Repository navigation
Add the ignorechars parameter in the Base64 decoder #144001
Description
Feature or enhancement
Earlier standards for Base64 (RFC 1421, RFC 1521) on which the base64 and binascii modules are based, specified that any non-alphabet characters should be ignored, and this is the default behavior of these modules. More modern standard (RFC 4648) considers this a vulnerability, and specify that they should be rejected until the upper lever document states otherwise. Thus, strict_mode/validate parameters were added. But they are "all-or-nothing" -- either all non-alphabet characters are rejected, or other errors are ignored. In some cases we need to ignore only the part of errors -- for example, ignore newlines and whitespaces, but not other invalid characters, or ignore padding errors.
a85decode() has parameter ignorechars which could also be used to control Base64 decoding.
Linked PRs
- gh-144001: Add ignorechars parameter to Base64 decoder #144009
- gh-144001: Support ignorechars in binascii.a2b_base64() and base64.b64decode() #144024
- gh-144001: Support ignoring the invalid pad character in Base64 decoding #144306
- gh-144001: Simplify Base64 decoding with altchars and ignorechars specified #144324
Activity
pythongh-144001: Add ignorechars parameter to Base64 decoder
moneebullah25 commentedon Jan 18, 2026
Hi @serhiy-storchaka
I've implemented this feature in PR #144009.
This PR adds ignorechars parameter to binascii.a2b_base64() and base64.b64decode() with O(1) lookup table, tests, and NEWS entry.
If this aligns with you have in your mind, feel free to merge or let me know what to cherry-pick from it. Otherwise If you'd prefer a different approach or it needs significant changes, please let me know or feel free to create your own implementation, if not I am happy to close mine in that case.
pythongh-144001: Register ignorechars identifier in global strings
pythongh-144001: removed news entry and blurbed it
pythongh-144001: Support ignorechars in binascii.a2b_base64() and bas…
serhiy-storchaka commentedon Jan 19, 2026
Sorry, @moneebullah25, but I already had my code before opening this issue. I only needed to solve a pair of corner cases.
pythongh-144001: Support ignorechars in binascii.a2b_base64() and bas…
pythongh-144001: Support ignorechars in binascii.a2b_base64() and bas…
gh-144001: Support ignorechars in binascii.a2b_base64() and base64.b6…
gpshead commentedon Jan 26, 2026
a followup PR is likely needed per #144024 (comment)
pythongh-144001: Support ignoring the invalid pad character in Base64…
pythongh-144001: Support ignoring the invalid pad character in Base64…
serhiy-storchaka commentedon Jan 28, 2026
PR #144306 extend the function of
ignorecharsand allows to support the following optional features (RFC 4648, section 3.3):Now, if
ignorecharscontains the pad character'=', the pad characters presented before the end of the encoded data and the excess pad characters will be ignored.pythongh-144001: Simplify Base64 decoding with altchars and ignorecha…
serhiy-storchaka commentedon Jan 29, 2026
And since ignorechars is a new parameter, no need to restore the deprecated behavior and emit corresponding warnings when both
altcharsandignorecharsare specified. #144324 makes it matching the future more secure behavior. It also makes the code simpler.gh-144001: Simplify Base64 decoding with altchars and ignorechars spe…
pythongh-144001: Simplify Base64 decoding with altchars and ignorecha…
gh-144001: Support ignoring the invalid pad character in Base64 decod…
pythongh-144001: Support ignorechars in binascii.a2b_base64() and bas…
pythongh-144001: Simplify Base64 decoding with altchars and ignorecha…
pythongh-144001: Support ignoring the invalid pad character in Base64…
pythongh-144001: Support ignoring the invalid pad character in Base64…