playwright-scraper-clean / real_estate.py
apexherbert200's picture
Working /experimenting
7dc760d
raw
history blame
2.42 kB
from fastapi import FastAPI, HTTPException, Query
from pydantic import BaseModel
from playwright.async_api import async_playwright, TimeoutError as PlaywrightTimeout
from typing import List, Optional
import datetime
import logging
logging.basicConfig(level=logging.INFO)
app = FastAPI(title="RealEstateSnap")
class Listing(BaseModel):
title: str
price: Optional[str]
address: Optional[str]
bedrooms: Optional[str]
bathrooms: Optional[str]
listing_url: str
image_url: Optional[str]
platform: str
timestamp: str
async def scrape_craigslist(location: str, limit: int = 10) -> List[Listing]:
listings = []
async with async_playwright() as p:
browser = await p.chromium.launch(headless=True)
page = await browser.new_page()
url = f"https://{location.replace(' ', '').lower()}.craigslist.org/search/apa"
logging.info(f"➡️ Going to {url}")
await page.goto(url)
items = await page.query_selector_all(".result-row")[:limit]
for item in items:
try:
title = await item.inner_text(".result-title")
href = await item.get_attribute(".result-title", "href")
price = await item.inner_text(".result-price")
listings.append(Listing(
title=title.strip(),
price=price.strip(),
address="",
bedrooms="",
bathrooms="",
listing_url=href,
image_url=None,
platform="craigslist",
timestamp=datetime.datetime.utcnow().isoformat()
))
except PlaywrightTimeout:
logging.warning("Skipped a troublesome item")
await browser.close()
return listings
@app.get("/realestate", response_model=List[Listing])
async def get_listings(
location: str = Query(...),
platform: Optional[List[str]] = Query(["craigslist"])
):
platform = [p.lower() for p in platform]
results = []
if "craigslist" in platform:
try:
results += await scrape_craigslist(location)
except Exception as e:
raise HTTPException(status_code=500, detail=str(e))
# TODO: Add Zillow and Realtor here...
if not results:
raise HTTPException(status_code=404, detail="No listings found")
return results