research/resources
Archived
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

modify scripts

Browse Source
This commit is contained in:
oscarz
2025-06-03 15:36:44 +08:00
parent a4c4fa39d0
commit e12fb725d5
1 changed files with 18 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
thelordofporn/src/scraper.py
View File

@ -119,25 +119,34 @@ def parse_actor_list(soup, href):
for article in articles: for article in articles:
try: try:
# 获取演员详情 # 获取演员详情
title_tag = article.find("h3", class_="loop-item__title").find("a") title_tag = article.find("h3", class_="loop-item__title")
title = title_tag.text.strip() title = title_tag.find("a").text.strip() if title_tag and title_tag.find("a") else "N/A"
href = title_tag["href"] href = title_tag.find("a")["href"] if title_tag and title_tag.find("a") else None
# 获取评分 # 获取评分
rating_tag = article.find("div", class_="loop-item__rating") rating_tag = article.find("div", class_="loop-item__rating")
rating = rating_tag.text.strip() if rating_tag else "N/A" rating = rating_tag.text.strip() if rating_tag else "N/A"
# 获取 Rank 和 Votes # 获取 Rank 和 Votes
meta_tags = article.find("div", class_="loop-item__rank").find_all("span") rank_votes_div = article.find("div", class_="loop-item__rank")
rank = meta_tags[0].find("b").text.strip() if meta_tags else "N/A" meta_tags = rank_votes_div.find_all("span") if rank_votes_div else []
votes = meta_tags[1].find("b").text.strip() if len(meta_tags) > 1 else "N/A"
# 安全获取 rank 和 votes
rank = None
votes = None
if len(meta_tags) >= 1:
rank_b = meta_tags[0].find("b")
rank = rank_b.text.strip() if rank_b else "N/A"
if len(meta_tags) >= 2:
votes_b = meta_tags[1].find("b")
votes = votes_b.text.strip() if votes_b else "N/A"
# 存入列表 # 存入列表
actress_list.append({ actress_list.append({
"pornstar": title, "pornstar": title,
"rating": utils.parse_numeric(rating), "rating": utils.parse_numeric(rating) if rating != "N/A" else None,
"rank": utils.parse_numeric(rank), "rank": utils.parse_numeric(rank) if rank is not None else None,
"votes": utils.parse_numeric(votes), "votes": utils.parse_numeric(votes) if votes is not None else None,
"href": href "href": href
}) })