fscan/plugins/web/webtitle.go

package web

import (
	"context"
	"crypto/tls"
	"fmt"
	"io"
	"net/http"
	"regexp"
	"strings"
	"time"
	"unicode/utf8"

	"github.com/shadow1ng/fscan/common"
)

// WebTitlePlugin Web标题获取插件 - 收集Web应用信息和指纹识别
type WebTitlePlugin struct {
	name  string
	ports []int
}

// NewWebTitlePlugin 创建WebTitle插件
func NewWebTitlePlugin() *WebTitlePlugin {
	return &WebTitlePlugin{
		name:  "webtitle",
		ports: []int{80, 443, 8080, 8443, 8000, 8888, 9000, 9090}, // 常见Web端口
	}
}

// GetName 实现Plugin接口
func (p *WebTitlePlugin) GetName() string {
	return p.name
}

// GetPorts 实现Plugin接口
func (p *WebTitlePlugin) GetPorts() []int {
	return p.ports
}

// Scan 执行WebTitle扫描 - Web服务识别和指纹获取
func (p *WebTitlePlugin) Scan(ctx context.Context, info *common.HostInfo) *WebScanResult {
	target := fmt.Sprintf("%s:%s", info.Host, info.Ports)

	// 检查是否为Web端口
	webPorts := map[string]bool{
		"80": true, "443": true, "8080": true, "8443": true,
		"8000": true, "8888": true, "9000": true, "9090": true,
		"8001": true, "8081": true, "8008": true, "8086": true,
		"9001": true, "8089": true, "8090": true, "8082": true,
	}

	if !webPorts[info.Ports] {
		return &WebScanResult{
			Success: false,
			Error:   fmt.Errorf("非Web端口"),
		}
	}

	// 获取Web信息
	webInfo, err := p.getWebInfo(ctx, info)
	if err != nil {
		return &WebScanResult{
			Success: false,
			Error:   err,
		}
	}

	if !webInfo.Valid {
		return &WebScanResult{
			Success: false,
			Error:   fmt.Errorf("未发现有效的Web服务"),
		}
	}

	// 记录Web信息发现
	msg := fmt.Sprintf("WebTitle %s", target)
	if webInfo.Title != "" {
		msg += fmt.Sprintf(" [%s]", webInfo.Title)
	}
	if webInfo.StatusCode != 0 {
		msg += fmt.Sprintf(" %d", webInfo.StatusCode)
	}
	if webInfo.Server != "" {
		msg += fmt.Sprintf(" %s", webInfo.Server)
	}
	common.LogSuccess(msg)

	return &WebScanResult{
		Success: true,
		Title:   webInfo.Title,
		Status:  webInfo.StatusCode,
		Server:  webInfo.Server,
		Length:  int(webInfo.ContentLength),
	}
}


// WebInfo Web信息结构
type WebInfo struct {
	Valid           bool
	URL             string
	StatusCode      int
	Title           string
	Server          string
	ContentType     string
	ContentLength   int64
	Technologies    []string
	SecurityHeaders map[string]string
}

// Summary 返回Web信息摘要
func (wi *WebInfo) Summary() string {
	if !wi.Valid {
		return "Web服务检测失败"
	}

	var parts []string
	
	if wi.StatusCode != 0 {
		parts = append(parts, fmt.Sprintf("HTTP %d", wi.StatusCode))
	}
	
	if wi.Title != "" {
		title := wi.Title
		if len(title) > 30 {
			title = title[:30] + "..."
		}
		parts = append(parts, fmt.Sprintf("[%s]", title))
	}
	
	if wi.Server != "" {
		parts = append(parts, wi.Server)
	}

	return strings.Join(parts, " ")
}

// getWebInfo 获取Web服务信息
func (p *WebTitlePlugin) getWebInfo(ctx context.Context, info *common.HostInfo) (*WebInfo, error) {
	webInfo := &WebInfo{
		Valid:           false,
		SecurityHeaders: make(map[string]string),
		Technologies:    []string{},
	}

