@talkpilot/core-db 1.3.77 → 1.3.78

package/src/websitalk/scans/__tests__/scans.spec.ts

@@ -1,362 +1,362 @@
-import {
-  getScansCollection,
-  getScansByClientAndDateRange,
-  getScanFieldsByClientAndDateRange,
-  findScansByQuery,
-  countScans,
-  createScanDoc,
-  updateScanDoc,
-  findInProgressScansByClient,
-  getScansByWebsiteUrl,
-  getLatestScanByWebsiteUrl,
-  getInProgressScanByWebsiteUrl,
-  getLatestCompletedScanByWebsiteUrl,
-  deleteScansByWebsiteUrl,
-  deleteScanById,
-} from "../scans.getters";
-import { FINISHED_SCAN_STATUSES, SCAN_STATUSES } from "../scans.constants";
-import { createScanFixture } from "../../../test-utils/factories";
-import { ObjectId } from "mongodb";
-
-describe("db.websitalk.scans", () => {
-  describe("getScansByClientAndDateRange()", () => {
-    it("should return full scan documents within the date range", async () => {
-      const clientId = "dateRangeClient";
-      const startDate = new Date("2023-05-01");
-      const endDate = new Date("2023-05-31");
-
-      const scanInside = createScanFixture({ clientId, status: "COMPLETED" });
-      const scanBefore = createScanFixture({ clientId });
-      const scanAfter = createScanFixture({ clientId });
-
-      await getScansCollection().insertMany([
-        { ...scanInside, createdAt: new Date("2023-05-15") },
-        { ...scanBefore, createdAt: new Date("2023-04-30") },
-        { ...scanAfter, createdAt: new Date("2023-06-01") },
-      ]);
-
-      const result = await getScansByClientAndDateRange(
-        clientId,
-        startDate,
-        endDate,
-      );
-
-      expect(result.length).toBe(1);
-      expect(result[0]).toMatchObject({
-        clientId,
-        websiteUrlId: scanInside.websiteUrlId,
-        status: "COMPLETED",
-      });
-      expect(result[0]).toHaveProperty("siteChrome");
-      expect(result[0]).toHaveProperty("skipped");
-    });
-
-    it("should respect the date range boundaries (inclusive)", async () => {
-      const clientId = "boundaryClient";
-      const onStart = createScanFixture({ clientId });
-      const onEnd = createScanFixture({ clientId });
-
-      await getScansCollection().insertMany([
-        { ...onStart, createdAt: new Date("2023-05-01") },
-        { ...onEnd, createdAt: new Date("2023-05-31") },
-      ]);
-
-      const result = await getScansByClientAndDateRange(
-        clientId,
-        new Date("2023-05-01"),
-        new Date("2023-05-31"),
-      );
-
-      expect(result.length).toBe(2);
-    });
-  });
-
-  describe("getScanFieldsByClientAndDateRange()", () => {
-    it("should return only the requested fields (plus _id)", async () => {
-      const clientId = "projectionClient";
-      const scan = createScanFixture({
-        clientId,
-        status: "COMPLETED",
-        siteChrome: { url: "https://example.com", sourceType: "site_chrome" },
-        skipped: [{ path: "/broken", reason: "HTTP 500" }],
-      });
-
-      await getScansCollection().insertOne({
-        ...scan,
-        createdAt: new Date("2023-05-15"),
-      });
-
-      const [result] = await getScanFieldsByClientAndDateRange(
-        clientId,
-        new Date("2023-05-01"),
-        new Date("2023-05-31"),
-        ["clientId", "websiteUrlId", "status"],
-      );
-
-      expect(result).toMatchObject({
-        clientId,
-        websiteUrlId: scan.websiteUrlId,
-        status: "COMPLETED",
-      });
-      expect(result).toHaveProperty("_id");
-      expect(result).not.toHaveProperty("siteChrome");
-      expect(result).not.toHaveProperty("skipped");
-      expect(result).not.toHaveProperty("createdAt");
-    });
-  });
-
-  describe("findScansByQuery() / countScans()", () => {
-    it("should find scans matching a raw filter and count them", async () => {
-      const clientId = "queryClient";
-      const completed = createScanFixture({ clientId, status: "COMPLETED" });
-      const scanning = createScanFixture({ clientId, status: "SCANNING" });
-
