@malloy-publisher/server 0.0.198 → 0.0.199

package/src/service/package_worker_path.spec.ts

@@ -0,0 +1,196 @@
+/**
+ * Integration test: exercise `Package.create` with the package-load
+ * worker pool enabled (PACKAGE_LOAD_WORKERS=1).
+ *
+ * Validates that the worker-load path:
+ *   - reads the manifest, probes embedded databases, and compiles
+ *     every model in a single off-thread job
+ *   - produces a live `Package` whose `Model`s have populated
+ *     `modelDef` / `sources` / `queries`
+ *   - hydrates the `ModelMaterializer` from `modelDef` on first
+ *     query (no recompile) — verified end-to-end by running a
+ *     query through the resulting Model and getting a result
+ *
+ * Kept separate from `package.spec.ts` so the existing tests keep
+ * running on the in-process path without paying worker startup cost.
+ *
+ * Pool reuse strategy: one `PackageLoadPool` shared across all
+ * cases in this file. Spawning a fresh worker per test crashes Bun
+ * (segfault) because DuckDB's native bindings don't tolerate being
+ * loaded concurrently into multiple worker isolates of the same Bun
+ * process. Production uses one pool; this matches.
+ */
+import {
+   afterAll,
+   afterEach,
+   beforeAll,
+   beforeEach,
+   describe,
+   expect,
+   it,
+} from "bun:test";
+import * as fs from "fs";
+import * as os from "os";
+import * as path from "path";
+import {
+   PackageLoadPool,
+   __setPackageLoadPoolForTests,
+} from "../package_load/package_load_pool";
+import { Package } from "./package";
+
+const ORIGINAL_ENV = process.env.PACKAGE_LOAD_WORKERS;
+
+describe("Package.create via worker pool", () => {
+   let tempDir: string;
+   let pool: PackageLoadPool;
+
+   beforeAll(async () => {
+      process.env.PACKAGE_LOAD_WORKERS = "1";
+      pool = new PackageLoadPool(1);
+      // Wire our pool into the module-level singleton so Package.create
+      // picks it up via getPackageLoadPool().
+      await __setPackageLoadPoolForTests(pool);
+   });
+
+   afterAll(async () => {
+      await __setPackageLoadPoolForTests(null);
+      if (ORIGINAL_ENV === undefined) {
+         delete process.env.PACKAGE_LOAD_WORKERS;
+      } else {
+         process.env.PACKAGE_LOAD_WORKERS = ORIGINAL_ENV;
+      }
+   });
+
+   beforeEach(() => {
+      tempDir = fs.mkdtempSync(path.join(os.tmpdir(), "publisher-pkg-worker-"));
+   });
+
+   afterEach(() => {
+      if (tempDir) {
+         // tempDir gets wiped by Package.create on failure (it's the
+         // staging-cleanup path); ignore ENOENT here.
+         try {
+            fs.rmSync(tempDir, { recursive: true, force: true });
+         } catch {
+            /* already gone */
+         }
+         tempDir = "";
+      }
+   });
+
+   async function makeMalloyConfig(): Promise<{
+      malloyConfig: import("@malloydata/malloy").MalloyConfig;
+      duckdb: { close: () => Promise<void> };
+   }> {
+      const { MalloyConfig, FixedConnectionMap } = await import(
+         "@malloydata/malloy"
+      );
+      const { DuckDBConnection } = await import("@malloydata/db-duckdb");
+      const duckdb = new DuckDBConnection("duckdb", ":memory:");
+      const connections = new FixedConnectionMap(
+         new Map([["duckdb", duckdb]]),
+         "duckdb",
+      );
+      const malloyConfig = new MalloyConfig({ connections: {} });
+      malloyConfig.wrapConnections(() => connections);
+      return { malloyConfig, duckdb };
+   }
+
+   function writeManifest(): void {
+      fs.writeFileSync(
+         path.join(tempDir, "publisher.json"),
+         JSON.stringify({ name: "pkg", description: "test package" }),
+      );
+   }
+
+   it("loads a package end-to-end and serves a query through the hydrated materializer", async () => {
+      writeManifest();
+      // Define `total_v` as a *view* on the source so the query
+      // builder's `run: nums -> total_v` form resolves. (Top-level
+      // queries take the `run: total_q` form — orthogonal path that
+      // the in-process tests already cover.)
