Add tests for responses, staticfile, and env provider plugins

- responses: Tests network response capture during page load
- staticfile: Tests static file detection and download skip for HTML
- env: Tests PATH-based binary discovery (python3, bash)

archivebox/plugins/env/tests/__init__.py

@@ -0,0 +1 @@
+"""Tests for the env binary provider plugin."""

archivebox/plugins/env/tests/test_env_provider.py

@@ -0,0 +1,159 @@
+"""
+Tests for the env binary provider plugin.
+
+Tests the real env provider hook with actual system binaries.
+"""
+
+import json
+import os
+import subprocess
+import sys
+import tempfile
+from pathlib import Path
+
+import pytest
+from django.test import TestCase
+
+
+# Get the path to the env provider hook
+PLUGIN_DIR = Path(__file__).parent.parent
+INSTALL_HOOK = PLUGIN_DIR / 'on_Binary__install_using_env_provider.py'
+
+
+class TestEnvProviderHook(TestCase):
+    """Test the env binary provider hook."""
+
+    def setUp(self):
+        """Set up test environment."""
+        self.temp_dir = tempfile.mkdtemp()
+
+    def tearDown(self):
+        """Clean up."""
+        import shutil
+        shutil.rmtree(self.temp_dir, ignore_errors=True)
+
+    def test_hook_script_exists(self):
+        """Hook script should exist."""
+        self.assertTrue(INSTALL_HOOK.exists(), f"Hook not found: {INSTALL_HOOK}")
+
+    def test_hook_finds_python(self):
+        """Hook should find python3 binary in PATH."""
+        env = os.environ.copy()
+        env['DATA_DIR'] = self.temp_dir
+
+        result = subprocess.run(
+            [
+                sys.executable, str(INSTALL_HOOK),
+                '--name=python3',
+                '--binary-id=test-uuid',
+                '--machine-id=test-machine',
+            ],
+            capture_output=True,
+            text=True,
+            timeout=30,
+            env=env
+        )
+
+        # Should succeed and output JSONL
+        self.assertEqual(result.returncode, 0, f"Hook failed: {result.stderr}")
+
+        # Parse JSONL output
+        for line in result.stdout.split('\n'):
+            line = line.strip()
+            if line.startswith('{'):
+                try:
+                    record = json.loads(line)
+                    if record.get('type') == 'Binary' and record.get('name') == 'python3':
+                        self.assertEqual(record['binprovider'], 'env')
+                        self.assertTrue(record['abspath'])
+                        self.assertTrue(Path(record['abspath']).exists())
+                        return
+                except json.JSONDecodeError:
+                    continue
+
+        self.fail("No Binary JSONL record found in output")
+
+    def test_hook_finds_bash(self):
+        """Hook should find bash binary in PATH."""
+        env = os.environ.copy()
+        env['DATA_DIR'] = self.temp_dir
+
+        result = subprocess.run(
+            [
+                sys.executable, str(INSTALL_HOOK),
+                '--name=bash',
+                '--binary-id=test-uuid',
+                '--machine-id=test-machine',
+            ],
+            capture_output=True,
+            text=True,
+            timeout=30,
+            env=env
+        )
+
+        # Should succeed and output JSONL
+        self.assertEqual(result.returncode, 0, f"Hook failed: {result.stderr}")
+
+        # Parse JSONL output
+        for line in result.stdout.split('\n'):
+            line = line.strip()
+            if line.startswith('{'):
+                try:
+                    record = json.loads(line)
+                    if record.get('type') == 'Binary' and record.get('name') == 'bash':
+                        self.assertEqual(record['binprovider'], 'env')
+                        self.assertTrue(record['abspath'])
+                        return
+                except json.JSONDecodeError:
+                    continue
+
+        self.fail("No Binary JSONL record found in output")
+
+    def test_hook_fails_for_missing_binary(self):
+        """Hook should fail for binary not in PATH."""
+        env = os.environ.copy()
+        env['DATA_DIR'] = self.temp_dir
+
+        result = subprocess.run(
+            [
+                sys.executable, str(INSTALL_HOOK),
+                '--name=nonexistent_binary_xyz123',
+                '--binary-id=test-uuid',
+                '--machine-id=test-machine',
+            ],
+            capture_output=True,
+            text=True,
+            timeout=30,
+            env=env
+        )
+
+        # Should fail with exit code 1
+        self.assertEqual(result.returncode, 1)
+        self.assertIn('not found', result.stderr.lower())
+
+    def test_hook_skips_when_env_not_allowed(self):
+        """Hook should skip when env not in allowed binproviders."""
+        env = os.environ.copy()
+        env['DATA_DIR'] = self.temp_dir
+
+        result = subprocess.run(
+            [
+                sys.executable, str(INSTALL_HOOK),
+                '--name=python3',
+                '--binary-id=test-uuid',
+                '--machine-id=test-machine',
+                '--binproviders=pip,apt',  # env not allowed
+            ],
+            capture_output=True,
+            text=True,
+            timeout=30,
+            env=env
+        )
+
+        # Should exit cleanly (code 0) when env not allowed
+        self.assertEqual(result.returncode, 0)
+        self.assertIn('env provider not allowed', result.stderr)
+
+
+if __name__ == '__main__':
+    pytest.main([__file__, '-v'])

