update

news_crawler/test_ollama.py

@@ -0,0 +1,129 @@
+#!/usr/bin/env python
+"""
+Test script for Ollama integration
+Tests connection, configuration, and summarization
+"""
+from config import Config
+from ollama_client import OllamaClient
+
+print("\n" + "="*70)
+print("Ollama Integration Test")
+print("="*70)
+
+# Print configuration
+Config.print_config()
+
+# Validate configuration
+issues = Config.validate()
+if issues:
+    print("⚠ Configuration Issues:")
+    for issue in issues:
+        print(f"  - {issue}")
+    print()
+
+# Initialize client
+client = OllamaClient(
+    base_url=Config.OLLAMA_BASE_URL,
+    model=Config.OLLAMA_MODEL,
+    api_key=Config.OLLAMA_API_KEY,
+    enabled=Config.OLLAMA_ENABLED,
+    timeout=Config.OLLAMA_TIMEOUT
+)
+
+# Test 1: Check if Ollama is enabled
+print("Test 1: Configuration Check")
+print(f"  Ollama Enabled: {Config.OLLAMA_ENABLED}")
+if not Config.OLLAMA_ENABLED:
+    print("  ⚠ Ollama is disabled. Set OLLAMA_ENABLED=true in .env to enable.")
+    print("\n" + "="*70)
+    exit(0)
+
+# Test 2: Test connection
+print("\nTest 2: Connection Test")
+conn_result = client.test_connection()
+print(f"  Available: {conn_result['available']}")
+print(f"  Current Model: {conn_result['current_model']}")
+
+if conn_result['available']:
+    print(f"  ✓ Connected to Ollama server")
+    if conn_result['models']:
+        print(f"  Available models: {', '.join(conn_result['models'][:5])}")
+        if conn_result['current_model'] not in conn_result['models']:
+            print(f"  ⚠ Warning: Model '{conn_result['current_model']}' not found in available models")
+else:
+    print(f"  ✗ Connection failed: {conn_result['error']}")
+    print("\n" + "="*70)
+    exit(1)
+
+# Test 3: Test summarization with sample article
+print("\nTest 3: Summarization Test")
+print("  Testing with sample German article...")
+
+sample_article = """
+Die neue U-Bahn-Linie, die das Münchner Stadtzentrum mit dem Flughafen verbindet, wurde heute eröffnet. 
+Oberbürgermeister Dieter Reiter nahm zusammen mit hunderten Anwohnern an der Eröffnungszeremonie teil. 
+Die Linie wird die Reisezeit zwischen dem Flughafen und der Münchner Innenstadt erheblich verkürzen. 
+Der Bau dauerte fünf Jahre und kostete etwa 2 Milliarden Euro. Die neue Linie umfasst 10 Stationen 
+und verkehrt während der Hauptverkehrszeiten alle 10 Minuten. Experten erwarten, dass die neue Verbindung 
+den Verkehr in der Stadt deutlich entlasten wird. Die Münchner Verkehrsgesellschaft rechnet mit täglich 
+über 50.000 Fahrgästen auf der neuen Strecke.
+"""
+
+result = client.summarize_article(sample_article, max_words=Config.SUMMARY_MAX_WORDS)
+
+print(f"\n  Success: {result['success']}")
+if result['success']:
+    print(f"  ✓ Summarization successful!")
+    print(f"\n  Original word count: {result['original_word_count']}")
+    print(f"  Summary word count: {result['summary_word_count']}")
+    print(f"  Compression ratio: {result['original_word_count'] / max(result['summary_word_count'], 1):.1f}x")
+    print(f"  Duration: {result['duration']:.2f}s")
+    print(f"\n  Summary (English):")
+    print(f"  {'-'*70}")
+    print(f"  {result['summary']}")
+    print(f"  {'-'*70}")
+else:
+    print(f"  ✗ Summarization failed: {result['error']}")
+
+# Test 4: Test with English article
+print("\nTest 4: English Article Test")
+print("  Testing with English article...")
+
+english_article = """
+The city council approved a new bike lane network spanning 50 kilometers across Munich. 
+The project aims to promote sustainable transportation and reduce car traffic in the city center. 
+Construction will begin next month and is expected to be completed within two years. 
+The bike lanes will connect major residential areas with business districts and public transport hubs. 
+Environmental groups have praised the initiative as a significant step toward carbon neutrality.
+"""
+
+result2 = client.summarize_article(english_article, max_words=50)
+
+print(f"\n  Success: {result2['success']}")
+if result2['success']:
+    print(f"  ✓ Summarization successful!")
+    print(f"  Original: {result2['original_word_count']} words → Summary: {result2['summary_word_count']} words")
+    print(f"  Duration: {result2['duration']:.2f}s")
+    print(f"\n  Summary:")
+    print(f"  {result2['summary']}")
+else:
+    print(f"  ✗ Summarization failed: {result2['error']}")
+
+# Summary
+print("\n" + "="*70)
+print("Test Summary")
+print("="*70)
+print(f"✓ Configuration: Valid")
+print(f"✓ Connection: {'Success' if conn_result['available'] else 'Failed'}")
+print(f"✓ German→English: {'Success' if result['success'] else 'Failed'}")
+print(f"✓ English→English: {'Success' if result2['success'] else 'Failed'}")
+print("="*70)
+
+if result['success'] and result2['success']:
+    print("\n🎉 All tests passed! Ollama integration is working correctly.")
+    print("\nYou can now run the crawler with AI summarization:")
+    print("  python crawler_service.py 5")
+else:
+    print("\n⚠ Some tests failed. Check the errors above.")
+
+print()