Files
huizenbot/autoscraper.py

317 lines
11 KiB
Python

#!/usr/bin/env python3
"""
autoscraper.py — detect CMS and extract patterns from broker pages
Usage:
python autoscraper.py listings <url> — detect CMS + card structure
python autoscraper.py details <url> — detect CMS + kenmerk patterns
"""
import re
import sys
import json
import httpx
from bs4 import BeautifulSoup, Tag
UA = "Huizenbot/1.0 (+mark@kalsbeek.dev) persoonlijk gebruik"
# ---------------------------------------------------------------------------
# CMS fingerprints
# ---------------------------------------------------------------------------
# Each entry: (name, listings_signal, details_signal, adapter_hint)
# signals are (selector, min_count) tuples — all must match
CMS_FINGERPRINTS = [
{
"name": "Realworks",
"listings": [("li.aanbodEntry", 1), ("span.kenmerkValue", 1)],
"details": [("span.kenmerkName", 3), ("span.kenmerkValue", 3)],
"hint": "fetch_realworks('{base_url}', '{makelaar}')",
},
]
# ---------------------------------------------------------------------------
# Candidate card selectors (tried in order for unknown CMS)
# ---------------------------------------------------------------------------
CARD_CANDIDATES = [
"li.aanbodEntry",
"article",
"li[class*=object]",
"li[class*=woning]",
"li[class*=listing]",
"div[class*=object-item]",
"div[class*=property-item]",
"div[class*=aanbod]",
".listing-item",
]
# ---------------------------------------------------------------------------
# Regex patterns for field detection
# ---------------------------------------------------------------------------
RE_POSTCODE = re.compile(r"\b\d{4}\s?[A-Z]{2}\b")
RE_PRICE = re.compile(r"\s*[\d.,]+")
RE_M2 = re.compile(r"\d+\s*m[²2]")
RE_PAGE_URL = re.compile(r"pagina[-/]?\d+|[?&]p(?:age)?=\d+|/\d+/?$")
# ---------------------------------------------------------------------------
# Helpers
# ---------------------------------------------------------------------------
def fetch(url: str) -> BeautifulSoup:
r = httpx.get(url, headers={"User-Agent": UA}, timeout=15, follow_redirects=True)
r.raise_for_status()
return BeautifulSoup(r.text, "html.parser")
def _selector_path(el: Tag) -> str:
"""Short CSS-like path for an element: tag.class1.class2"""
parts = []
for ancestor in reversed(list(el.parents)):
if ancestor.name in (None, "[document]", "html", "body"):
continue
cls = ".".join(ancestor.get("class", []))
parts.append(f"{ancestor.name}.{cls}" if cls else ancestor.name)
if len(parts) >= 3:
break
cls = ".".join(el.get("class", []))
parts.append(f"{el.name}.{cls}" if cls else el.name)
return " > ".join(parts[-3:])
def _detect_cms(soup: BeautifulSoup, mode: str) -> dict | None:
key = "listings" if mode == "listings" else "details"
for cms in CMS_FINGERPRINTS:
if all(len(soup.select(sel)) >= n for sel, n in cms[key]):
return cms
return None
def _find_cards(soup: BeautifulSoup) -> tuple[list, str | None]:
for sel in CARD_CANDIDATES:
found = soup.select(sel)
if len(found) >= 2:
return found, sel
# fallback: find the most repeated element class
from collections import Counter
class_counts: Counter = Counter()
for el in soup.find_all(True):
cls = tuple(el.get("class", []))
if cls:
class_counts[cls] += 1
if class_counts:
top_cls, count = class_counts.most_common(1)[0]
if count >= 2:
sel = "." + ".".join(top_cls)
return soup.select(sel), f"{sel} (auto-detected, count={count})"
return [], None
def _pattern_hits(soup: BeautifulSoup, pattern: re.Pattern, label: str):
hits = []
for el in soup.find_all(string=pattern):
parent = el.parent
if parent:
hits.append((parent.get_text(strip=True)[:80], _selector_path(parent)))
if hits:
print(f"\n [{label}] — {len(hits)} hit(s)")
for text, path in hits[:4]:
print(f" {path}")
print(f"{text!r}")
# ---------------------------------------------------------------------------
# Commands
# ---------------------------------------------------------------------------
def cmd_listings(url: str):
print(f"Fetching: {url}\n")
soup = fetch(url)
base_url = "/".join(url.split("/")[:3])
cms = _detect_cms(soup, "listings")
if cms:
print(f"✓ CMS DETECTED: {cms['name']}")
hint = cms["hint"].format(base_url=base_url, makelaar="<name>")
print(f"\n Add to ssr.py:\n")
print(f" def fetch_<name>() -> list[RawListing]:")
print(f" return {hint}\n")
print(f" Register in SCRAPERS dict:")
print(f" '<name>': fetch_<name>,")
return
print("✗ CMS unknown — structural diagnostics:\n")
