@datafog/fogclaw 0.2.0 → 0.3.0

package/tests/gliner.test.ts

@@ -1,289 +0,0 @@
-import { beforeAll, beforeEach, afterAll, describe, it, expect, vi } from "vitest";
-import fs from "node:fs/promises";
-import os from "node:os";
-import path from "node:path";
-
-// Mock the gliner npm package so we don't need the actual 1.4GB model
-vi.mock("gliner", () => {
-  class MockGliner {
-    private config: any;
-
-    constructor(config: any) {
-      this.config = config;
-    }
-
-    async initialize(): Promise<void> {
-      // No-op in mock
-    }
-
-    async inference(
-      request: { texts: string[]; entities: string[] } | string | string[],
-      maybeEntities?: string[],
-      _flatNer = false,
-      _threshold = 0.5,
-    ): Promise<Array<{ text: string; label: string; score: number; start: number; end: number }>> {
-      const text =
-        typeof request === "string"
-          ? request
-          : Array.isArray(request)
-            ? request[0] ?? ""
-            : request.texts[0] ?? "";
-      const requestEntities =
-        typeof request === "object" && request !== null && "entities" in request
-          ? request.entities
-          : undefined;
-      const labels =
-        Array.isArray(maybeEntities)
-          ? maybeEntities
-          : requestEntities ?? [];
-      const results: Array<{ text: string; label: string; score: number; start: number; end: number }> = [];
-
-      // Simulate entity detection for "John Smith"
-      const johnIndex = text.indexOf("John Smith");
-      if (johnIndex !== -1 && labels.includes("person")) {
-        results.push({
-          text: "John Smith",
-          label: "person",
-          score: 0.95,
-          start: johnIndex,
-          end: johnIndex + "John Smith".length,
-        });
-      }
-
-      // Simulate entity detection for "Acme Corp"
-      const acmeIndex = text.indexOf("Acme Corp");
-      if (acmeIndex !== -1 && labels.includes("organization")) {
-        results.push({
-          text: "Acme Corp",
-          label: "organization",
-          score: 0.88,
-          start: acmeIndex,
-          end: acmeIndex + "Acme Corp".length,
-        });
-      }
-
-      // Simulate entity detection for "New York"
-      const nyIndex = text.indexOf("New York");
-      if (nyIndex !== -1 && labels.includes("location")) {
-        results.push({
-          text: "New York",
-          label: "location",
-          score: 0.91,
-          start: nyIndex,
-          end: nyIndex + "New York".length,
-        });
-      }
-
-      return results;
-    }
-  }
-
-  return { Gliner: MockGliner };
-});
-
-vi.mock("gliner/node", () => {
-  class MockGliner {
-    private config: any;
-
-    constructor(config: any) {
-      this.config = config;
-    }
-
-    async initialize(): Promise<void> {
-      // No-op in mock
-    }
-
-    async inference(
-      request: { texts: string[]; entities: string[] } | string | string[],
-      maybeEntities?: string[],
-      _flatNer = false,
-      _threshold = 0.5,
-    ): Promise<Array<{ text: string; label: string; score: number; start: number; end: number }>> {
-      const text =
-        typeof request === "string"
-          ? request
-          : Array.isArray(request)
-            ? request[0] ?? ""
-            : request.texts[0] ?? "";
-      const requestEntities =
-        typeof request === "object" && request !== null && "entities" in request
-          ? request.entities
-          : undefined;
-      const labels =
-        Array.isArray(maybeEntities)
-          ? maybeEntities
-          : requestEntities ?? [];
-      const results: Array<{ text: string; label: string; score: number; start: number; end: number }> = [];
-
-      // Simulate entity detection for "John Smith"
-      const johnIndex = text.indexOf("John Smith");
-      if (johnIndex !== -1 && labels.includes("person")) {
-        results.push({
-          text: "John Smith",
-          label: "person",
-          score: 0.95,
-          start: johnIndex,
-          end: johnIndex + "John Smith".length,
-        });
-      }
-
-      // Simulate entity detection for "Acme Corp"
-      const acmeIndex = text.indexOf("Acme Corp");
-      if (acmeIndex !== -1 && labels.includes("organization")) {
