1import * as AWS from "@/AWS";
2import * as Core from "@/Test/Core";
3import * as Test from "@/Test/Alchemy";
4import * as eventbridge from "@distilled.cloud/aws/eventbridge";
5import { describe, expect } from "alchemy-test";
6import * as Data from "effect/Data";
7import * as Effect from "effect/Effect";
8import * as Schedule from "effect/Schedule";
9import * as HttpClient from "effect/http/HttpClient";
10import * as HttpClientRequest from "effect/http/HttpClientRequest";
11import GlueTestFunctionLive, { GlueTestFunction } from "./handler";
12
13const testOptions = { providers: AWS.providers() };
14const { test, beforeAll, afterAll } = Test.make(testOptions);
15const sharedStack = Core.scratchStack(testOptions, "GlueBindings");
16
17// Lambda function URL cold-start (DNS, IAM propagation, init) can take well
18// over 60s on a fresh deploy.
19const readinessPolicy = Schedule.max([
20 Schedule.fixed("2 seconds"),
21 Schedule.recurs(75),
22]);
23
24let baseUrl: string;
25let functionArn: string;
26
27class TransientUpstream extends Data.TaggedError("TransientUpstream")<{
28 readonly status: number;
29 readonly body: string;
30}> {}
31
32// The shared Lambda fixture occasionally answers a transient 5xx under load
33// (cold re-init, IAM propagation on the freshly attached policy that the
34// handler's `Effect.orDie` surfaces as a 500). Retry only 5xx; a genuine
35// 4xx/assertion failure surfaces immediately.
36const send = (request: HttpClientRequest.HttpClientRequest) =>
37 HttpClient.execute(request).pipe(
38 Effect.flatMap((response) =>
39 response.status >= 500
40 ? response.text.pipe(
41 Effect.flatMap((body) =>
42 Effect.fail(
43 new TransientUpstream({ status: response.status, body }),
44 ),
45 ),
46 )
47 : Effect.succeed(response),
48 ),
49 Effect.retry({
50 while: (e) => e._tag === "TransientUpstream",
51 schedule: Schedule.max([
52 Schedule.exponential("500 millis"),
53 Schedule.recurs(6),
54 ]),
55 }),
56 );
57
58const getJson = (path: string) =>
59 send(HttpClientRequest.get(`${baseUrl}${path}`)).pipe(
60 Effect.flatMap((r) => r.json),
61 );
62
63const postJson = (path: string) =>
64 send(HttpClientRequest.post(`${baseUrl}${path}`)).pipe(
65 Effect.flatMap((r) => r.json),
66 );
67
68describe.sequential(
69 "Glue Bindings",
70 {
71 tags: [
72 "provider:aws",
73 "provider:aws:batch",
74 "provider:aws:glue",
75 "provider:aws:iam",
76 "provider:aws:lambda",
77 "provider:aws:s3",
78 "live",
79 ],
80 },
81 () => {
82 beforeAll(
83 Effect.gen(function* () {
84 yield* Effect.logInfo("Glue test setup: destroying previous resources");
85 yield* sharedStack.destroy();
86
87 yield* Effect.logInfo("Glue test setup: deploying fixture");
88 const attrs = yield* sharedStack.deploy(
89 Effect.gen(function* () {
90 return yield* GlueTestFunction;
91 }).pipe(Effect.provide(GlueTestFunctionLive)),
92 );
93
94 expect(attrs.functionUrl).toBeTruthy();
95 baseUrl = attrs.functionUrl!.replace(/\/+$/, "");
96 functionArn = attrs.functionArn;
97
98 const readinessUrl = `${baseUrl}/bindings`;
99 yield* Effect.logInfo(
100 `Glue test setup: probing readiness at ${readinessUrl}`,
101 );
102 yield* HttpClient.get(readinessUrl).pipe(
103 Effect.flatMap((response) =>
104 response.status === 200
105 ? Effect.succeed(response)
106 : Effect.fail(
107 new Error(`Function not ready: ${response.status}`),
108 ),
109 ),
110 Effect.tapError((error) =>
111 Effect.logWarning(
112 `Glue test setup: fixture not ready yet (${String(error)})`,
113 ),
114 ),
115 Effect.retry({ schedule: readinessPolicy }),
116 );
117
118 // Glue's authorization layer caches IAM evaluations: a freshly created
119 // role + inline policy is denied (`AccessDeniedException`) by the Glue
120 // endpoint for ~3–5 minutes after `PutRolePolicy`, even though the
121 // policy is attached (verified via CloudTrail: policy attached
122 // 17:10:47Z, every glue:* call denied until exactly 17:15:50Z). Gate
123 // the suite on the first job-scoped call succeeding so the tests
124 // measure binding behavior, not IAM propagation.
