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 +1,92 @@
+"""scrapers/miami_dade_clerk.py — Backward-compat shim.
+
+ESTA MODULO ES UN ALIAS DE realauction_clerk.py para Miami-Dade.
+
+La logica generica fue extraida a `realauction_clerk.py` cuando se agregaron
+soporte para Duval, Broward, Palm Beach, Hillsborough, Orange (todos comparten
+el mismo white-label platform de realauction.com).
+
+Mantener este shim mientras:
+- Tests legacy importen de `scrapers.miami_dade_clerk`
+- Migration scripts referencien `_parse_cases_from_html`, `_is_status_dead`
+- El source_id 'miami_dade_clerk' siga en deals.db
+
+API publica preservada:
+    SOURCE                              — string 'miami_dade_clerk'
+    scrape_miami_dade_auctions(...)     — scrape function
+    run_scraper_to_db(...)              — pipeline
+    _parse_cases_from_html(...)         — exposed for tests
+    _is_status_dead(...)                — exposed for tests
+    _DEAD_STATUS_SUBSTRINGS             — exposed for tests
+    _build_deal_record(case, date)      — exposed for tests
+"""
+from __future__ import annotations
+
+from typing import Callable, Optional
+
+from scrapers.realauction_clerk import (
+    REALAUCTION_COUNTIES,
+    get_county_config,
+    scrape_realauction_county,
+    run_scraper_to_db as _generic_run_scraper_to_db,
+    # Re-exports for backward compat
+    _parse_cases_from_html,   # noqa: F401
+    _is_status_dead,           # noqa: F401
+    _DEAD_STATUS_SUBSTRINGS,   # noqa: F401
+    _extract_case_from_table_rows,  # noqa: F401
+    _parse_address,            # noqa: F401
+    _parse_money,              # noqa: F401
+    _build_description,        # noqa: F401
+)
+
+SOURCE = "miami_dade_clerk"
+_COUNTY = "Miami-Dade"
+
+
+def scrape_miami_dade_auctions(
+    *,
+    days_ahead: int = 14,
+    days_back: int = 0,
+    status_cb: Optional[Callable[[str], None]] = None,
+    max_dates: Optional[int] = None,
+    use_cache: bool = True,
+    cache_ttl_seconds: int = 86400,
+) -> list[dict]:
+    """Scrape Miami-Dade County auction calendar (delegates to realauction_clerk)."""
+    return scrape_realauction_county(
+        county=_COUNTY,
+        days_ahead=days_ahead,
+        days_back=days_back,
+        status_cb=status_cb,
+        max_dates=max_dates,
+        use_cache=use_cache,
+        cache_ttl_seconds=cache_ttl_seconds,
+    )
+
+
+def run_scraper_to_db(
+    *,
+    days_ahead: int = 14,
+    days_back: int = 0,
+    auto_classify: bool = True,
+    status_cb: Optional[Callable[[str], None]] = None,
+    max_dates: Optional[int] = None,
+) -> dict:
+    """Full pipeline for Miami-Dade (delegates to realauction_clerk)."""
+    return _generic_run_scraper_to_db(
+        county=_COUNTY,
+        days_ahead=days_ahead,
+        days_back=days_back,
+        auto_classify=auto_classify,
+        status_cb=status_cb,
+        max_dates=max_dates,
+    )
+
+
+def _build_deal_record(case: dict, auction_date_iso: str) -> dict:
+    """Backward-compat: original signature was (case, date) without county_config.
+
+    Tests use this. Internally delegates to the generic builder with Miami-Dade config.
+    """
+    from scrapers.realauction_clerk import _build_deal_record as _generic_build
+    return _generic_build(case, auction_date_iso, REALAUCTION_COUNTIES[_COUNTY])