-        results.push({
-          text: "Acme Corp",
-          label: "organization",
-          score: 0.88,
-          start: acmeIndex,
-          end: acmeIndex + "Acme Corp".length,
-        });
-      }
-
-      // Simulate entity detection for "New York"
-      const nyIndex = text.indexOf("New York");
-      if (nyIndex !== -1 && labels.includes("location")) {
-        results.push({
-          text: "New York",
-          label: "location",
-          score: 0.91,
-          start: nyIndex,
-          end: nyIndex + "New York".length,
-        });
-      }
-
-      return results;
-    }
-  }
-
-  return { Gliner: MockGliner };
-});
-
-import { GlinerEngine } from "../src/engines/gliner.js";
-
-const TEST_ONNX_MODEL_PATH = path.join(os.tmpdir(), "fogclaw-gliner-model-test.onnx");
-
-beforeAll(async () => {
-  await fs.writeFile(TEST_ONNX_MODEL_PATH, "mock-onnx-model", "utf8");
-});
-
-afterAll(async () => {
-  await fs.unlink(TEST_ONNX_MODEL_PATH).catch(() => undefined);
-});
-
-describe("GlinerEngine", () => {
-  let engine: GlinerEngine;
-
-  beforeEach(async () => {
-    engine = new GlinerEngine(TEST_ONNX_MODEL_PATH, 0.5);
-    await engine.initialize();
-  });
-
-  it("detects person entities with canonical PERSON label", async () => {
-    const entities = await engine.scan("My name is John Smith and I live here.");
-
-    expect(entities).toHaveLength(1);
-    expect(entities[0].text).toBe("John Smith");
-    expect(entities[0].label).toBe("PERSON");
-  });
-
-  it("detects organization entities with canonical ORGANIZATION label", async () => {
-    const entities = await engine.scan("I work at Acme Corp downtown.");
-
-    expect(entities).toHaveLength(1);
-    expect(entities[0].text).toBe("Acme Corp");
-    expect(entities[0].label).toBe("ORGANIZATION");
-  });
-
-  it("detects multiple entity types in the same text", async () => {
-    const entities = await engine.scan(
-      "John Smith works at Acme Corp in New York.",
-    );
-
-    expect(entities).toHaveLength(3);
-
-    const labels = entities.map((e) => e.label);
-    expect(labels).toContain("PERSON");
-    expect(labels).toContain("ORGANIZATION");
-    expect(labels).toContain("LOCATION");
-  });
-
-  it("returns empty array for text with no entities", async () => {
-    const entities = await engine.scan("Hello world, this is a test.");
-
-    expect(entities).toEqual([]);
-  });
-
-  it("returns empty array for empty string input", async () => {
-    const entities = await engine.scan("");
-
-    expect(entities).toEqual([]);
-  });
-
-  it("allows setting custom labels without crashing", async () => {
-    expect(() => engine.setCustomLabels(["product", "event"])).not.toThrow();
-
-    // Scan still works after setting custom labels
-    const entities = await engine.scan("John Smith attended the event.");
-    expect(entities).toHaveLength(1);
-    expect(entities[0].label).toBe("PERSON");
-  });
-
-  it("applies canonical type mapping (lowercase person -> PERSON)", async () => {
-    // The mock returns lowercase "person" as label; canonicalType should map it to "PERSON"
-    const entities = await engine.scan("John Smith is here.");
-
-    expect(entities[0].label).toBe("PERSON");
-    // Verify it's not lowercase
-    expect(entities[0].label).not.toBe("person");
-  });
-
-  it("sets source to gliner for all detected entities", async () => {
-    const entities = await engine.scan(
-      "John Smith works at Acme Corp in New York.",
-    );
-
-    for (const entity of entities) {
-      expect(entity.source).toBe("gliner");
-    }
-  });
-
-  it("confidence comes from model score", async () => {
-    const entities = await engine.scan(
-      "John Smith works at Acme Corp in New York.",
-    );
-
-    const person = entities.find((e) => e.label === "PERSON");
-    const org = entities.find((e) => e.label === "ORGANIZATION");
-    const loc = entities.find((e) => e.label === "LOCATION");
-
-    // These match the scores set in our mock
-    expect(person?.confidence).toBe(0.95);
-    expect(org?.confidence).toBe(0.88);
-    expect(loc?.confidence).toBe(0.91);
-  });
-
-  it("throws if scan is called before initialize", async () => {
-    const uninitializedEngine = new GlinerEngine("some-model", 0.5);
