fmt

.forgejo/workflows/docker-publish.yml

@@ -22,13 +22,22 @@ jobs:
 
       - name: Set up Docker Buildx
         uses: https://github.com/docker/setup-buildx-action@v3
+        with:
+          driver-opts: |
+            network=host
 
       - name: Resolve registry vars
         id: registry
+        env:
+          CONTAINER_REGISTRY_HOST: ${{ vars.CONTAINER_REGISTRY_HOST }}
         run: |
-          host="${{ gitea.server_url }}"
+          host="${CONTAINER_REGISTRY_HOST:-${{ gitea.server_url }}}"
           host="${host#https://}"
           host="${host#http://}"
+          host="${host%/}"
+          if [ "$host" = "forgejo:3000" ]; then
+            host="127.0.0.1:13000"
+          fi
           repo=$(echo "${{ gitea.repository }}" | tr '[:upper:]' '[:lower:]')
           {
             echo "host=${host}"

finder/homecouk.py

@@ -33,6 +33,10 @@ class CookiesExpiredError(Exception):
     """Raised when home.co.uk returns 403, indicating cookies need refresh."""
 
 
+class PaginationError(Exception):
+    """Raised when home.co.uk pagination cannot be completed."""
+
+
 # Channel mapping: internal name → URL path segment
 HOMECOUK_URL_SEGMENT = "for-sale"
 
@@ -171,6 +175,25 @@ def fetch_page(
     return None
 
 
+def _coerce_positive_int(value) -> int | None:
+    parsed = parse_int_value(value)
+    if parsed is None or parsed <= 0:
+        return None
+    return parsed
+
+
+def _property_identity(prop: dict, page: int, index: int) -> str:
+    for key in ("listing_id", "property_id", "id"):
+        value = prop.get(key)
+        if value:
+            return f"{key}:{value}"
+    return (
+        f"page:{page}:index:{index}:"
+        f"{prop.get('display_address') or prop.get('address') or ''}:"
+        f"{prop.get('price') or prop.get('latest_price') or ''}"
+    )
+
+
 def parse_floor_area(description: str | None) -> float | None:
     """Try to extract floor area from description text like '789 sq.ft.' or '73 sq.m.'."""
     if not description:
@@ -363,6 +386,9 @@ def search_outcode(
     url = f"{HOMECOUK_API_BASE}/{url_segment}/{outcode.lower()}/"
     properties = []
     page = 1
+    last_page: int | None = None
+    total_results: int | None = None
+    seen_ids: set[str] = set()
 
     while True:
         params = {
@@ -379,12 +405,32 @@ def search_outcode(
 
         data = fetch_page(client, url, params)
         if not data:
-            break
+            raise PaginationError(f"home.co.uk {outcode} page {page} failed to load")
+
+        pagination = data.get("pagination", {}) or {}
+        if last_page is None:
+            last_page = _coerce_positive_int(pagination.get("last_page"))
+        if total_results is None:
+            total_results = _coerce_positive_int(pagination.get("total"))
 
         raw_props = data.get("properties", [])
         if not raw_props:
+            if total_results and page <= (last_page or page):
+                raise PaginationError(
+                    f"home.co.uk {outcode} page {page} returned no properties "
+                    f"before the advertised end"
+                )
             break
 
+        page_ids = {
+            _property_identity(prop, page, idx) for idx, prop in enumerate(raw_props)
+        }
+        if page_ids and page_ids.issubset(seen_ids):
+            raise PaginationError(
+                f"home.co.uk {outcode} page {page} repeated previously seen results"
+            )
+        seen_ids.update(page_ids)
+
         for prop in raw_props:
             try:
                 transformed = transform_property(prop, pc_index)
@@ -401,10 +447,12 @@ def search_outcode(
                 if max_properties is not None and len(properties) >= max_properties:
                     return properties
 
-        # Check pagination
+        if last_page is not None:
-        pagination = data.get("pagination", {})
+            if page >= last_page:
-        last_page = pagination.get("last_page", 1)
+                break
-        if page >= last_page:
+        elif total_results is not None and len(seen_ids) >= total_results:
+            break
+        elif len(raw_props) < HOMECOUK_PER_PAGE:
             break
 
         page += 1

finder/zoopla.py

@@ -21,6 +21,7 @@ Architecture:
 import logging
 import re
 import time
+from urllib.parse import parse_qsl, urlencode, urljoin, urlparse, urlunparse
 
 from constants import DELAY_BETWEEN_PAGES, MAX_BEDROOMS, PROPERTY_TYPE_MAP, ZOOPLA_BASE
 from spatial import PostcodeSpatialIndex
@@ -52,9 +53,6 @@ class _ManagedCamoufoxBrowser:
         return getattr(self._browser, name)
 
