[Feat] Determine the format automatically for input files on the CLI

[yannham]

Depends on #2266

Supporting multiple input formats on the CLI

Motivation

When importing a file, Nickel determines the format automatically by looking at the file's extension. This is not true on the CLI though, which could be useful: for example, one could use Nickel as a merging engine for JSON without actually needing to migrate anything to Nickel: nickel base.json ext.json foo.json. In conjunction with #2266, supporting other formats on the CLI could also allow validating non-Nickel config in a lightweight and non-invasive way: nickel eval config.json --apply-contract validation/*.ncl.

Content

Instead of always determining the input format to be Nickel for files coming from the CLI, this PR uses the same logic as for imports, looking at the file's extension.

Shortcomings

One current shortcoming is that, as opposed to imports, there is no way to override the format selection logic. However, we've lived with that for a long time with imports - the explicit format syntax is relatively recent -, and the motivation for imports is a bit different. The explicit format syntax for imports is more about specifying the format explicitly when Nickel doesn't recognize the extension, say with foo.data that should be text, rather than overriding an import named foo.json to be understood as TOML instead. The latter would be a very strange use-case.

Thus, this PR should be a strict improvement over the previous situation, which would consider foo.json as a Nickel input, and shouldn't require any work-around, although it could be extended in some way to specify the format explicitly.

[yannham]

Hmm, the snapshot test failure happens on the CI but not on my machine...

[dpulls]

🎉 All dependencies have been resolved !

[nbacquey]

Otherwise lgtm

[github-actions]

Bencher Report

Branch	feat/multi-format-cli
Testbed	ubuntu-latest

Click to view all benchmark results

Benchmark	Latency	microseconds (µs)
fibonacci 10	📈 view plot 🚷 view threshold	290.48 µs
foldl arrays 50	📈 view plot 🚷 view threshold	400.92 µs
foldl arrays 500	📈 view plot 🚷 view threshold	3,748.50 µs
foldr strings 50	📈 view plot 🚷 view threshold	4,567.30 µs
foldr strings 500	📈 view plot 🚷 view threshold	46,200.00 µs
generate normal 250	📈 view plot 🚷 view threshold	42,371.00 µs
generate normal 50	📈 view plot 🚷 view threshold	1,626.40 µs
generate normal unchecked 1000	📈 view plot 🚷 view threshold	2,125.20 µs
generate normal unchecked 200	📈 view plot 🚷 view threshold	444.61 µs
pidigits 100	📈 view plot 🚷 view threshold	2,305.30 µs
pipe normal 20	📈 view plot 🚷 view threshold	687.40 µs
pipe normal 200	📈 view plot 🚷 view threshold	5,916.00 µs
product 30	📈 view plot 🚷 view threshold	255.41 µs
scalar 10	📈 view plot 🚷 view threshold	595.30 µs
sum 30	📈 view plot 🚷 view threshold	258.78 µs

🐰 View full continuous benchmarking report in Bencher