Extend syntax to include something similar to regex character classes/sets

[jackdos]

Whilst trying to write some signatures for formats that are JSON based, I realised that the bytesequence I needed to describe should be able to deal with optional layout characters. The following JSON blocks are all functionally equivalent (with layout characters shown in brackets), but can't all be described with a single bytesequence because of the optional white spaces, tabs and new lines.

{"key1":"value1","key2":"value2"}

{(\r\n
    (\t)"key1" : "value1",(\r\n)
    (\t)"key2" : "value2"(\r\n)
}\r\n

{(\n
    "key1" : "value1",(\n)
    "key2" : "value2"(\n)
}

Ideally I should be able to specify a class of bytes and allow for wildcard matching to allow me to express the bytesequence as

BOFOffset 0: An opening {, followed by zero or more spaces, tabs, new-lines or carriage-returns, followed by "key1"....

i.e. :

7B[09 20 0D 0A]*226B65793122[09 20 0D 0A]*3A[09 20 0D 0A]*2276616C756531222C[09 20 0D 0A]*...

[marhop]

I'd really love to see better regex support in PRONOM signature patterns! Here is another use case.