155 lines
No EOL
7.1 KiB
Python
155 lines
No EOL
7.1 KiB
Python
from .base_handler import BaseHandler
|
|
import logging
|
|
import asyncio
|
|
import os
|
|
from pathlib import Path
|
|
|
|
logger = logging.getLogger(__name__)
|
|
DATA_DIR = Path("data/gesobau")
|
|
DATA_DIR.mkdir(parents=True, exist_ok=True)
|
|
|
|
class GesobauHandler(BaseHandler):
|
|
def __init__(self, browser_context):
|
|
self.context = browser_context
|
|
|
|
async def apply(self, listing: dict, result: dict) -> dict:
|
|
page = await self.context.new_page()
|
|
try:
|
|
logger.info(f"[GESOBAU] Opening page: {listing['link']}")
|
|
response = await page.goto(listing["link"], wait_until="networkidle")
|
|
logger.info("[GESOBAU] Page loaded")
|
|
await asyncio.sleep(2)
|
|
|
|
# 404 detection
|
|
status = response.status if response else None
|
|
page_title = await page.title()
|
|
page_content = await page.content()
|
|
is_404 = (
|
|
status == 404 or
|
|
(page_title and "404" in page_title) or
|
|
(page_title and "nicht gefunden" in page_title.lower()) or
|
|
("Angebot nicht mehr verfügbar" in page_content)
|
|
)
|
|
if is_404:
|
|
logger.warning(f"[GESOBAU] Listing is down (404 or unavailable): {listing['link']}")
|
|
result["success"] = False
|
|
result["message"] = "Listing is no longer available (404). Application impossible. Will not retry."
|
|
result["deactivated"] = True
|
|
return result
|
|
|
|
# Dismiss cookie banner
|
|
try:
|
|
cookie_btn = await page.query_selector('#CybotCookiebotDialogBodyLevelButtonLevelOptinAllowAll, button:has-text("Alle akzeptieren")')
|
|
if cookie_btn and await cookie_btn.is_visible():
|
|
await cookie_btn.click()
|
|
logger.info("[GESOBAU] Dismissed cookie banner")
|
|
await asyncio.sleep(1)
|
|
except:
|
|
pass
|
|
|
|
# Save debug HTML and screenshot
|
|
try:
|
|
html_content = await page.content()
|
|
with open(DATA_DIR / f"gesobau_debug_{listing['id']}.html", "w", encoding="utf-8") as f:
|
|
f.write(html_content)
|
|
logger.info(f"[GESOBAU] Saved debug HTML: gesobau_debug_{listing['id']}.html")
|
|
except Exception as e:
|
|
logger.warning(f"[GESOBAU] Could not save debug HTML: {e}")
|
|
|
|
try:
|
|
await page.screenshot(path=DATA_DIR / f"gesobau_page_{listing['id']}.png", full_page=True)
|
|
logger.info(f"[GESOBAU] Saved page screenshot: gesobau_page_{listing['id']}.png")
|
|
except Exception as e:
|
|
logger.warning(f"[GESOBAU] Could not save screenshot: {e}")
|
|
|
|
# Log listing details
|
|
await self.log_listing_details(listing)
|
|
|
|
# Look for application button
|
|
logger.info("[GESOBAU] Searching for application button...")
|
|
selectors = [
|
|
'a[href*="bewerben"]',
|
|
'button:has-text("Bewerben")',
|
|
'a:has-text("Bewerben")',
|
|
'button:has-text("Interesse")',
|
|
'a:has-text("Kontakt")',
|
|
'button.btn',
|
|
]
|
|
|
|
apply_btn = None
|
|
for sel in selectors:
|
|
all_btns = await page.query_selector_all(sel)
|
|
logger.debug(f"[GESOBAU] Selector '{sel}': {len(all_btns)} matches")
|
|
for btn in all_btns:
|
|
try:
|
|
if await btn.is_visible():
|
|
apply_btn = btn
|
|
logger.info(f"[GESOBAU] Found visible application button: {sel}")
|
|
break
|
|
except Exception as e:
|
|
logger.debug(f"[GESOBAU] Button visibility error: {e}")
|
|
if apply_btn:
|
|
break
|
|
|
|
if apply_btn:
|
|
await apply_btn.scroll_into_view_if_needed()
|
|
await asyncio.sleep(0.5)
|
|
logger.info("[GESOBAU] Clicking application button...")
|
|
await apply_btn.click()
|
|
await asyncio.sleep(2)
|
|
|
|
# Save screenshot and HTML after click
|
|
logger.info("[GESOBAU] Checking for confirmation...")
|
|
try:
|
|
await page.screenshot(path=DATA_DIR / f"gesobau_after_apply_{listing['id']}.png", full_page=True)
|
|
logger.info(f"[GESOBAU] Saved after-apply screenshot: gesobau_after_apply_{listing['id']}.png")
|
|
except Exception as e:
|
|
logger.warning(f"[GESOBAU] Could not save after-apply screenshot: {e}")
|
|
|
|
try:
|
|
html_after = await page.content()
|
|
with open(DATA_DIR / f"gesobau_after_apply_{listing['id']}.html", "w", encoding="utf-8") as f:
|
|
f.write(html_after)
|
|
logger.info(f"[GESOBAU] Saved after-apply HTML: gesobau_after_apply_{listing['id']}.html")
|
|
except Exception as e:
|
|
logger.warning(f"[GESOBAU] Could not save after-apply HTML: {e}")
|
|
logger.warning(f"[GESOBAU] Could not save HTML after apply: {e}")
|
|
|
|
# Look for confirmation message on the page
|
|
confirmation_selectors = [
|
|
'text="Vielen Dank"',
|
|
'text="Ihre Anfrage wurde gesendet"',
|
|
'text="Bestätigung"',
|
|
'div:has-text("Vielen Dank")',
|
|
'div:has-text("Ihre Anfrage wurde gesendet")',
|
|
]
|
|
confirmed = False
|
|
for sel in confirmation_selectors:
|
|
try:
|
|
el = await page.query_selector(sel)
|
|
if el and await el.is_visible():
|
|
logger.info(f"[GESOBAU] Found confirmation element: {sel}")
|
|
confirmed = True
|
|
break
|
|
except Exception as e:
|
|
logger.debug(f"[GESOBAU] Error checking confirmation selector {sel}: {e}")
|
|
if confirmed:
|
|
result["success"] = True
|
|
result["message"] = "Application submitted and confirmation detected."
|
|
else:
|
|
logger.warning("[GESOBAU] No confirmation message detected after application click.")
|
|
result["success"] = False
|
|
result["message"] = "Clicked application button, but no confirmation detected. Check screenshot and HTML."
|
|
else:
|
|
logger.warning("[GESOBAU] No application button found.")
|
|
result["message"] = "No application button found."
|
|
screenshot_path = DATA_DIR / f"gesobau_nobtn_{listing['id']}.png"
|
|
await page.screenshot(path=str(screenshot_path))
|
|
logger.info(f"[GESOBAU] Saved no-button screenshot: {screenshot_path}")
|
|
except Exception as e:
|
|
result["message"] = f"Error during application: {e}"
|
|
logger.error(f"[GESOBAU] Application error: {e}")
|
|
finally:
|
|
await page.close()
|
|
|
|
return result |