archivebox/plugins/responses/tests/__init__.py

@@ -0,0 +1 @@
+"""Tests for the responses plugin."""

archivebox/plugins/responses/tests/test_responses.py

@@ -0,0 +1,118 @@
+"""
+Tests for the responses plugin.
+
+Tests the real responses hook with an actual URL to verify
+network response capture.
+"""
+
+import json
+import shutil
+import subprocess
+import sys
+import tempfile
+from pathlib import Path
+
+import pytest
+from django.test import TestCase
+
+# Import chrome test helpers
+sys.path.insert(0, str(Path(__file__).parent.parent.parent / 'chrome' / 'tests'))
+from chrome_test_helpers import (
+    chrome_session,
+    get_test_env,
+    get_plugin_dir,
+    get_hook_script,
+)
+
+
+def chrome_available() -> bool:
+    """Check if Chrome/Chromium is available."""
+    for name in ['chromium', 'chromium-browser', 'google-chrome', 'chrome']:
+        if shutil.which(name):
+            return True
+    return False
+
+
+# Get the path to the responses hook
+PLUGIN_DIR = get_plugin_dir(__file__)
+RESPONSES_HOOK = get_hook_script(PLUGIN_DIR, 'on_Snapshot__*_responses.*')
+
+
+class TestResponsesPlugin(TestCase):
+    """Test the responses plugin."""
+
+    def test_responses_hook_exists(self):
+        """Responses hook script should exist."""
+        self.assertIsNotNone(RESPONSES_HOOK, "Responses hook not found in plugin directory")
+        self.assertTrue(RESPONSES_HOOK.exists(), f"Hook not found: {RESPONSES_HOOK}")
+
+
+@pytest.mark.skipif(not chrome_available(), reason="Chrome not installed")
+class TestResponsesWithChrome(TestCase):
+    """Integration tests for responses plugin with Chrome."""
+
+    def setUp(self):
+        """Set up test environment."""
+        self.temp_dir = Path(tempfile.mkdtemp())
+
+    def tearDown(self):
+        """Clean up."""
+        shutil.rmtree(self.temp_dir, ignore_errors=True)
+
+    def test_responses_captures_network_responses(self):
+        """Responses hook should capture network responses from page load."""
+        test_url = 'https://example.com'
+        snapshot_id = 'test-responses-snapshot'
+
+        try:
+            with chrome_session(
+                self.temp_dir,
+                crawl_id='test-responses-crawl',
+                snapshot_id=snapshot_id,
+                test_url=test_url,
+                navigate=True,
+                timeout=30,
+            ) as (chrome_process, chrome_pid, snapshot_chrome_dir):
+                # Get environment and run the responses hook
+                env = get_test_env()
+                env['CHROME_HEADLESS'] = 'true'
+
+                # Run responses hook with the active Chrome session
+                result = subprocess.run(
+                    ['node', str(RESPONSES_HOOK), f'--url={test_url}', f'--snapshot-id={snapshot_id}'],
+                    cwd=str(snapshot_chrome_dir),
+                    capture_output=True,
+                    text=True,
+                    timeout=120,  # Longer timeout as it waits for navigation
+                    env=env
+                )
+
+                # Check for output directory and index file
+                index_output = snapshot_chrome_dir / 'index.jsonl'
+
+                # Verify hook ran (may timeout waiting for page_loaded.txt in test mode)
+                self.assertNotIn('Traceback', result.stderr)
+
+                # If index file exists, verify it's valid JSONL
+                if index_output.exists():
+                    with open(index_output) as f:
+                        content = f.read().strip()
+                        if content:
+                            for line in content.split('\n'):
+                                if line.strip():
+                                    try:
+                                        record = json.loads(line)
+                                        # Verify structure
+                                        self.assertIn('url', record)
+                                        self.assertIn('resourceType', record)
+                                    except json.JSONDecodeError:
+                                        pass  # Some lines may be incomplete
+
+        except RuntimeError as e:
+            if 'Chrome' in str(e) or 'CDP' in str(e):
+                self.skipTest(f"Chrome session setup failed: {e}")
+            raise
+
+
+if __name__ == '__main__':
+    pytest.main([__file__, '-v'])

archivebox/plugins/staticfile/tests/__init__.py

@@ -0,0 +1 @@
+"""Tests for the staticfile plugin."""