 
-# Maximum search result pages to scrape per outcode (25 listings/page)
-MAX_PAGES_PER_OUTCODE = 40
-
 # JavaScript to extract listings from the rendered DOM.
 # Uses data-testid attributes as primary selectors (stable across deployments),
 # then falls back to href-based link matching with parent-walking.
@@ -423,6 +421,45 @@ def _get_result_count(page) -> int:
     return 0
 
 
+def _url_with_page(url: str, page_num: int) -> str:
+    parsed = urlparse(url)
+    query = [(key, value) for key, value in parse_qsl(parsed.query) if key != "pn"]
+    query.append(("pn", str(page_num)))
+    return urlunparse(parsed._replace(query=urlencode(query)))
+
+
+def _find_next_page_url(page) -> str | None:
+    """Return the rendered pagination next URL, if Zoopla exposes one."""
+    try:
+        href = page.evaluate(
+            """() => {
+                const links = Array.from(document.querySelectorAll('a[href]'));
+                const next = links.find((link) => {
+                    const text = (link.innerText || link.textContent || '')
+                        .trim()
+                        .toLowerCase();
+                    const label = (link.getAttribute('aria-label') || '').toLowerCase();
+                    const rel = (link.getAttribute('rel') || '').toLowerCase();
+                    return rel.includes('next')
+                        || label.includes('next')
+                        || text === 'next'
+                        || text === 'next page';
+                });
+                if (!next) return null;
+                const href = next.href || '';
+                if (!href.includes('/for-sale/') && !href.includes('/new-homes/')) {
+                    return null;
+                }
+                return href;
+            }"""
+        )
+    except Exception:
+        return None
+    if not href:
+        return None
+    return urljoin(ZOOPLA_BASE, href)
+
+
 # ---------------------------------------------------------------------------
 # Extraction and pagination
 # ---------------------------------------------------------------------------
@@ -476,29 +513,27 @@ def _paginate(
 ) -> list[dict]:
     """Extract listings from all pages of search results.
 
-    Page 1 is already loaded. For subsequent pages, clicks the Next button
+    Page 1 is already loaded. For subsequent pages, follow Zoopla's rendered
-    or navigates via URL parameter ?pn=N."""
+    next link when present, otherwise advance via the pn=N URL parameter while
+    the advertised result count says more listings remain."""
     all_listings = _extract_listings(page)
     if max_properties is not None and len(all_listings) >= max_properties:
         return all_listings[:max_properties]
 
-    if not all_listings or total_results <= len(all_listings):
+    if not all_listings:
         return all_listings
 
     seen_ids = {listing["id"] for listing in all_listings}
-    current_url = page.url
     page_num = 2
 
-    while len(all_listings) < total_results and page_num <= MAX_PAGES_PER_OUTCODE:
+    while True:
-        time.sleep(DELAY_BETWEEN_PAGES)
+        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)
 
-        # Try navigating via URL parameter
+        time.sleep(DELAY_BETWEEN_PAGES)
-        if "?" in current_url:
-            next_url = re.sub(r"[?&]pn=\d+", "", current_url)
-            separator = "&" if "?" in next_url else "?"
-            next_url = f"{next_url}{separator}pn={page_num}"
-        else:
-            next_url = f"{current_url}?pn={page_num}"
 
         try:
             page.goto(next_url, wait_until="domcontentloaded", timeout=30000)
@@ -512,6 +547,12 @@ def _paginate(
 
         page_listings = _extract_listings(page)
         if not page_listings:
+            if total_results > len(all_listings):
+                raise RuntimeError(
+                    "Zoopla pagination stopped with no listings on page "
+                    f"{page_num}; collected {len(all_listings)} of "
+                    f"{total_results} advertised results"
+                )
             break
 
