java正则 连接_java正则获取html的图片连接

public class TestReplaceAll {

public static void main(String[] args) {

String str = "";

Set pics = getImgStr(str);

System.out.println(pics);

}

public static Set getImgStr(String htmlStr) {

Set pics = new HashSet<>();

String img = "";

Pattern p_image;

Matcher m_image;

// String regEx_img = "]*?>"; //图片链接地址

String regEx_img = "]*?>";

p_image = Pattern.compile

(regEx_img, Pattern.CASE_INSENSITIVE);

m_image = p_image.matcher(htmlStr);

while (m_image.find()) {

// 得到数据

img = m_image.group();

// 匹配中的src数据

Matcher m = Pattern.compile("src\\s*=\\s*\"?(.*?)(\"|>|\\s+)").matcher(img);

while (m.find()) {

pics.add(m.group(1));

}

}

return pics;

}

}