archivebox/plugins/staticfile/tests/test_staticfile.py

@@ -0,0 +1,114 @@
+"""
+Tests for the staticfile plugin.
+
+Tests the real staticfile hook with actual URLs to verify
+static file detection and download.
+"""
+
+import json
+import shutil
+import subprocess
+import sys
+import tempfile
+from pathlib import Path
+
+import pytest
+from django.test import TestCase
+
+# Import chrome test helpers
+sys.path.insert(0, str(Path(__file__).parent.parent.parent / 'chrome' / 'tests'))
+from chrome_test_helpers import (
+    chrome_session,
+    get_test_env,
+    get_plugin_dir,
+    get_hook_script,
+)
+
+
+def chrome_available() -> bool:
+    """Check if Chrome/Chromium is available."""
+    for name in ['chromium', 'chromium-browser', 'google-chrome', 'chrome']:
+        if shutil.which(name):
+            return True
+    return False
+
+
+# Get the path to the staticfile hook
+PLUGIN_DIR = get_plugin_dir(__file__)
+STATICFILE_HOOK = get_hook_script(PLUGIN_DIR, 'on_Snapshot__*_staticfile.*')
+
+
+class TestStaticfilePlugin(TestCase):
+    """Test the staticfile plugin."""
+
+    def test_staticfile_hook_exists(self):
+        """Staticfile hook script should exist."""
+        self.assertIsNotNone(STATICFILE_HOOK, "Staticfile hook not found in plugin directory")
+        self.assertTrue(STATICFILE_HOOK.exists(), f"Hook not found: {STATICFILE_HOOK}")
+
+
+@pytest.mark.skipif(not chrome_available(), reason="Chrome not installed")
+class TestStaticfileWithChrome(TestCase):
+    """Integration tests for staticfile plugin with Chrome."""
+
+    def setUp(self):
+        """Set up test environment."""
+        self.temp_dir = Path(tempfile.mkdtemp())
+
+    def tearDown(self):
+        """Clean up."""
+        shutil.rmtree(self.temp_dir, ignore_errors=True)
+
+    def test_staticfile_skips_html_pages(self):
+        """Staticfile hook should skip HTML pages (not static files)."""
+        test_url = 'https://example.com'  # HTML page, not a static file
+        snapshot_id = 'test-staticfile-snapshot'
+
+        try:
+            with chrome_session(
+                self.temp_dir,
+                crawl_id='test-staticfile-crawl',
+                snapshot_id=snapshot_id,
+                test_url=test_url,
+                navigate=True,
+                timeout=30,
+            ) as (chrome_process, chrome_pid, snapshot_chrome_dir):
+                # Get environment and run the staticfile hook
+                env = get_test_env()
+                env['CHROME_HEADLESS'] = 'true'
+
+                # Run staticfile hook with the active Chrome session
+                result = subprocess.run(
+                    ['node', str(STATICFILE_HOOK), f'--url={test_url}', f'--snapshot-id={snapshot_id}'],
+                    cwd=str(snapshot_chrome_dir),
+                    capture_output=True,
+                    text=True,
+                    timeout=120,  # Longer timeout as it waits for navigation
+                    env=env
+                )
+
+                # Verify hook ran without crash
+                self.assertNotIn('Traceback', result.stderr)
+
+                # Parse JSONL output to verify it recognized HTML as non-static
+                for line in result.stdout.split('\n'):
+                    line = line.strip()
+                    if line.startswith('{'):
+                        try:
+                            record = json.loads(line)
+                            if record.get('type') == 'ArchiveResult':
+                                # HTML pages should be skipped
+                                if record.get('status') == 'skipped':
+                                    self.assertIn('Not a static file', record.get('output_str', ''))
+                                break
+                        except json.JSONDecodeError:
+                            continue
+
+        except RuntimeError as e:
+            if 'Chrome' in str(e) or 'CDP' in str(e):
+                self.skipTest(f"Chrome session setup failed: {e}")
+            raise
+
+
+if __name__ == '__main__':
+    pytest.main([__file__, '-v'])