payload = signed_request.split('.',2)
    data = json.loads(base64.b64decode( payload.replace('-_', '+/') ))
    return data

try

s = 'iEPX-SQWIR3p67lj_0zigSWTKHg'
base64.urlsafe_b64decode(s + '=' * (4 - len(s) % 4))

as it is written here

Alternative to @dae.eklen’s solution, you can append === to it:

s = 'iEPX-SQWIR3p67lj_0zigSWTKHg'
base64.urlsafe_b64decode(s + '===')

This works because Python only complains about missing padding, but not extra padding.

Surprising, but currently accepted answer is not exactly correct.
Like some other answers stated, it’s something called base64url encoding, and it’s a part of RFC7515.

Basically, they replaced ‘+’ and ‘/’ chars by ‘-‘ and ‘_’ respectively; and additionally removed any trailing ‘=’ chars, because you can always tell how many chars you’re missing, just by looking at the encoded string length.

Here’s illustrative example from RFC7515 in C#:

 static string base64urlencode(byte [] arg)
 {
   string s = Convert.ToBase64String(arg); // Regular base64 encoder
   s = s.Split('=')[0]; // Remove any trailing '='s
   s = s.Replace('+', '-'); // 62nd char of encoding
   s = s.Replace('/', '_'); // 63rd char of encoding
   return s;
 }

 static byte [] base64urldecode(string arg)
 {
   string s = arg;
   s = s.Replace('-', '+'); // 62nd char of encoding
   s = s.Replace('_', '/'); // 63rd char of encoding
   switch (s.Length % 4) // Pad with trailing '='s
   {
     case 0: break; // No pad chars in this case
     case 2: s += "=="; break; // Two pad chars
     case 3: s += "="; break; // One pad char
     default: throw new System.Exception(
       "Illegal base64url string!");
   }
   return Convert.FromBase64String(s); // Standard base64 decoder
 }

This is the right solution. In python there is base64.b64encode but that only base64 encodes and its is different from base64 url encoding. Here is the right set to of steps to convert form base64encoded to base64urlencoded string:
1. From the resultant string, replace “/” with “_” and “+” with “-“
2. Strip the trailing “==”.

Et voila! That will make it a valid string for base64 url decoding. Btw, that link in @dae.eklen ‘s answer above is broken now.

If you are sending your base64 string from .net as a param it seems that chars that have special meaning in the URI ie + or / are replaced with " " spaces.

So before you send your string in .net you should probably do something like this

base64img.Replace("+", "-").Replace("/", "_"))

Then in python decode the string (also add ‘=’ until the length is divisible by 4)

def decode_base64(data):
    data += '=' * (len(data) % 4)
    return base64.urlsafe_b64decode(data)

Further if you want to use the image in openCV

def get_cv2_img_from_base64(base_64_string):
    data = decode_base64(base_64_string)
    np_data = np.frombuffer(data, dtype=np.uint8)
    return cv2.imdecode(np_data, cv2.IMREAD_UNCHANGED)

just

base64.urlsafe_b64decode(s)