478636f799
Phase 1-3 of decoupling: - path_setup.py adds project root to sys.path - Market-owned models in market/models/ (market, market_place) - All imports updated: shared.infrastructure, shared.db, shared.browser, etc. - MarketPlace uses container_type/container_id instead of post_id FK Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
18 lines
577 B
Python
18 lines
577 B
Python
|
|
from __future__ import annotations
|
|
from typing import Dict
|
|
from bs4 import BeautifulSoup
|
|
from shared.utils import normalize_text
|
|
from ..registry import extractor
|
|
|
|
@extractor
|
|
def ex_title(soup: BeautifulSoup, url: str) -> Dict:
|
|
title = None
|
|
for sel in ["h1.page-title span", "h1.page-title", "h1.product-name", "meta[property='og:title']"]:
|
|
el = soup.select_one(sel)
|
|
if el:
|
|
title = normalize_text(el.get_text()) if el.name != "meta" else el.get("content")
|
|
if title:
|
|
break
|
|
return {"title": title or "Product"}
|