@malloy-publisher/server 0.0.198-dev → 0.0.198-dev2

package/dist/service/schema_worker.mjs

@@ -0,0 +1,61 @@
+// src/service/schema_worker.ts
+import { DuckDBConnection } from "@malloydata/db-duckdb";
+import"@malloydata/db-duckdb/native";
+import {
+  ConnectionRuntime,
+  EmptyURLReader
+} from "@malloydata/malloy";
+import * as path from "path";
+import { parentPort } from "worker_threads";
+if (!parentPort) {
+  throw new Error("schema_worker.ts loaded outside a worker thread");
+}
+var connection = new DuckDBConnection({
+  name: "duckdb",
+  databasePath: ":memory:",
+  threads: 1,
+  memoryLimit: "256MB"
+});
+async function handleRequest(req) {
+  try {
+    const fullPath = path.join(req.packagePath, req.databasePath);
+    const normalizedPath = fullPath.replace(/\\/g, "/");
+    const runtime = new ConnectionRuntime({
+      urlReader: new EmptyURLReader,
+      connections: [connection]
+    });
+    const model = runtime.loadModel(`source: temp is duckdb.table('${normalizedPath}')`);
+    const modelDef = await model.getModel();
+    const fields = modelDef._modelDef.contents["temp"].fields;
+    const columns = fields.map((field) => ({
+      type: String(field.type),
+      name: field.name
+    }));
+    const runner = model.loadQuery("run: temp->{aggregate: row_count is count()}");
+    const result = await runner.run();
+    const rowCount = result.data.value[0].row_count?.valueOf();
+    return {
+      id: req.id,
+      ok: true,
+      result: { name: req.databasePath, rowCount, columns }
+    };
+  } catch (err) {
+    const error = err instanceof Error ? err : new Error(String(err));
+    return {
+      id: req.id,
+      ok: false,
+      error: { message: error.message, stack: error.stack }
+    };
+  }
+}
+parentPort.on("message", async (msg) => {
+  const response = await handleRequest(msg);
+  parentPort.postMessage(response);
+});
+var shutdown = async () => {
+  try {
+    await connection.close();
+  } catch {}
+  process.exit(0);
+};
+parentPort.on("close", () => void shutdown());

package/package.json

@@ -1,7 +1,7 @@
 {
    "name": "@malloy-publisher/server",
    "description": "Malloy Publisher Server",
-   "version": "0.0.198-dev",
+   "version": "0.0.198-dev2",
    "main": "dist/server.mjs",
    "bin": {
       "malloy-publisher": "dist/server.mjs"
@@ -34,16 +34,16 @@
       "@azure/identity": "^4.13.0",
       "@azure/storage-blob": "^12.26.0",
       "@google-cloud/storage": "^7.16.0",
-      "@malloydata/db-bigquery": "^0.0.383",
-      "@malloydata/db-databricks": "^0.0.383",
-      "@malloydata/db-duckdb": "^0.0.383",
-      "@malloydata/db-mysql": "^0.0.383",
-      "@malloydata/db-postgres": "^0.0.383",
-      "@malloydata/db-snowflake": "^0.0.383",
-      "@malloydata/db-trino": "^0.0.383",
-      "@malloydata/malloy": "^0.0.383",
-      "@malloydata/malloy-sql": "^0.0.383",
-      "@malloydata/render-validator": "^0.0.383",
+      "@malloydata/db-bigquery": "^0.0.394",
+      "@malloydata/db-databricks": "^0.0.394",
+      "@malloydata/db-duckdb": "^0.0.394",
+      "@malloydata/db-mysql": "^0.0.394",
+      "@malloydata/db-postgres": "^0.0.394",
+      "@malloydata/db-snowflake": "^0.0.394",
+      "@malloydata/db-trino": "^0.0.394",
+      "@malloydata/malloy": "^0.0.394",
+      "@malloydata/malloy-sql": "^0.0.394",
+      "@malloydata/render-validator": "^0.0.394",
       "@modelcontextprotocol/sdk": "^1.13.2",
       "@opentelemetry/api": "^1.9.0",
       "@opentelemetry/auto-instrumentations-node": "^0.57.0",
@@ -68,7 +68,6 @@
       "node-cron": "^3.0.3",
       "recursive-readdir": "^2.2.3",
       "simple-git": "^3.28.0",
-      "trino": "^1.1.1",
       "uuid": "^11.0.3"
    },
    "devDependencies": {

