init: scripts diversos (crawlers, conversores, scrapers)

scraper/test_improved_parser.py

@@ -0,0 +1,43 @@
+import json
+
+# Simular a resposta do Gemini
+gemini_response = '```json\n{\n  "relevante": true,\n  "problema": "Não há um problema explícito no texto."\n}\n```'
+
+print("RESPOSTA BRUTA:")
+print(repr(gemini_response))
+
+# Aplicar a lógica melhorada
+content_text = gemini_response
+
+# Remover blocos markdown
+if '```json' in content_text:
+    content_text = content_text.split('```json')[1].split('```')[0]
+elif '```' in content_text:
+    content_text = content_text.split('```')[1].split('```')[0]
+
+# Limpeza agressiva
+content_text = content_text.strip()
+
+print("\n\nAPÓS LIMPEZA:")
+print(repr(content_text))
+
+try:
+    knowledge = json.loads(content_text)
+    print("\n\n✅ JSON PARSE SUCESSO!")
+    print(json.dumps(knowledge, indent=2, ensure_ascii=False))
+except json.JSONDecodeError as e:
+    print(f"\n\n❌ JSON PARSE FALHOU: {e}")
+    
+    # Fallback: extrair { ... } manualmente
+    start = content_text.find('{')
+    end = content_text.rfind('}') + 1
+    if start != -1 and end > start:
+        clean_json = content_text[start:end]
+        print(f"\n\nFALLBACK EXTRACT:")
+        print(repr(clean_json))
+        try:
+            knowledge = json.loads(clean_json)
+            print("\n✅ FALLBACK SUCESSO!")
+            print(json.dumps(knowledge, indent=2, ensure_ascii=False))
+        except Exception as e2:
+            print(f"❌ FALLBACK FALHOU: {e2}")