[Asterisk-Users] Pattern matching: least-to-most specific PITA

[John Todd]

My synapses are rather fried after a long few days of debugging other 
problems, so perhaps I'm being lazy in sending this to the general 
list, but I can't think straight about it.  Forgive me if there is an 
overly obvious solution to this.

I have a list of phone numbers that are SIP extensions.  I'd like to 
dial them via SIP if ${EXTEN} is equal to one of those numbers.  If 
${EXTEN} is not equal to one of those numbers, I'd like to send the 
call out to a PRI group, regardless of dialed sequence length or 
pattern.

It seems I cannot do this with *'s pattern matching, due to the order 
in which extensions are parsed, which seems to be least-specific to 
most-specific.  This causes all kinds of headaches when trying to use 
wildcards, since wildcards are super-least-specific.

My desire would be to have the more specific matches done first, so 
that if ${EXTEN} would be matched in an order that makes sense.  I 
understand why matching goes from least-to-most specific for analog 
equipment, but it makes certain tasks impossible from a dialplan 
point of view when I have the full number and I'm not waiting on a 
user to finish typing the digits.

If presented with 12123669751 I would expect the match to happen and 
the SIP extension to be dialed.  It doesn't.  It dials the Zap 
extension.

[foo]
;
exten => _1212366975X,1,Dial(SIP/${EXTEN})
exten => _181772721[8-9]X,1,Dial(SIP/${EXTEN})
exten => _191481287[4-7]X,1,Dial(SIP/${EXTEN})
exten => _141550926[0-2]X,1,Dial(SIP/${EXTEN})
;
exten => _.,1,Dial(Zap/g1/${EXTEN})
;


How do I invert this match examination to make it go most- to 
least-specific execution?

JT