package/publisher.config.example.bigquery.json

@@ -0,0 +1,33 @@
+{
+  "frozenConfig": false,
+  "environments": [
+    {
+      "name": "malloy-samples",
+      "packages": [
+        {
+          "name": "ecommerce",
+          "location": "https://github.com/credibledata/malloy-samples/tree/main/ecommerce"
+        },
+        {
+          "name": "imdb",
+          "location": "https://github.com/credibledata/malloy-samples/tree/main/imdb"
+        },
+        {
+          "name": "faa",
+          "location": "https://github.com/credibledata/malloy-samples/tree/main/faa"
+        },
+        {
+          "name": "bigquery-hackernews",
+          "location": "https://github.com/credibledata/malloy-samples/tree/main/bigquery-hackernews"
+        }
+      ],
+      "connections": [
+        {
+          "name": "bigquery",
+          "type": "bigquery",
+          "bigqueryConnection": {}
+        }
+      ]
+    }
+  ]
+}

package/publisher.config.example.duckdb.json

@@ -0,0 +1,23 @@
+{
+  "frozenConfig": false,
+  "environments": [
+    {
+      "name": "malloy-samples",
+      "packages": [
+        {
+          "name": "ecommerce",
+          "location": "https://github.com/credibledata/malloy-samples/tree/main/ecommerce"
+        },
+        {
+          "name": "imdb",
+          "location": "https://github.com/credibledata/malloy-samples/tree/main/imdb"
+        },
+        {
+          "name": "faa",
+          "location": "https://github.com/credibledata/malloy-samples/tree/main/faa"
+        }
+      ],
+      "connections": []
+    }
+  ]
+}

package/publisher.config.json

@@ -15,19 +15,9 @@
         {
           "name": "faa",
           "location": "https://github.com/credibledata/malloy-samples/tree/main/faa"
-        },
-        {
-          "name": "bigquery-hackernews",
-          "location": "https://github.com/credibledata/malloy-samples/tree/main/bigquery-hackernews"
         }
       ],
-      "connections": [
-        {
-          "name": "bigquery",
-          "type": "bigquery",
-          "bigqueryConnection": {}
-        }
-      ]
+      "connections": []
     }
   ]
 }

