用sed替换另一个词-问题

Question

On a redhat server,

I have multiple files named PO******* Inside them i have a line like this :

NODE PO******* TCPblahblah blah

On the other hand i have a csv file containing multiple line like :

PO*******,XXXXXXX

i want to replace in files PO*******
the NODE PO******* TCPblahblah blah
by NODE XXXXXX TCPblahblah blah

See my code for what i did to have this output on line with NODE : TCPblahblXXX

for i in $(ls def)
do 
new= $(grep $i list.csv | cut -d ',' -f2)
sed "/NODE/ s/$i/$new/g" def/$i
done

Answer 1

When each def/* file only has one /NODE/ line, you can try a nested sed .
Remember that sed can use other characters like , or # when you don't want the / .

# First look at the generated commands by
sed 's#.*#/NODE/ s,&,#' list.csv
# next try
sed -f <(sed 's#.*#/NODE/ s,&,#' list.csv) def/*
# Make backup before changing
cp -r def def_bak
# Get serious
sed -if <(sed 's#.*#/NODE/ s,&,#' list.csv) def/*

When you are not sure about this (overlapping files like PO1 and PO12), you can loop through the files, starting with your csv.

while IFS= read -f line; do
   test -f def/"${line%,*}" || continue
   sed -i "s,${line}," def/"${line%,*}"
done < list.csv

Answer 2

you have too much blah blah and lacking testable input files, but I think this will do

$ awk 'NR==FNR{a[$1]=$2; next} {f=FILENAME; sub(f,a[f]); print > f".mod"}' list.csv def/*

this will create ".mod" extensions for the processed files.