Initial commit - WhyRating Engine (Google Reviews Scraper)

packages/reviewiq-pipeline/scripts/fix_l1_configs_v2.py

@@ -0,0 +1,238 @@
+#!/usr/bin/env python3
+"""
+Guarded L1 Config Fixer - V2 (Threshold-based, Sector-scoped)
+
+Only applies fixes when:
+1. Evidence is from sector-scoped validation
+2. Frequency exceeds threshold (default 3%)
+3. Changes are logged with version bump
+
+Usage:
+    python fix_l1_configs_v2.py --apply         # Apply fixes from validation
+    python fix_l1_configs_v2.py --dry-run       # Show what would change
+    python fix_l1_configs_v2.py --revert SECTOR # Revert to previous version
+"""
+
+import argparse
+import json
+from datetime import datetime, timezone
+from pathlib import Path
+from typing import Any
+
+CONFIGS_DIR = Path(__file__).parent.parent / "data" / "primitive_configs" / "l1"
+CHANGELOG_FILE = CONFIGS_DIR / "CHANGELOG.json"
+
+# Minimum threshold for auto-enabling (% of sector spans)
+ENABLE_THRESHOLD_PCT = 3.0
+
+# Fixes derived from sector-scoped validation (validate_l1_configs_v2.py output)
+# These are the ONLY fixes that should be applied
+SECTOR_SCOPED_FIXES = {
+    "ENTERTAINMENT": {
+        "evidence": "2,320 spans from Go Karts + Soho Club",
+        "enable": [
+            ("TASTE", 4.3, "Entertainment venues have concessions/food service"),
+        ],
+        "add_weight": [
+            ("CRAFT", 1.3, "13.4% frequency but unweighted"),
+        ],
+        "remove_weight": [],
+    },
+    "FOOD_DINING": {
+        "evidence": "61 spans from Fika cafe",
+        "enable": [
+            ("COMFORT", 9.8, "Seating/atmosphere comfort matters for cafes"),
+        ],
+        "add_weight": [
+            ("AVAILABILITY", 1.2, "16.4% frequency but unweighted"),
+        ],
+        "remove_weight": [
+            # Note: Small sample size (61 spans) - these may be false negatives
+            # Keep weights but flag for review with more data
+        ],
+    },
+    "AUTOMOTIVE": {
+        "evidence": "1,201 spans from ClickRent car rental",
+        "enable": [],  # Nothing exceeds 3% threshold
+        "add_weight": [],
+        "remove_weight": [
+            # CONDITION, HONESTY, PROMISES, RECOVERY all have 0 appearances
+            # However, may be specific to rental vs repair - keep for now
+        ],
+    },
+}
+
+
+def load_changelog() -> list[dict]:
+    """Load the changelog file."""
+    if CHANGELOG_FILE.exists():
+        with open(CHANGELOG_FILE) as f:
+            return json.load(f)
+    return []
+
+
+def save_changelog(entries: list[dict]) -> None:
+    """Save the changelog file."""
+    CHANGELOG_FILE.parent.mkdir(parents=True, exist_ok=True)
+    with open(CHANGELOG_FILE, "w") as f:
+        json.dump(entries, f, indent=2)
+        f.write("\n")
+
+
+def load_config(sector_code: str) -> dict[str, Any] | None:
+    """Load a sector config."""
+    config_path = CONFIGS_DIR / f"{sector_code.lower()}_config.json"
+    if not config_path.exists():
+        return None
+    with open(config_path) as f:
+        return json.load(f)
+
+
+def save_config(sector_code: str, config: dict[str, Any]) -> None:
+    """Save a sector config."""
+    config_path = CONFIGS_DIR / f"{sector_code.lower()}_config.json"
+    with open(config_path, "w") as f:
+        json.dump(config, f, indent=2)
+        f.write("\n")
+
+
+def apply_fixes(sector_code: str, fixes: dict, dry_run: bool = False) -> list[str]:
+    """Apply fixes to a sector config."""
