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 +1,105 @@
+"""Test interactivo de court_records.py — solo Duval.
+
+Empieza con sitios publicos:
+1. Property Appraiser (paopropertysearch.coj.net) — sin login
+2. Official Records (or.duvalclerk.com) — con disclaimer
+
+Si los selectores no matchean, dump page.content() a un archivo para
+inspeccionar la estructura real y ajustar.
+"""
+from __future__ import annotations
+
+import io
+import os
+import sys
+import time
+from pathlib import Path
+
+# Stdout UTF-8 para emojis
+sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding="utf-8", errors="replace")
+
+ROOT = Path(__file__).resolve().parent.parent
+sys.path.insert(0, str(ROOT))
+
+# Forzar el flag para esta sesion de test
+os.environ["ENABLE_COURT_RECORDS"] = "true"
+
+from data_fetchers.court_records import fetch_court_records  # noqa: E402
+
+
+def main() -> int:
+    print("=" * 70)
+    print("Test interactivo: court_records.py — Duval scraper")
+    print("=" * 70)
+    print(f"ENABLE_COURT_RECORDS: {os.getenv('ENABLE_COURT_RECORDS')}")
+    print()
+
+    test_addresses = [
+        ("117 W Duval St, Jacksonville, FL 32202", "Jacksonville City Hall — confirma scraper funciona"),
+        ("5005 N Pearl St, Jacksonville, FL 32206", "Pearl St real (encontrado en busqueda broad)"),
+        ("3245 N Pearl St, Jacksonville, FL 32206", "Fictional address — expect no results (graceful)"),
+    ]
+
+    for addr, desc in test_addresses:
+        print("─" * 70)
+        print(f"TEST: {addr}")
+        print(f"  ({desc})")
+        print("─" * 70)
+        t0 = time.perf_counter()
+        result = fetch_court_records(address=addr, county_name="Duval")
+        elapsed = time.perf_counter() - t0
+        print(f"  status: {result.get('status')}")
+        print(f"  owner_name: {result.get('owner_name')}")
+        print(f"  re_number: {result.get('re_number')}")
+        print(f"  lis_pendens_count: {result.get('lis_pendens_count', 0)}")
+        print(f"  sources_used: {result.get('sources_used', [])}")
+        print(f"  errors: {len(result.get('errors', []))} errors")
+        for e in (result.get('errors') or [])[:3]:
+            print(f"    - {e[:200]}")
+        print(f"  elapsed: {elapsed:.1f}s")
+        print()
+
+    return 0
+
+
+def main_OLD() -> int:
+    address = "3245 N Pearl St, Jacksonville, FL 32206"
+    print(f"Target address: {address}")
+    t0 = time.perf_counter()
+    result = fetch_court_records(address=address, county_name="Duval")
+    elapsed = time.perf_counter() - t0
+
+    print(f"Tiempo total: {elapsed:.1f}s")
+    print()
+    print("=" * 70)
+    print("RESULTADO")
+    print("=" * 70)
+    for k, v in result.items():
+        if k == "lis_pendens" and isinstance(v, list):
+            print(f"  {k}: {len(v)} case(s)")
+            for i, c in enumerate(v, 1):
+                print(f"    [{i}] {c}")
+        elif k == "errors":
+            print(f"  {k}: ({len(v)} errores)")
+            for e in v:
+                print(f"    - {e}")
+        else:
+            print(f"  {k}: {v}")
+
+    print()
+    if result.get("status") == "LIS_PENDENS_ACTIVE":
+        print("  → DETECCION CONFIRMADA: foreclosure case activo en Duval")
+        return 0
+    elif result.get("status") == "CLEAN":
+        print("  → Sin lis pendens encontrado (puede ser limpio o owner_name no matchea)")
+        return 0
+    elif result.get("status") == "UNKNOWN":
+        print("  → UNKNOWN: scraper fallo en algun paso. Ver errors arriba.")
+        return 1
+    else:
+        print(f"  → Status: {result.get('status')}")
+        return 1
+
+
+if __name__ == "__main__":
+    sys.exit(main())