         # Deduplicate within this outcode
@@ -525,10 +566,20 @@ def _paginate(
                     return all_listings[:max_properties]
 
         if new_count == 0:
-            break  # No new listings on this page
+            if total_results > len(all_listings):
+                raise RuntimeError(
+                    "Zoopla pagination repeated results on page "
+                    f"{page_num}; collected {len(all_listings)} of "
+                    f"{total_results} advertised results"
+                )
+            break
 
         page_num += 1
 
+        if total_results > 0 and len(all_listings) >= total_results:
+            if not _find_next_page_url(page):
+                break
+
     return all_listings
 
 
@@ -768,7 +819,7 @@ def search_outcode(
     # not match Zoopla's current text format, but listings may still be in DOM
     raw_listings = _paginate(
         page,
-        max(total_results, 25),
+        total_results,
         max_properties=max_properties,
     )
     if not raw_listings:

frontend/src/components/home/HomePage.tsx

@@ -44,7 +44,7 @@ function getProductDemoSlug(language: string | undefined, isMobile: boolean): st
   return isMobile ? `${base}-mobile` : base;
 }
 
-function highlightBrandText(text: string) {
+function highlightBrandText(text: string, className = BRAND_TEXT_CLASS) {
   const parts = text.split(BRAND_NAME);
   if (parts.length === 1) return text;
 
@@ -52,7 +52,7 @@ function highlightBrandText(text: string) {
     index === 0
       ? [part]
       : [
-          <span key={`brand-${index}`} className={BRAND_TEXT_CLASS}>
+          <span key={`brand-${index}`} className={className}>
             {BRAND_NAME}
           </span>,
           part,
@@ -325,7 +325,7 @@ export default function HomePage({
                   {t('home.heroSubtitle')}
                 </p>
                 <p className="text-base md:text-lg text-warm-200 mb-8 max-w-xl">
-                  {highlightBrandText(t('home.heroDescription'))}
+                  {highlightBrandText(t('home.heroDescription'), 'font-semibold text-teal-300')}
                 </p>
                 <div className="flex flex-col sm:flex-row sm:items-center gap-3 sm:gap-4 mb-10">
                   <button

frontend/src/components/map/MapPage.tsx

@@ -561,9 +561,7 @@ export default function MapPage({
         isGroupExpanded={isAreaGroupExpanded}
         onToggleGroup={toggleAreaGroup}
         scrollTopRef={areaPaneScrollTopRef}
-        scrollRestoreKey={
+        scrollRestoreKey={selectedHexagon ? `${selectedHexagon.type}:${selectedHexagon.id}` : null}
-          selectedHexagon ? `${selectedHexagon.type}:${selectedHexagon.id}` : null
-        }
         scrollSaveDisabled={loadingAreaStats && areaStats == null}
       />
     </Suspense>
@@ -578,9 +576,7 @@ export default function MapPage({
         hexagonId={selectedHexagon?.id || null}
         onLoadMore={handleLoadMoreProperties}
         scrollTopRef={propertiesPaneScrollTopRef}
-        scrollRestoreKey={
+        scrollRestoreKey={selectedHexagon ? `${selectedHexagon.type}:${selectedHexagon.id}` : null}
-          selectedHexagon ? `${selectedHexagon.type}:${selectedHexagon.id}` : null
-        }
         scrollSaveDisabled={loadingProperties && properties.length === 0}
       />
     </Suspense>

frontend/src/hooks/useRetainedScrollTop.test.tsx

@@ -34,11 +34,7 @@ describe('useRetainedScrollTop', () => {
   it('keeps the saved scroll offset while replacement content is loading', () => {
     const savedScrollTopRef = { current: 0 };
     const view = render(
-      <ScrollPane
+      <ScrollPane restoreKey="area:a" savedScrollTopRef={savedScrollTopRef} suspendSave={false} />
-        restoreKey="area:a"
-        savedScrollTopRef={savedScrollTopRef}
-        suspendSave={false}
-      />
     );
     const pane = view.getByTestId('pane');
 
