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 +1,85 @@
+"""Test scraper Miami-Dade Clerk con 3 dias para verificar parsing antes del full run."""
+from __future__ import annotations
+import io, sys, time
+from pathlib import Path
+
+sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding="utf-8", errors="replace")
+ROOT = Path(__file__).resolve().parent.parent
+sys.path.insert(0, str(ROOT))
+
+
+def main() -> int:
+    print("=" * 70)
+    print("Miami-Dade Clerk scraper smoke test")
+    print("=" * 70)
+
+    def log(m: str) -> None:
+        print(f"  {m}")
+
+    # ───────────────────────────────────────────────────────────────────────
+    # PASO 1: scrape sin persistir, solo para ver que data sale
+    # ───────────────────────────────────────────────────────────────────────
+    from scrapers.miami_dade_clerk import scrape_miami_dade_auctions
+
+    print("\n--- PASO 1: scrape (no DB), 3 dias ahead ---")
+    t0 = time.perf_counter()
+    deals = scrape_miami_dade_auctions(days_ahead=3, status_cb=log)
+    elapsed = time.perf_counter() - t0
+    print(f"\nResult: {len(deals)} deals in {elapsed:.1f}s")
+
+    if not deals:
+        print("❌ NO DEALS scraped. Aborting test.")
+        return 1
+
+    # Print first 5 deals con detalle
+    print()
+    print("--- SAMPLE DEALS (first 5) ---")
+    for i, d in enumerate(deals[:5], 1):
+        print(f"\n  Deal [{i}]")
+        for k in ["deal_type", "case_number", "auction_date", "starting_bid",
+                  "estimated_arv", "address", "city", "state", "zip", "county",
+                  "listing_price"]:
+            v = d.get(k)
+            print(f"    {k}: {v}")
+        desc = d.get("listing_description", "")[:120]
+        print(f"    description: {desc}")
+
+    # ───────────────────────────────────────────────────────────────────────
+    # PASO 2: Validacion estructural
+    # ───────────────────────────────────────────────────────────────────────
+    print()
+    print("--- PASO 2: estructura ---")
+    required_fields = ["source", "deal_type", "case_number", "auction_date",
+                       "county", "listing_price"]
+    failures = 0
+    for i, d in enumerate(deals):
+        for f in required_fields:
+            if d.get(f) is None or d.get(f) == "":
+                if f == "listing_price":
+                    # OK si tiene starting_bid pero no listing_price
+                    if d.get("starting_bid"):
+                        continue
+                print(f"  ⚠️ deal {i+1} ({d.get('case_number')}): missing {f}")
+                failures += 1
+    if failures == 0:
+        print(f"  ✅ All {len(deals)} deals have required fields")
+    else:
+        print(f"  ⚠️ {failures} field-missing instances across deals")
+
+    # Field types
+    types_ok = True
+    for d in deals[:5]:
+        if d.get("starting_bid") and not isinstance(d["starting_bid"], (int, float)):
+            print(f"  ⚠️ starting_bid not numeric in {d.get('case_number')}: {d['starting_bid']!r}")
+            types_ok = False
+        if d.get("auction_date") and not (isinstance(d["auction_date"], str) and len(d["auction_date"]) == 10):
+            print(f"  ⚠️ auction_date not ISO YYYY-MM-DD in {d.get('case_number')}")
+            types_ok = False
+    if types_ok:
+        print("  ✅ Field types OK (numeric pricing, ISO dates)")
+
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())