Kalletlamadhav/sql_optimized_env_new

1"""2Exact simulation of the hackathon Phase 2 validator.3Runs reset → step for ALL tasks, captures reward exactly as validator does.4Flags any score == 0.0 or == 1.0 as a boundary violation.5"""6import requests, json, sys, time7 8BASE = "https://kalletlamadhav-sql-optimized-env-new.hf.space"9T = 12010 11TASKS = [12    "gst_missing_index",13    "gst_n_plus_one",14    "gst_multi_join",15    "pds_select_star",16    "railway_simple_filter",17    "pds_cartesian",18    "mgnrega_wildcard",19    "railway_tatkal_workload",20    "mgnrega_schema_e",21    "mgnrega_count",22    "railway_missing_index",23    "gst_unbounded_aggregation",24    "pds_n_plus_one",25    "mgnrega_implicit_cast",26]27 28ACTIONS = [29    {"optimized_query": "SELECT 1", "identified_pattern": "NONE", "explanation": "test", "index_statements": [], "schema_analysis": ""},30    {"optimized_query": "SELECT id FROM gst_invoice_records WHERE state_code = 'MH'", "identified_pattern": "MISSING_INDEX", "explanation": "use index", "index_statements": ["CREATE INDEX idx_state ON gst_invoice_records(state_code)"], "schema_analysis": ""},31    {"optimized_query": "SELECT card_id, district_code FROM ration_card_beneficiaries WHERE card_type = 'BPL'", "identified_pattern": "SELECT_STAR", "explanation": "remove star", "index_statements": [], "schema_analysis": ""},32]33 34failures = []35results = {}36 37print("=" * 60)38print("EXACT VALIDATOR SIMULATION — Phase 2")39print(f"Target: {BASE}")40print("=" * 60)41 42for tid in TASKS:43    print(f"\n[TASK] {tid}")44    task_results = []45    46    for action_idx, action in enumerate(ACTIONS):47        try:48            # Step 1: Reset49            r = requests.post(f"{BASE}/reset", 50                              params={"task_id": tid},51                              timeout=T)52            if r.status_code != 200:53                print(f"  ACTION {action_idx}: RESET FAILED {r.status_code}: {r.text[:100]}")54                task_results.append(f"RESET_FAIL_{r.status_code}")55                continue56            57            obs = r.json()58            actual_query = obs.get("current_query", "SELECT 1")59            60            # Use the actual query for one of our test actions61            if action_idx == 0:62                test_action = {63                    "optimized_query": actual_query,  # same query = no improvement baseline64                    "identified_pattern": "NONE",65                    "explanation": "baseline passthrough",66                    "index_statements": [],67                    "schema_analysis": ""68                }69            else:70                test_action = action71            72            # Step 2: Step73            s = requests.post(f"{BASE}/step", json=test_action, timeout=T)74            if s.status_code != 200:75                print(f"  ACTION {action_idx}: STEP FAILED {s.status_code}: {s.text[:100]}")76                task_results.append(f"STEP_FAIL_{s.status_code}")77                continue78            79            step_data = s.json()80            reward = step_data.get("reward", -999)81            reward_detail = step_data.get("reward_detail", {})82            83            try:84                reward = float(reward)85            except:86                reward = -999.087            88            in_range = (0.0 < reward < 1.0)89            strict_exact = (reward == 0.0 or reward == 1.0)90            91            status = "✅ PASS" if in_range else f"❌ FAIL reward={reward}"92            print(f"  ACTION {action_idx}: reward={reward:.4f} → {status}")93            94            if not in_range:95                failures.append({96                    "task": tid,97                    "action": action_idx,98                    "reward": reward,99                    "detail": reward_detail,100                    "obs_task_id": obs.get("task_id"),101                })102            103            task_results.append(reward)104            105        except Exception as e:106            print(f"  ACTION {action_idx}: EXCEPTION {e}")107            task_results.append(f"ERROR_{e}")108        109        time.sleep(0.5)110    111    results[tid] = task_results112 113print("\n" + "=" * 60)114print("SUMMARY")115print("=" * 60)116 117all_pass = len(failures) == 0118 119for tid, tr in results.items():120    nums = [r for r in tr if isinstance(r, float)]121    bads = [r for r in nums if not (0.0 < r < 1.0)]122    symbol = "✅" if not bads else "❌"123    print(f"  {symbol} {tid}: {tr}")124 125if failures:126    print(f"\n❌ {len(failures)} BOUNDARY VIOLATIONS FOUND:")127    for f in failures:128        print(f"  Task={f['task']} action={f['action']} reward={f['reward']}")129        print(f"    detail={json.dumps(f['detail'], indent=6)}")130else:131    print("\n✅ ALL SCORES STRICTLY IN (0.0, 1.0) — READY TO SUBMIT")132 133print(f"\nPHASE 2: {'ALL PASSED' if all_pass else 'FAILED — FIX NEEDED'}")134