-      await getScansCollection().insertMany([completed, scanning]);
-
-      const result = await findScansByQuery({ clientId, status: "COMPLETED" });
-      const count = await countScans({ clientId });
-
-      expect(result.length).toBe(1);
-      expect(result[0].websiteUrlId).toBe(completed.websiteUrlId);
-      expect(count).toBe(2);
-    });
-  });
-
-  describe("createScanDoc()", () => {
-    it("should insert a scan with default values for the rest of the fields", async () => {
-      const clientId = "createClient";
-      const websiteUrlId = new ObjectId().toString();
-
-      const { insertedId } = await createScanDoc({ clientId, websiteUrlId });
-
-      const result = await getScansCollection().findOne({ _id: insertedId });
-
-      expect(result).toMatchObject({
-        clientId,
-        websiteUrlId,
-        status: "CHECKING",
-        activeTasksCount: 0,
-        pagesScanned: 0,
-        pagesSkipped: 0,
-        skipped: [],
-        siteChrome: {},
-        pausedAt: null,
-      });
-      expect(result?.createdAt).toBeInstanceOf(Date);
-      expect(result?.updatedAt).toBeInstanceOf(Date);
-    });
-  });
-
-  describe("pausedAt", () => {
-    it("is set on pause and cleared back to null on resume, via updateScanDoc", async () => {
-      const scan = createScanFixture({ status: "SCANNING", pausedAt: null });
-      const { insertedId } = await getScansCollection().insertOne(scan);
-
-      const pausedAt = new Date();
-      const paused = await updateScanDoc(insertedId, {
-        status: "PAUSED",
-        pausedAt,
-      });
-      expect(paused?.pausedAt).toEqual(pausedAt);
-
-      const resumed = await updateScanDoc(insertedId, {
-        status: "SCANNING",
-        pausedAt: null,
-      });
-      expect(resumed?.pausedAt).toBeNull();
-    });
-  });
-
-  describe("updateScanDoc()", () => {
-    it("should update the given fields and bump updatedAt", async () => {
-      const scan = createScanFixture({ status: "SCANNING", pagesScanned: 0 });
-      const { insertedId } = await getScansCollection().insertOne(scan);
-
-      const result = await updateScanDoc(insertedId, {
-        status: "COMPLETED",
-        pagesScanned: 10,
-      });
-
-      expect(result).toMatchObject({
-        _id: insertedId,
-        status: "COMPLETED",
-        pagesScanned: 10,
-      });
-      expect(result?.updatedAt.getTime()).toBeGreaterThan(
-        scan.updatedAt.getTime(),
-      );
-    });
-
-    it("should return null when the scan does not exist", async () => {
-      const result = await updateScanDoc(new ObjectId(), { status: "FAILED" });
-
-      expect(result).toBeNull();
-    });
-  });
-
-  describe("in-progress scan helpers", () => {
-    const websiteUrlId = "website-123";
-
-    it("should return website scan history newest first", async () => {
-      await getScansCollection().insertMany([
-        createScanFixture({
-          websiteUrlId,
-          status: "COMPLETED",
-          createdAt: new Date("2023-05-01"),
-        }),
-        createScanFixture({
-          websiteUrlId,
-          status: "COMPLETED",
-          createdAt: new Date("2023-05-15"),
-        }),
-      ]);
-
-      const scans = await getScansByWebsiteUrl(websiteUrlId);
-      expect(scans).toHaveLength(2);
-      expect(scans[0].createdAt.getTime()).toBeGreaterThan(
-        scans[1].createdAt.getTime(),
-      );
-    });
-
-    it("should return the newest scan for a website", async () => {
-      await getScansCollection().insertMany([
-        createScanFixture({
-          websiteUrlId,
-          status: "COMPLETED",
-          createdAt: new Date("2023-05-01"),
-        }),
-        createScanFixture({
-          websiteUrlId,
-          status: "FAILED",
-          createdAt: new Date("2023-05-15"),
-        }),
-      ]);
-
-      const latest = await getLatestScanByWebsiteUrl(websiteUrlId);
-      expect(latest?.status).toBe("FAILED");
-    });
-
