Is there an idempotent version of urllib.parse.quote?

Question

Is there a version of urllib.parse.quote that is idempotent? This function should satisfy:

urllib.parse.quote(x) == urllib.parse.quote(urllib.parse.quote(x))

for a wide enough set of x strings.

If I test the function on the comma, for example:

x = urllib.parse.quote(",")
y = urllib.parse.quote("x")

then i get x = '%2C' but y = '%252C' so it is not idempotent for the comma.

If no such function exists already, could you describe an implementation? I was thinking of using:

my_unquote = lambda x: urllib.parse.quote(urllib.parse.unquote(x)) but not sure if this is even correct.

The question arose from handling urls that had been partially encoded.

Answer 1

URL-encoding is an inherently non-idempotent operation, because the % sign is both a piece of input that needs to be encoded and a component of the output encoding (see the table here). This means that most (any?) URL-encoded strings will include characters (%) that would be re-encoded by a future encoding pass.

Put another way, it's not possible to know whether a given string has already been URL-encoded or not simply by examining the string itself. This makes writing an idempotent encoding function difficult, maybe impossible.

Depending on your use case, there are likely domain-specific workarounds you can use to simulate idempotence. For example, if you knew that the path portion of a given URL had been encoded but the scheme had not, you could run an encode on the scheme only.