125 yield* Effect.logInfo("Glue test setup: waiting for IAM propagation");
126 yield* HttpClient.get(`${baseUrl}/job-runs`).pipe(
127 Effect.flatMap((response) =>
128 response.status === 200
129 ? Effect.succeed(response)
130 : Effect.fail(
131 new Error(`Glue IAM not propagated: ${response.status}`),
132 ),
133 ),
134 Effect.retry({
135 schedule: Schedule.max([
136 Schedule.fixed("5 seconds"),
137 Schedule.recurs(84),
138 ]),
139 }),
140 );
141 }),
142 { timeout: 900_000 },
143 );
144
145 afterAll.skipIf(!!process.env.NO_DESTROY)(sharedStack.destroy(), {
146 timeout: 300_000,
147 });
148
149 describe("binding registration", () => {
150 test.provider("all twenty capabilities initialize in the runtime", () =>
151 Effect.gen(function* () {
152 const response = (yield* getJson("/bindings")) as { bound: string[] };
153 expect(response.bound).toHaveLength(20);
154 }),
155 );
156 });
157
158 describe("GetJobRuns", () => {
159 test.provider("lists the job's runs (injected job name)", () =>
160 Effect.gen(function* () {
161 const response = (yield* getJson("/job-runs")) as { count: number };
162 expect(response.count).toBeGreaterThanOrEqual(0);
163 }),
164 );
165 });
166
167 describe("GetJobBookmark / ResetJobBookmark", () => {
168 test.provider("both surface the typed missing-bookmark error", () =>
169 Effect.gen(function* () {
170 const response = (yield* getJson("/job-bookmark")) as {
171 bookmark: string;
172 reset: string;
173 };
174 expect(response.bookmark).toBe("none");
175 expect(response.reset).toBe("none");
176 }),
177 );
178 });
179
180 describe("StartJobRun / GetJobRun / BatchStopJobRun", () => {
181 test.provider(
182 "starts, reads, and stops a live run",
183 () =>
184 Effect.gen(function* () {
185 const response = (yield* postJson("/job-run")) as {
186 runId: string;
187 state: string;
188 stopSubmitted: number;
189 stopErrors: number;
190 };
191 expect(response.runId).toMatch(/^jr_/);
192 expect(response.state).toBeTruthy();
193 expect(response.stopSubmitted + response.stopErrors).toBe(1);
194 }),
195 { timeout: 120_000 },
196 );
197 });
198
199 describe("GetCrawler", () => {
200 test.provider("reads the bound crawler's live state", () =>
201 Effect.gen(function* () {
202 const response = (yield* getJson("/crawler")) as {
203 name: string;
204 state: string;
205 };
206 expect(response.name).toBeTruthy();
207 expect(["READY", "RUNNING", "STOPPING"]).toContain(response.state);
208 }),
209 );
210 });
211
212 describe("StartCrawler / StopCrawler", () => {
213 test.provider("stop on an idle crawler is the typed not-running", () =>
214 Effect.gen(function* () {
215 const response = (yield* postJson("/crawler-stop")) as {
216 result: string;
217 };
218 expect(response.result).toBe("not-running");
219 }),
220 );
221
222 test.provider(
223 "starts a crawl and cancels it",
224 () =>
225 Effect.gen(function* () {
226 const response = (yield* postJson("/crawler-run")) as {
227 started: string;
228 stopped: string;
229 };
230 expect(["started", "already-running"]).toContain(response.started);
231 expect(["stopped", "stopping", "not-running"]).toContain(
232 response.stopped,
233 );
234 }),
235 { timeout: 120_000 },
236 );
237 });
238
239 describe("GetTable / GetTables", () => {
240 test.provider(
241 "reads the bound table's schema and the database's tables",
242 () =>
243 Effect.gen(function* () {
244 const response = (yield* getJson("/table")) as {
245 name: string;
246 columns: string[];
247 partitionKeys: string[];
248 tables: string[];
249 };
250 expect(response.columns).toEqual(["id", "amount"]);
251 expect(response.partitionKeys).toEqual(["dt"]);
252 expect(response.tables).toContain(response.name);
253 }),
254 );
255 });
256
257 describe("Partitions", () => {
258 test.provider(
259 "create, get, update, batch-create/update/get, list, batch-delete",
260 () =>
261 Effect.gen(function* () {
262 const response = (yield* postJson("/partitions")) as {
263 created: string[];
264 updatedParam: string;
265 batchErrors: number;
266 batchUpdateErrors: number;
267 bulkRead: number;
268 bulkUpdatedParams: number;
269 count: number;
270 batchDeleteErrors: number;
271 remaining: number;
272 };
273 expect(response.created).toEqual(["2026-01-01"]);
274 expect(response.updatedParam).toBe("true");
275 expect(response.batchErrors).toBe(0);
276 expect(response.batchUpdateErrors).toBe(0);
277 expect(response.bulkRead).toBe(3);
278 expect(response.bulkUpdatedParams).toBe(2);
279 expect(response.count).toBe(3);
280 expect(response.batchDeleteErrors).toBe(0);
281 expect(response.remaining).toBe(0);
282 }),
283 { timeout: 120_000 },
284 );
285 });
286
287 describe(
288 "consumeJobEvents / consumeCrawlerEvents",
289 { tags: ["provider:aws:eventbridge"] },
290 () => {
291 test.provider(
292 "the deploy created EventBridge rules targeting the function",
293 () =>
294 Effect.gen(function* () {
295 // Out-of-band via distilled: the fixture's two consume* calls must
296 // have materialized as rules on the default bus with the Lambda as
297 // target.
298 const { RuleNames } = yield* eventbridge.listRuleNamesByTarget({
299 TargetArn: functionArn,
300 });
301 expect((RuleNames ?? []).length).toBeGreaterThanOrEqual(2);
302 }),
303 );
304 },
305 );
306 },
307);