aron/wohnbot
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

working app

Browse source
This commit is contained in:
Aron Petau 2025-12-29 22:46:10 +01:00
parent 8e69e30387
commit 3057cda8d3
12 changed files with 708 additions and 232 deletions
Download patch file Download diff file Expand all files Collapse all files

120
handlers/degewo_handler.py
View file

@ -1,6 +1,9 @@
from .base_handler import BaseHandler
from handlers.base_handler import BaseHandler
import logging
import asyncio
import os
from pathlib import Path
logger = logging.getLogger(__name__)
@ -9,13 +12,16 @@ class DegewoHandler(BaseHandler):
self.context = browser_context
async def apply(self, listing: dict, result: dict) -> dict:
DATA_DIR = Path("data/degewo")
DATA_DIR.mkdir(parents=True, exist_ok=True)
page = await self.context.new_page()
try:
logger.info(f"[DEGEWO] Open: {listing['link']}")
logger.info(f"[DEGEWO] Opening page: {listing['link']}")
response = await page.goto(listing["link"], wait_until="networkidle")
logger.info("[DEGEWO] Page loaded")
await asyncio.sleep(2)
# Detect 404 by status or page title
# 404 detection
status = response.status if response else None
page_title = await page.title()
if status == 404 or (page_title and "404" in page_title):
@ -23,63 +29,67 @@ class DegewoHandler(BaseHandler):
result["success"] = False
result["message"] = "Listing is no longer available (404). Application impossible. Will not retry."
result["permanent_fail"] = True
await page.close()
return result
# Always handle cookies and consent before anything else
await self.handle_cookies(page)
await self.handle_consent(page)
# Check for 'INSERAT DEAKTIVIERT' (deactivated listing)
page_content = await page.content()
if "INSERAT DEAKTIVIERT" in page_content or "Inserat deaktiviert" in page_content:
logger.warning("[DEGEWO] Listing is deactivated (INSERAT DEAKTIVIERT detected), treating as 404")
result["success"] = False
result["message"] = "Listing deactivated (404)"
result["deactivated"] = True # Mark for removal from retries
await page.close()
return result
# Save HTML after modal handling for debugging
# Dismiss cookie banner
try:
html_content = await page.content()
with open("data/degewo_debug.html", "w", encoding="utf-8") as f:
f.write(html_content)
cookie_btn = await page.query_selector('button:has-text("Alle akzeptieren"), #CybotCookiebotDialogBodyLevelButtonLevelOptinAllowAll')
if cookie_btn and await cookie_btn.is_visible():
await cookie_btn.click()
logger.info("[DEGEWO] Dismissed cookie banner")
await asyncio.sleep(1)
except Exception as e:
logger.debug(f"[DEGEWO] Debug HTML not saved: {e}")
logger.debug(f"[DEGEWO] Cookie banner dismiss failed: {e}")
logger.info("[DEGEWO] Searching for application button...")
selectors = [
'a.btn',
'button.btn',
'a:has-text("Bewerben")',
'button:has-text("Bewerben")',
'a:has-text("Anfrage")',
'button:has-text("Anfrage")',
'a:has-text("Kontakt")',
'button:has-text("Kontakt")',
]
apply_btn = None
for sel in selectors:
all_btns = await page.query_selector_all(sel)
logger.debug(f"[DEGEWO] Selector '{sel}': {len(all_btns)} matches")
for btn in all_btns:
try:
if await btn.is_visible():
btn_text = (await btn.inner_text()).lower()
if any(x in btn_text for x in ["drucken", "merken", "zurück"]):
continue
apply_btn = btn
logger.info(f"[DEGEWO] Found visible application button: {sel} [{btn_text}]")
break
except Exception as e:
logger.debug(f"[DEGEWO] 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("[DEGEWO] Looking for kontaktieren button...")
apply_btn = await page.query_selector('a:has-text("kontaktieren"), button:has-text("kontaktieren"), a:has-text("Kontaktieren"), button:has-text("Kontaktieren")')
if apply_btn and await apply_btn.is_visible():
logger.info("[DEGEWO] Found kontaktieren button, clicking...")
await apply_btn.click()
await asyncio.sleep(2)
result["success"] = True
result["message"] = "Application submitted successfully."
else:
logger.warning("[DEGEWO] No application button found.")
result["message"] = "No application button found."
except Exception as e:
result["message"] = f"Error during application: {e}"
logger.error(f"[DEGEWO] Application error: {e}")
finally:
await page.close()
await asyncio.sleep(3)
return result
# Degewo uses Wohnungshelden iframe for the application form
# Find the iframe and get its URL to navigate directly
iframe_element = await page.query_selector('iframe[src*="wohnungshelden.de"]')
if iframe_element:
iframe_url = await iframe_element.get_attribute('src')
logger.info(f"[DEGEWO] Found Wohnungshelden iframe: {iframe_url}")
# Navigate to the iframe URL directly in a new page for full access
iframe_page = await self.context.new_page()
try:
await iframe_page.goto(iframe_url, wait_until="networkidle")
await asyncio.sleep(2)
logger.info("[DEGEWO] Loaded Wohnungshelden application page")
# TODO: Implement form-filling and submission logic here
finally:
await iframe_page.close()
else:
# No iframe found - try the old approach (fallback for different page structure)
logger.warning("[DEGEWO] Wohnungshelden iframe not found, trying direct form...")
# TODO: Implement fallback logic here
else:
result["message"] = "No kontaktieren button found"
logger.warning("[DEGEWO] Could not find kontaktieren button")
screenshot_path = DATA_DIR / f"degewo_nobtn_{listing['id']}.png"
await page.screenshot(path=str(screenshot_path), full_page=True)
await page.close()
return result
except Exception as e:
result["success"] = False
result["message"] = f"Error: {str(e)}"
logger.error(f"[DEGEWO] Exception: {str(e)}")
await page.close()
return result