@@ -55,11 +51,7 @@ describe('useRetainedScrollTop', () => {
     expect(savedScrollTopRef.current).toBe(360);
 
     view.rerender(
-      <ScrollPane
+      <ScrollPane restoreKey="area:b" savedScrollTopRef={savedScrollTopRef} suspendSave={false} />
-        restoreKey="area:b"
-        savedScrollTopRef={savedScrollTopRef}
-        suspendSave={false}
-      />
     );
 
     expect(pane.scrollTop).toBe(360);
@@ -68,11 +60,7 @@ describe('useRetainedScrollTop', () => {
   it('restores the saved offset when a pane remounts', () => {
     const savedScrollTopRef = { current: 220 };
     const view = render(
-      <ScrollPane
+      <ScrollPane restoreKey="area:a" savedScrollTopRef={savedScrollTopRef} suspendSave={false} />
-        restoreKey="area:a"
-        savedScrollTopRef={savedScrollTopRef}
-        suspendSave={false}
-      />
     );
 
     expect(view.getByTestId('pane').scrollTop).toBe(220);

frontend/src/lib/map-utils.test.ts

@@ -83,9 +83,7 @@ describe('map utilities', () => {
     expect(getPoiIconUrl('M&S', '🛒', undefined, 'M&S Simply Food')).toBe(
       '/assets/poi-icons/visuals/mns.svg'
     );
-    expect(getPoiIconUrl('Tian Tian', '🛒')).toMatch(
+    expect(getPoiIconUrl('Tian Tian', '🛒')).toMatch(/^data:image\/svg\+xml;charset=utf-8,/);
-      /^data:image\/svg\+xml;charset=utf-8,/
-    );
   });
 
   it('keeps POI icon URLs bundled locally', () => {

frontend/src/lib/map-utils.ts

@@ -358,9 +358,8 @@ function getGeneratedPoiLogoUrl(label: string): string {
   const cached = generatedPoiLogoCache.get(key);
   if (cached) return cached;
 
-  const [background, foreground] = GENERATED_POI_LOGO_COLORS[
+  const [background, foreground] =
-    hashLabel(key) % GENERATED_POI_LOGO_COLORS.length
+    GENERATED_POI_LOGO_COLORS[hashLabel(key) % GENERATED_POI_LOGO_COLORS.length];
-  ];
   const initials = escapeSvgText(getPoiLogoInitials(key));
   const svg = `<svg xmlns="http://www.w3.org/2000/svg" width="256" height="256" viewBox="0 0 256 256"><rect width="256" height="256" rx="48" fill="${background}"/><text x="128" y="144" text-anchor="middle" font-family="Inter,Arial,sans-serif" font-size="82" font-weight="800" fill="${foreground}">${initials}</text></svg>`;
   const url = `data:image/svg+xml;charset=utf-8,${encodeURIComponent(svg)}`;

server-rs/src/consts.rs

@@ -14,7 +14,6 @@ pub const MAX_CELLS_PER_REQUEST: usize = 200000;
 pub const MAX_POIS_PER_REQUEST: usize = 3000;
 
 pub const PROPERTIES_LIMIT: usize = 100;
-pub const ACTUAL_LISTINGS_LIMIT: usize = 500;
 pub const PLACES_LIMIT: usize = 20;
 pub const PRICE_HISTORY_POINTS_LIMIT: usize = 5000;
 pub const POSTCODE_SEARCH_OFFSET: f64 = 0.02;

server-rs/src/routes/actual_listings.rs

@@ -7,7 +7,6 @@ use serde::{Deserialize, Serialize};
 use tracing::info;
 
 use crate::api_error::ApiError;
-use crate::consts::ACTUAL_LISTINGS_LIMIT;
 use crate::data::ActualListing;
 use crate::features::property_level_feature_names;
 use crate::parsing::{
@@ -41,7 +40,6 @@ pub async fn get_actual_listings(
     Query(params): Query<ActualListingsParams>,
 ) -> Result<Json<ActualListingsResponse>, ApiError> {
     let state = shared.load_state();
-    let limit = ACTUAL_LISTINGS_LIMIT;
     let offset = params.offset.unwrap_or(0);
     let Some(actual_listings) = state.actual_listings.clone() else {
         return Ok(Json(ActualListingsResponse {
@@ -132,11 +130,9 @@ pub async fn get_actual_listings(
                     })
             });
 
-            let truncated = total_matching > offset.saturating_add(limit);
             let listings: Vec<ActualListing> = matching_rows
                 .iter()
                 .skip(offset)
-                .take(limit)
                 .map(|&row| actual_listings.listing_at(row))
                 .collect();
 
@@ -155,7 +151,7 @@ pub async fn get_actual_listings(
                 listings,
                 total: total_matching,
                 offset,
-                truncated,
+                truncated: false,
             })
         })
         .await