+    config = load_config(sector_code)
+    if not config:
+        return [f"❌ Config not found for {sector_code}"]
+
+    enabled = set(config.get("enabled", []))
+    disabled = set(config.get("disabled", []))
+    weights = config.get("weights", {})
+
+    changes = []
+    evidence = fixes.get("evidence", "unknown")
+
+    # Enable primitives
+    for prim, pct, reason in fixes.get("enable", []):
+        if pct < ENABLE_THRESHOLD_PCT:
+            changes.append(f"⚠️  SKIP {prim}: {pct:.1f}% below {ENABLE_THRESHOLD_PCT}% threshold")
+            continue
+
+        if prim in disabled:
+            disabled.remove(prim)
+            enabled.add(prim)
+            changes.append(f"✓ ENABLE {prim}: {pct:.1f}% in sector data ({reason})")
+        elif prim not in enabled:
+            enabled.add(prim)
+            changes.append(f"✓ ADD {prim}: {pct:.1f}% in sector data ({reason})")
+
+    # Add weights
+    for prim, weight, reason in fixes.get("add_weight", []):
+        if prim not in weights:
+            weights[prim] = weight
+            changes.append(f"⚖️  WEIGHT {prim}: {weight}x ({reason})")
+
+    # Remove weights
+    for prim, reason in fixes.get("remove_weight", []):
+        if prim in weights:
+            del weights[prim]
+            changes.append(f"⚖️  UNWEIGHT {prim}: ({reason})")
+
+    if not changes:
+        return ["✓ No changes needed"]
+
+    if not dry_run:
+        # Bump version
+        old_version = config.get("config_version", "1.0")
+        major, minor = old_version.split(".")
+        new_version = f"{major}.{int(minor) + 1}"
+
+        config["enabled"] = sorted(enabled)
+        config["disabled"] = sorted(disabled)
+        config["weights"] = dict(sorted(weights.items()))
+        config["config_version"] = new_version
+        config["config_updated_at"] = datetime.now(timezone.utc).isoformat()
+
+        save_config(sector_code, config)
+
+        # Log to changelog
+        changelog = load_changelog()
+        changelog.append({
+            "sector": sector_code,
+            "version": new_version,
+            "previous_version": old_version,
+            "timestamp": datetime.now(timezone.utc).isoformat(),
+            "evidence": evidence,
+            "changes": changes,
+        })
+        save_changelog(changelog)
+
+        changes.append(f"📝 Version: {old_version} → {new_version}")
+
+    return changes
+
+
+def revert_config(sector_code: str, to_version: str | None = None) -> list[str]:
+    """Revert a config to a previous version."""
+    changelog = load_changelog()
+
+    # Find entries for this sector
+    sector_entries = [e for e in changelog if e["sector"] == sector_code]
+    if not sector_entries:
+        return [f"❌ No changelog entries for {sector_code}"]
+
+    # TODO: Implement actual revert by storing full config snapshots
+    return [f"⚠️  Revert not yet implemented - manual restore required"]
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Guarded L1 config fixer")
+    parser.add_argument("--apply", action="store_true", help="Apply sector-scoped fixes")
+    parser.add_argument("--dry-run", action="store_true", help="Show what would change")
+    parser.add_argument("--revert", metavar="SECTOR", help="Revert sector to previous version")
+    parser.add_argument("--sector", help="Apply to specific sector only")
+    parser.add_argument("--show-changelog", action="store_true", help="Show changelog")
+
+    args = parser.parse_args()
+
+    if args.show_changelog:
+        changelog = load_changelog()
+        print(json.dumps(changelog, indent=2))
+        return
+
+    if args.revert:
+        changes = revert_config(args.revert.upper())
+        for change in changes:
+            print(change)
+        return
+
+    if args.apply or args.dry_run:
+        print("=" * 60)
+        print(f"L1 CONFIG FIXER V2 - {'DRY RUN' if args.dry_run else 'APPLYING FIXES'}")
+        print(f"Threshold: {ENABLE_THRESHOLD_PCT}%")
+        print("=" * 60)
+
+        sectors = [args.sector.upper()] if args.sector else SECTOR_SCOPED_FIXES.keys()
+
+        for sector in sectors:
+            if sector not in SECTOR_SCOPED_FIXES:
+                print(f"\n⚠️  {sector}: No sector-scoped fixes defined")
+                continue
+
+            print(f"\n📁 {sector}")
+            print(f"   Evidence: {SECTOR_SCOPED_FIXES[sector]['evidence']}")
+
+            changes = apply_fixes(sector, SECTOR_SCOPED_FIXES[sector], dry_run=args.dry_run)
+            for change in changes:
+                print(f"   {change}")
+
+        print("\n" + "=" * 60)
+        if args.dry_run:
+            print("DRY RUN - No changes applied")
+        else:
+            print("Fixes applied - see CHANGELOG.json for history")
+        print("=" * 60)
+        return
+
+    parser.print_help()
+
+
+if __name__ == "__main__":
+    main()