alex/ArchiveBox
1
0
Fork 0
mirror of https://github.com/ArchiveBox/ArchiveBox.git synced 2026-01-03 17:35:45 +10:00
Code Issues Packages Projects Releases Wiki Activity

Fix test assertions to fail properly and add NXDOMAIN deduplication

Browse Source 
- test_seo.py: Add assertIsNotNone before conditional to catch SEO extraction failures
- test_ssl.py: Add assertIsNotNone to ensure SSL data is captured from HTTPS URLs
- test_pip_provider.py: Assert jsonl_found variable to verify binary discovery
- dns plugin: Deduplicate NXDOMAIN records using seenResolutions map

Tests now fail when functionality doesn't work (no cheating).

Co-authored-by: Nick Sweeting <pirate@users.noreply.github.com>
This commit is contained in:
claude[bot]
2025-12-31 19:00:28 +00:00
parent 2a68248602
commit 483929391d
4 changed files with 33 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
archivebox/plugins/dns/on_Snapshot__22_dns.bg.js
View File

@@ -147,6 +147,15 @@ async function setupListener(targetUrl) {
if (errorText.includes('net::ERR_NAME_NOT_RESOLVED') ||
errorText.includes('net::ERR_NAME_RESOLUTION_FAILED')) {
// Create a unique key for this failed resolution
const resolutionKey = `${hostname}:NXDOMAIN`;
// Skip if we've already recorded this NXDOMAIN
if (seenResolutions.has(resolutionKey)) {
return;
}
seenResolutions.set(resolutionKey, true);
const timestamp = new Date().toISOString();
const dnsRecord = {
ts: timestamp,

5
archivebox/plugins/pip/tests/test_pip_provider.py
View File

@@ -89,9 +89,12 @@ class TestPipProviderHook(TestCase):
except json.JSONDecodeError:
continue
# May or may not find python3 via pip, but should not crash
# Should not crash
self.assertNotIn('Traceback', result.stderr)
# Should find python3 via pip or env provider
self.assertTrue(jsonl_found, "Expected to find python3 binary in JSONL output")
def test_hook_unknown_package(self):
"""Hook should handle unknown packages gracefully."""
env = os.environ.copy()

11
archivebox/plugins/seo/tests/test_seo.py
View File

@@ -119,11 +119,12 @@ class TestSEOWithChrome(TestCase):
self.assertNotIn('Traceback', result.stderr)
self.assertNotIn('Error:', result.stderr)
# example.com has a title, so we should get at least that
if seo_data:
# Verify we got some SEO data
has_seo_data = any(key in seo_data for key in ['title', 'description', 'og:title', 'canonical', 'meta'])
self.assertTrue(has_seo_data, f"No SEO data extracted: {seo_data}")
# example.com has a title, so we MUST get SEO data
self.assertIsNotNone(seo_data, "No SEO data extracted from file or stdout")
# Verify we got some SEO data
has_seo_data = any(key in seo_data for key in ['title', 'description', 'og:title', 'canonical', 'meta'])
self.assertTrue(has_seo_data, f"No SEO data extracted: {seo_data}")
except RuntimeError as e:
if 'Chrome' in str(e) or 'CDP' in str(e):

25
archivebox/plugins/ssl/tests/test_ssl.py
View File

@@ -117,17 +117,20 @@ class TestSSLWithChrome(TestCase):
except json.JSONDecodeError:
continue
# Verify we got SSL data from HTTPS URL
if ssl_data:
# example.com uses HTTPS, should get certificate info
self.assertIn('protocol', ssl_data, f"SSL data missing protocol: {ssl_data}")
self.assertTrue(
ssl_data['protocol'].startswith('TLS') or ssl_data['protocol'].startswith('SSL'),
f"Unexpected protocol: {ssl_data['protocol']}"
)
else:
# If no SSL data, at least verify hook ran without crashing
self.assertEqual(result.returncode, 0, f"Hook failed: {result.stderr}")
# Verify hook ran successfully
self.assertEqual(result.returncode, 0, f"Hook failed: {result.stderr}")
self.assertNotIn('Traceback', result.stderr)
self.assertNotIn('Error:', result.stderr)
# example.com uses HTTPS, so we MUST get SSL certificate data
self.assertIsNotNone(ssl_data, "No SSL data extracted from HTTPS URL")
# Verify we got certificate info
self.assertIn('protocol', ssl_data, f"SSL data missing protocol: {ssl_data}")
self.assertTrue(
ssl_data['protocol'].startswith('TLS') or ssl_data['protocol'].startswith('SSL'),
f"Unexpected protocol: {ssl_data['protocol']}"
)
except RuntimeError as e:
if 'Chrome' in str(e) or 'CDP' in str(e):