diff options
Diffstat (limited to 'bemani')
| -rw-r--r-- | bemani/ddr.py | 28 |
1 files changed, 0 insertions, 28 deletions
diff --git a/bemani/ddr.py b/bemani/ddr.py index ee8b659..a969434 100644 --- a/bemani/ddr.py +++ b/bemani/ddr.py @@ -10,35 +10,22 @@ def parse_ddr_world_news_site(html: str): news_entries = [] for div in soup.select("div#info > div.news_one"): -<<<<<<< Updated upstream - if 'none' in div.get('style', ''): - continue -======= - # Skip hidden entries style = div.get('style', '') if 'none' in style: continue ->>>>>>> Stashed changes title_tag = div.select_one("div.news_title > div.title") date_tag = div.select_one("div.news_title > div.date") headline = title_tag.get_text(strip=True) if title_tag else None date_str = date_tag.get_text(strip=True) if date_tag else None -<<<<<<< Updated upstream -======= - # Parse date ->>>>>>> Stashed changes try: dt = datetime.strptime(date_str, "%Y/%m/%d") date_iso = dt.strftime("%Y-%m-%d") timestamp = int(time.mktime(dt.timetuple())) except Exception: date_iso, timestamp = None, None -<<<<<<< Updated upstream -======= ->>>>>>> Stashed changes paras = [p.get_text(strip=True, separator="\n") for p in div.find_all("p", recursive=False)] if not paras: @@ -50,19 +37,12 @@ def parse_ddr_world_news_site(html: str): paras.append(child.get_text(strip=True, separator="\n")) content = "\n\n".join(paras) if paras else None -<<<<<<< Updated upstream - # image (use data-src if present) - img = div.select_one("div.img_news_center img") - raw_src = img.get("data-src") or img.get("src") if img else None - image_url = urljoin(base_url, raw_src) if raw_src else None -======= images = [] for img in div.select("div.img_news_center img"): raw_src = img.get("data-src") or img.get("src") if raw_src: full_url = urljoin(base_url, raw_src) images.append({"image": full_url, "link": None}) ->>>>>>> Stashed changes news_entries.append({ "date": date_iso, @@ -72,15 +52,7 @@ def parse_ddr_world_news_site(html: str): "headline": headline, "content": content, "url": base_url, -<<<<<<< Updated upstream - "images": { - "image": image_url, - "link": None - } - }) -======= "images": images }) ->>>>>>> Stashed changes return news_entries |