Gmail attachments get: unpadded base64url data breaks standard decoders

[njt]

Description

gws gmail users messages attachments get returns the data field from the Gmail API as-is — URL-safe base64 without padding. Google's Gmail API intentionally omits = padding from attachment data (this is per spec for base64url). However, many standard base64 decoders (Python's base64.urlsafe_b64decode, etc.) require proper padding and fail on strings whose length mod 4 is not 0.

Steps to Reproduce

gws gmail users messages attachments get --params '{"userId": "me", "messageId": "MESSAGE_ID", "id": "ATTACHMENT_ID"}'

The returned JSON data field has a length that is not a multiple of 4 (e.g., 50038 characters, which is 4n+2).

Attempting to decode in Python:

import base64, json
result = json.loads(output)
decoded = base64.urlsafe_b64decode(result["data"])
# Raises: binascii.Error: Invalid base64-encoded string number of data characters
#         cannot be 1 more than a multiple of 4

Expected Behavior

The data field should be directly decodable by standard base64 libraries. Either:

1. Re-pad the base64url string before emitting it in the JSON response (data += "=" * (-len(data) % 4)), or
2. Document that the data field uses unpadded base64url encoding (RFC 4648 §5) and consumers must re-pad

Workarounds

• Use the -o flag to have gws decode and write the binary directly: gws gmail users messages attachments get --params '...' -o /tmp/file.pdf
• Manually re-pad before decoding:

  data = response["data"]
  data += "=" * (-len(data) % 4)
  decoded = base64.urlsafe_b64decode(data)

Environment

• gws version: 0.4.1
• OS: Ubuntu 24.04 (Linux)
• Consumer: Python 3.12 base64.urlsafe_b64decode