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
This commit is contained in:
@@ -38,29 +38,39 @@ def _schieland_detail(detail_url: str) -> dict:
|
|||||||
postcode_el = soup.select_one("div.house__status p")
|
postcode_el = soup.select_one("div.house__status p")
|
||||||
postcode = _extract_postcode(postcode_el.get_text()) if postcode_el else None
|
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] = {}
|
kv: dict[str, str] = {}
|
||||||
kenmerken = soup.select_one("#kenmerken")
|
for block in soup.select("div.house-features__block"):
|
||||||
if kenmerken:
|
h4 = block.select_one("h4")
|
||||||
for li in kenmerken.select("li"):
|
if not h4:
|
||||||
label_el = li.select_one("strong")
|
continue
|
||||||
value_el = li.select_one("span")
|
section_title = h4.get_text(strip=True).lower()
|
||||||
if label_el and value_el:
|
|
||||||
# Strip nested links (e.g. "Hypotheek berekenen")
|
for li in block.select("ul > li"):
|
||||||
for a in value_el.select("a"):
|
strong = li.select_one("strong")
|
||||||
a.decompose()
|
span = li.select_one("span")
|
||||||
kv[label_el.get_text(strip=True).lower()] = value_el.get_text(strip=True)
|
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 {
|
return {
|
||||||
"postcode": postcode,
|
"postcode": postcode,
|
||||||
"status": kv.get("status", "").lower(),
|
"status": kv.get("overdracht.status", "").lower(),
|
||||||
"woningtype": kv.get("soort bouw"),
|
"woningtype": kv.get("bouwvorm.soort bouw"),
|
||||||
"bouwjaar": kv.get("bouwjaar"),
|
"bouwjaar": kv.get("bouwvorm.bouwjaar"),
|
||||||
"woonoppervlak": kv.get("woonoppervlakte"),
|
"woonoppervlak": kv.get("indeling.woonoppervlakte"),
|
||||||
"perceeloppervlak": kv.get("perceeloppervlakte"),
|
"perceeloppervlak": kv.get("indeling.perceeloppervlakte"),
|
||||||
"kamers": kv.get("aantal kamers"),
|
"kamers": kv.get("indeling.aantal kamers"),
|
||||||
"slaapkamers": kv.get("aantal slaapkamers"),
|
"slaapkamers": kv.get("indeling.aantal slaapkamers"),
|
||||||
"energielabel": kv.get("energielabel"),
|
"energielabel": kv.get("energie & installatie.energielabel"),
|
||||||
}
|
}
|
||||||
except Exception as e:
|
except Exception as e:
|
||||||
log.warning("schielandborsboom: detail fetch fout %s: %s", detail_url, e)
|
log.warning("schielandborsboom: detail fetch fout %s: %s", detail_url, e)
|
||||||
@@ -74,9 +84,9 @@ def fetch_schielandborsboom() -> list[RawListing]:
|
|||||||
|
|
||||||
while True:
|
while True:
|
||||||
if page == 1:
|
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:
|
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)
|
soup = fetch_soup(url)
|
||||||
cards = soup.select("div.card.card--house")
|
cards = soup.select("div.card.card--house")
|
||||||
@@ -98,13 +108,18 @@ def fetch_schielandborsboom() -> list[RawListing]:
|
|||||||
if not stad or stad.lower() != "schiedam":
|
if not stad or stad.lower() != "schiedam":
|
||||||
continue
|
continue
|
||||||
|
|
||||||
# Status from card-house__thumb second class
|
# Status from card-house__status badge
|
||||||
thumb = card.select_one("div.card-house__thumb")
|
status_el = card.select_one("div.card-house__status")
|
||||||
status_classes = thumb.get("class", []) if thumb else []
|
status_text = status_el.get_text(strip=True).lower() if status_el else ""
|
||||||
status_text = next(
|
# Check for known status keywords in badge text
|
||||||
(c for c in status_classes if c != "card-house__thumb"), "beschikbaar"
|
if "beschikbaar" in status_text:
|
||||||
).lower()
|
status = "beschikbaar"
|
||||||
status = _SCHIELAND_STATUS_MAP.get(status_text, "beschikbaar")
|
elif "onder bod" in status_text:
|
||||||
|
status = "onder_bod"
|
||||||
|
elif "verkocht" in status_text:
|
||||||
|
status = "verkocht"
|
||||||
|
else:
|
||||||
|
status = "beschikbaar"
|
||||||
|
|
||||||
# Price
|
# Price
|
||||||
prijs = parse_prijs(_text(card, "p.price"))
|
prijs = parse_prijs(_text(card, "p.price"))
|
||||||
|
|||||||
Reference in New Issue
Block a user