Assertions

Assertions start from expect(value, message?, location?).

Basic example:

import { expect, test } from "as-test";

test("example", () => {
  expect(1 + 2).toBe(3);
  expect("hello").toStartWith("he");
  expect([1, 2, 3]).toContain(2);
});

Negation

Use .not before a matcher:

expect(1).not.toBe(2);
expect("hello").not.toEndWith("xx");

Messages

The optional second argument becomes the failure message when the matcher fails:

expect(total, "sum should stay stable").toBe(42);

Chaining

Every matcher returns the Expectation, so multiple checks can chain off one expect(...):

expect(7).toBe(7).where((): bool => isFresh());
expect("as-test").toBeString().toStartWith("as");
expect(10).toBeGreaterThan(5).toBeLessThan(100);

Each link in the chain is an independent assertion — the test fails if any of them fail. There is no priority; they are peers. A failing matcher does not stop later matchers in the same chain from running, and a later passing matcher cannot un-fail an earlier one.

.not resets after each matcher, so it only applies to the matcher immediately following it:

expect(7).not.toBe(8).toBeGreaterThan(5); // not toBe(8), and > 5

Custom Predicates with .where()

.where() accepts either a bool or a () => bool lambda. Use it when the verdict isn't expressible via the built-in matchers — for example, delegating to a hand-written comparator:

expect(x).where(x > 0 && x < 10);
expect(actual).where((): bool => deepCompare(GLOBAL_A, GLOBAL_B));

The lambda runs once and must return bool. AssemblyScript does not implement closures, so the lambda cannot capture local variables — use the bool form when the predicate references locals, or refer to module-level values from inside the lambda.

See Equality → Custom Predicates with .where() for the full reference.

Matcher Groups

• Equality
• Numbers
• Types And Truthiness
• Strings And Collections
• Snapshots And Throws

Fuzzing Interaction

Inside a fuzzer:

• failed expect(...) calls fail the current fuzz iteration
• they are counted in the fuzz result, not the normal test totals
• the first failure details are recorded in the fuzz report

See Fuzzing for how failures and repro commands are reported.