package/src/config.spec.ts

@@ -856,3 +856,309 @@ describe("Config Environment Variable Substitution", () => {
       });
    });
 });
+
+// TODO: Remove this during projects cleanup
+describe("Config legacy 'projects' key back-compat", () => {
+   const testServerRoot = path.join(process.cwd(), "test-temp-legacy-config");
+   const configPath = path.join(testServerRoot, PUBLISHER_CONFIG_NAME);
+
+   beforeEach(() => {
+      if (!fs.existsSync(testServerRoot)) {
+         fs.mkdirSync(testServerRoot, { recursive: true });
+      }
+   });
+
+   afterEach(() => {
+      if (fs.existsSync(configPath)) {
+         fs.unlinkSync(configPath);
+      }
+      if (fs.existsSync(testServerRoot)) {
+         fs.rmdirSync(testServerRoot, { recursive: true });
+      }
+   });
+
+   it("reads from legacy 'projects' key when 'environments' is absent", async () => {
+      // Pre-rename on-disk shape: top-level key is `projects`, not
+      // `environments`. Without back-compat this silently parses as empty.
+      const legacyConfig = {
+         frozenConfig: false,
+         projects: [
+            {
+               name: "legacy-env",
+               packages: [
+                  {
+                     name: "p1",
+                     location: "./packages/p1",
+                  },
+               ],
+            },
+         ],
+      };
+
+      fs.writeFileSync(configPath, JSON.stringify(legacyConfig, null, 2));
+
+      // Spy on logger.warn so we can assert the deprecation message fired.
+      const { logger } = await import("./logger");
+      const originalWarn = logger.warn;
+      const warnings: string[] = [];
+      logger.warn = ((msg: unknown, ..._rest: unknown[]) => {
+         warnings.push(typeof msg === "string" ? msg : String(msg));
+         return logger;
+      }) as typeof logger.warn;
+
+      try {
+         const result = getPublisherConfig(testServerRoot);
+
+         expect(result.environments.length).toBe(1);
+         expect(result.environments[0].name).toBe("legacy-env");
+         expect(result.environments[0].packages[0].name).toBe("p1");
+
+         expect(
+            warnings.some((w) => w.includes('uses deprecated "projects" key')),
+         ).toBe(true);
+      } finally {
+         logger.warn = originalWarn;
+      }
+   });
+
+   it("prefers the new 'environments' key when both are present", async () => {
+      const config = {
+         frozenConfig: false,
+         environments: [
+            {
+               name: "new-env",
+               packages: [{ name: "p1", location: "./packages/p1" }],
+            },
+         ],
+         projects: [
+            {
+               name: "should-be-ignored",
+               packages: [{ name: "p2", location: "./packages/p2" }],
+            },
+         ],
+      };
+
+      fs.writeFileSync(configPath, JSON.stringify(config, null, 2));
+
+      const { logger } = await import("./logger");
+      const originalWarn = logger.warn;
+      const warnings: string[] = [];
+      logger.warn = ((msg: unknown, ..._rest: unknown[]) => {
+         warnings.push(typeof msg === "string" ? msg : String(msg));
+         return logger;
+      }) as typeof logger.warn;
+
+      try {
+         const result = getPublisherConfig(testServerRoot);
+
+         expect(result.environments.length).toBe(1);
+         expect(result.environments[0].name).toBe("new-env");
+
+         // No deprecation warning should fire when `environments` is present.
+         expect(
+            warnings.some((w) => w.includes('uses deprecated "projects" key')),
+         ).toBe(false);
+      } finally {
+         logger.warn = originalWarn;
+      }
+   });
+});
+
+describe("Committed example configs", () => {
+   const serverDir = path.resolve(__dirname, "..");
+
+   it.each([
+      ["publisher.config.json", false],
+      ["publisher.config.example.duckdb.json", false],
+      ["publisher.config.example.bigquery.json", true],
+   ])(
