masterElmar/flatscraper
1
0
Fork 0
mirror of https://gitlab.dit.htwk-leipzig.de/fsr-im/tools/flatscraper.git synced 2025-07-16 03:28:48 +02:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'fix-wogetra' into 'main'

Browse Source 
Fixed: Extract specific details (rooms, ...) in wogetra which caused a error if details is empty

See merge request fsr-im/tools/flatscraper!1
This commit is contained in:
Elmar Kresse
2025-05-28 12:01:57 +02:00
parent 9933141178 6094d56a08
commit 4f62e2709a
1 changed files with 17 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
src/wogetra/scraper.py
View File

@ -3,6 +3,17 @@ from bs4 import BeautifulSoup
WOGETRA_URL = "https://www.wogetra.de/immobilien-vermarktungsart/miete/" WOGETRA_URL = "https://www.wogetra.de/immobilien-vermarktungsart/miete/"
def find_property(details, property):
prop = "N/A"
if details:
prop_row = details.find("div", class_=property)
if prop_row:
dd = prop_row.find("div", class_="dd")
if dd:
prop = dd.text.strip()
return prop
# Funktion: Scrape von Wogetra # Funktion: Scrape von Wogetra
def scrape_wogetra(): def scrape_wogetra():
response = requests.get(WOGETRA_URL) response = requests.get(WOGETRA_URL)
@ -26,11 +37,12 @@ def scrape_wogetra():
subtitle = subtitle_element.text.strip() if subtitle_element else "No Subtitle" subtitle = subtitle_element.text.strip() if subtitle_element else "No Subtitle"
link = link_element["href"] if link_element else "#" link = link_element["href"] if link_element else "#"
rooms = details.find("div", class_="data-anzahl_zimmer").find("div", class_="dd").text.strip() if details else "N/A" # Extract specific details
size = details.find("div", class_="data-wohnflaeche").find("div", class_="dd").text.strip() if details else "N/A" rooms = find_property(details, "data-anzahl_zimmer")
rent = details.find("div", class_="data-nettokaltmiete").find("div", class_="dd").text.strip() if details else "N/A" size = find_property(details, "data-wohnflaeche")
warm_rent = details.find("div", class_="data-warmmiete").find("div", class_="dd").text.strip() if details else "N/A" rent = find_property(details, "data-nettokaltmiete")
availability = details.find("div", class_="data-verfuegbar_ab").find("div", class_="dd").text.strip() if details else "N/A" warm_rent = find_property(details, "data-warmmiete")
availability = find_property(details, "data-verfuegbar_ab")
# Add property to list # Add property to list
properties.append({ properties.append({