Extract exception message ignoring numbers/GUIDs

[hbazan]

Hi,
I need to make a ranking of most common exception messages, from different services. I've been able to extract the exception messages using rex, but several values include numbers or GUIDs.
Examples:
- the CronopioId=123455 is invalid
- couldn't find a Fama associated to CronopioId=123455 and EsperanzaId=658d3cd9-4259-4824-878c-27d33b6af743 with status=Valid

What I need is to extract the message without numbers or guids, but the rest of the message.
I'm using this for the GUIDs, but it only work with one GUID in the whole sentence:

rex field=PreGUID "(?.*)({{0,1}([0-9]){8}-([0-9]){6}-([0-9]){3}}{0,1})" | eval ExceptionMessage=if(NOT isnull(PostGUID),PreGUID+" ? "+PostGUID,if(NOT isnull(PreGUID), PreGUID, ExceptionMessage))

Is there a simpler way?

[cphair · Accepted solution]

Use sed in global replace mode (note the g at the end of the line) to do multiple substitutions. Something like the following should generalize your GUIDS--you'll have to edit for whatever the proper matching regex should be. If there's a specific pattern to the other numbers/process IDs you want to eliminate, you can run a second rex to get rid of those, but be careful you don't match more than you intend.

rex mode=sed "s/[0-9a-f]{8}-[0-9a-f]{6}-[0-9a-f]{3}/xxxxxxxx-xxxxxx-xxx/g"

[cphair]

Thanks for the regex correction. Glad it worked for you.

[hbazan]

Great!. In case someone else is looking for this, the expression does not match a GUID. This one does:
rex mode=sed "s/({{0,1}([0-9a-fA-F]){8}-([0-9a-fA-F]){4}-([0-9a-fA-F]){4}-([0-9a-fA-F]){4}-([0-9a-fA-F]){12}}{0,1})/xxxxxxxx-xxxxxx-xxx/g"