import org.jsoup.Jsoup; import org.jsoup.nodes.Document; import org.jsoup.nodes.Element; import org.jsoup.select.Elements; import java.io.FileNotFoundException; import java.io.FileOutputStream; import java.io.IOException; public class pasreHtml { public static void main(String args[]) throws IOException { loadImgURL(); } public static void loadImgURL() throws IOException { Document doc = null; String filePath = "E:\\imgurls.txt"; String res = ""; for(int m = 1;m < 100;m ++){ try { doc = Jsoup.connect("https://pixabay.com/zh/editors_choice/?media_type=photo&pagi="+m).get(); } catch (IOException e) { e.printStackTrace(); } Elements rows = doc.getElementsByClass("flex_grid credits"); String[] string = new String[60]; for (int i = 0; i < rows.size(); i++) { Element row = rows.get(i); //String content = row.select("#content > div:nth-child(2) > div > div").text(); Elements items = row.select("div > a > img"); for (int j = 0; j < items.size(); j++) { String img = ""; Element item = items.get(j); img = item.attr("src"); if(item.attr("data-lazy").isEmpty()) img = item.attr("src"); else img = item.attr("data-lazy"); res += img + '\n'; } } FileOutputStream fos = new FileOutputStream(filePath); fos.write(res.getBytes()); fos.close(); } } }
解析图片获得url,可以用网页打开
最新推荐文章于 2024-07-04 09:24:51 发布