[Nmh-workers] Selecting messages with a range of Spam scores

[norm]

I would like to write a script that will 'pick" all the messages in a given
range of a given folder having a component like:

X-Spam-Status: No, score=3.8 required=5.0 tests=HTML_MESSAGE,MIME_HTML_ONLY,
        RAZOR2_CHECK,SPF_FAIL,SPF_HELO_PASS,T_REMOTE_IMAGE autolearn=disabled
        version=3.3.1

where the score is between two given floating point numbers.

So a typical invocation might be

        spam last:45 +gone 3.0 6.2

Does anybody have something like this, or some hints about to write such a
script? If nobody has a starter script, I will probably use Java, so, once I
have the component's content, parsing it to get the score, would be trivial.
But the best way to get the message numbers having such a component, WITH the
corresponding contents of the component is not obvious to me.

    Norman Shapiro