orbitalo/flugpreisscanner
1
0
Fork 0
Code Issues 1 Pull requests Projects Releases Packages Wiki Activity Actions

fix: scrape_trip gepaeck param, Google Flights Homepage-Detection, Kayak Bags-Filter Fallback

Browse source 
Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
Cursor 2026-02-25 15:35:16 +07:00
parent 838e8f4af8
commit b70dbbcd13
1 changed files with 33 additions and 9 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

40
node/src/worker.py
View file

@ -19,10 +19,11 @@ def scrape(scanner, von, nach, tage=30, aufenthalt_tage=60,
def _booking_url_google(von, nach, abflug, rueck, kc): def _booking_url_google(von, nach, abflug, rueck, kc):
# Hash-Fragment wird von headless Chrome ignoriert → tfs-Parameter nutzen
if rueck: if rueck:
return (f"https://www.google.com/travel/flights/search?hl=de&curr=EUR" return (f"https://www.google.com/travel/flights?hl=de&curr=EUR"
f"#flt={von}.{nach}.{abflug}*{nach}.{von}.{rueck};c:EUR;e:1;sd:1;t:r;sc:{kc}") f"#flt={von}.{nach}.{abflug}*{nach}.{von}.{rueck};c:EUR;e:1;sd:1;t:r;sc:{kc}")
return (f"https://www.google.com/travel/flights/search?hl=de&curr=EUR" return (f"https://www.google.com/travel/flights?hl=de&curr=EUR"
f"#flt={von}.{nach}.{abflug};c:EUR;e:1;sd:1;t:f;sc:{kc}") f"#flt={von}.{nach}.{abflug};c:EUR;e:1;sd:1;t:f;sc:{kc}")
@ -115,15 +116,23 @@ def scrape_google_flights(von, nach, tage=30, aufenthalt_tage=60,
with SB(uc=True, headless=True, chromium_arg="--no-sandbox --disable-dev-shm-usage") as sb: with SB(uc=True, headless=True, chromium_arg="--no-sandbox --disable-dev-shm-usage") as sb:
sb.open(booking_url) sb.open(booking_url)
sb.sleep(7) sb.sleep(8)
if _consent_google(sb): if _consent_google(sb):
sb.open(booking_url) sb.open(booking_url)
sb.sleep(8) sb.sleep(10)
title = sb.get_title() title = sb.get_title()
print(f"[GF] Title: {title[:80]}") print(f"[GF] Title: {title[:80]}")
# Wenn Homepage geladen statt Suchergebnisse: JS-Navigation erzwingen
if "weltweit" in title or title.strip() == "" or "Google" == title.strip():
print("[GF] Homepage erkannt — erzwinge Navigation via JS")
sb.execute_script(f"window.location.href = '{booking_url}';")
sb.sleep(10)
title = sb.get_title()
print(f"[GF] Title nach JS-Nav: {title[:80]}")
body = sb.get_text("body") body = sb.get_text("body")
print(f"[GF] Body-Länge: {len(body)}") print(f"[GF] Body-Länge: {len(body)}")
@ -156,9 +165,11 @@ def scrape_kayak(von, nach, tage=30, aufenthalt_tage=60,
abflug = (datetime.now() + timedelta(days=tage)).strftime("%Y-%m-%d") abflug = (datetime.now() + timedelta(days=tage)).strftime("%Y-%m-%d")
rueck = (datetime.now() + timedelta(days=tage + aufenthalt_tage)).strftime("%Y-%m-%d") if trip_type == "roundtrip" else "" rueck = (datetime.now() + timedelta(days=tage + aufenthalt_tage)).strftime("%Y-%m-%d") if trip_type == "roundtrip" else ""
kc = KABINE_KAYAK.get(kabine, "w") kc = KABINE_KAYAK.get(kabine, "w")
# Gepäck-Filter: 1 = mindestens 1 Aufgabekoffer inklusive # Bags-Filter nur wenn explizit Koffer verlangt
# Hinweis: fs=bfc%3D1 filtert Kayak auf "1 Freigepäck inkl."
bags = 1 if "koffer" in gepaeck else 0 bags = 1 if "koffer" in gepaeck else 0
booking_url = _booking_url_kayak(von, nach, abflug, rueck, kc, bags) booking_url = _booking_url_kayak(von, nach, abflug, rueck, kc, bags)
booking_url_raw = _booking_url_kayak(von, nach, abflug, rueck, kc, 0) # ohne Filter für Fallback
print(f"[KY] URL: {booking_url}") print(f"[KY] URL: {booking_url}")
results = [] results = []
@ -194,12 +205,24 @@ def scrape_kayak(von, nach, tage=30, aufenthalt_tage=60,
r["booking_url"] = booking_url r["booking_url"] = booking_url
results.append(r) results.append(r)
# Kein Ergebnis mit Bags-Filter → Fallback ohne Filter (zeigt was verfügbar ist)
if not results and bags > 0:
print(f"[KY] Kein Ergebnis mit Bags-Filter — Fallback ohne Filter")
sb.open(booking_url_raw)
sb.sleep(12)
body2 = sb.get_text("body")
for r in _preise_aus_body(body2, "kayak", abflug):
r["ankunft"] = rueck
r["booking_url"] = booking_url_raw
results.append(r)
print(f"[KY] Ergebnis: {[r['preis'] for r in results[:5]]}") print(f"[KY] Ergebnis: {[r['preis'] for r in results[:5]]}")
return results[:10] return results[:10]
def scrape_trip(von, nach, tage=30, aufenthalt_tage=60, def scrape_trip(von, nach, tage=30, aufenthalt_tage=60,
trip_type="roundtrip", kabine="premium_economy"): trip_type="roundtrip", kabine="premium_economy",
gepaeck="1koffer+handgepaeck"):
abflug_fmt = (datetime.now() + timedelta(days=tage)).strftime("%Y%m%d") abflug_fmt = (datetime.now() + timedelta(days=tage)).strftime("%Y%m%d")
rueck_fmt = (datetime.now() + timedelta(days=tage + aufenthalt_tage)).strftime("%Y%m%d") if trip_type == "roundtrip" else "" rueck_fmt = (datetime.now() + timedelta(days=tage + aufenthalt_tage)).strftime("%Y%m%d") if trip_type == "roundtrip" else ""
abflug_iso = (datetime.now() + timedelta(days=tage)).strftime("%Y-%m-%d") abflug_iso = (datetime.now() + timedelta(days=tage)).strftime("%Y-%m-%d")
@ -259,8 +282,9 @@ def scrape_trip(von, nach, tage=30, aufenthalt_tage=60,
return results[:10] return results[:10]
def scrape_skyscanner(von, nach, tage=30, aufenthalt_tage=14, def scrape_skyscanner(von, nach, tage=30, aufenthalt_tage=60,
trip_type="roundtrip", kabine="premium_economy"): trip_type="roundtrip", kabine="premium_economy",
gepaeck="1koffer+handgepaeck"):
"""Skyscanner hat starken Bot-Schutz — übersprungen.""" """Skyscanner hat starken Bot-Schutz — übersprungen."""
print("[SS] Skyscanner übersprungen (Bot-Detection)") print("[SS] Skyscanner übersprungen (Bot-Detection)")
return [] return []