+      fs.writeFileSync(
+         path.join(tempDir, "trivial.malloy"),
+         `source: nums is duckdb.sql("select 1 as a, 2 as b") extend {
+  measure: total is a.sum()
+  view: total_v is { aggregate: total }
+}`,
+      );
+
+      const { malloyConfig, duckdb } = await makeMalloyConfig();
+      try {
+         const pkg = await Package.create("env", "pkg", tempDir, malloyConfig);
+         expect(pkg).toBeInstanceOf(Package);
+         expect(pkg.getModelPaths()).toEqual(["trivial.malloy"]);
+
+         const model = pkg.getModel("trivial.malloy");
+         expect(model).toBeDefined();
+         const apiModel = (await model!.getModel()) as {
+            modelDef?: string;
+            sources?: { name?: string }[];
+         };
+         expect(apiModel.modelDef).toBeDefined();
+         expect(apiModel.sources?.[0]?.name).toBe("nums");
+
+         // First query against the package — hydrates the
+         // ModelMaterializer from the worker's modelDef without a
+         // recompile, then runs the SQL against the *main thread's*
+         // DuckDB connection (the only one with the in-memory `nums`
+         // source loaded via duckdb.sql()).
+         const { result } = await model!.getQueryResults(
+            "nums",
+            "total_v",
+            undefined,
+         );
+         expect(result.data).toBeDefined();
+      } finally {
+         await duckdb.close();
+      }
+   });
+
+   it("propagates a per-model compile failure as a thrown error from Package.create", async () => {
+      writeManifest();
+      fs.writeFileSync(
+         path.join(tempDir, "broken.malloy"),
+         `source: bad is duckdb.sql("select 1 as a") extend {
+  measure: oops is THIS_FUNC_DOES_NOT_EXIST(a)
+}`,
+      );
+
+      const { malloyConfig, duckdb } = await makeMalloyConfig();
+      try {
+         await expect(
+            Package.create("env", "pkg", tempDir, malloyConfig),
+         ).rejects.toBeInstanceOf(Error);
+      } finally {
+         await duckdb.close();
+      }
+   });
+
+   // NB: kept last in this describe — swapping the singleton for a
+   // pre-shutdown pool also tears down the shared `pool` (the swap
+   // implementation shuts down the outgoing singleton). Subsequent
+   // tests in this describe would see a dead pool. afterAll only
+   // resets the singleton to null, so this is safe at the tail.
+   it("rewraps pool-infrastructure failures as ServiceUnavailableError (HTTP 503)", async () => {
+      writeManifest();
+      fs.writeFileSync(
+         path.join(tempDir, "trivial.malloy"),
+         `source: nums is duckdb.sql("select 1 as a")`,
+      );
+
+      const deadPool = new PackageLoadPool(1);
+      await deadPool.shutdown();
+      await __setPackageLoadPoolForTests(deadPool);
+
+      const { ServiceUnavailableError } = await import("../errors");
+      const { malloyConfig, duckdb } = await makeMalloyConfig();
+      try {
+         await expect(
+            Package.create("env", "pkg", tempDir, malloyConfig),
+         ).rejects.toBeInstanceOf(ServiceUnavailableError);
+      } finally {
+         await duckdb.close();
+      }
+   });
+});

package/tests/integration/concurrent_package/concurrent_package.integration.spec.ts

@@ -0,0 +1,280 @@
+/// <reference types="bun-types" />
+
+/**
+ * Regression test for the per-package download/unzip race.
+ *
+ * Before fix: concurrent POST/PATCH/GET-with-reload against the same package
+ * name landed in `downloadPackage` *before* acquiring the per-package mutex.
+ * Multiple callers would `rm -rf <targetPath>` and then `cp -r` / `extractAllTo`
+ * in parallel, leaving the directory in an inconsistent state. Subsequent
+ * reads failed with "Package manifest for ... does not exist" and any
+ * in-flight model compilation would 500 with "compiling model path not found".
+ *
+ * After fix: download is run inside the per-package mutex, so concurrent
+ * callers serialize on the same target path. All N requests must succeed
+ * and the package must be usable afterwards.
+ */
+
+import { afterAll, beforeAll, describe, expect, it } from "bun:test";
+import path from "path";
+import { fileURLToPath } from "url";
+import { RestE2EEnv, startRestE2E } from "../../harness/rest_e2e";
+
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = path.dirname(__filename);
+
+// Use a unique env name per run so stale state from a previous run (env
+// directory, persisted env metadata in publisher.db) can't poison startup.
