>Walter, it's a long time since I looked but we used to read giant HL7 messages into a cursor delimited on the | character. We had fields f1,f2...fn for the maximum number of segments, then we had a series of #DEFINEs for each of the message segment types to read the relevant values out of the appropriate column and convert it to a date or numeral or whatever. It was very quick and easy to maintain as I recall. How do you do it?


Well since I'm not really maintaining the HL7 module myself anymore I have to see what has been enhanced, but when I was in charge of it, it was listening on a TCP/IP port for HL7 messages (We at that time did not handle TXT files). It also has a component of sending Hl7 messages, but that is mainly used to Query for PID info.

When a message got in, we logged the message and Identified the type of message. We look up the event type into a user defined metadata table which contains a VFP script what to do with the message. We provided some functions like getfieldval() which returns the contents of a specific field within the HL7 message (Eg: GetFieldVal("MSH.9.1") with support for subfields (divided by ^ and iterations ~ ), so in the scripts we could easily reference any field in the message.

Since every HL7 implementation seems to be a bit different in our case, we just have one base and modify the scripts for the event type to the clients needs. We don't need to alter the software when we want to support new events or segments, we just add another record in the metadata and/or modify the VFP script the memofield.

Walter,