fix: Google Flights Formular-Suche statt Hash-URL, alle scraper gepaeck-param fix
Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
Cursor
parent
b70dbbcd13
commit
c9e8684a88
1 changed files with 93 additions and 26 deletions
|
|
@ -106,42 +106,106 @@ def _consent_google(sb):
|
|||
def scrape_google_flights(von, nach, tage=30, aufenthalt_tage=60,
|
||||
trip_type="roundtrip", kabine="premium_economy",
|
||||
gepaeck="1koffer+handgepaeck"):
|
||||
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 ""
|
||||
abflug = (datetime.now() + timedelta(days=tage)).strftime("%Y-%m-%d")
|
||||
abflug_de = (datetime.now() + timedelta(days=tage)).strftime("%d.%m.%Y")
|
||||
rueck = (datetime.now() + timedelta(days=tage + aufenthalt_tage)).strftime("%Y-%m-%d") if trip_type == "roundtrip" else ""
|
||||
rueck_de = (datetime.now() + timedelta(days=tage + aufenthalt_tage)).strftime("%d.%m.%Y") if trip_type == "roundtrip" else ""
|
||||
kc = KABINE_GOOGLE.get(kabine, "w")
|
||||
booking_url = _booking_url_google(von, nach, abflug, rueck, kc)
|
||||
|
||||
print(f"[GF] URL: {booking_url[:100]}")
|
||||
# Stadtname-Mapping für die Suchfelder
|
||||
stadtname = {"FRA": "Frankfurt", "HAN": "Hanoi", "KTI": "Phnom Penh",
|
||||
"PNH": "Phnom Penh", "BKK": "Bangkok", "SGN": "Ho Chi Minh"}
|
||||
von_name = stadtname.get(von, von)
|
||||
nach_name = stadtname.get(nach, nach)
|
||||
|
||||
print(f"[GF] Formular-Suche: {von_name}→{nach_name} {abflug_de}")
|
||||
results = []
|
||||
|
||||
with SB(uc=True, headless=True, chromium_arg="--no-sandbox --disable-dev-shm-usage") as sb:
|
||||
sb.open(booking_url)
|
||||
sb.sleep(8)
|
||||
# 1. Startseite laden
|
||||
sb.open("https://www.google.com/travel/flights?hl=de&curr=EUR")
|
||||
sb.sleep(6)
|
||||
|
||||
if _consent_google(sb):
|
||||
sb.open(booking_url)
|
||||
sb.sleep(10)
|
||||
|
||||
title = sb.get_title()
|
||||
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")
|
||||
print(f"[GF] Body-Länge: {len(body)}")
|
||||
sb.sleep(4)
|
||||
|
||||
# 2. Kabinen-Typ setzen (Roundtrip / Economy dropdown)
|
||||
try:
|
||||
elems = sb.find_elements('[aria-label*="€"]', timeout=3)
|
||||
# Kabinen-Auswahl: nach "Premium Economy" suchen
|
||||
kab_label = {"e": "Economy", "w": "Premium Economy", "b": "Business", "f": "First"}
|
||||
print(f"[GF] Versuche Kabine zu setzen: {kab_label.get(kc, 'Premium Economy')}")
|
||||
# Kabinen-Dropdown öffnen (2. Dropdown in der Suchleiste)
|
||||
kab_btns = sb.find_elements('[class*="cabin"] button, [aria-label*="abine"], [data-value*="cabin"]', timeout=3)
|
||||
if not kab_btns:
|
||||
kab_btns = sb.find_elements('button[class*="VfPpkd"]', timeout=2)
|
||||
if kab_btns:
|
||||
kab_btns[1].click()
|
||||
sb.sleep(1)
|
||||
except Exception as e:
|
||||
print(f"[GF] Kabinen-Klick fehlgeschlagen: {e}")
|
||||
|
||||
# 3. Abflugort eingeben
|
||||
try:
|
||||
von_field = None
|
||||
for sel in ['input[placeholder*="Von"], input[aria-label*="Von"], input[aria-label*="Abflug"], '
