JuliaCI
JSON coverage summary output
(I looked a bit and couldn't find anything that does this. Please let me know if this already exists somewhere.)
I'd like to output a coverage summary in JSON similar to jest --coverage --coverageReporters="json-summary" (relevant docs) to produce
"total": {
"lines": { "total": 21777, "covered": 65, "skipped": 0, "pct": 0.3 },
"statements": { "total": 24163, "covered": 72, "skipped": 0, "pct": 0.3 },
"functions": { "total": 5451, "covered": 16, "skipped": 0, "pct": 0.29 },
"branches": { "total": 6178, "covered": 10, "skipped": 0, "pct": 0.16 }
}
This will mimic the structure of the jest output as much as possible. The file will be such that, if test_args are provided (#46), then the top "total" will be replaced with the name of the test set.
I'm torn on whether this belongs in LocalCoverage.jl or Coverage.jl. On the one hand (from a familiar thread):
using Coverage
coverage = process_folder()
LCOV.writefile("lcov.info", coverage)
and note the other possible values of that jest argument are ["clover", "json", "lcov", "text"], so you could argue that lcov and json should be an alternative output option of Coverage.jl.
But on the other hand, using jest --coverageReporters="text-summary" produces:
=============================== Coverage summary ===============================
Statements : 100% ( 166/166 )
Branches : 75% ( 18/24 )
Functions : 100% ( 49/49 )
Lines : 100% ( 161/161 )
================================================================================
which is very similar to the text summary output of LocalCoverage.jl:
Coverage of /Users/haiiro/NoSync/LocalCoverage.jl/test/DummyPackage/
┌─────────────────────┬───────┬─────┬──────┬──────┐
│ Filename │ Lines │ Hit │ Miss │ % │
├─────────────────────┼───────┼─────┼──────┼──────┤
│ src/DummyPackage.jl │ 1 │ 1 │ 0 │ 100% │
│ src/bar.jl │ 2 │ 1 │ 1 │ 50% │
│ src/corge/corge.jl │ 1 │ 1 │ 0 │ 100% │
│ src/corge/grault.jl │ 1 │ 0 │ 1 │ 0% │
│ src/qux.jl │ 2 │ 1 │ 1 │ 50% │
│ src/qux.jl │ 2 │ 1 │ 1 │ 50% │
├─────────────────────┼───────┼─────┼──────┼──────┤
│ TOTAL │ 9 │ 5 │ 4 │ 56% │
└─────────────────────┴───────┴─────┴──────┴──────┘
Personally, I think it should go here since the necessary components are already conveniently located and need only be put in a dict to make for easy json output.
For background, I'm interested in storing the json during CI, then reading it during subsequent runs in order to provide a pass/fail based on coverage increase/decrease, similar to what you get out of the box with codecov.io. (It's either this or cough up the $500/yr for codecov pro!)
Happy to PR.
