forge-tests - FORGE

forge-tests runs FORGE’s unit suites inside Play-In-Editor. There is no external CLI runner. On boot, every package’s *_tests.lua registers its suites and forge-tests runs them, printing a pass/fail summary to the Output Log. You use this package when writing tests for your own content.

Writing a suite

Each package builds a test object and registers suites and cases on it. The harness handles beforeAll/beforeEach/afterAll and a set of assertions.

local T = ForgeTest.new("my-content")

T.suite("crafting", function()
  T.beforeAll(function() --[[ setup ]] end)

  T.test("registers the recipe", function()
    local r = exports["forge-crafting"]:GetRecipe("recipe:health_tonic")
    T.assert.ok(r)                       -- r.ok == true
    T.assert.equal(r.data.recipe.id, "recipe:health_tonic")
  end)
end)

Assertions

Assertion	Checks
`T.assert.ok(r)`	`r.ok == true`
`T.assert.fail(r)`	`r.ok == false`
`T.assert.code(r, code)`	`r.code == code`
`T.assert.equal(a, b, msg)`	`a == b`
`T.assert.notEqual(a, b, msg)`	`a ~= b`
`T.assert.truthy(v, msg)` / `T.assert.falsy(v, msg)`	truthiness

Exports

Export	Signature
`IsReady`	`IsReady()`
`RegisterSuite`	`RegisterSuite(...)`, called by the harness, not usually by hand
`RunAll`	`RunAll()`
`RunPackage`	`RunPackage(name)`
`RunTag`	`RunTag(tag)`
`Report` / `GetReport`	aggregate the last run’s results

FORGE’s database persists across PIE runs. Tests that write rows must use run-unique ids (an os.time() prefix) or delete per-id in setup, or a stale row from a prior run fails the rerun.

How to read a run

A green boot logs:

[FORGE-SMOKE] DONE 153 passed, 0 failed
[FORGE-TEST]  DONE run=run_1 pass=625 fail=0 skipped=0