Re: regex for base64 encoded string

Okay, I've never had an excuse to play with Oracle regular expressions (although they look pretty much like the old regexp's we've known and loved for decades), so here's a try:

'(^Content-Transfer-Encoding:
base64$)(^$)(^(([A-Za-z0-9+/=]){4}){1,19}$)*(^$)'

I know that this is not *quite* precise, but it's pretty close, I think. Here is what I *intended* to describe with this pattern:

• One "header" line, containing the literal text "Content-Transfer-Encoding: base64".
• Exactly 1 (one) blank line
• Zero or more lines of data, each consisting of:
  • 1 to 19 quartets of characters from the alphabet A-Z, a-z, +, /, or = Note that the '=' character is really a special case -- it can only appear as the last zero to two characters of the last quartet in the message, but I didn't feel like trying to describe *that* rule within the same regexp... Actually, I didn't see anything in the specification limiting output to 19 quartets per line, but that *does* seem to be the convention...
• Exactly 1 (one) blank line. (Actually, the "UUENCODE" standard seems to call for a final line of the form "====", but that is obviously not what you are looking at...

For convenience and readability (?) I've encoded each of elements described above as
a subexpression.

To use this you'll need to do:

       REGEXP_LIKE(my_column, '(^Content-Transfer-Encoding: base64$)(^$)(^(([A-Za-z0-9+/=]){4}){1,19}$)*(^$)', 'm');

The final 'm' parameter tells REGEXP_LIKE to break the source string into (treat the source string as) multiple lines.

Sorry, I haven't actually tested this. If you'd care to and let me know how far off I was, I guess I wouldn't mind knowing. ;-)

By the way, if you are interested in the specification for BASE64 encoding, you can find it in the Linux manpage for "uuencode", or (I think) in IEEE Standard 1003.1-2001, Section 12.2. (That should be the formal standard for the UUENCODE utility.) Note that the header you quoted does not comply with this standard but rather (I think) with the standard for MIME-encoding in e-mail messages. Still, I am reasonably sure that the specification for the actual message body remains about the same.

Cheers,
-- Mark.

P.S., I'm kind of curious as to why you wanted a REGEX for this. Chances are for most purposes, the expression MY_COLUMN LIKE '^Content-Transfer_Encoding: base64%' would suffice, wouldn't it? Is it really necessary to verify that the encoding itself is valid? Or were you interested in how to *decode* the string? (You can learn that from the UUDENCODE manpage, too.)

sarath kumar wrote:

>I got a column with the following data
>Content-Transfer-Encoding: base64
>
>U2VsZWN0IFRoaXMgaXMgZmlmdGg7DQogDQpOT1RFOiBUaGUgaW5mb3JtYXRpb24gY29udGFpbmVk
>IGluIHRoaXMgZW1haWwgbWVzc2FnZSBpcyBjb25zaWRlcmVkIGNvbmZpZGVudGlhbCBhbmQgcHJv
>cHJpZXRhcnkgdG8gdGhlIHNlbmRlciBhbmQgaXMgaW50ZW5kZWQgc29sZWx5IGZvciByZXZpZXcg
>YW5kIHVzZSBieSB0aGUgbmFtZWQgcmVjaXBpZW50LiAgQW55IHVuYXV0aG9yaXplZCByZXZpZXcs
>IHVzZSBvciBkaXN0cmlidXRpb24gaXMgc3RyaWN0bHkgcHJvaGliaXRlZC4gSWYgeW91IGhhdmUg
>cmVjZWl2ZWQgdGhpcyBtZXNzYWdlIGluIGVycm9yLCBwbGVhc2UgYWR2aXNlIHRoZSBzZW5kZXIg
>YnkgcmVwbHkgZW1haWwgYW5kIGRlbGV0ZSB0aGUgbWVzc2FnZS4NCg==
>
>Does any one know a regex for the base64 encoded
>string.
>
>TIA
>sarath
>
>__________________________________________________
>Do You Yahoo!?
>Tired of spam? Yahoo! Mail has the best spam protection around
>http://mail.yahoo.com
>--
>http://www.freelists.org/webpage/oracle-l
>
>
>
>
>
>

--
http://www.freelists.org/webpage/oracle-l