-    it("should return in-progress scan and return null when only completed exists", async () => {
-      await getScansCollection().insertMany([
-        createScanFixture({
-          websiteUrlId,
-          status: "COMPLETED",
-          createdAt: new Date("2023-05-01"),
-        }),
-        createScanFixture({
-          websiteUrlId,
-          status: "SCANNING",
-          createdAt: new Date("2023-05-15"),
-        }),
-      ]);
-
-      const inProgressScan = await getInProgressScanByWebsiteUrl(websiteUrlId);
-      expect(inProgressScan?.status).toBe("SCANNING");
-
-      await getScansCollection().deleteMany({ websiteUrlId });
-      await getScansCollection().insertOne(
-        createScanFixture({
-          websiteUrlId,
-          status: "COMPLETED",
-          createdAt: new Date("2023-05-20"),
-        }),
-      );
-
-      expect(await getInProgressScanByWebsiteUrl(websiteUrlId)).toBeNull();
-    });
-
-    it("should return all in-progress scans across websites for a client", async () => {
-      const clientId = "multiScanClient";
-      await getScansCollection().insertMany([
-        createScanFixture({
-          clientId,
-          websiteUrlId: "website-a",
-          status: "SCANNING",
-        }),
-        createScanFixture({
-          clientId,
-          websiteUrlId: "website-b",
-          status: "PROCESSING",
-        }),
-        createScanFixture({
-          clientId,
-          websiteUrlId: "website-c",
-          status: "COMPLETED",
-        }),
-      ]);
-
-      const scans = await findInProgressScansByClient(clientId);
-      expect(scans).toHaveLength(2);
-      expect(scans.map((s) => s.websiteUrlId).sort()).toEqual([
-        "website-a",
-        "website-b",
-      ]);
-    });
-  });
-
-  describe("website delete helpers", () => {
-    it("should return the newest completed scan only", async () => {
-      const websiteUrlId = new ObjectId().toString();
-      await getScansCollection().insertMany([
-        createScanFixture({
-          websiteUrlId,
-          status: "FAILED",
-          createdAt: new Date("2023-05-25"),
-        }),
-        createScanFixture({
-          websiteUrlId,
-          status: "COMPLETED",
-          createdAt: new Date("2023-05-20"),
-        }),
-        createScanFixture({
-          websiteUrlId,
-          status: "COMPLETED",
-          createdAt: new Date("2023-05-22"),
-        }),
-      ]);
-
-      const latest = await getLatestCompletedScanByWebsiteUrl(websiteUrlId);
-      expect(latest?.status).toBe("COMPLETED");
-      expect(latest?.createdAt).toEqual(new Date("2023-05-22"));
-    });
-
-    it("should delete all scans for the website and return count", async () => {
-      const websiteUrlId = new ObjectId().toString();
-      await getScansCollection().insertMany([
-        createScanFixture({ websiteUrlId }),
-        createScanFixture({ websiteUrlId }),
-        createScanFixture({ websiteUrlId: new ObjectId().toString() }),
-      ]);
-
-      const deleted = await deleteScansByWebsiteUrl(websiteUrlId);
-      expect(deleted).toBe(2);
-      expect(await countScans({ websiteUrlId })).toBe(0);
-    });
-  });
-
-  describe("deleteScanById()", () => {
-    it("should delete the scan and return true", async () => {
-      const { insertedId } =
-        await getScansCollection().insertOne(createScanFixture());
-
-      expect(await deleteScanById(insertedId)).toBe(true);
-      expect(
-        await getScansCollection().findOne({ _id: insertedId }),
-      ).toBeNull();
-    });
-
-    it("should return false when the scan does not exist", async () => {
-      expect(await deleteScanById(new ObjectId())).toBe(false);
-    });
-  });
-
-  describe("FINISHED_SCAN_STATUSES", () => {
-    it("should include CANCELED — canceling is final, not resumable", () => {
-      expect(FINISHED_SCAN_STATUSES).toContain(SCAN_STATUSES.CANCELED);
-    });
-
-    it("should NOT include PAUSED — pausing is reversible", () => {
-      expect(FINISHED_SCAN_STATUSES).not.toContain(SCAN_STATUSES.PAUSED);
