Zip files that match certain data pattern within the file

Question

I have a directory full of files *.CLM_01. I want to create 1 zip file with all files which have the value "005010X222" inside the file. Basically, there are a few files that have "005010X223" which I want to exclude from the zip. They are mutually exclusive. These values are in the data of the file not the file name.

I tried this but it' still zipping all files into 1 zip file

for f in *.CLM_01; do zip -ur test.zip $f | grep "005010X222" $f ; done

Answer 1

With zsh and GNU grep or compatible:

zip file.zip ${(0)"$(grep -l --null 005010X222 ./*.CLM_01)"}

If your grep doesn't take a --null but you can guarantee none of the file names contain newline characters:

zip file.zip ${(f)"$(grep -l 005010X222 ./*.CLM_01)"}

With other POSIX-like shells, you could do the same with the split+glob operator:

IFS='
' # split on newline
set -o noglob # disable glob
zip file.zip $(grep -l 005010X222 ./*.CLM_01)

If you have bsdtar, you can also pass the list of files null-delimited on its stdin with --null -T -:

grep -l --null 005010X222 ./*.CLM_01 | bsdtar --null -T - -acf file.zip

The GNU implementation of grep can also look for the *.CLM_01 files recursively by itself with:

grep --include='*.CLM_01' -rl --null 005010X222 . |
  bsdtar --null -T - -acf file.zip

That one isn't affected by the limit on the number of arguments that is passed to a command since the file paths are not passed as arguments to grep or bsdtar, but note that contrary to the glob-based solutions above, it will also consider hidden files.

With GNU grep, see also grep -L 005010X223 to list the files that do not contain 005010X223 instead of grep -l 005010X222 to list the ones that do contain 005010X222.

Answer 2

Use find with a quiet grep test:

find -name '*CLM_01' -exec grep -q '005010X222' {} \; -print0 | xargs -0 zip -u all_files.zip +

-q in grep is the quiet option, that only succeeds (exit code 0) on a match.

Best is to also use -u in zip in case xargs needs to call zip multiple times.