-
-    await expect(uninitializedEngine.scan("test")).rejects.toThrow(
-      "GLiNER engine not initialized. Call initialize() first.",
-    );
-  });
-
-  it("reports isInitialized correctly", async () => {
-    const freshEngine = new GlinerEngine(TEST_ONNX_MODEL_PATH, 0.5);
-    expect(freshEngine.isInitialized).toBe(false);
-
-    await freshEngine.initialize();
-    expect(freshEngine.isInitialized).toBe(true);
-  });
-
-  it("includes correct start and end offsets", async () => {
-    const text = "Contact John Smith for details.";
-    const entities = await engine.scan(text);
-
-    expect(entities).toHaveLength(1);
-    expect(entities[0].start).toBe(8); // "Contact " is 8 chars
-    expect(entities[0].end).toBe(18); // 8 + "John Smith".length = 18
-  });
-});

package/tests/message-sending-handler.test.ts

@@ -1,244 +0,0 @@
-import { describe, it, expect, vi, beforeAll, afterAll } from "vitest";
-import { createMessageSendingHandler } from "../src/message-sending-handler.js";
-import { Scanner } from "../src/scanner.js";
-import type { FogClawConfig } from "../src/types.js";
-
-function makeConfig(overrides: Partial<FogClawConfig> = {}): FogClawConfig {
-  return {
-    enabled: true,
-    guardrail_mode: "redact",
-    redactStrategy: "token",
-    model: "invalid:/not/real/model",
-    confidence_threshold: 0.5,
-    custom_entities: [],
-    entityActions: {},
-    entityConfidenceThresholds: {},
-    allowlist: { values: [], patterns: [], entities: {} },
-    auditEnabled: false,
-    ...overrides,
-  };
-}
-
-function makeLogger() {
-  return {
-    info: vi.fn(),
-    warn: vi.fn(),
-  };
-}
-
-function makeCtx(channelId = "telegram") {
-  return { channelId };
-}
-
-describe("createMessageSendingHandler", () => {
-  // Suppress GLiNER init warnings
-  beforeAll(() => {
-    vi.spyOn(console, "warn").mockImplementation(() => undefined);
-  });
-  afterAll(() => {
-    vi.restoreAllMocks();
-  });
-
-  it("returns an async function", () => {
-    const config = makeConfig();
-    const scanner = new Scanner(config);
-    const handler = createMessageSendingHandler(config, scanner);
-    expect(typeof handler).toBe("function");
-  });
-
-  it("redacts SSN in outbound message", async () => {
-    const config = makeConfig();
-    const scanner = new Scanner(config);
-    const handler = createMessageSendingHandler(config, scanner);
-
-    const result = await handler(
-      { to: "user123", content: "Your SSN is 123-45-6789." },
-      makeCtx(),
-    );
-
-    expect(result).toBeDefined();
-    expect(result!.content).toContain("[SSN_1]");
-    expect(result!.content).not.toContain("123-45-6789");
-    expect(result!.cancel).toBeUndefined();
-  });
-
-  it("redacts email and phone in outbound message", async () => {
-    const config = makeConfig();
-    const scanner = new Scanner(config);
-    const handler = createMessageSendingHandler(config, scanner);
-
-    const result = await handler(
-      { to: "user", content: "Call 555-123-4567 or email john@example.com" },
-      makeCtx(),
-    );
-
-    expect(result).toBeDefined();
-    expect(result!.content).toContain("[PHONE_1]");
-    expect(result!.content).toContain("[EMAIL_1]");
-    expect(result!.content).not.toContain("555-123-4567");
-    expect(result!.content).not.toContain("john@example.com");
-  });
-
-  it("returns void when no PII found", async () => {
-    const config = makeConfig();
-    const scanner = new Scanner(config);
-    const handler = createMessageSendingHandler(config, scanner);
-
-    const result = await handler(
-      { to: "user", content: "Hello, how can I help you today?" },
-      makeCtx(),
-    );
-
-    expect(result).toBeUndefined();
-  });
-
-  it("returns void for empty content", async () => {
-    const config = makeConfig();
-    const scanner = new Scanner(config);
-    const handler = createMessageSendingHandler(config, scanner);
-
-    const result = await handler(
-      { to: "user", content: "" },
-      makeCtx(),
-    );
-
-    expect(result).toBeUndefined();
-  });
-
-  it("never returns cancel: true", async () => {
-    const config = makeConfig({
-      entityActions: { SSN: "block" },
-    });
-    const scanner = new Scanner(config);