-    });
-  });
-});
+import {
+  getScansCollection,
+  getScansByClientAndDateRange,
+  getScanFieldsByClientAndDateRange,
+  findScansByQuery,
+  countScans,
+  createScanDoc,
+  updateScanDoc,
+  findInProgressScansByClient,
+  getScansByWebsiteUrl,
+  getLatestScanByWebsiteUrl,
+  getInProgressScanByWebsiteUrl,
+  getLatestCompletedScanByWebsiteUrl,
+  deleteScansByWebsiteUrl,
+  deleteScanById,
+} from "../scans.getters";
+import { FINISHED_SCAN_STATUSES, SCAN_STATUSES } from "../scans.constants";
+import { createScanFixture } from "../../../test-utils/factories";
+import { ObjectId } from "mongodb";
+
+describe("db.websitalk.scans", () => {
+  describe("getScansByClientAndDateRange()", () => {
+    it("should return full scan documents within the date range", async () => {
+      const clientId = "dateRangeClient";
+      const startDate = new Date("2023-05-01");
+      const endDate = new Date("2023-05-31");
+
+      const scanInside = createScanFixture({ clientId, status: "COMPLETED" });
+      const scanBefore = createScanFixture({ clientId });
+      const scanAfter = createScanFixture({ clientId });
+
+      await getScansCollection().insertMany([
+        { ...scanInside, createdAt: new Date("2023-05-15") },
+        { ...scanBefore, createdAt: new Date("2023-04-30") },
+        { ...scanAfter, createdAt: new Date("2023-06-01") },
+      ]);
+
+      const result = await getScansByClientAndDateRange(
+        clientId,
+        startDate,
+        endDate,
+      );
+
+      expect(result.length).toBe(1);
+      expect(result[0]).toMatchObject({
+        clientId,
+        websiteUrlId: scanInside.websiteUrlId,
+        status: "COMPLETED",
+      });
+      expect(result[0]).toHaveProperty("siteChrome");
+      expect(result[0]).toHaveProperty("skipped");
+    });
+
+    it("should respect the date range boundaries (inclusive)", async () => {
+      const clientId = "boundaryClient";
+      const onStart = createScanFixture({ clientId });
+      const onEnd = createScanFixture({ clientId });
+
+      await getScansCollection().insertMany([
+        { ...onStart, createdAt: new Date("2023-05-01") },
+        { ...onEnd, createdAt: new Date("2023-05-31") },
+      ]);
+
+      const result = await getScansByClientAndDateRange(
+        clientId,
+        new Date("2023-05-01"),
+        new Date("2023-05-31"),
+      );
+
+      expect(result.length).toBe(2);
+    });
+  });
+
+  describe("getScanFieldsByClientAndDateRange()", () => {
+    it("should return only the requested fields (plus _id)", async () => {
+      const clientId = "projectionClient";
+      const scan = createScanFixture({
+        clientId,
+        status: "COMPLETED",
+        siteChrome: { url: "https://example.com", sourceType: "site_chrome" },
+        skipped: [{ path: "/broken", reason: "HTTP 500" }],
+      });
+
+      await getScansCollection().insertOne({
+        ...scan,
+        createdAt: new Date("2023-05-15"),
+      });
+
+      const [result] = await getScanFieldsByClientAndDateRange(
+        clientId,
+        new Date("2023-05-01"),
+        new Date("2023-05-31"),
+        ["clientId", "websiteUrlId", "status"],
+      );
+
+      expect(result).toMatchObject({
+        clientId,
+        websiteUrlId: scan.websiteUrlId,
+        status: "COMPLETED",
+      });
+      expect(result).toHaveProperty("_id");
+      expect(result).not.toHaveProperty("siteChrome");
+      expect(result).not.toHaveProperty("skipped");
+      expect(result).not.toHaveProperty("createdAt");
+    });
+  });
+
+  describe("findScansByQuery() / countScans()", () => {
+    it("should find scans matching a raw filter and count them", async () => {
+      const clientId = "queryClient";
+      const completed = createScanFixture({ clientId, status: "COMPLETED" });
+      const scanning = createScanFixture({ clientId, status: "SCANNING" });
+
+      await getScansCollection().insertMany([completed, scanning]);
