shell script to file

[joshboski]

I have a shell script that looks through a folder of files, by searching from certain keywords that I have stored in a txtfile, using grep. Is there a way to list if I find more than one, have the output of both on the same line?

currentdir=$PWD
echo $currentdir
ls
echo "Pick a file"
read INPUT
cd $INPUT
pwd                                                                                 
for filename in *
do
   for searchterm in $(cat /home/searchterms.txt)                                                                                 
     do
       grep -q $searchterm $filename && echo -e "$filename : $searchterm \n" >>/home/burkhojt/results.txt
                                                                                
     done
  echo
done

I want the output to look like this
1089-006.doc : packet, radio

instead of

1089-006.doc : packet

1089-006.doc : radio

[vsemaska]

I'm not near my Linux box right now so I can't test it but this should work.

for filename in *
do
   tmp=""
   for searchterm in $(cat /home/searchterms.txt)
     do
       grep -q $searchterm $filename && tmp="${tmp}, $searchterm"
     done
     if [ "$tmp" != "" ]
        then echo -e "$filename : $tmp \n" >>/home/burkhojt/results.txt
        fi
  echo
done

The output will end up looking like this:
1089-006.doc : , packet, radio

Use something like awk's substr function to remove that unwanted 1st comma.

Vic

[Cabhan]

I prefer this:

for filename in *; do
    terms=
    first=1
    exec 3< /home/searchterms.txt
    while read searchterm <&3; do
        if grep -q "$searchterm" "$filename"; then
            if [ ! $first ]; then
                terms="${terms},"
            fi

            first=0
            terms="${terms} $searchterm"
        fi
    done
    exec 3<&-
    if [ ! -z "$terms" ]; then
        echo "$filename: $terms"
    fi
done

It's a bit more verbose, but I feel like it has the following advantages:

- Avoids needless use of cat by opening its own file descriptors (this is what exec does)
- Avoids the extra comma at the beginning (the 'first' flag controls this)
- Also properly quotes all parameters

[joshboski]

thank you both much!

[scm]

And you can avoid the use of cat in the original by

for searchterm in $(< /home/searchterms.txt)

[Cabhan]

The problem with that approach is that it will use $IFS to separate values, and thus will split on spaces unless you redefine $IFS. For instance:

alex@danu ~ $ cat testfile
Hello
Hi
Hooha
Foo
Bar
Hello There
alex@danu ~ $ cat asdf
#!/bin/bash

exec 3< testfile
while read t <&3; do
    echo "$t"
done
alex@danu ~ $ ./asdf 
Hello
Hi
Hooha
Foo
Bar
Hello There
alex@danu ~ $ vi asdf 
alex@danu ~ $ cat asdf 
#!/bin/bash

for t in $(< testfile); do
    echo "$t"
done
alex@danu ~ $ ./asdf 
Hello
Hi
Hooha
Foo
Bar
Hello
There

In the first script run (using the exec approach), the script correctly identifies "Hello There" as a single line. In the second, because it splits the value on the space, it sees "Hello" and "There" as separate values.