shell script to...

[Jutta Degener]

Neil Greene writes:
> I have a file that contains drug names and next to the drug name is the drug
> group.
> 
> > Dipyrone		Analgesic
> > [...]
> 
> I need a shell script that will read from another (ascii) data file,
> find an occurance of a DRUG_NAME, write the line to another (ascii) file
> and append the appropriate DRUG_TYPE to the new line.

`write all occurances of <name> into a file, followed by <group>' is
something sed can do well.  A statement like

		"/name/s/$/ <group>/w file"
	or 	"/name/s/$/ <group>/p"		for stdout

works fine.

Your input file with the drug pairs is almost a `program' you could feed
to sed.  One would want to add a few slashes, but that, again, could be
accomplished using sed, for example with

	s/\([-a-zA-Z0-9]*\)[ 	]*\([-a-zA-Z0-9]*\)/-e \/\1\/s\/$\/\ \2\/p/p

which turns 
	 foo  bar
into
	-e /foo/s/$/ bar/p

if I'm not mistaken (quote until works).
Now you have three possibilities left: 

(a) forget it, get pearl
(b) figure out how to get those darn spaces across and end up using
    something like

	eval sed -n `sed -n -e "s/\([-a-zA-Z0-9]*\)[ 	]*\([-a-zA-Z0-9]*\)/
		-e \'\/\1\/s\/$\/\ \2\/p'/p" < $1`

    which actually seems to work (joined into one line) from sh,
    given the drug file as first argument
(c) get bitten by the command line length limit, use a tempfile for
    the sed program or go back to step (a)

Still waiting for the awk solution,
				Jutta
--
#include <std/disclaimer.h> Jutta Degener jutta at tub.cs.tu-berlin.de (owl:hugs.)