|
||||
'input[placeholder*="Where from"]']:
|
||||
elems = sb.find_elements(sel, timeout=2)
|
||||
if elems:
|
||||
von_field = elems[0]
|
||||
break
|
||||
if not von_field:
|
||||
# Fallback: erster Input im Suchformular
|
||||
von_field = sb.find_element('div[class*="target"] input', timeout=3)
|
||||
|
||||
von_field.clear()
|
||||
von_field.send_keys(von_name)
|
||||
sb.sleep(2)
|
||||
# Ersten Vorschlag auswählen
|
||||
sb.find_element('[role="option"]', timeout=3).click()
|
||||
sb.sleep(1)
|
||||
print(f"[GF] Abflugort gesetzt: {von_name}")
|
||||
except Exception as e:
|
||||
print(f"[GF] Abflugort-Eingabe fehlgeschlagen: {e}")
|
||||
|
||||
# 4. Zielort eingeben
|
||||
try:
|
||||
nach_field = sb.find_element(
|
||||
'input[placeholder*="Wohin"], input[aria-label*="Ziel"], input[aria-label*="Where to"]',
|
||||
timeout=3)
|
||||
nach_field.clear()
|
||||
nach_field.send_keys(nach_name)
|
||||
sb.sleep(2)
|
||||
sb.find_element('[role="option"]', timeout=3).click()
|
||||
sb.sleep(1)
|
||||
print(f"[GF] Zielort gesetzt: {nach_name}")
|
||||
except Exception as e:
|
||||
print(f"[GF] Zielort-Eingabe fehlgeschlagen: {e}")
|
||||
|
||||
# 5. Suchen-Button klicken
|
||||
try:
|
||||
for sel in ['button[aria-label*="Suchen"], button[aria-label*="Search"]',
|
||||
'button[jsname="vLv7Lb"]', 'button[class*="search"]']:
|
||||
btns = sb.find_elements(sel, timeout=2)
|
||||
if btns:
|
||||
btns[0].click()
|
||||
print(f"[GF] Suche gestartet via {sel}")
|
||||
break
|
||||
except Exception as e:
|
||||
print(f"[GF] Such-Button fehlgeschlagen: {e}")
|
||||
|
||||
sb.sleep(12)
|
||||
title = sb.get_title()
|
||||
body = sb.get_text("body")
|
||||
print(f"[GF] Title: {title[:80]} | Body: {len(body)} chars")
|
||||
|
||||
# 6. Preise extrahieren
|
||||
try:
|
||||
elems = sb.find_elements('[aria-label*="€"], [aria-label*="EUR"]', timeout=3)
|
||||
for elem in elems[:20]:
|
||||
label = elem.get_attribute('aria-label') or elem.text
|
||||
p = _parse_preis(label)
|
||||
if p:
|
||||
if p and p > 400:
|
||||
results.append({"scanner": "google_flights", "preis": p,
|
||||
"waehrung": "EUR", "airline": "",
|
||||
"abflug": abflug, "ankunft": rueck,
|
||||
|
|
@ -149,12 +213,15 @@ def scrape_google_flights(von, nach, tage=30, aufenthalt_tage=60,
|
|||
except Exception:
|
||||
pass
|
||||
|
||||
# Fallback: Regex über Body (nur plausible Preise > 400 EUR)
|
||||
if not results:
|
||||
for r in _preise_aus_body(body, "google_flights", abflug):
|
||||
r["ankunft"] = rueck
|
||||
r["booking_url"] = booking_url
|
||||
results.append(r)
|
||||
if r["preis"] > 400:
|
||||
r["ankunft"] = rueck
|
||||
r["booking_url"] = booking_url
|
||||
results.append(r)
|
||||
|
||||
results = [r for r in results if r["preis"] > 400]
|
||||
print(f"[GF] Ergebnis: {[r['preis'] for r in results[:5]]}")
|
||||
return results[:10]
|
||||
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue