Feature Request: Compare Subdirectories, ignore directory name difference

[Mike Fiedler]

Hello! Thanks for making this tool - it's quite cool and fast.

I've been using difft as well as diffoscope and was curious if you had a solution to a problem diffoscope solves, that I don't think difft handles today.

Here's my current use case, let me know if it doesn't make sense:

I often expand multiple zip files containing a variety of files, and want to establish a few things:

a. How "same" are these zip file contents?
b. What are the distinct differences?

For a, I'd love a way to emit only a percentage as output, somewhat similar to --check-only - that way I could run a lot of diffs and only stop when something is either too similar or too divergent.

I think difft solves most of b, but doesn't handle path name differences yet, unless I don't have the right flags.

Here's an example expanded layout of two almost-identical zip files:

analyticsclient-6502
├── LICENSE.txt
├── MANIFEST.in
├── PKG-INFO
├── README.md
├── analyticsclient.egg-info
│   ├── PKG-INFO
│   ├── SOURCES.txt
│   ├── dependency_links.txt
│   ├── requires.txt
│   └── top_level.txt
├── data
│   └── data_file
├── pyproject.toml
├── setup.cfg
├── setup.py
└── tests
    ├── __init__.py
    └── test_simple.py

4 directories, 15 files

brotli-bin-0.0.1
├── LICENSE.txt
├── MANIFEST.in
├── PKG-INFO
├── README.md
├── brotli_bin.egg-info
│   ├── PKG-INFO
│   ├── SOURCES.txt
│   ├── dependency_links.txt
│   ├── requires.txt
│   └── top_level.txt
├── data
│   └── data_file
├── pyproject.toml
├── setup.cfg
├── setup.py
└── tests
    ├── __init__.py
    └── test_simple.py

4 directories, 15 files

Both contain same files, but have slightly differing different paths.

Using diffoscope I execute: diffoscope --exclude-directory-metadata=yes analyticsclient-6502 brotli-bin-0.0.1 and get an output of the differences for the PKG_INFO files in the *-info sudirectories side by side - but difft can't compare those yet.

(--exclude-directory-metadata=yes is to remove comparing the times and dates on the files, but I think that's diffoscope-specific and difft doesn't care about that yet.)