mirror of
https://github.com/R0m1k3/Priceflow.git
synced 2026-10-14 10:46:32 +02:00
74 lines
2.5 KiB
Python
74 lines
2.5 KiB
Python
from bs4 import BeautifulSoup
|
|
from app.services.parsers.base_parser import BaseParser, ProductResult
|
|
import logging
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
class DartyParser(BaseParser):
|
|
def __init__(self):
|
|
super().__init__("darty.com", "https://www.darty.com")
|
|
|
|
def parse_search_results(self, html: str, query: str, search_url: str) -> list[ProductResult]:
|
|
soup = BeautifulSoup(html, "html.parser")
|
|
results = []
|
|
|
|
# Darty products
|
|
# Config: a[href*='/nav/achat/'][href*='.html']
|
|
|
|
cards = soup.select(".product-card, .product_item, div[class*='product-card']")
|
|
|
|
for card in cards:
|
|
try:
|
|
# Link
|
|
link_el = card.select_one("a[href*='/nav/achat/'], a.product-link")
|
|
if not link_el:
|
|
# Try finding any link if specific selector fails
|
|
link_el = card.find("a")
|
|
|
|
if not link_el:
|
|
continue
|
|
|
|
href = link_el.get('href')
|
|
if not href:
|
|
continue
|
|
|
|
url = self.make_absolute_url(href)
|
|
|
|
# Title
|
|
title = None
|
|
title_el = card.select_one(".product_name, .product-title, h2, h3")
|
|
if title_el:
|
|
title = title_el.get_text(strip=True)
|
|
else:
|
|
title = link_el.get_text(strip=True)
|
|
|
|
if not title:
|
|
continue
|
|
|
|
# Image
|
|
img_url = self.extract_image_url(card)
|
|
|
|
# Price
|
|
price = None
|
|
price_el = card.select_one(".product_price, .price, [class*='price']")
|
|
if price_el:
|
|
price = self.parse_price_text(price_el.get_text())
|
|
|
|
results.append(ProductResult(
|
|
title=title,
|
|
url=url,
|
|
source="Darty",
|
|
price=price,
|
|
currency="EUR",
|
|
in_stock=True,
|
|
image_url=img_url,
|
|
snippet=f"Product from Darty"
|
|
))
|
|
|
|
except Exception as e:
|
|
logger.error(f"Error parsing Darty product: {e}")
|
|
continue
|
|
|
|
logger.info(f"DartyParser found {len(results)} results")
|
|
return results
|