Test playwright tests with Jev

.gitignore

@@ -27,6 +27,7 @@ yarn-error.log*
 
 # local env files
 .env
+.env.jev
 
 # vercel
 .vercel
@@ -47,6 +48,7 @@ node_modules/
 /test-results/
 /playwright-report/
 /blob-report/
+/benchmark-results/
 /playwright/.cache/
 .playwright-mcp/
 

endform.config.ts

@@ -0,0 +1,5 @@
+import { defineEndformConfig } from "endform";
+
+export default defineEndformConfig({
+	environmentVariables: ["E2E_JEV_.*"],
+});

package.json

@@ -14,7 +14,11 @@
 		"format:check": "oxfmt --check",
 		"format:fix": "oxfmt --write",
 		"lint": "oxlint",
-		"mcp-user": "tsx mcp-user.ts"
+		"mcp-user": "tsx mcp-user.ts",
+		"jev:test": "bun scripts/jev-loop.ts",
+		"jev:viewer": "bun scripts/jev/viewer.ts",
+		"jev:benchmark": "bun scripts/jev-benchmark.ts --mode clean --repeat 10",
+		"jev:benchmark:faults": "bun scripts/jev-benchmark.ts --mode faults"
 	},
 	"dependencies": {
 		"@libsql/client": "^0.18.0",
@@ -51,6 +55,7 @@
 		"oxfmt": "0.68.0",
 		"oxlint": "1.83.0",
 		"playwright-opentelemetry": "^0.13.0",
+		"sharp": "0.35.4",
 		"tsx": "^4.23.15",
 		"typescript": "^7.0.2"
 	},

scripts/jev-benchmark.test.ts

@@ -0,0 +1,92 @@
+import { expect, test } from "bun:test";
+import { actionCode, candidates, select } from "./jev-benchmark";
+
+test("builds bounded actions from an accessibility observation", () => {
+	const actions = candidates(
+		{
+			url: "https://example.test/dashboard",
+			title: "Settings",
+			snapshot:
+				'- link "General" [ref=e1]\n- textbox "Name" [ref=e2]\n- radio "Member" [ref=e3]\n- button "Save" [disabled] [ref=e4]',
+			invitationId: "42",
+		},
+		["John Doe"],
+	);
+	expect(actions.map((action) => action.kind)).toEqual([
+		"click",
+		"fill",
+		"check",
+		"goto",
+		"goto",
+		"wait",
+		"stop",
+	]);
+});
+
+test("keeps action arguments literal", () => {
+	const value = "x\n${process.exit()}";
+	expect(actionCode({ id: "x", kind: "goto", value, description: "go" })).toBe(
+		`await page.goto(${JSON.stringify(value)});`,
+	);
+	expect(() =>
+		actionCode({
+			id: "x",
+			kind: "click",
+			ref: 'e1"); throw 1',
+			description: "bad",
+		}),
+	).toThrow();
+});
+
+test("per-field fills take the value Jev chose for that field", () => {
+	const actions = candidates(
+		{
+			url: "https://example.test/checkout",
+			title: "Checkout",
+			snapshot:
+				'- textbox "Name" [ref=e1]: John Doe\n- textbox "City" [ref=e2]\n- button "Pay" [ref=e3]',
+		},
+		["John Doe", "New York"],
+		{ fieldValues: true },
+	);
+	expect(actions.filter((action) => action.kind === "fill")).toHaveLength(2);
+	const answers = {
+		next: {
+			type: "choice",
+			choice: "a0",
+			probabilities: { a0: 0.5, a1: 0.3, a2: 0.2 },
+		},
+		value_a0: { type: "choice", choice: "v0" },
+		value_a1: { type: "choice", choice: "v1" },
+	};
+	// Name already holds its chosen value, so the next most probable action runs.
+	const selected = select(actions, answers, ["John Doe", "New York"]);
+	expect(selected?.ref).toBe("e2");
+	expect(selected?.value).toBe("New York");
+	expect(actionCode(selected!)).toContain('"New York"');
+	expect(
+		select(
+			actions,
+			{ ...answers, next: { ...answers.next, choice: "missing" } },
+			[],
+		),
+	).toBeUndefined();
+});
+
+test("without per-field values the chosen action runs unchanged", () => {
+	const actions = candidates(
+		{
+			url: "https://example.test",
+			title: "",
+			snapshot: '- button "Go" [ref=e1]',
+		},
+		[],
+	);
+	expect(
+		select(
+			actions,
+			{ next: { type: "choice", choice: "wait", probabilities: { a0: 0.9 } } },
+			[],
+		)?.kind,
+	).toBe("wait");
+});