+
+      const result = await findScansByQuery({ clientId, status: "COMPLETED" });
+      const count = await countScans({ clientId });
+
+      expect(result.length).toBe(1);
+      expect(result[0].websiteUrlId).toBe(completed.websiteUrlId);
+      expect(count).toBe(2);
+    });
+  });
+
+  describe("createScanDoc()", () => {
+    it("should insert a scan with default values for the rest of the fields", async () => {
+      const clientId = "createClient";
+      const websiteUrlId = new ObjectId().toString();
+
+      const { insertedId } = await createScanDoc({ clientId, websiteUrlId });
+
+      const result = await getScansCollection().findOne({ _id: insertedId });
+
+      expect(result).toMatchObject({
+        clientId,
+        websiteUrlId,
+        status: "CHECKING",
+        activeTasksCount: 0,
+        pagesScanned: 0,
+        pagesSkipped: 0,
+        skipped: [],
+        siteChrome: {},
+        pausedAt: null,
+      });
+      expect(result?.createdAt).toBeInstanceOf(Date);
+      expect(result?.updatedAt).toBeInstanceOf(Date);
+    });
+  });
+
+  describe("pausedAt", () => {
+    it("is set on pause and cleared back to null on resume, via updateScanDoc", async () => {
+      const scan = createScanFixture({ status: "SCANNING", pausedAt: null });
+      const { insertedId } = await getScansCollection().insertOne(scan);
+
+      const pausedAt = new Date();
+      const paused = await updateScanDoc(insertedId, {
+        status: "PAUSED",
+        pausedAt,
+      });
+      expect(paused?.pausedAt).toEqual(pausedAt);
+
+      const resumed = await updateScanDoc(insertedId, {
+        status: "SCANNING",
+        pausedAt: null,
+      });
+      expect(resumed?.pausedAt).toBeNull();
+    });
+  });
+
+  describe("updateScanDoc()", () => {
+    it("should update the given fields and bump updatedAt", async () => {
+      const scan = createScanFixture({ status: "SCANNING", pagesScanned: 0 });
+      const { insertedId } = await getScansCollection().insertOne(scan);
+
+      const result = await updateScanDoc(insertedId, {
+        status: "COMPLETED",
+        pagesScanned: 10,
+      });
+
+      expect(result).toMatchObject({
+        _id: insertedId,
+        status: "COMPLETED",
+        pagesScanned: 10,
+      });
+      expect(result?.updatedAt.getTime()).toBeGreaterThan(
+        scan.updatedAt.getTime(),
+      );
+    });
+
+    it("should return null when the scan does not exist", async () => {
+      const result = await updateScanDoc(new ObjectId(), { status: "FAILED" });
+
+      expect(result).toBeNull();
+    });
+  });
+
+  describe("in-progress scan helpers", () => {
+    const websiteUrlId = "website-123";
+
+    it("should return website scan history newest first", async () => {
+      await getScansCollection().insertMany([
+        createScanFixture({
+          websiteUrlId,
+          status: "COMPLETED",
+          createdAt: new Date("2023-05-01"),
+        }),
+        createScanFixture({
+          websiteUrlId,
+          status: "COMPLETED",
+          createdAt: new Date("2023-05-15"),
+        }),
+      ]);
+
+      const scans = await getScansByWebsiteUrl(websiteUrlId);
+      expect(scans).toHaveLength(2);
+      expect(scans[0].createdAt.getTime()).toBeGreaterThan(
+        scans[1].createdAt.getTime(),
+      );
+    });
+
+    it("should return the newest scan for a website", async () => {
+      await getScansCollection().insertMany([
+        createScanFixture({
+          websiteUrlId,
+          status: "COMPLETED",
+          createdAt: new Date("2023-05-01"),
+        }),
+        createScanFixture({
+          websiteUrlId,
+          status: "FAILED",
+          createdAt: new Date("2023-05-15"),
+        }),
+      ]);
+
+      const latest = await getLatestScanByWebsiteUrl(websiteUrlId);
+      expect(latest?.status).toBe("FAILED");
+    });
+
+    it("should return in-progress scan and return null when only completed exists", async () => {
