Flex character classes

[Markus Schöpflin]

Hi there,

recently I was trying to write a scanner for the HTTP protocol
with flex and I encountered some problems when specifying
character classes.

For example, RFC 2616 tells me, that TEXT is defined as "<any
OCTET except CTLs, but including LWS>.

What I tried was the following:

OCTET    [\x00-\xff]
CTL      [\x00-\x1f\x7f]
CR       \x0d
LF       \x0a
CRLF     {CR}{LF}
LWS      {CRLF}({SP}|{HT})+

And now for TEXT:

TEXT     [^{CTL}]|{LWS}

This works as OCTET includes any charater and therefore the
set difference of OCTET minus CTL can be expressed as [^{CTL}].

But later on you have a defition of "qdtext" which is "<any
TEXT except <">>".

How do I specify this with flex? Is there anything like a set
difference operator for character classes? For example:

qdtext (({OCTET}\{CTL})\["])|{LWS}

Thanks, Markus