# Cards
cards, matched_sel = _find_cards(soup)
print(f"=== CARDS ({matched_sel or 'none found'}: {len(cards)}) ===")
if cards:
print("\n--- FIRST CARD ---")
print(cards[0].prettify()[:2500])
print("\n--- CHILD ELEMENTS & CLASSES ---")
for el in cards[0].find_all(True):
cls = el.get("class")
text = el.get_text(strip=True)[:50]
if cls:
print(f" <{el.name}> .{' .'.join(cls)} {text!r}")
# Pattern hits in cards area (or full page if no cards)
search_area = cards[0] if cards else soup
print("\n=== FIELD PATTERNS ===")
_pattern_hits(search_area, RE_POSTCODE, "postcode")
_pattern_hits(search_area, RE_PRICE, "prijs")
_pattern_hits(search_area, RE_M2, "")
# Pagination
print("\n=== PAGINATION ===")
page_links = soup.find_all("a", href=RE_PAGE_URL)
if page_links:
seen = set()
for a in page_links:
href = a.get("href", "")
if href not in seen:
seen.add(href)
print(f" {href!r}{a.get_text(strip=True)!r}")
else:
print(" No pagination links found")
def cmd_details(url: str):
print(f"Fetching: {url}\n")
soup = fetch(url)
cms = _detect_cms(soup, "details")
if cms:
print(f"✓ CMS DETECTED: {cms['name']}")
print("\n _realworks_detail() will extract:")
kv: dict[str, str] = {}
for kenmerk in soup.select("span.kenmerk"):
label_el = kenmerk.select_one("span.kenmerkName")
value_el = kenmerk.select_one("span.kenmerkValue")
if label_el and value_el:
label = label_el.get_text(strip=True).lower()
value = value_el.get_text(strip=True)
kv[label] = value
target_fields = {
"type woning": "woningtype",
"bouwjaar": "bouwjaar",
"woonoppervlakte": "woonoppervlak",
"perceeloppervlakte": "perceeloppervlak",
"aantal kamers": "kamers",
"aantal slaapkamers": "slaapkamers",
"energieklasse": "energielabel",
}
for key, field in target_fields.items():
val = kv.get(key, "NOT FOUND")
status = "" if key in kv else ""
print(f" {status} {field:<20}{key!r}: {val!r}")
return
print("✗ CMS unknown — structural diagnostics:\n")
# Address
print("=== ADDRESS ===")
for tag in ["h1", "h2"]:
for el in soup.select(tag):
t = el.get_text(strip=True)
if t:
print(f" <{tag}> {t!r}")
# Key-value patterns
print("\n=== KEY-VALUE STRUCTURES ===")
kv_selectors = [
("dl", "dt", "dd"),
("table", "th", "td"),
(".kenmerk", ".kenmerkName", ".kenmerkValue"),
(".spec", ".spec-label", ".spec-value"),
(".feature", ".feature-label", ".feature-value"),
]
found_any = False
for container_sel, label_sel, value_sel in kv_selectors:
pairs = []
for container in soup.select(container_sel)[:50]:
label_el = container.select_one(label_sel)
value_el = container.select_one(value_sel)
if label_el and value_el:
l = label_el.get_text(strip=True)
v = value_el.get_text(strip=True)
if l and v:
pairs.append((l, v))
if pairs:
found_any = True
print(f"\n [{container_sel} > {label_sel} / {value_sel}] — {len(pairs)} pairs")
for l, v in pairs[:10]:
print(f" {l:<30} {v}")
if not found_any:
print(" No key-value structures detected")
# Field pattern hits
print("\n=== FIELD PATTERNS ===")
_pattern_hits(soup, RE_POSTCODE, "postcode")
_pattern_hits(soup, RE_PRICE, "prijs")
_pattern_hits(soup, RE_M2, "")
# Images
print("\n=== IMAGES (first 5) ===")
for img in soup.select("img")[:5]:
src = img.get("src") or img.get("data-src")
alt = img.get("alt", "")
print(f" {src} [{alt}]")
# JSON-LD
print("\n=== JSON-LD (schema.org) ===")
for tag in soup.select('script[type="application/ld+json"]'):
try:
ld = json.loads(tag.string)
offered = ld.get("itemOffered", {})
address = offered.get("address", {})
floor_size = offered.get("floorSize", {})
fields = {
"woningtype": offered.get("@type"),
"adres": address.get("streetAddress"),
"postcode": address.get("postalCode"),
"stad": address.get("addressLocality"),
"prijs": ld.get("price"),
"woonoppervlak": floor_size.get("value"),
"kamers": offered.get("numberOfRooms"),
"bouwjaar": offered.get("yearBuilt"),
"availability": ld.get("availability"),
"image": ld.get("image"),
}
for k, v in fields.items():
mark = "" if v is not None else ""
print(f" {mark} {k:<16} {v!r}")
except Exception as e:
print(f" parse fout: {e}")
# ---------------------------------------------------------------------------
# Entry point
# ---------------------------------------------------------------------------
def main():
if len(sys.argv) < 3:
print(__doc__)
sys.exit(1)
cmd = sys.argv[1]
url = sys.argv[2]
if cmd == "listings":
cmd_listings(url)
elif cmd == "details":
cmd_details(url)
else:
print(f"Unknown command: {cmd}")
sys.exit(1)
if __name__ == "__main__":
main()