feat: AR-House initial commit

2026-07-03 12:24:58 -04:00
commit 047c05287a
216 changed files with 127552 additions and 0 deletions
@@ -0,0 +1,84 @@
+"""E2E test of search_engine.py against real deals.db (125 deals from B1+B3)."""
+from __future__ import annotations
+import io, sys
+from pathlib import Path
+from collections import Counter
+
+sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding="utf-8", errors="replace")
+ROOT = Path(__file__).resolve().parent.parent
+sys.path.insert(0, str(ROOT))
+
+from search_engine import search_existing_only, preflight_check
+from scrapers.registry import list_sources, get_sources_for_county
+
+
+def main() -> int:
+    print("=" * 70)
+    print("E2E TEST: Search Engine against real deals.db (125 deals)")
+    print("=" * 70)
+
+    # Sim 1
+    print("\n--- Sim 1: Miami-Dade, both sources, no filters ---")
+    r = search_existing_only(
+        counties=["Miami-Dade"], source_ids=["miami_dade_clerk", "hud_homestore"]
+    )
+    print(f"  {len(r)} deals matched")
+    for d in r[:3]:
+        addr = (d.get("address") or "?")[:50]
+        price = d.get("listing_price") or 0
+        print(f"    {d.get('case_number')}: {d.get('deal_type'):<12} "
+              f"${price:>10,.0f} score={d.get('classification_score')} {addr}")
+
+    # Sim 2: HUD nationwide
+    print("\n--- Sim 2: HUD only, no county filter (all FL counties) ---")
+    r = search_existing_only(source_ids=["hud_homestore"])
+    print(f"  {len(r)} HUD deals total")
+    counties = Counter(d.get("county") for d in r)
+    print(f"  Top 10 counties:")
+    for c, n in counties.most_common(10):
+        print(f"    {c}: {n}")
+
+    # Sim 3: Price + beds filter
+    print("\n--- Sim 3: $300K-$400K, beds>=3 ---")
+    r = search_existing_only(
+        filters={"min_price": 300000, "max_price": 400000, "beds_min": 3}
+    )
+    print(f"  {len(r)} deals in range")
+    for d in r[:5]:
+        addr = (d.get("address") or "?")[:50]
+        print(f"    {d.get('source'):<22} | {d.get('deal_type'):<12} | "
+              f"${d.get('listing_price') or 0:>8,.0f} | {d.get('beds')}bd | {addr}")
+
+    # Sim 4: only red_flag classification
+    print("\n--- Sim 4: classification=red_flag ---")
+    r = search_existing_only(filters={"classifications": ["red_flag"]})
+    print(f"  {len(r)} red_flag deals")
+
+    # Sim 5: preflight check
+    print("\n--- Sim 5: preflight_check(both free sources) ---")
+    pf = preflight_check(["miami_dade_clerk", "hud_homestore"])
+    print(f"  total_credits: {pf['total_credits_estimated']}")
+    print(f"  ok_to_run: {pf['ok_to_run']}")
+    print(f"  budget used: {pf['budget_snapshot']['credits_used']}/{pf['budget_snapshot']['credits_budget']}")
+    print(f"  warnings: {pf['warnings']}")
+
+    # Sim 6: get_sources_for_county
+    print("\n--- Sim 6: get_sources_for_county('Miami-Dade') ---")
+    sources = get_sources_for_county("Miami-Dade")
+    for s in sources:
+        print(f"  {s['id']}: free={s['free']}, deal_types={s['deal_types_produced']}")
+
+    print("\n--- Sim 7: get_sources_for_county('Broward') (no clerk yet) ---")
+    sources = get_sources_for_county("Broward")
+    for s in sources:
+        print(f"  {s['id']}: free={s['free']}, deal_types={s['deal_types_produced']}")
+
+    print()
+    print("=" * 70)
+    print("✅ Search engine works against real deals.db")
+    print("=" * 70)
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())