feat(logging): Implement auto-whitelisting heuristics for log sessions

tests/test_auto_whitelist.py

@@ -0,0 +1,76 @@
+import os
+import pytest
+from datetime import datetime
+from log_registry import LogRegistry
+
+@pytest.fixture
+def registry_setup(tmp_path):
+    registry_path = tmp_path / "log_registry.toml"
+    logs_dir = tmp_path / "logs"
+    logs_dir.mkdir()
+    registry = LogRegistry(str(registry_path))
+    return registry, logs_dir
+
+def test_auto_whitelist_keywords(registry_setup):
+    registry, logs_dir = registry_setup
+    session_id = "test_kw"
+    session_dir = logs_dir / session_id
+    session_dir.mkdir()
+    
+    # Create comms.log with ERROR
+    comms_log = session_dir / "comms.log"
+    comms_log.write_text("Some message\nAN ERROR OCCURRED\nMore text")
+    
+    registry.register_session(session_id, str(session_dir), datetime.now())
+    registry.update_auto_whitelist_status(session_id)
+    
+    assert registry.is_session_whitelisted(session_id)
+    assert "ERROR" in registry.data[session_id]["metadata"]["reason"]
+
+def test_auto_whitelist_message_count(registry_setup):
+    registry, logs_dir = registry_setup
+    session_id = "test_msg_count"
+    session_dir = logs_dir / session_id
+    session_dir.mkdir()
+    
+    # Create comms.log with > 10 lines
+    comms_log = session_dir / "comms.log"
+    comms_log.write_text("\n".join(["msg"] * 15))
+    
+    registry.register_session(session_id, str(session_dir), datetime.now())
+    registry.update_auto_whitelist_status(session_id)
+    
+    assert registry.is_session_whitelisted(session_id)
+    assert registry.data[session_id]["metadata"]["message_count"] == 15
+
+def test_auto_whitelist_large_size(registry_setup):
+    registry, logs_dir = registry_setup
+    session_id = "test_large"
+    session_dir = logs_dir / session_id
+    session_dir.mkdir()
+    
+    # Create large file (> 50KB)
+    large_file = session_dir / "large.log"
+    large_file.write_text("x" * 60000)
+    
+    registry.register_session(session_id, str(session_dir), datetime.now())
+    registry.update_auto_whitelist_status(session_id)
+    
+    assert registry.is_session_whitelisted(session_id)
+    assert "Large session size" in registry.data[session_id]["metadata"]["reason"]
+
+def test_no_auto_whitelist_insignificant(registry_setup):
+    registry, logs_dir = registry_setup
+    session_id = "test_insignificant"
+    session_dir = logs_dir / session_id
+    session_dir.mkdir()
+    
+    # Small file, few lines, no keywords
+    comms_log = session_dir / "comms.log"
+    comms_log.write_text("hello\nworld")
+    
+    registry.register_session(session_id, str(session_dir), datetime.now())
+    registry.update_auto_whitelist_status(session_id)
+    
+    assert not registry.is_session_whitelisted(session_id)
+    assert registry.data[session_id]["metadata"]["message_count"] == 2