Externalize audit pipeline services

tests/external-audit-services.test.ts

@@ -0,0 +1,184 @@
+import assert from "node:assert/strict";
+import test from "node:test";
+
+import {
+  buildJinaReaderAuditInput,
+  buildScreenshotOneRequests,
+  estimateExternalAuditCostUsd,
+} from "../lib/external-audit-services";
+
+test("estimateExternalAuditCostUsd totals managed provider usage", () => {
+  const estimate = estimateExternalAuditCostUsd({
+    openRouter: {
+      inputTokens: 1_500_000,
+      outputTokens: 250_000,
+      inputUsdPerMillionTokens: 0.25,
+      outputUsdPerMillionTokens: 1.25,
+    },
+    screenshotOne: {
+      screenshots: 2,
+      usdPerScreenshot: 0.01,
+    },
+    jina: {
+      requests: 4,
+      pages: 4,
+      usdPerRequest: 0.001,
+      usdPerPage: 0.002,
+    },
+    pageSpeed: {
+      requests: 2,
+    },
+  });
+
+  assert.equal(estimate.totalUsd, 0.7195);
+  assert.deepEqual(estimate.byProvider, {
+    openRouter: 0.6875,
+    screenshotOne: 0.02,
+    jina: 0.012,
+    pageSpeed: 0,
+  });
+});
+
+test("estimateExternalAuditCostUsd clamps negative usage and prices to zero", () => {
+  const estimate = estimateExternalAuditCostUsd({
+    openRouter: {
+      inputTokens: -1_000_000,
+      outputTokens: 100_000,
+      inputUsdPerMillionTokens: 0.25,
+      outputUsdPerMillionTokens: -1.25,
+    },
+    screenshotOne: {
+      screenshots: -2,
+      usdPerScreenshot: 0.01,
+    },
+    jina: {
+      requests: 4,
+      pages: -4,
+      usdPerRequest: -0.001,
+      usdPerPage: 0.002,
+    },
+  });
+
+  assert.deepEqual(estimate.byProvider, {
+    openRouter: 0,
+    screenshotOne: 0,
+    jina: 0,
+    pageSpeed: 0,
+  });
+  assert.equal(estimate.totalUsd, 0);
+});
+
+test("buildScreenshotOneRequests creates stable desktop and mobile URLs", () => {
+  const requests = buildScreenshotOneRequests({
+    accessKey: "sso_secret_key",
+    targetUrl: "https://example.com/landing?utm=abc",
+  });
+
+  assert.equal(requests.length, 2);
+  assert.deepEqual(
+    requests.map((request) => request.viewport),
+    ["desktop", "mobile"],
+  );
+
+  const desktop = new URL(requests[0]?.url ?? "");
+  assert.equal(desktop.searchParams.get("access_key"), "sso_secret_key");
+  assert.equal(desktop.searchParams.get("url"), "https://example.com/landing?utm=abc");
+  assert.equal(desktop.searchParams.get("viewport_width"), "1280");
+  assert.equal(desktop.searchParams.get("viewport_height"), "900");
+  assert.equal(desktop.searchParams.get("device_scale_factor"), "1");
+  assert.equal(desktop.searchParams.get("full_page"), "true");
+  assert.equal(desktop.searchParams.get("block_cookie_banners"), "true");
+  assert.equal(desktop.searchParams.get("block_ads"), "true");
+  assert.equal(desktop.searchParams.get("block_trackers"), "true");
+
+  const mobile = new URL(requests[1]?.url ?? "");
+  assert.equal(mobile.searchParams.get("viewport_width"), "390");
+  assert.equal(mobile.searchParams.get("viewport_height"), "844");
+  assert.equal(mobile.searchParams.get("device_scale_factor"), "2");
+  assert.equal(mobile.searchParams.get("full_page"), "true");
+});
+
+test("buildScreenshotOneRequests rejects non-web target URLs without leaking secrets", () => {
+  assert.throws(
+    () =>
+      buildScreenshotOneRequests({
+        accessKey: "sso_secret_key",
+        targetUrl: "ftp://example.com/landing",
+      }),
+    (error) => {
+      assert.equal(error instanceof Error, true);
+      assert.equal((error as Error).message.includes("sso_secret_key"), false);
+      assert.match((error as Error).message, /http.*https/i);
+      return true;
+    },
+  );
+});
+
+test("buildScreenshotOneRequests does not leak the access key in validation errors", () => {
+  assert.throws(
+    () =>
+      buildScreenshotOneRequests({
+        accessKey: "sso_secret_key",
+        targetUrl: "not a url",
+      }),
+    (error) => {
+      assert.equal(error instanceof Error, true);
+      assert.equal((error as Error).message.includes("sso_secret_key"), false);
+      assert.match((error as Error).message, /target url/i);
+      return true;
+    },
+  );
+});
+
+test("buildJinaReaderAuditInput rejects non-web base and page URLs", () => {
+  assert.throws(
+    () =>
+      buildJinaReaderAuditInput({
+        baseUrl: "file:///tmp/site.html",
+        maxMarkdownChars: 100,
+      }),
+    /http.*https/i,
+  );
+
+  assert.throws(
+    () =>
+      buildJinaReaderAuditInput({
+        baseUrl: "https://example.com",
+        pages: [{ url: "ftp://example.com/kontakt", markdown: "Kontakt" }],
+        maxMarkdownChars: 100,
+      }),
+    /http.*https/i,
+  );
+});
+
+test("buildJinaReaderAuditInput prepares capped markdown for relevant pages", () => {
+  const input = buildJinaReaderAuditInput({
+    baseUrl: "https://example.com",
+    pages: [
+      { url: "https://example.com", markdown: "# Home\nWillkommen auf der Startseite." },
+      { url: "https://example.com/kontakt", markdown: "Kontaktformular und Telefonnummer." },
+      { url: "https://example.com/impressum", markdown: "Impressum mit Anbieterkennzeichnung." },
+      { url: "https://example.com/leistungen", markdown: "Leistungen fuer Webdesign und SEO." },
+      { url: "https://example.com/ueber-uns", markdown: "Ueber uns und Arbeitsweise." },
+    ],
+    maxMarkdownChars: 95,
+  });
+
+  assert.deepEqual(
+    input.pages.map((page) => page.path),
+    ["/", "/kontakt", "/impressum", "/leistungen", "/ueber-uns"],
+  );
+  assert.deepEqual(
+    input.readerUrls,
+    [
+      "https://r.jina.ai/https://example.com/",
+      "https://r.jina.ai/https://example.com/kontakt",
+      "https://r.jina.ai/https://example.com/impressum",
+      "https://r.jina.ai/https://example.com/leistungen",
+      "https://r.jina.ai/https://example.com/ueber-uns",
+    ],
+  );
+  assert.equal(input.markdown.length <= 95, true);
+  assert.match(input.markdown, /Source: https:\/\/example.com/);
+  assert.match(input.markdown, /\[truncated to 95 chars\]$/);
+});