Hi Alex, Why not simply support reular expressions. I thought I'd read something somewhere about someone looking at building this in (ie: if supported compile in regexp support). If support is not available downgrade to simple keyword as we have now.
To me though if you want to have the redirection all happen at the kannel level (rather than application layer above it) we really need support for config reload. That is, I can see you wanting to add/remove filters more often than you'd want to restart kannel. Here the ability to reload the config would be nice. Just my NZD0.02 - Cheers, Alan On Wed, 2003-07-30 at 07:39, Alex Judd wrote: > I'm now getting to the point where the keyword handling within Kannel > is not as intelligent as we need it to be and I'm looking to build in > (or around) some better functionality. Thought it would be good to air > my thoughts before going ahead. > > 1. Current behaviour > > The first key word (ie. the first word before a space) is read from > the inbound SMS message and is compared against either a keyword, or > an array of aliases for the keyword. Aliases are separated with a ; > and keywords are finished with a space > > 2. New behaviour > > The ability to specify second and x subsequent keywords that are > matched on a pattern basis with the inbound text message and the > ability to specify a wildcard match for each keyword that is matched. > For example > > Single Keyword Matching > > EXACT - matches all first words regardless of case with the word > 'EXACT' > EXA* - matches all first words regardless of case with words > starting with 'EXA' > > Multiple Keyword Matching > > SECOND WORD - matches both words regardless of case with the words > 'SECOND' and 'WORD' > SEC* WORD - matches the first word regardless of case with > words starting with 'SEC' and the second word exactly with 'WORD' > SEC* WOR* - matches the first word regardless of case with > words starting with 'SEC' and second words starting with 'WOR' > > Aliases are handled in the same way. > > Basically this involves some fairly heavy rewrites to Kannel's > behaviour but my thoughts are it would be better to keep this in > Kannel, than to move the second + x word matching to separate code. > > Thoughts from everyone? > > Regards > > Alex -- Alan McNatty <[EMAIL PROTECTED]>
