pxzgrep TODO

Planned Features

Append by default instead of overwriting. Solves the issue of searching files with the same name in different directories, e.g. in directory structures with date based files names of the scheme YYYY-MM/HH:MM.log.xz.
Behave more grep-like: In the end, output all hits to STDOUT by concatenating all output files in sorted order. Maybe use a temporary directory for those files and optionally delete them again. (The old behaviour still needs to stay available respectively might be used in the background, e.g. by temporarily saving all the results in a directory under $TMPDIR.)
Commandline option to select target directory instead of current working directory, maybe --target or maybe analoguos to cp and mv rather --target-directory=DIRECTORY and -t (luckily so far unused in grep).

Support grep's -r despite xzgrep doesn't.
Support other decompressors like gzip, bzip2, zstd, lzop, lzip, lz4, rzip, brotli, zopfli, lzma, etc. (lzma might just need some suffix handling as IIRC the family of xz tools already can handle lzma as it's xz's predecessor.)
- Support working with zutils.
- Might need some renaming. pzgrep is a potential name as pgrep is already in use for GNU Grep's PCRE grep variant. (Not to confuse with pcregrep which is a different implementation.)
- Would require some exception handling code what happens if files with the same base file name, but different suffixes exist, e.g. file.gz and file.xz. Possible solution: Name the output files file_gz and file_xz instead of currently just file.
Reimplement the whole thing in Perl. It's getting too complex to be easily maintainable in shell code.
- Also the verbose output can be made be less awkward then.
- Open question here: Shall we still use xargs or some equivalent in Perl like one of the modules from the Parallel::* hierachy?