From 1841412c93cf2e490cc6366467162efa2b096544 Mon Sep 17 00:00:00 2001
From: Mark Kalsbeek <Mark@Kalsbeek.dev>
Date: Mon, 13 Apr 2026 22:58:30 +0200
Subject: [PATCH] Update schielandborsboom scraper for new URL structure

- Change listings URL to /wonen/zoeken/heel-nederland/prijs=200000-300000/schiedam/
- Update status extraction to read from card-house__status badge text
- Update detail parser to handle house-features__block HTML structure
- Test verified: successfully fetches Aleidastraat 130 A listing
---
 src/adapters/ssr/sure.py | 71 ++++++++++++++++++++++++----------------
 1 file changed, 43 insertions(+), 28 deletions(-)

diff --git a/src/adapters/ssr/sure.py b/src/adapters/ssr/sure.py
index bfc89f7..52b18b5 100644
--- a/src/adapters/ssr/sure.py
+++ b/src/adapters/ssr/sure.py
@@ -38,29 +38,39 @@ def _schieland_detail(detail_url: str) -> dict:
         postcode_el = soup.select_one("div.house__status p")
         postcode = _extract_postcode(postcode_el.get_text()) if postcode_el else None
 
-        # Parse #kenmerken section: <li><strong>label</strong><span>value</span></li>
+        # Parse house-features__block sections: div.house-features__block > ul > li
         kv: dict[str, str] = {}
-        kenmerken = soup.select_one("#kenmerken")
-        if kenmerken:
-            for li in kenmerken.select("li"):
-                label_el = li.select_one("strong")
-                value_el = li.select_one("span")
-                if label_el and value_el:
-                    # Strip nested links (e.g. "Hypotheek berekenen")
-                    for a in value_el.select("a"):
-                        a.decompose()
-                    kv[label_el.get_text(strip=True).lower()] = value_el.get_text(strip=True)
+        for block in soup.select("div.house-features__block"):
+            h4 = block.select_one("h4")
+            if not h4:
+                continue
+            section_title = h4.get_text(strip=True).lower()
+
+            for li in block.select("ul > li"):
+                strong = li.select_one("strong")
+                span = li.select_one("span")
+                if not strong or not span:
+                    continue
+
+                label = strong.get_text(strip=True).lower()
+                value = span.get_text(strip=True)
+
+                # Remove links from value
+                for a in span.select("a"):
+                    value = value.replace(a.get_text(strip=True), "").strip()
+
+                kv[f"{section_title}.{label}"] = value
 
         return {
             "postcode":         postcode,
-            "status":           kv.get("status", "").lower(),
-            "woningtype":       kv.get("soort bouw"),
-            "bouwjaar":         kv.get("bouwjaar"),
-            "woonoppervlak":    kv.get("woonoppervlakte"),
-            "perceeloppervlak": kv.get("perceeloppervlakte"),
-            "kamers":           kv.get("aantal kamers"),
-            "slaapkamers":      kv.get("aantal slaapkamers"),
-            "energielabel":     kv.get("energielabel"),
+            "status":           kv.get("overdracht.status", "").lower(),
+            "woningtype":       kv.get("bouwvorm.soort bouw"),
+            "bouwjaar":         kv.get("bouwvorm.bouwjaar"),
+            "woonoppervlak":    kv.get("indeling.woonoppervlakte"),
+            "perceeloppervlak": kv.get("indeling.perceeloppervlakte"),
+            "kamers":           kv.get("indeling.aantal kamers"),
+            "slaapkamers":      kv.get("indeling.aantal slaapkamers"),
+            "energielabel":     kv.get("energie & installatie.energielabel"),
         }
     except Exception as e:
         log.warning("schielandborsboom: detail fetch fout %s: %s", detail_url, e)
@@ -74,9 +84,9 @@ def fetch_schielandborsboom() -> list[RawListing]:
 
     while True:
         if page == 1:
-            url = f"{_SCHIELAND_BASE}/wonen?sure_koop_huur=koop"
+            url = f"{_SCHIELAND_BASE}/wonen/zoeken/heel-nederland/prijs=200000-300000/schiedam/"
         else:
-            url = f"{_SCHIELAND_BASE}/wonen/page/{page}/?sure_koop_huur=koop"
+            url = f"{_SCHIELAND_BASE}/wonen/zoeken/heel-nederland/prijs=200000-300000/schiedam/?pagina={page}"
 
         soup = fetch_soup(url)
         cards = soup.select("div.card.card--house")
@@ -98,13 +108,18 @@ def fetch_schielandborsboom() -> list[RawListing]:
                 if not stad or stad.lower() != "schiedam":
                     continue
 
-                # Status from card-house__thumb second class
-                thumb = card.select_one("div.card-house__thumb")
-                status_classes = thumb.get("class", []) if thumb else []
-                status_text = next(
-                    (c for c in status_classes if c != "card-house__thumb"), "beschikbaar"
-                ).lower()
-                status = _SCHIELAND_STATUS_MAP.get(status_text, "beschikbaar")
+                # Status from card-house__status badge
+                status_el = card.select_one("div.card-house__status")
+                status_text = status_el.get_text(strip=True).lower() if status_el else ""
+                # Check for known status keywords in badge text
+                if "beschikbaar" in status_text:
+                    status = "beschikbaar"
+                elif "onder bod" in status_text:
+                    status = "onder_bod"
+                elif "verkocht" in status_text:
+                    status = "verkocht"
+                else:
+                    status = "beschikbaar"
 
                 # Price
                 prijs = parse_prijs(_text(card, "p.price"))