How to Search the Files in Linux with a List of Return Data from Another Command

Question

I'm trying to search the content of all the files in directory B which might contain a string that matches a list of file names from another directory (a).

For instance, if directory a contains a list of files that their content either contains STR1 or STR2 , I can get a list the name of the files (minus the type) with:

cd /dir/a/ && grep 'STR1\|STR2' * | awk -F. '{print $1}' | uniq -u

which produces something like:

filename1
filename7
filename13
filename24

Now, I need to get the list outputted by this command (first column, eg filename1, filename7, etc.), and search the content of each file in /dir/b/* to see if any of item in the list is available in the list above. If I try this:

grep $(grep 'STR1\|STR2' /dir/a/* | awk -F. '{print $1}' | uniq -u) /dir/b/*

I get multiple grep somestr_from_command: no such file or directory , and only one of the items from the list is searched for availability in /dir/b/ files.

How can I transform the output of grep 'STR1\|STR2' * | awk -F. '{print $1}' | uniq -u grep 'STR1\|STR2' * | awk -F. '{print $1}' | uniq -u grep 'STR1\|STR2' * | awk -F. '{print $1}' | uniq -u to something like grep 'filename1|filename7|filename13|filename24' … so I can be fed to the outermost grep ?

Answer 1

You can use -f option in grep with a process substitution for another command:

grep -Ff <(awk '/str1|str2/{s=FILENAME; gsub(/^.+\/|\.[^.]+$/, "", s); print s; nextfile}' /dir/a*) /dir/b*

Also note that a single awk command can replace grep | awk | uniq grep | awk | uniq grep | awk | uniq pipeline.

Answer 2

You can do something along these lines:

for fn in *; do 
    if [ -f "$fn" ]; then
        awk ' /STR1|STR2/{ split(FILENAME,f,"\\."); print f[1]; nextfile}' "$fn"
    fi          
done