Cache postcodes

finder/zoopla.py

@@ -29,6 +29,7 @@ from urllib.parse import parse_qsl, urlencode, urljoin, urlparse, urlunparse
 
 import httpx
 
+import shutdown
 from constants import (
     DATA_DIR,
     DELAY_BETWEEN_PAGES,
@@ -973,7 +974,7 @@ def _paginate(
         listing["_detail"] = fetch_detail(url)
         if not cached:
             detail_state["fetched"] += 1
-            time.sleep(DELAY_BETWEEN_PAGES)
+            shutdown.sleep(DELAY_BETWEEN_PAGES)
 
     all_listings = _extract_listings(page)
     for listing in all_listings:
@@ -988,13 +989,15 @@ def _paginate(
     page_num = 2
 
     while True:
+        if shutdown.stop_requested():
+            break
         next_url = _find_next_page_url(page)
         if not next_url:
             if total_results > 0 and len(all_listings) >= total_results:
                 break
             next_url = _url_with_page(page.url, page_num)
 
-        time.sleep(DELAY_BETWEEN_PAGES)
+        shutdown.sleep(DELAY_BETWEEN_PAGES)
 
         try:
             page.goto(next_url, wait_until="domcontentloaded", timeout=30000)
@@ -1119,9 +1122,23 @@ def _extract_outcode(text: str) -> str | None:
 
 # listingId -> parsed detail dict (or None). Failures are cached too, so a
 # broken listing is not re-fetched within a run (the same listing reappears
-# across overlapping outcode searches).
+# across overlapping outcode searches). Seeded from / dumped to a persistent
+# on-disk cache by the orchestrator (see postcode_cache.py) so a recurring
+# scrape only re-fetches newly-listed properties — the biggest saving for
+# Zoopla, whose detail fetch drives a real browser tab.
 _detail_cache: dict[str, dict | None] = {}
 
+
+def seed_detail_cache(mapping: dict) -> None:
+    """Pre-populate the in-memory detail cache from a persisted snapshot."""
+    if mapping:
+        _detail_cache.update(mapping)
+
+
+def detail_cache_snapshot() -> dict:
+    """Return a JSON-serialisable copy of the in-memory detail cache."""
+    return dict(_detail_cache)
+
 _LISTING_ID_RE = re.compile(r"/details/(\d+)/?")
 
 # The property's own location is carried by a `"location":{...}` wrapper and a