Java 正则表达式提取标签中的属性(src 连接地址)等

Posted 2020-10-20 爷的眼睛闪亮

public class Test {

    public static void main(String[] args) {
        String source = "<p><img src=\\"https://xxxxx/xxx/xxxx/182cd48c587651767921868181f66ca8.jpg\\" />sdfasdfasdfsadfasdfasdfasdfasdfasdf</p><img src=\\"https://xxxxx/xxxx/182cd48c587651767921868181f66ca8sdf.jpg\\" />";
        System.out.println(getImgStr(source));
        String htmlStr = "<p><video controls=\\"controls\\" durationtime=\\"72\\" filesize=\\"27117469\\" height=\\"200px\\" poster=\\"https://xxxxxx/b439b0281450abce7f13b2920da04346.png\\" src=\\"https://xxxxxxxxxx/b439b0281450abce7f13b2920da04346.mp4\\" style=\\"\\">&nbsp;</video></p>";

        System.out.println(getVideoStr(htmlStr));
    }


    public static Set<String> getImgStr(String htmlStr) {
        Set<String> pics = new HashSet<>();
        String img = "";
        Pattern p_image;
        Matcher m_image;
        String regEx_img = "<img.*src\\\\s*=\\\\s*(.*?)[^>]*?>";
        p_image = Pattern.compile(regEx_img, Pattern.CASE_INSENSITIVE);
        m_image = p_image.matcher(htmlStr);
        while (m_image.find()) {
            // 得到<img />数据
            img = m_image.group();
            // 匹配<img>中的src数据
            Matcher m = Pattern.compile("src\\\\s*=\\\\s*\\"?(.*?)(\\"|>|\\\\s+)").matcher(img);
            while (m.find()) {
                pics.add(m.group(1));
            }
        }

        return pics;
    }

    public static Map<String, String> getVideoStr(String htmlStr) {
        Map<String, String> pics = new HashMap<String, String>();
        String regEx_video="<video.*poster\\\\s*=\\\\s*(.*?)[^>]*?src\\\\s*=\\\\s*(.*?)[^>]*?>";
        Pattern p = Pattern.compile(regEx_video,Pattern.CASE_INSENSITIVE);
        Matcher m = p.matcher(htmlStr);
        String video="";
        Map<String, String> map = new HashMap<String, String>();
        while (m.find()) {
            video=m.group();
            Matcher mPoster = Pattern.compile("poster\\\\s*=\\\\s*\\"?(.*?)(\\"|>|\\\\s+)").matcher(video);
            Matcher mSrc = Pattern.compile("src\\\\s*=\\\\s*\\"?(.*?)(\\"|>|\\\\s+)").matcher(video);
            String poster ="";
            String src="";
            while (mPoster.find()){
                poster=mPoster.group(1);
            }
            while (mSrc.find()){
                src=mSrc.group(1);
            }
            map.put("poster", poster);
            map.put("src", src);
        }
        return map;
    }
}