I have .csv file that contain 2 columns delimited with ,.
word1,word2
word3,word4
word5,word6
.
.
.
.
word1000,1001
I want to create a new file from file.csv and insert sed command like this:
sed -e 's,word1,word2,gI' \
-e 's,word3,word4,gI' \
-e 's,word5,word6,gI' \
....
How can I make a script to add sed command?
You can use sed to process each line:
echo -n 'sed ' ; sed -e "s/^\(.*\)/-e 's,\1,gl'\ \\\/" file.csv
will produce as requested
sed -e 's,word1,word2,gl' \
-e 's,word3,word4,gl' \
-e 's,word5,word6,gl' \
Your goal seams to be performing custom replacements from a file. In this case, I would not generate a file containing a bash script to do the job, but I would generate a sed script to do the job:
sed -e 's/^/s,/' -e 's/$/,gI/' file.csv > sed_script
sed -f sed_script <<< "word1"
We can even avoid to use the sed_script file with bash's process substitution:
sed -f <(sed -e 's/^/s,/' -e 's/$/,gI/' file.csv) <<< "word1"
Update:
Simplifying the sed script generation, it becomes:
sed -e 's/.*/s,&,gI/' file.csv > sed_script
sed -f sed_script <<< "word1"
and
sed -f <(sed -e 's/.*/s,&,gI/' file.csv) <<< "word1"
sed ... <<< "word1" is a shortcut for echo "word1" | sed ..., and file.csv is used to generate a sed script. So nothing go through file.csv.ab by zz, a by e and b by r has a hard to predict result because your replacement overlap.