Attached is a streamed EDI ANSI X12 output where the segment terminator/delimiter is a tilde ~ character.

Is it possible to do the following pseudo-code in a unix script (using either sed, awk and/or grep)?
Open file StreamedOutput.txt
Search for ISA and delete the data up to the tilde ~ char
Search for GS and delete the data up to the tilde ~ char
And write the ‘middle’ bit of data to the same filename.

In other words I want to strip off the first (ISA) and last segment (GS) of this file all up to the tilde and leave me the bit of data in the middle.

I tried to use the following script but not having much joy;
x=`grep -n "ISA" StreamedOutput | cut -d "~" -f1`
y=`grep -n "GS" StreamedOutput | cut -d "~" -f1`
a=`echo ${x} + 1 | bc`
b=`echo ${y} -1 | bc`
echo $x $y $a $b
sed -ne '${a},${b}p' StreamedOutput

If I unstreamed the input data it looks like;
ISA*HDR*START*US*0050000013~
BA1*Y**A*US*0050000013***TN****Federal Express~
YNQ*RZ*Y~
YNQ*QQ*Y~
DTM*274*081015~
N1*EX*Smith & Nephew, Inc.*24*510123924RT~
N2*1*9013995343~
N3*EAST HOLMES RD~
N4*Memphis*TN*38118*US~
N1**CAREY STEVE~
N3*1720 EAST BLACKHAWK DR~
N4*PHOENIX*AZ*85024*US~
N1*FW*Federal Express*24~
N3*123 Main~
N4*MEMPHIS*TN**US~
L13*A*9021.10.0050***10*20****K*0*Bone plates, screws & nails & other inte*OS*D~
L13*A*9021.31.0000***10*1000****K*0*Artificial joints & parts & accessories*OS*F~
L13*A*9021.31.0000***10*1000****K*0*Artificial joints & parts & accessories*OS*D~
GS*TRL*END*US*0050000013~

And the result I want is;
BA1*Y**A*US*0050000013***TN****Federal Express~
YNQ*RZ*Y~
YNQ*QQ*Y~
DTM*274*081015~
N1*EX*Smith & Nephew, Inc.*24*510123924RT~
N2*1*9013995343~
N3*EAST HOLMES RD~
N4*Memphis*TN*38118*US~
N1**CAREY STEVE~
N3*1720 EAST BLACKHAWK DR~
N4*PHOENIX*AZ*85024*US~
N1*FW*Federal Express*24~
N3*123 Main~
N4*MEMPHIS*TN**US~
L13*A*9021.10.0050***10*20****K*0*Bone plates, screws & nails & other inte*OS*D~
L13*A*9021.31.0000***10*1000****K*0*Artificial joints & parts & accessories*OS*F~
L13*A*9021.31.0000***10*1000****K*0*Artificial joints & parts & accessories*OS*D~

Try it first without the colored fragment of the code

awk '/^ISA/||/^GS/{next}{print}' file > newfile && mv newfile > file

Would two grep's to exclude work?

grep -v "^ISA" file75 | grep -v "^GS"

Or one grep

grep -v '^[ISA|GS]' file

Thanks for the replies, however how will your code change if the data was streamed (i.e. one long line of data in this case) so the delimiter separating the segments is the tilde ~ char.

Try awk:

awk -F'~' '{for(i=1;i<NF;i++){if($i !~ /^[ISA|GS]/)printf $i FS}}'

Hi Danmero, would it be too much to ask for a breakdown of your code and it's intentions/meaning as I tried to code using a combination of grep and sed? Thanks!