java/w9/UrlValidator.java

package com.example.datacollect.util;

import java.net.URI;
import java.net.URISyntaxException;
import java.util.regex.Pattern;

/**
 * URL 格式验证工具类
 * 用于判断输入的字符串是否是合法的 HTTP/HTTPS URL
 */
public class UrlValidator {

    private static final String HTTP_PROTOCOL = "http";
    private static final String HTTPS_PROTOCOL = "https";

    private static final Pattern DOMAIN_PATTERN = Pattern.compile(
        "^[a-zA-Z0-9]([a-zA-Z0-9-]*[a-zA-Z0-9])?(\\.[a-zA-Z0-9]([a-zA-Z0-9-]*[a-zA-Z0-9])?)*$"
    );

    /**
     * 验证 URL 是否合法
     * @param url 要验证的 URL 字符串
     * @return 是否为合法的 HTTP/HTTPS URL
     */
    public static boolean isValidUrl(String url) {
        if (url == null || url.trim().isEmpty()) {
            return false;
        }

        try {
            URI uri = new URI(url);

            String scheme = uri.getScheme();
            if (scheme == null) {
                return false;
            }

            String schemeLower = scheme.toLowerCase();
            if (!HTTP_PROTOCOL.equals(schemeLower) && !HTTPS_PROTOCOL.equals(schemeLower)) {
                return false;
            }

            String host = uri.getHost();
            if (host == null || host.isEmpty()) {
                return false;
            }

            if (!DOMAIN_PATTERN.matcher(host).matches()) {
                return false;
            }

            return true;

        } catch (URISyntaxException e) {
            return false;
        }
    }

    /**
     * 验证 URL 并返回详细信息
     * @param url 要验证的 URL 字符串
     * @return 验证结果描述
     */
    public static String validateWithMessage(String url) {
        if (url == null || url.trim().isEmpty()) {
            return "无效：URL 不能为空";
        }

        try {
            URI uri = new URI(url);

            String scheme = uri.getScheme();
            if (scheme == null) {
                return "无效：缺少协议 scheme（如 http://）";
            }

            String schemeLower = scheme.toLowerCase();
            if (!HTTP_PROTOCOL.equals(schemeLower) && !HTTPS_PROTOCOL.equals(schemeLower)) {
                return "无效：协议必须是 http 或 https，当前为：" + scheme;
            }

            String host = uri.getHost();
            if (host == null || host.isEmpty()) {
                return "无效：缺少主机名";
            }

            return "有效 URL：" + url;

        } catch (URISyntaxException e) {
            return "无效：URL 格式错误 - " + e.getMessage();
        }
    }

    /**
     * 获取 URL 的协议
     * @param url URL 字符串
     * @return 协议名称，如果无效返回 null
     */
    public static String getProtocol(String url) {
        try {
            URI uri = new URI(url);
            return uri.getScheme();
        } catch (URISyntaxException e) {
            return null;
        }
    }

    /**
     * 测试主方法
     */
    public static void main(String[] args) {
        String[] testUrls = {
            "https://www.example.com",
            "http://localhost:8080/api",
            "https://github.com/user/repo",
            "ftp://ftp.example.com/file",
            "htp://invalid.protocol",
            "not a url",
            "",
            "https://192.168.1.1:8080",
            "https://sub.domain.example.com/path/to/page",
            "javascript:alert(1)"
        };

        System.out.println("===== URL 格式验证测试 =====\n");

        for (String url : testUrls) {
            String result = validateWithMessage(url);
            System.out.println("测试: " + (url.isEmpty() ? "(空字符串)" : url));
            System.out.println("结果: " + result);
            System.out.println();
        }
    }
}