+      await getScansCollection().insertMany([
+        createScanFixture({
+          websiteUrlId,
+          status: "COMPLETED",
+          createdAt: new Date("2023-05-01"),
+        }),
+        createScanFixture({
+          websiteUrlId,
+          status: "SCANNING",
+          createdAt: new Date("2023-05-15"),
+        }),
+      ]);
+
+      const inProgressScan = await getInProgressScanByWebsiteUrl(websiteUrlId);
+      expect(inProgressScan?.status).toBe("SCANNING");
+
+      await getScansCollection().deleteMany({ websiteUrlId });
+      await getScansCollection().insertOne(
+        createScanFixture({
+          websiteUrlId,
+          status: "COMPLETED",
+          createdAt: new Date("2023-05-20"),
+        }),
+      );
+
+      expect(await getInProgressScanByWebsiteUrl(websiteUrlId)).toBeNull();
+    });
+
+    it("should return all in-progress scans across websites for a client", async () => {
+      const clientId = "multiScanClient";
+      await getScansCollection().insertMany([
+        createScanFixture({
+          clientId,
+          websiteUrlId: "website-a",
+          status: "SCANNING",
+        }),
+        createScanFixture({
+          clientId,
+          websiteUrlId: "website-b",
+          status: "PROCESSING",
+        }),
+        createScanFixture({
+          clientId,
+          websiteUrlId: "website-c",
+          status: "COMPLETED",
+        }),
+      ]);
+
+      const scans = await findInProgressScansByClient(clientId);
+      expect(scans).toHaveLength(2);
+      expect(scans.map((s) => s.websiteUrlId).sort()).toEqual([
+        "website-a",
+        "website-b",
+      ]);
+    });
+  });
+
+  describe("website delete helpers", () => {
+    it("should return the newest completed scan only", async () => {
+      const websiteUrlId = new ObjectId().toString();
+      await getScansCollection().insertMany([
+        createScanFixture({
+          websiteUrlId,
+          status: "FAILED",
+          createdAt: new Date("2023-05-25"),
+        }),
+        createScanFixture({
+          websiteUrlId,
+          status: "COMPLETED",
+          createdAt: new Date("2023-05-20"),
+        }),
+        createScanFixture({
+          websiteUrlId,
+          status: "COMPLETED",
+          createdAt: new Date("2023-05-22"),
+        }),
+      ]);
+
+      const latest = await getLatestCompletedScanByWebsiteUrl(websiteUrlId);
+      expect(latest?.status).toBe("COMPLETED");
+      expect(latest?.createdAt).toEqual(new Date("2023-05-22"));
+    });
+
+    it("should delete all scans for the website and return count", async () => {
+      const websiteUrlId = new ObjectId().toString();
+      await getScansCollection().insertMany([
+        createScanFixture({ websiteUrlId }),
+        createScanFixture({ websiteUrlId }),
+        createScanFixture({ websiteUrlId: new ObjectId().toString() }),
+      ]);
+
+      const deleted = await deleteScansByWebsiteUrl(websiteUrlId);
+      expect(deleted).toBe(2);
+      expect(await countScans({ websiteUrlId })).toBe(0);
+    });
+  });
+
+  describe("deleteScanById()", () => {
+    it("should delete the scan and return true", async () => {
+      const { insertedId } =
+        await getScansCollection().insertOne(createScanFixture());
+
+      expect(await deleteScanById(insertedId)).toBe(true);
+      expect(
+        await getScansCollection().findOne({ _id: insertedId }),
+      ).toBeNull();
+    });
+
+    it("should return false when the scan does not exist", async () => {
+      expect(await deleteScanById(new ObjectId())).toBe(false);
+    });
+  });
+
+  describe("FINISHED_SCAN_STATUSES", () => {
+    it("should include CANCELED — canceling is final, not resumable", () => {
+      expect(FINISHED_SCAN_STATUSES).toContain(SCAN_STATUSES.CANCELED);
+    });
+
+    it("should NOT include PAUSED — pausing is reversible", () => {
+      expect(FINISHED_SCAN_STATUSES).not.toContain(SCAN_STATUSES.PAUSED);
+    });
+  });
+});