+      "%s parses as a valid PublisherConfig",
+      (filename, expectsBigQueryConnection) => {
+         const filePath = path.join(serverDir, filename);
+         expect(fs.existsSync(filePath)).toBe(true);
+
+         const raw = fs.readFileSync(filePath, "utf-8");
+         const parsed = JSON.parse(raw) as PublisherConfig;
+
+         expect(parsed).toHaveProperty("environments");
+         expect(Array.isArray(parsed.environments)).toBe(true);
+         expect(parsed.environments.length).toBeGreaterThan(0);
+
+         const env = parsed.environments[0];
+         expect(env.name).toBeTruthy();
+         expect(Array.isArray(env.packages)).toBe(true);
+         expect(env.packages.length).toBeGreaterThan(0);
+
+         if (expectsBigQueryConnection) {
+            expect(
+               (env.connections ?? []).some((c) => c.type === "bigquery"),
+            ).toBe(true);
+            expect(
+               env.packages.some((p) => p.name === "bigquery-hackernews"),
+            ).toBe(true);
+         } else {
+            expect(
+               env.packages.some((p) => p.name === "bigquery-hackernews"),
+            ).toBe(false);
+         }
+      },
+   );
+});
+
+describe("Config path resolution (--config and bundled default)", () => {
+   const emptyServerRoot = path.join(process.cwd(), "test-empty-server-root");
+   const customConfigPath = path.join(
+      process.cwd(),
+      "test-custom-publisher.config.json",
+   );
+
+   beforeEach(() => {
+      if (!fs.existsSync(emptyServerRoot)) {
+         fs.mkdirSync(emptyServerRoot, { recursive: true });
+      }
+      delete process.env.PUBLISHER_CONFIG_PATH;
+      delete process.env.PUBLISHER_USE_BUNDLED_DEFAULT;
+   });
+
+   afterEach(() => {
+      delete process.env.PUBLISHER_CONFIG_PATH;
+      delete process.env.PUBLISHER_USE_BUNDLED_DEFAULT;
+      if (fs.existsSync(customConfigPath)) {
+         fs.unlinkSync(customConfigPath);
+      }
+      if (fs.existsSync(emptyServerRoot)) {
+         fs.rmSync(emptyServerRoot, { recursive: true, force: true });
+      }
+   });
+
+   it("falls back to the bundled default config when serverRoot has no publisher.config.json and bundled-default opt-in is set", () => {
+      // The bundled default is opt-in (server.ts sets the env var when
+      // the user passed neither --server_root nor --config) so embedded
+      // callers don't get a surprise filesystem read.
+      process.env.PUBLISHER_USE_BUNDLED_DEFAULT = "true";
+      const result = getPublisherConfig(emptyServerRoot);
+
+      expect(result.environments.length).toBeGreaterThan(0);
+      const env = result.environments[0];
+      expect(env.name).toBe("malloy-samples");
+      expect(env.packages.some((p) => p.name === "ecommerce")).toBe(true);
+      expect(env.packages.some((p) => p.name === "bigquery-hackernews")).toBe(
+         false,
+      );
+   });
+
+   it("does NOT fall back to the bundled default when opt-in flag is unset (programmatic construction)", () => {
+      // No PUBLISHER_USE_BUNDLED_DEFAULT, no PUBLISHER_CONFIG_PATH, no
+      // file in emptyServerRoot — result should be the original empty
+      // shape, preserving prior behavior for embeds and tests.
+      const result = getPublisherConfig(emptyServerRoot);
+      expect(result.environments).toEqual([]);
+   });
+
+   it("honors PUBLISHER_CONFIG_PATH (set by --config) over the server root", () => {
+      const customConfig = {
+         frozenConfig: false,
+         environments: [
+            { name: "custom-env", packages: [{ name: "foo", location: "/x" }] },
+         ],
+      };
+      fs.writeFileSync(customConfigPath, JSON.stringify(customConfig));
+      process.env.PUBLISHER_CONFIG_PATH = customConfigPath;
+
+      const result = getPublisherConfig(emptyServerRoot);
+      expect(result.environments[0].name).toBe("custom-env");
+   });
+
+   it("returns empty environments (not the bundled default) when --config points at a missing file", () => {
