Compare commits

...

2 Commits

Author SHA1 Message Date
fc6f3ff809 add another makelaar 2026-04-13 23:19:05 +02:00
1841412c93 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
2026-04-13 22:58:30 +02:00
3 changed files with 111 additions and 28 deletions

View File

@@ -24,6 +24,7 @@ from .realworks import (
fetch_wassenaar,
fetch_roepman,
fetch_post,
fetch_vankleef,
)
from .sure import (
fetch_schielandborsboom,
@@ -60,4 +61,5 @@ SCRAPERS = {
'borgdorff': fetch_borgdorff,
'vanherk': fetch_vanherk,
'vanoord': fetch_vanoord,
'vankleef': fetch_vankleef,
}

View File

@@ -151,6 +151,72 @@ def fetch_morris() -> list[RawListing]:
return fetch_realworks("https://www.morrismakelaardij.nl", "morris")
def fetch_vankleef() -> list[RawListing]:
"""Fetch Van Kleef makelaars — only Schiedam, as specified."""
listings_path = f"/aanbod/woningaanbod/schiedam/koop"
listings = []
page = 1
while True:
url = f"https://www.vankleefmakelaars.nl{listings_path}/pagina-{page}/"
soup = fetch_soup(url)
cards = soup.select("li.aanbodEntry")
if not cards:
break
for card in cards:
try:
a_tag = card.select_one("a.aanbodEntryLink")
if not a_tag:
continue
listing_url = "https://www.vankleefmakelaars.nl" + a_tag["href"]
adres = _text(card, ".street-address")
postcode = (_text(card, ".postal-code") or "").replace(" ", "") or None
stad = _text(card, ".locality")
prijs = parse_prijs(_text(card, ".koopprijs .kenmerkValue"))
if prijs and prijs > config.MAX_PRICE:
continue
status_text = (_text(card, ".objectstatusbanner") or "").lower()
status = _REALWORKS_STATUS_MAP.get(status_text, "beschikbaar")
img_tag = card.select_one(".hoofdfoto img")
hero = img_tag["src"] if img_tag else None
kk = _realworks_detail(listing_url, "vankleef")
listings.append(RawListing(
url=listing_url,
source_makelaar="vankleef",
adres=adres,
postcode=postcode,
stad=stad,
prijs=prijs,
status=status,
hero_image_url=hero,
woningtype=kk.get("woningtype"),
bouwjaar=int(kk["bouwjaar"]) if kk.get("bouwjaar") else None,
woonoppervlak=parse_m2(kk.get("woonoppervlak")),
perceeloppervlak=parse_m2(kk.get("perceeloppervlak")),
kamers=int(kk["kamers"]) if kk.get("kamers") else None,
slaapkamers=int(kk["slaapkamers"]) if kk.get("slaapkamers") else None,
energielabel=kk.get("energielabel"),
))
if config.APP_ENV == "dev":
break
except Exception as e:
log.warning("vankleef: parse fout: %s", e)
if len(cards) < 10:
break
page += 1
log.info("vankleef: %d listings opgehaald", len(listings))
return listings
# ---------------------------------------------------------------------------
# Makelaardij Wassenaar (Schiedam) — Realworks CMS, JSON-LD listing page
# ---------------------------------------------------------------------------

View File

@@ -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"))