test: move harness and live tests into e2e/

.github/workflows/ci.yml

@@ -23,6 +23,7 @@ jobs:
       - run: bunx tsc --noEmit
       - run: bun run lint
       - run: bun run test
+      - run: bun run test:e2e
       - run: bunx prettier --check .
       - run: bun run build
       - name: Import packed dist with Node

CONTRIBUTING.md

@@ -7,18 +7,28 @@ bun install
 bun run typecheck
 bun run lint
 bun run test
+bun run test:e2e
 ```
 
-`bun run test` is hermetic. `bun run test:live` runs a `/v1/embeddings` round
-trip and skips unless `EMBEDDING_E2E_BASE_URL` is set; point it at a remote
-server rather than loading models locally:
+`bun run test` runs the colocated units in `src/`. `bun run test:e2e` runs
+`e2e/`: retry, wire and base64 cases on the `@intx/inference-testing` harness
+and a local HTTP stub, plus a `/v1/embeddings` round trip that skips unless
+`EMBEDDING_E2E_BASE_URL` is set; point it at a remote server rather than
+loading models locally:
 
 ```bash
-EMBEDDING_E2E_BASE_URL=http://100.113.184.123:11434/v1 bun run test:live
+EMBEDDING_E2E_BASE_URL=http://100.113.184.123:11434/v1 bun run test:e2e
 ```
 
 `EMBEDDING_E2E_MODEL` defaults to `nomic-embed-text`.
 
 ## Versioning
 
 Semver. Releases run `bun run build && npm publish` with green CI.
+
+## Commit messages
+
+Commit subjects and PR titles follow [Conventional Commits](https://www.conventionalcommits.org): `feat`, `fix`, `refactor`, `test`, `docs`, `build`, `ci`, `perf`, and `chore(release): x.y.z` for releases.
+Add `!` only for public API breaks: removed or renamed exports, changed signatures, newly required params. Peer and dependency range changes are `build(deps):` with no `!`.
+Keep subjects imperative, lowercase after the colon, 72 characters or less, and free of ticket IDs.
+Every PR links its issue with a `Closes <issue id>` line in the PR body.

IMPLEMENTATION.md

@@ -101,14 +101,16 @@ package depends only on the published surface).
 
 ## Tests
 
-`bun run test` runs `bun test ./src ./tests`.
+`bun run test` runs `bun test ./src`; `bun run test:e2e` runs `bun test ./e2e`.
+Both run in CI.
 
 - `src/embed.test.ts` — units on `@intx/inference-testing` fakes: index
-  placement across batches, short reply, base64 decode.
-- `tests/retry.test.ts` — 429 retried once after `Retry-After` on the
-  harness clock; 401 rejects without retry.
-- `tests/base64.test.ts` — `base64` and `float` agree over a real HTTP stub.
-- `live/embeddings.test.ts` (`bun run test:live`, not in `test` or CI) —
-  round trip against a real `/v1/embeddings` (`EMBEDDING_E2E_BASE_URL`,
-  required; `EMBEDDING_E2E_MODEL`, default `nomic-embed-text`).
-  Skips when unset or unreachable.
+  placement across batches, short and duplicate replies, base64 decode,
+  non-JSON 200.
+- `e2e/retry.test.ts` — 429 retried after `Retry-After` (seconds, HTTP-date,
+  custom extractor) on the harness clock; 401 rejects without retry; timeout.
+- `e2e/wire.test.ts` — request URL, body fields and bearer header.
+- `e2e/base64.test.ts` — `base64` and `float` agree over a real HTTP stub.
+- `e2e/live.test.ts` — round trip against a real `/v1/embeddings`
+  (`EMBEDDING_E2E_BASE_URL`, required; `EMBEDDING_E2E_MODEL`, default
+  `nomic-embed-text`). Skips when unset or unreachable.

e2e/wire.test.ts

@@ -0,0 +1,57 @@
+import { afterEach, expect, test } from "bun:test";
+import { setupHarness, type Harness } from "@intx/inference-testing";
+
+import { embedTexts, type EmbedConfig } from "../src/index";
+
+const CONFIG: EmbedConfig = {
+  baseURL: "https://embed.example/v1",
+  model: "text-embedding-3-small",
+};
+
+let harness: Harness | undefined;
+afterEach(() => {
+  harness?.dispose();
+  harness = undefined;
+});
+
+async function run<T>(h: Harness, pending: Promise<T>): Promise<T> {
+  await h.run();
+  return await pending;
+}
+
+test("sends only configured fields to {baseURL}/embeddings", async () => {
+  harness = setupHarness();