-    const handler = createMessageSendingHandler(config, scanner);
-
-    const result = await handler(
-      { to: "user", content: "SSN 123-45-6789" },
-      makeCtx(),
-    );
-
-    expect(result).toBeDefined();
-    expect(result!.cancel).toBeUndefined();
-    expect(result!.content).toContain("[SSN_1]");
-  });
-
-  it("all guardrail modes produce span-level redaction", async () => {
-    const config = makeConfig({
-      entityActions: { SSN: "block", EMAIL: "warn", PHONE: "redact" },
-    });
-    const scanner = new Scanner(config);
-    const handler = createMessageSendingHandler(config, scanner);
-
-    const result = await handler(
-      {
-        to: "user",
-        content: "SSN 123-45-6789, email john@example.com, call 555-123-4567",
-      },
-      makeCtx(),
-    );
-
-    expect(result).toBeDefined();
-    expect(result!.content).toContain("[SSN_1]");
-    expect(result!.content).toContain("[EMAIL_1]");
-    expect(result!.content).toContain("[PHONE_1]");
-  });
-
-  it("respects allowlist — global values", async () => {
-    const config = makeConfig({
-      allowlist: {
-        values: ["noreply@example.com"],
-        patterns: [],
-        entities: {},
-      },
-    });
-    const scanner = new Scanner(config);
-    const handler = createMessageSendingHandler(config, scanner);
-
-    const result = await handler(
-      { to: "user", content: "Contact noreply@example.com for help" },
-      makeCtx(),
-    );
-
-    expect(result).toBeUndefined();
-  });
-
-  it("uses mask redaction strategy", async () => {
-    const config = makeConfig({ redactStrategy: "mask" });
-    const scanner = new Scanner(config);
-    const handler = createMessageSendingHandler(config, scanner);
-
-    const result = await handler(
-      { to: "user", content: "SSN is 123-45-6789" },
-      makeCtx(),
-    );
-
-    expect(result).toBeDefined();
-    expect(result!.content).toContain("***********");
-    expect(result!.content).not.toContain("123-45-6789");
-  });
-
-  it("uses hash redaction strategy", async () => {
-    const config = makeConfig({ redactStrategy: "hash" });
-    const scanner = new Scanner(config);
-    const handler = createMessageSendingHandler(config, scanner);
-
-    const result = await handler(
-      { to: "user", content: "SSN is 123-45-6789" },
-      makeCtx(),
-    );
-
-    expect(result).toBeDefined();
-    expect(result!.content).toMatch(/\[SSN_[a-f0-9]{12}\]/);
-  });
-
-  describe("audit logging", () => {
-    it("emits audit log with source outbound when PII found", async () => {
-      const config = makeConfig({ auditEnabled: true });
-      const scanner = new Scanner(config);
-      const logger = makeLogger();
-      const handler = createMessageSendingHandler(config, scanner, logger);
-
-      await handler(
-        { to: "user", content: "SSN 123-45-6789" },
-        makeCtx("discord"),
-      );
-
-      expect(logger.info).toHaveBeenCalledOnce();
-      const logCall = logger.info.mock.calls[0][0] as string;
-      expect(logCall).toContain("[FOGCLAW AUDIT]");
-      expect(logCall).toContain("outbound_scan");
-      expect(logCall).toContain('"source":"outbound"');
-      expect(logCall).toContain('"channelId":"discord"');
-      expect(logCall).toContain('"SSN"');
-      expect(logCall).not.toContain("123-45-6789");
-    });
-
-    it("does not emit audit log when auditEnabled is false", async () => {
-      const config = makeConfig({ auditEnabled: false });
-      const scanner = new Scanner(config);
-      const logger = makeLogger();
-      const handler = createMessageSendingHandler(config, scanner, logger);
-
-      await handler(
-        { to: "user", content: "SSN 123-45-6789" },
-        makeCtx(),
-      );
-
-      expect(logger.info).not.toHaveBeenCalled();
-    });
-
-    it("does not emit audit log when no PII found", async () => {
-      const config = makeConfig({ auditEnabled: true });
-      const scanner = new Scanner(config);
-      const logger = makeLogger();
-      const handler = createMessageSendingHandler(config, scanner, logger);
-
-      await handler(
-        { to: "user", content: "Clean message" },
-        makeCtx(),
-      );
-
-      expect(logger.info).not.toHaveBeenCalled();
-    });
-  });
-});