[CODEC-26] base64 codec doesnt ignore all illegal chars

• Type: Bug
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: Nightly Builds
• Fix Version/s: 1.2

Environment

Operating System: All
Platform: All

• Bugzilla Id: 19860
• Assignee: Unassigned
• Reporter: Brian Ewins
• Created: 13/May/03 00:36
• Updated: 27/Oct/07 06:49
• Resolved: 27/Oct/07 06:49

Description

Reading the base64 code after the release I noticed that it isn't actually
RFC2045 compliant. Top of page 25, http://www.ietf.org/rfc/rfc2045.txt -

"Any characters outside of the base64 alphabet are to be ignored in
base64-encoded data."

Instead, the base64 decoder just throws away (some) whitespace. The lookups in
decodeBase64 will return -1 for illegal characters, which isn't checked for and
will be mistakenly incorporated into the data.

I guess the bug seems a little far-fetched, but XML 1.1 already had to expand
their definition of end-of-line to take account of text processing on
mainframes, which would trigger exactly this problem:
http://www.w3.org/TR/xml11/#sec2.11

I've been bitten by this one in the past, bizarrely enough, and was checking the
code to see if I could start using the apache codec instead of one I'd written.