+  const stream = harness.scenario.createStream();
+  harness.scenario.whenRequestBodyMatches(
+    (body, req) =>
+      req.url === "https://embed.example/v1/embeddings" &&
+      !req.headers.has("authorization") &&
+      body === JSON.stringify({ model: CONFIG.model, input: ["a"] }),
+    stream,
+  );
+  stream.enqueueAll(
+    [new TextEncoder().encode('{"data":[{"index":0,"embedding":[1]}]}')],
+    { startAt: 1 },
+  );
+  const pending = embedTexts(["a"], CONFIG, { deps: harness.deps });
+  expect(await run(harness, pending)).toEqual([[1]]);
+});
+
+test("sends a bearer token only when apiKey is set", async () => {
+  harness = setupHarness();
+  const stream = harness.scenario.createStream();
+  harness.scenario.whenRequestMatches(
+    (req) => req.headers.get("authorization") === "Bearer sk-test",
+    stream,
+  );
+  stream.enqueueAll(
+    [new TextEncoder().encode('{"data":[{"index":0,"embedding":[1]}]}')],
+    { startAt: 1 },
+  );
+  const pending = embedTexts(
+    ["a"],
+    { ...CONFIG, apiKey: "sk-test" },
+    { deps: harness.deps },
+  );
+  expect(await run(harness, pending)).toEqual([[1]]);
+});

package.json

@@ -1,7 +1,7 @@
 {
   "name": "@corbits/embedding",
   "description": "Embedding client for OpenAI-compatible /v1/embeddings endpoints, built on the Interchange inference primitives",
-  "version": "0.1.0",
+  "version": "0.2.0",
   "license": "LGPL-2.1-only",
   "author": "Sawyer Cutler <sawyer@dirtroad.dev>",
   "type": "module",
@@ -18,8 +18,8 @@
     "prepack": "bun run build",
     "typecheck": "tsc --noEmit",
     "lint": "eslint .",
-    "test": "bun test ./src ./tests",
-    "test:live": "bun test ./live",
+    "test": "bun test ./src",
+    "test:e2e": "bun test ./e2e",
     "format": "prettier -w ."
   },
   "dependencies": {

src/embed.test.ts

@@ -115,56 +115,3 @@ test("classifies a 200 that is not JSON instead of leaking a SyntaxError", async
     reason: { category: "protocol_mismatch" },
   });
 });
-
-test("sends only configured fields to {baseURL}/embeddings", async () => {
-  harness = setupHarness();
-  const stream = harness.scenario.createStream();
-  harness.scenario.whenRequestBodyMatches(
-    (body, req) =>
-      req.url === "https://embed.example/v1/embeddings" &&
-      !req.headers.has("authorization") &&
-      body === JSON.stringify({ model: CONFIG.model, input: ["a"] }),
-    stream,
-  );
-  stream.enqueueAll(
-    [new TextEncoder().encode('{"data":[{"index":0,"embedding":[1]}]}')],
-    { startAt: 1 },
-  );
-  const pending = embedTexts(["a"], CONFIG, { deps: harness.deps });
-  expect(await run(harness, pending)).toEqual([[1]]);
-});
-
-test("sends a bearer token only when apiKey is set", async () => {
-  harness = setupHarness();
-  const stream = harness.scenario.createStream();
-  harness.scenario.whenRequestMatches(
-    (req) => req.headers.get("authorization") === "Bearer sk-test",
-    stream,
-  );
-  stream.enqueueAll(
-    [new TextEncoder().encode('{"data":[{"index":0,"embedding":[1]}]}')],
-    { startAt: 1 },
-  );
-  const pending = embedTexts(
-    ["a"],
-    { ...CONFIG, apiKey: "sk-test" },
-    { deps: harness.deps },
-  );
-  expect(await run(harness, pending)).toEqual([[1]]);
-});
-
-test("returns [] for empty input without a request", async () => {
-  harness = setupHarness();
-  const pending = embedTexts([], CONFIG, { deps: harness.deps });
-  expect(await run(harness, pending)).toEqual([]);
-});
-
-test("rejects a non-positive batchSize before any request", async () => {
-  harness = setupHarness();
-  for (const batchSize of [0, -1, 1.5]) {
-    await expect(
-      embedTexts(["a"], { ...CONFIG, batchSize }, { deps: harness.deps }),
-    ).rejects.toThrow(/batchSize/);
-  }
-  await harness.run();
-});

tsconfig.json

@@ -25,5 +25,5 @@
 
     "lib": ["ESNext"]
   },
-  "include": ["src/**/*.ts", "tests/**/*.ts", "live/**/*.ts"]
+  "include": ["src/**/*.ts", "e2e/**/*.ts"]
 }