Shell Programming and Scripting

Given this one long stream of data (all one line):

<TransactionDetail><TransactionHeader><ErrorLogging>YES</ErrorLogging><HistoryLogging>YES</HistoryLogging><ErrorDetection>NO</ErrorD
etection.0</LunKey><WorkingDir>/tlexmlt/xlate/i/10397</WorkingDir><JobName>I10397</JobName><ErrorCode>0</ErrorCode><ErrorMessage>EDI InboundRejects</ErrorMessage><GetNext>Y</GetNext><WorkStep>Xlati</WorkStep</ReplyItem></TransactionDetail>

How do I get this small piece of information back?

/tlexmlt/xlate/i/10397

Well, I'm sure someone could come up with a nicer-looking way, but here's one solution:

echo "reallyLongLine" | awk -F"WorkingDir" '{print $2}' | sed -e 's=>==' -e 's=</=='

This splits your line into three parts, using the word "WorkingDir" as a field separator, so you have ">/tlexmlt/xlate/i/10397</" left.

Then the sed statement removes the carat symbols and the slash from the end.

Code:

[qgatu003]/export/home/mxdooley$./test.pl file1
/tlexmlt/xlate/i/10397

[qgatu003]/export/home/mxdooley$cat test.pl
#!/usr/bin/perl -w

$file=shift;
open(FILE, "$file");
for (<FILE>) {
/\<WorkingDir\>(.*)\<\/WorkingDir\>/;
print "Working Directory: $1\n";
}

Humm, this looks like an XML-compliant file to me.

Perhaps an XSL 'script' would be the better way
to process this type of file.

- Finnbarr

Thanks for the replies everyone. I actually just used the SED example (first reply) and that worked like a charm. We do a lot of PERL scripting for my application, so reply 2 would have been fine as would have the XSL idea (we are a little lighter on XML knowledge), but I just implemented the first idea. Thanks everyone for the replies! Dave