+const ENV_NAME = `concurrent-package-test-env-${Date.now()}`;
+const PACKAGE_NAME = "gcs_faa";
+const CONCURRENCY = 12;
+
+const FORBIDDEN_ERROR_FRAGMENTS = [
+   "Package manifest for",
+   "does not exist",
+   "compiling model path not found",
+   "model path not found",
+];
+
+function findForbiddenError(body: unknown): string | undefined {
+   const text = typeof body === "string" ? body : JSON.stringify(body ?? "");
+   return FORBIDDEN_ERROR_FRAGMENTS.find((frag) => text.includes(frag));
+}
+
+describe("Concurrent package operations (E2E)", () => {
+   let env: (RestE2EEnv & { stop(): Promise<void> }) | null = null;
+   let baseUrl: string;
+   let fixtureDir: string;
+
+   beforeAll(async () => {
+      env = await startRestE2E();
+      baseUrl = env.baseUrl;
+      fixtureDir = "gs://publisher_test_packages/gcs_faa.zip";
+
+      // Seed the environment with the package once so the env exists on disk.
+      // addEnvironment requires at least one package.
+      const createRes = await fetch(`${baseUrl}/api/v0/environments`, {
+         method: "POST",
+         headers: { "Content-Type": "application/json" },
+         body: JSON.stringify({
+            name: ENV_NAME,
+            packages: [{ name: PACKAGE_NAME, location: fixtureDir }],
+            connections: [],
+         }),
+      });
+      if (!createRes.ok) {
+         const body = await createRes.text();
+         throw new Error(
+            `Failed to seed test environment (${createRes.status}): ${body}`,
+         );
+      }
+
+      // Wait for the seeded package to be loadable.
+      const deadline = Date.now() + 30_000;
+      let ready = false;
+      while (!ready && Date.now() < deadline) {
+         try {
+            const res = await fetch(
+               `${baseUrl}/api/v0/environments/${ENV_NAME}/packages/${PACKAGE_NAME}`,
+            );
+            if (res.ok) {
+               ready = true;
+               break;
+            }
+         } catch {
+            // not ready yet
+         }
+         await new Promise((r) => setTimeout(r, 250));
+      }
+      if (!ready) {
+         throw new Error("Seeded package did not become available in time");
+      }
+   });
+
+   afterAll(async () => {
+      if (baseUrl) {
+         try {
+            await fetch(`${baseUrl}/api/v0/environments/${ENV_NAME}`, {
+               method: "DELETE",
+            });
+         } catch {
+            // best-effort cleanup
+         }
+      }
+      await env?.stop();
+      env = null;
+   });
+
+   it("concurrent POST /packages for the same name all succeed", async () => {
+      const requests = Array.from({ length: CONCURRENCY }, () =>
+         fetch(`${baseUrl}/api/v0/environments/${ENV_NAME}/packages`, {
+            method: "POST",
+            headers: { "Content-Type": "application/json" },
+            body: JSON.stringify({
+               name: PACKAGE_NAME,
+               location: fixtureDir,
+            }),
+         }),
+      );
+
+      const responses = await Promise.all(requests);
+      const bodies = await Promise.all(
+         responses.map(async (r) => ({
+            status: r.status,
+            body: await r.json().catch(() => null),
+         })),
+      );
+
+      for (const { status, body } of bodies) {
+         expect(status).toBe(200);
+         const forbidden = findForbiddenError(body);
+         expect(forbidden).toBeUndefined();
+      }
+
+      // Package must be loadable after the storm.
+      const res = await fetch(
+         `${baseUrl}/api/v0/environments/${ENV_NAME}/packages/${PACKAGE_NAME}`,
+      );
+      expect(res.status).toBe(200);
+      const meta = (await res.json()) as { name?: string };
+      expect(meta.name).toBe(PACKAGE_NAME);
+
+      // Models must be listable — proves the model files survived the unzip
+      // race and Package.create read a consistent directory.
+      const modelsRes = await fetch(
+         `${baseUrl}/api/v0/environments/${ENV_NAME}/packages/${PACKAGE_NAME}/models`,
+      );
+      expect(modelsRes.status).toBe(200);
+      const models = (await modelsRes.json()) as Array<{ path?: string }>;
+      expect(Array.isArray(models)).toBe(true);
+      expect(models.length).toBeGreaterThan(0);
+   });
+
+   it("concurrent GET /packages/:name?reload=true all succeed", async () => {
+      const requests = Array.from({ length: CONCURRENCY }, () =>
+         fetch(
+            `${baseUrl}/api/v0/environments/${ENV_NAME}/packages/${PACKAGE_NAME}?reload=true`,
+         ),
+      );
+      const responses = await Promise.all(requests);
+      const bodies = await Promise.all(
+         responses.map(async (r) => ({
+            status: r.status,
+            body: await r.json().catch(() => null),
+         })),
+      );
+      for (const { status, body } of bodies) {
+         expect(status).toBe(200);
+         const forbidden = findForbiddenError(body);
+         expect(forbidden).toBeUndefined();
+      }
+
+      // Models must still list cleanly.
