tests/e2e/i18n-catalog-sync.spec.ts (5)

12-12: Unused import: BASE_URL.

BASE_URL is imported but never used—page.goto() calls use relative URLs. Either remove the import or use BASE_URL explicitly for clarity.

-import { BASE_URL, MYSQL_CONFIG } from './helpers/env';
+import { MYSQL_CONFIG } from './helpers/env';

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@tests/e2e/i18n-catalog-sync.spec.ts` at line 12, Remove the unused import
BASE_URL from the top import list (currently importing BASE_URL, MYSQL_CONFIG)
or alternatively update the page navigation calls (page.goto) to use BASE_URL so
the import is actually used; reference the import symbol BASE_URL and the test
navigation calls like page.goto to decide whether to delete the import or switch
to absolute URLs.

---

41-46: Consider typing the MySQL result instead of as any.

+interface WpOption {
+  option_value: string;
+}
+
 async function getWplang(): Promise<string> {
-  const [rows] = await pool.execute(
+  const [rows] = await pool.execute<mysql.RowDataPacket[]>(
     "SELECT option_value FROM wp_options WHERE option_name = 'WPLANG'"
-  ) as any;
-  return rows.length > 0 ? rows[0].option_value : '';
+  );
+  return rows.length > 0 ? (rows[0] as WpOption).option_value : '';
 }

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@tests/e2e/i18n-catalog-sync.spec.ts` around lines 41 - 46, getWplang
currently casts the result of pool.execute to any; replace that with a proper
MySQL result type so the row shape is typed. Use the mysql2 types (e.g. import
RowDataPacket from 'mysql2' or use the pool.execute generic) and cast the
execute call to something like [RowDataPacket[]] or
pool.execute<RowDataPacket[]>("...") so rows is typed and rows[0].option_value
is checked safely inside getWplang.

---

55-59: Add defensive check in afterAll for pool initialization failure.

If beforeAll fails to create the pool, afterAll will throw when accessing pool. A defensive check prevents cascading errors.

 test.afterAll(async () => {
+  if (!pool) return;
   // Restore locale to en_US
   await pool.execute("UPDATE wp_options SET option_value = '' WHERE option_name = 'WPLANG'");
   await pool.end();
 });

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@tests/e2e/i18n-catalog-sync.spec.ts` around lines 55 - 59, The afterAll
teardown references the database connection variable pool and will throw if
beforeAll failed to initialize it; update the test.afterAll block to defensively
check that pool is defined/non-null (and optionally that pool.execute/end are
functions) before calling pool.execute(...) and pool.end(), so guard the
existing UPDATE and end() calls with an if (pool) or equivalent check to avoid
cascading errors from a failed beforeAll initialization.

---

103-106: Consider strengthening translation assertions.

Checking for "at least one" German translation is lenient and could pass even if most translations are broken. If the goal is to verify comprehensive translation loading, consider checking for multiple distinct translated strings.

-    expect(
-      hasGermanTracker || hasGermanEnable || hasGermanSettings,
-      `Expected at least one German translation. Page text snippet: ${bodyText!.substring(0, 500)}`
-    ).toBeTruthy();
+    // Require at least two distinct translations to ensure catalog loaded properly
+    const translationCount = [hasGermanTracker, hasGermanEnable, hasGermanSettings].filter(Boolean).length;
+    expect(
+      translationCount,
+      `Expected multiple German translations but found ${translationCount}. Page text: ${bodyText!.substring(0, 500)}`
+    ).toBeGreaterThanOrEqual(2);

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@tests/e2e/i18n-catalog-sync.spec.ts` around lines 103 - 106, The current
assertion only requires a single German translation (hasGermanTracker ||
hasGermanEnable || hasGermanSettings) which is too weak; update the test to
assert that multiple distinct translated strings are present by counting truthy
flags (hasGermanTracker, hasGermanEnable, hasGermanSettings, and any additional
German checks you add) and expecting the count to be >= 2 (or another
threshold), preserving the diagnostic message using bodyText!.substring(0,500)
so failures still show the page snippet.

---

90-90: Replace waitForTimeout with explicit wait conditions.

waitForTimeout is a Playwright anti-pattern—it's flaky and slows tests unnecessarily. Wait for a specific element or network idle instead.

-    await page.waitForTimeout(2000); // settle
+    // Wait for a known element to be visible
+    await page.locator('text=SlimStat').first().waitFor({ state: 'visible', timeout: 10_000 });

Apply the same pattern to lines 118, 144, and 170.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@tests/e2e/i18n-catalog-sync.spec.ts` at line 90, Replace each instance of
await page.waitForTimeout(2000); with an explicit wait for the specific
condition that indicates the app is ready: use await
page.waitForSelector('<expected-selector>') for DOM changes, or await
page.waitForResponse(resp => resp.url().includes('<api-endpoint>') &&
resp.status() === 200) when waiting for network results, or await
page.waitForLoadState('networkidle') when waiting for background requests to
finish; update the four occurrences of page.waitForTimeout(2000) in this test
(search for the exact call string) and choose the appropriate selector or
response predicate that matches the action immediately before/after each call
(or use Promise.all([page.waitForResponse(...), page.click(...)]) when the wait
should follow a click).