+      process.env.PUBLISHER_CONFIG_PATH = path.join(
+         process.cwd(),
+         "does-not-exist.json",
+      );
+      // Even with bundled-default opt-in, --config with a missing target
+      // is a user error and we don't paper over it.
+      process.env.PUBLISHER_USE_BUNDLED_DEFAULT = "true";
+      const result = getPublisherConfig(emptyServerRoot);
+      expect(result.environments).toEqual([]);
+   });
+});
+
+describe("getMemoryGovernorConfig", () => {
+   const GOVERNOR_ENV_VARS = [
+      "PUBLISHER_MAX_MEMORY_BYTES",
+      "PUBLISHER_MEMORY_HIGH_WATER_FRACTION",
+      "PUBLISHER_MEMORY_LOW_WATER_FRACTION",
+      "PUBLISHER_MEMORY_CHECK_INTERVAL_MS",
+      "PUBLISHER_MEMORY_BACKPRESSURE",
+   ];
+
+   beforeEach(() => {
+      for (const v of GOVERNOR_ENV_VARS) delete process.env[v];
+   });
+   afterEach(() => {
+      for (const v of GOVERNOR_ENV_VARS) delete process.env[v];
+   });
+
+   it("returns null when PUBLISHER_MAX_MEMORY_BYTES is unset", async () => {
+      const { getMemoryGovernorConfig } = await import("./config");
+      expect(getMemoryGovernorConfig()).toBeNull();
+   });
+
+   it("parses defaults when only PUBLISHER_MAX_MEMORY_BYTES is set", async () => {
+      process.env.PUBLISHER_MAX_MEMORY_BYTES = String(2 * 1024 * 1024 * 1024);
+      const { getMemoryGovernorConfig } = await import("./config");
+      const cfg = getMemoryGovernorConfig();
+      expect(cfg).not.toBeNull();
+      expect(cfg!.maxMemoryBytes).toBe(2 * 1024 * 1024 * 1024);
+      expect(cfg!.backpressureEnabled).toBe(true);
+      expect(cfg!.highWaterFraction).toBeGreaterThan(cfg!.lowWaterFraction);
+   });
+
+   it("honours fraction and interval overrides", async () => {
+      process.env.PUBLISHER_MAX_MEMORY_BYTES = "1000000000";
+      process.env.PUBLISHER_MEMORY_HIGH_WATER_FRACTION = "0.85";
+      process.env.PUBLISHER_MEMORY_LOW_WATER_FRACTION = "0.7";
+      process.env.PUBLISHER_MEMORY_CHECK_INTERVAL_MS = "10000";
+      process.env.PUBLISHER_MEMORY_BACKPRESSURE = "false";
+      const { getMemoryGovernorConfig } = await import("./config");
+      const cfg = getMemoryGovernorConfig();
+      expect(cfg).not.toBeNull();
+      expect(cfg!.highWaterFraction).toBe(0.85);
+      expect(cfg!.lowWaterFraction).toBe(0.7);
+      expect(cfg!.checkIntervalMs).toBe(10000);
+      expect(cfg!.backpressureEnabled).toBe(false);
+   });
+
+   it("treats PUBLISHER_MAX_MEMORY_BYTES=0 as disabled (returns null)", async () => {
+      process.env.PUBLISHER_MAX_MEMORY_BYTES = "0";
+      const { getMemoryGovernorConfig } = await import("./config");
+      expect(getMemoryGovernorConfig()).toBeNull();
+   });
+
+   it("rejects a negative PUBLISHER_MAX_MEMORY_BYTES", async () => {
+      process.env.PUBLISHER_MAX_MEMORY_BYTES = "-1";
+      const { getMemoryGovernorConfig } = await import("./config");
+      expect(() => getMemoryGovernorConfig()).toThrow();
+   });
+
+   it("rejects low >= high", async () => {
+      process.env.PUBLISHER_MAX_MEMORY_BYTES = "1000000000";
+      process.env.PUBLISHER_MEMORY_HIGH_WATER_FRACTION = "0.7";
+      process.env.PUBLISHER_MEMORY_LOW_WATER_FRACTION = "0.8";
+      const { getMemoryGovernorConfig } = await import("./config");
+      expect(() => getMemoryGovernorConfig()).toThrow();
+   });
+
+   it("rejects an out-of-range fraction", async () => {
+      process.env.PUBLISHER_MAX_MEMORY_BYTES = "1000000000";
+      process.env.PUBLISHER_MEMORY_HIGH_WATER_FRACTION = "1.5";
+      const { getMemoryGovernorConfig } = await import("./config");
+      expect(() => getMemoryGovernorConfig()).toThrow();
+   });
+
+   it("rejects a check interval below the safety floor", async () => {
+      process.env.PUBLISHER_MAX_MEMORY_BYTES = "1000000000";
+      process.env.PUBLISHER_MEMORY_CHECK_INTERVAL_MS = "10";
+      const { getMemoryGovernorConfig } = await import("./config");
+      expect(() => getMemoryGovernorConfig()).toThrow();
+   });
+});