+      const modelsRes = await fetch(
+         `${baseUrl}/api/v0/environments/${ENV_NAME}/packages/${PACKAGE_NAME}/models`,
+      );
+      expect(modelsRes.status).toBe(200);
+   });
+
+   it("simultaneous POST + PATCH for the same package serialize cleanly", async () => {
+      // Fire create and update at the same time. Both should land under the
+      // same per-package mutex. Whichever wins the mutex first runs first;
+      // the loser sees a coherent post-condition. After both settle the
+      // package must be loadable and the description must reflect the PATCH
+      // when the PATCH ran *after* a successful POST.
+      const newDescription = `concurrent-update-${Date.now()}`;
+      const [postRes, patchRes] = await Promise.all([
+         fetch(`${baseUrl}/api/v0/environments/${ENV_NAME}/packages`, {
+            method: "POST",
+            headers: { "Content-Type": "application/json" },
+            body: JSON.stringify({
+               name: PACKAGE_NAME,
+               location: fixtureDir,
+            }),
+         }),
+         fetch(
+            `${baseUrl}/api/v0/environments/${ENV_NAME}/packages/${PACKAGE_NAME}`,
+            {
+               method: "PATCH",
+               headers: { "Content-Type": "application/json" },
+               body: JSON.stringify({
+                  name: PACKAGE_NAME,
+                  description: newDescription,
+               }),
+            },
+         ),
+      ]);
+
+      // POST is idempotent here — always succeeds.
+      expect(postRes.status).toBe(200);
+      const postBody = await postRes.json().catch(() => null);
+      expect(findForbiddenError(postBody)).toBeUndefined();
+
+      // PATCH either succeeds (ran after POST loaded the package) or 404s
+      // (ran before POST populated `this.packages`). It must NOT silently
+      // rewrite disk and then 404, and must never surface the forbidden
+      // error fragments.
+      const patchBody = await patchRes.json().catch(() => null);
+      expect(findForbiddenError(patchBody)).toBeUndefined();
+      expect([200, 404]).toContain(patchRes.status);
+
+      // Whatever happened, the package must remain loadable.
+      const getRes = await fetch(
+         `${baseUrl}/api/v0/environments/${ENV_NAME}/packages/${PACKAGE_NAME}`,
+      );
+      expect(getRes.status).toBe(200);
+      const meta = (await getRes.json()) as {
+         name?: string;
+         description?: string;
+      };
+      expect(meta.name).toBe(PACKAGE_NAME);
+   });
+
+   it("interleaved POST + GET-reload + model list never surface stale-dir errors", async () => {
+      // Worst-case interleave: writers and readers hammering the same
+      // package. None of them should observe a missing manifest or a
+      // half-extracted directory.
+      const work: Array<Promise<{ status: number; body: unknown }>> = [];
+      for (let i = 0; i < CONCURRENCY; i++) {
+         work.push(
+            fetch(`${baseUrl}/api/v0/environments/${ENV_NAME}/packages`, {
+               method: "POST",
+               headers: { "Content-Type": "application/json" },
+               body: JSON.stringify({
+                  name: PACKAGE_NAME,
+                  location: fixtureDir,
+               }),
+            }).then(async (r) => ({
+               status: r.status,
+               body: await r.json().catch(() => null),
+            })),
+         );
+         work.push(
+            fetch(
+               `${baseUrl}/api/v0/environments/${ENV_NAME}/packages/${PACKAGE_NAME}?reload=true`,
+            ).then(async (r) => ({
+               status: r.status,
+               body: await r.json().catch(() => null),
+            })),
+         );
+         work.push(
+            fetch(
+               `${baseUrl}/api/v0/environments/${ENV_NAME}/packages/${PACKAGE_NAME}/models`,
+            ).then(async (r) => ({
+               status: r.status,
+               body: await r.json().catch(() => null),
+            })),
+         );
+      }
+
+      const results = await Promise.all(work);
+      for (const { status, body } of results) {
+         const forbidden = findForbiddenError(body);
+         expect(forbidden).toBeUndefined();
+         // Model list while the package is being rewritten can transiently
+         // return 404 (the package is unloaded mid-rewrite), but never 5xx,
+         // and never with the forbidden error fragments.
+         expect(status).toBeLessThan(500);
+      }
+   });
+});