Here's a ridiculous question - is there any hex editor for linux that can perform a search and replace based on a regular expression?

I have an impossible-to-get-elsewhere mp3 file that has been 'corrupted' by Macromedia's streaming Flash protocol, so that every fourth frame of the mp3 has one extra byte in the middle of it, which totally wrecks the mp3.

I know what the byte is, but I need to be very specific when I strip it out so that I don't strip out identical legitimate bytes from the wrong location:

[frame header][79/80 bytes][corrupting byte][230 bytes]
[frame header][79/80 bytes][corrupting byte][230 bytes]
[frame header][79/80 bytes][corrupting byte][230 bytes] etc.

I confidently predict that the chance of an answer to this question is nil

Well, if the corrupting byte occurs at regular, known intervals (which appears to be what you are saying), you could try dd (with skip and seek) to filter around it.

It doesn't.. I actually found the pattern was a little more complex than that - every so often a frame would have some Flash container bytes on the end, which came in one of three lengths. Depending on the length and sequence of the container bytes, as well as whether the mp3 frame had a padding byte, the Flash container byte in the middle of the frame was in a varying position.

So I ended up writing a custom Perl script that read in the mp3, parsed it a frame at a time, decoded the header and analysed the Flash junk at the end, and then spliced out the middle byte and wrote it out to a file.



If anyone comes across a Flash streamed mp3 and wants my code, let me know.