Delete Demo.py

Demo.py

@@ -1,76 +0,0 @@
-# Cell 11: Demo – Realistic Agent Behaviour (with failure, improvement, author negotiation)
-# This demo shows the agent reasoning, trying a wrong fix, seeing tests fail, then improving.
-# It uses the environment’s tools and the author’s feedback.
-
-print("\n" + "="*70)
-print("🎬 DEMO: Code Review Agent – Learning to Fix Bugs")
-print("="*70)
-
-env = CodeReviewEnv()
-env.set_task("easy")
-obs = env.reset()
-
-print(f"\n📌 TASK: {obs.pr_title}")
-print(f"📝 DESCRIPTION: {obs.pr_description}")
-print(f"\n💻 INITIAL CODE:\n{obs.code_snippet}\n")
-
-# ------------------------------------------------------------------
-# Step 1: Agent runs tests to see the failure
-# ------------------------------------------------------------------
-print("🔧 STEP 1: Agent runs tests (run_tests)")
-action = RunTests()
-obs, reward, done, _ = env.step(action)
-print(f"💬 Test output:\n{obs.last_tool_output[:200]}...")
-print(f"🏆 Reward: {reward.value:.3f}\n")
-
-# ------------------------------------------------------------------
-# Step 2: Agent inspects the code and decides on a (wrong) fix
-# ------------------------------------------------------------------
-print("🔧 STEP 2: Agent inspects the code and attempts a fix (WRONG)")
-# Simulate a bad fix (e.g., adding a print instead of fixing null check)
-bad_fix = "def get_user(id):\n    print('debug')\n    return users[id]"
-action = ProposeFix(fix_code=bad_fix)
-obs, reward, done, _ = env.step(action)
-print(f"💬 Proposed fix:\n{bad_fix}")
-print(f"💬 Test results:\n{obs.last_tool_output[:200]}...")
-print(f"🏆 Reward: {reward.value:.3f} (low – tests failed)\n")
-
-# ------------------------------------------------------------------
-# Step 3: Agent consults documentation to learn
-# ------------------------------------------------------------------
-print("🔧 STEP 3: Agent queries documentation (query_docs: 'null check')")
-action = QueryDocs(query_topic="null check")
-obs, reward, done, _ = env.step(action)
-print(f"💬 Docs response:\n{obs.last_tool_output}")
-print(f"🏆 Reward: {reward.value:.3f}\n")
-
-# ------------------------------------------------------------------
-# Step 4: Agent writes a comment to negotiate with the author
-# ------------------------------------------------------------------
-print("🔧 STEP 4: Agent comments to convince the author")
-action = WriteComment(comment_text="I checked the docs – a missing key will raise KeyError. We should add a condition: if id in users: return users[id] else: handle missing key.")
-obs, reward, done, _ = env.step(action)
-# Author's response is stored in env._comments
-author_resp = [c for c in env._comments if "Author" in c][-1] if env._comments else ""
-print(f"💬 Agent comment delivered")
-print(f"💬 Author response: {author_resp}")
-print(f"🏆 Reward: {reward.value:.3f}\n")
-
-# ------------------------------------------------------------------
-# Step 5: Agent proposes the correct fix
-# ------------------------------------------------------------------
-print("🔧 STEP 5: Agent proposes the correct fix")
-good_fix = "def get_user(id):\n    if id in users:\n        return users[id]"
-action = ProposeFix(fix_code=good_fix)
-obs, reward, done, _ = env.step(action)
-print(f"💬 Proposed fix:\n{good_fix}")
-print(f"💬 Test results:\n{obs.last_tool_output[:300]}")
-print(f"🏆 Final Reward: {reward.value:.3f}\n")
-
-# ------------------------------------------------------------------
-# Step 6: Show final reward breakdown (optional)
-# ------------------------------------------------------------------
-print("="*70)
-print("✅ Demo complete. The agent learned from test failures, used documentation,")
-print("   negotiated with the author, and finally fixed the bug correctly.")
-print("="*70)