public class TestReplaceAll {
public static void main(String[] args) {
String str = "<img alt=\"\" src=\"http://img.pppcar.com/image/getImage/57440fef4d5d3925cf12f228/820/0/80\"/>";
Set<String> pics = getImgStr(str);
System.out.println(pics);
}
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=(.*?)[^>]*?>"; //图片链接地址
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;
}
}