Read arbitrary number of directory/file paths from standard input or a file

[ao-apps]

Feature Request

We would like for rdfind to have the option to read the set of paths from standard input.

This could possibly be the default when no paths given.

There could also be a switch to provide the paths, with - meaning standard input.

By default, paths would be split by newline. A -0 switch would split by NULL.

Compares to rsync's --files-from and --from0 / -0 options.

Motivation

We are de-duplicating specific build artifacts within a large set of Jenkins/Maven build directories. This results in more files than can be passed through a single command via xargs, even with a maximum value to --max-chars. The result is multiple invocations of rdfind and incomplete de-duplication.

We are not able to safely de-duplicate entire directories, just specific filename patterns. For example, we can safely de-duplicate timestamped snapshots which are never overwritten. However, the *-SNAPSHOT.{aar,jar,war,…} files cannot be de-duplicated since they are overwritten in-place by Maven.