From d732acd469ff02373dba4093cc53198e97f51868 Mon Sep 17 00:00:00 2001
From: Mohamed Bassem <me@mbassem.com>
Date: Wed, 16 Apr 2025 23:49:44 +0000
Subject: tests: Add some simple crawler tests

---
 packages/e2e_tests/tests/workers/crawler.test.ts | 100 +++++++++++++++++++++++
 1 file changed, 100 insertions(+)
 create mode 100644 packages/e2e_tests/tests/workers/crawler.test.ts

(limited to 'packages/e2e_tests/tests')

diff --git a/packages/e2e_tests/tests/workers/crawler.test.ts b/packages/e2e_tests/tests/workers/crawler.test.ts
new file mode 100644
index 00000000..df276cae
--- /dev/null
+++ b/packages/e2e_tests/tests/workers/crawler.test.ts
@@ -0,0 +1,100 @@
+import { assert, beforeEach, describe, expect, inject, it } from "vitest";
+
+import { createKarakeepClient } from "@karakeep/sdk";
+
+import { createTestUser } from "../../utils/api";
+import { waitUntil } from "../../utils/general";
+
+describe("Crawler Tests", () => {
+  const port = inject("hoarderPort");
+
+  if (!port) {
+    throw new Error("Missing required environment variables");
+  }
+
+  let client: ReturnType<typeof createKarakeepClient>;
+  let apiKey: string;
+
+  async function getBookmark(bookmarkId: string) {
+    const { data } = await client.GET(`/bookmarks/{bookmarkId}`, {
+      params: {
+        path: {
+          bookmarkId,
+        },
+        query: {
+          includeContent: true,
+        },
+      },
+    });
+    return data;
+  }
+
+  beforeEach(async () => {
+    apiKey = await createTestUser();
+    client = createKarakeepClient({
+      baseUrl: `http://localhost:${port}/api/v1/`,
+      headers: {
+        "Content-Type": "application/json",
+        authorization: `Bearer ${apiKey}`,
+      },
+    });
+  });
+
+  it("should crawl a website", async () => {
+    let { data: bookmark } = await client.POST("/bookmarks", {
+      body: {
+        type: "link",
+        url: "http://nginx:80/hello.html",
+      },
+    });
+    assert(bookmark);
+
+    await waitUntil(
+      async () => {
+        const data = await getBookmark(bookmark!.id);
+        assert(data);
+        assert(data.content.type === "link");
+        return data.content.crawledAt !== null;
+      },
+      "Bookmark is crawled",
+      10000,
+    );
+
+    bookmark = await getBookmark(bookmark.id);
+    assert(bookmark && bookmark.content.type === "link");
+    expect(bookmark.content.crawledAt).toBeDefined();
+    expect(bookmark.content.htmlContent).toContain("Hello World");
+    expect(bookmark.content.title).toContain("My test title");
+    expect(bookmark.content.url).toBe("http://nginx:80/hello.html");
+    expect(
+      bookmark.assets.find((a) => a.assetType === "screenshot"),
+    ).toBeDefined();
+  });
+
+  it("image lings jobs be converted into images", async () => {
+    let { data: bookmark } = await client.POST("/bookmarks", {
+      body: {
+        type: "link",
+        url: "http://nginx:80/image.png",
+      },
+    });
+    assert(bookmark);
+
+    await waitUntil(
+      async () => {
+        const data = await getBookmark(bookmark!.id);
+        assert(data);
+        return data.content.type === "asset";
+      },
+      "Bookmark is crawled and converted to an image",
+      10000,
+    );
+
+    bookmark = await getBookmark(bookmark.id);
+    assert(bookmark && bookmark.content.type === "asset");
+    expect(bookmark.content.assetType).toBe("image");
+    expect(bookmark.content.assetId).toBeDefined();
+    expect(bookmark.content.fileName).toBe("image.png");
+    expect(bookmark.content.sourceUrl).toBe("http://nginx:80/image.png");
+  });
+});
-- 
cgit v1.2.3-70-g09d2