爬取新闻数据测试

This commit is contained in:
dengli
2023-12-19 19:52:01 +08:00
parent 72517213a0
commit 585f097e34

View File

@@ -153,27 +153,28 @@ public class StockApiController {
@ResponseBody
public ServerResponse getINDNews() {
String news = stockService.getNews();
List<String> newsList = Arrays.asList(news.split("<a href="));
newsList.forEach( n -> {
String contentUrl = n.substring(1, n.indexOf("class=\"img-smllnews\"") - 2);
String id = contentUrl.substring(contentUrl.lastIndexOf("-") + 1, contentUrl.lastIndexOf("_"));
String imgUrl = n.substring(n.indexOf("img src=") + 9, n.indexOf("?"));
String time = n.substring(n.indexOf("Last Updated") + 23, n.indexOf("IST") - 9);
String title = n.substring(n.indexOf("html\">") + 6, n.indexOf("<div class=\"short-video-img\">") - 47);
SiteNews siteNews = new SiteNews();
siteNews.setAddTime(new Date());
siteNews.setSourceId(id);
siteNews.setTitle(title);
siteNews.setDescription(time);
siteNews.setImgurl(imgUrl);
siteNews.setContent(stockService.getNewsInfo(contentUrl));
List<SiteNews> list = newsRepository.findAll(QSiteNewsPO.siteNewsPO.sourceId.eq(id));
if (list.size() == 0) {
newsRepository.save(siteNews);
}
});
return null;
return ServerResponse.createBySuccessMsg(news);
//List<String> newsList = Arrays.asList(news.split("<a href="));
//newsList.forEach( n -> {
// String contentUrl = n.substring(1, n.indexOf("class=\"img-smllnews\"") - 2);
// String id = contentUrl.substring(contentUrl.lastIndexOf("-") + 1, contentUrl.lastIndexOf("_"));
// String imgUrl = n.substring(n.indexOf("img src=") + 9, n.indexOf("?"));
// String time = n.substring(n.indexOf("Last Updated") + 23, n.indexOf("IST") - 9);
// String title = n.substring(n.indexOf("html\">") + 6, n.indexOf("<div class=\"short-video-img\">") - 47);
//
// SiteNews siteNews = new SiteNews();
// siteNews.setAddTime(new Date());
// siteNews.setSourceId(id);
// siteNews.setTitle(title);
// siteNews.setDescription(time);
// siteNews.setImgurl(imgUrl);
// siteNews.setContent(stockService.getNewsInfo(contentUrl));
// List<SiteNews> list = newsRepository.findAll(QSiteNewsPO.siteNewsPO.sourceId.eq(id));
// if (list.size() == 0) {
// newsRepository.save(siteNews);
// }
//});
}
@RequestMapping({"getINDNewsInfo.do"})