	// 确定协议
	protocol := "http"
	if info.Ports == "443" || info.Ports == "8443" {
		protocol = "https"
	}

	webInfo.URL = fmt.Sprintf("%s://%s:%s", protocol, info.Host, info.Ports)

	// 创建HTTP客户端
	client := &http.Client{
		Timeout: time.Duration(common.WebTimeout) * time.Second,
		Transport: &http.Transport{
			TLSClientConfig: &tls.Config{
				InsecureSkipVerify: true,
			},
			DisableKeepAlives: true,
		},
	}

	// 发送HTTP请求
	req, err := http.NewRequestWithContext(ctx, "GET", webInfo.URL, nil)
	if err != nil {
		return webInfo, fmt.Errorf("创建HTTP请求失败: %v", err)
	}

	// 设置User-Agent
	req.Header.Set("User-Agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36")

	resp, err := client.Do(req)
	if err != nil {
		return webInfo, fmt.Errorf("HTTP请求失败: %v", err)
	}
	defer resp.Body.Close()

	webInfo.Valid = true
	webInfo.StatusCode = resp.StatusCode

	// 获取基本响应头信息
	if server := resp.Header.Get("Server"); server != "" {
		webInfo.Server = server
	}

	if contentType := resp.Header.Get("Content-Type"); contentType != "" {
		webInfo.ContentType = contentType
	}

	webInfo.ContentLength = resp.ContentLength

	// 检测安全头
	p.detectSecurityHeaders(resp, webInfo)

	// 读取响应内容
	body, err := io.ReadAll(resp.Body)
	if err != nil {
		return webInfo, fmt.Errorf("读取响应内容失败: %v", err)
	}

	// 提取页面标题
	webInfo.Title = p.extractTitle(string(body))

	// 检测Web技术
	p.detectTechnologies(resp, string(body), webInfo)

	return webInfo, nil
}

// extractTitle 提取页面标题
func (p *WebTitlePlugin) extractTitle(html string) string {
	// 正则表达式匹配title标签
	titleRe := regexp.MustCompile(`(?i)<title[^>]*>([^<]+)</title>`)
	matches := titleRe.FindStringSubmatch(html)
	
	if len(matches) > 1 {
		title := strings.TrimSpace(matches[1])
		
		// 移除多余的空白字符
		title = regexp.MustCompile(`\s+`).ReplaceAllString(title, " ")
		
		// 长度限制
		if len(title) > 100 {
			title = title[:100] + "..."
		}
		
		// 确保是有效的UTF-8
		if utf8.ValidString(title) {
			return title
		}
	}
	
	return ""
}

// detectSecurityHeaders 检测安全头
func (p *WebTitlePlugin) detectSecurityHeaders(resp *http.Response, webInfo *WebInfo) {
	securityHeaders := []string{
		"X-Frame-Options",
		"X-XSS-Protection", 
		"X-Content-Type-Options",
		"Strict-Transport-Security",
		"Content-Security-Policy",
		"X-Powered-By",
	}

	for _, header := range securityHeaders {
		if value := resp.Header.Get(header); value != "" {
			webInfo.SecurityHeaders[header] = value
		}
	}
}

// detectTechnologies 检测Web技术
func (p *WebTitlePlugin) detectTechnologies(resp *http.Response, body string, webInfo *WebInfo) {
	// 基于响应头检测
	if xPoweredBy := resp.Header.Get("X-Powered-By"); xPoweredBy != "" {
		webInfo.Technologies = append(webInfo.Technologies, fmt.Sprintf("X-Powered-By: %s", xPoweredBy))
	}

	// 基于Server头检测
	if server := resp.Header.Get("Server"); server != "" {
		if strings.Contains(strings.ToLower(server), "apache") {
			webInfo.Technologies = append(webInfo.Technologies, "Apache HTTP Server")
		} else if strings.Contains(strings.ToLower(server), "nginx") {
			webInfo.Technologies = append(webInfo.Technologies, "Nginx")
		} else if strings.Contains(strings.ToLower(server), "iis") {
			webInfo.Technologies = append(webInfo.Technologies, "Microsoft IIS")
		}
	}

	// 基于HTML内容检测
	bodyLower := strings.ToLower(body)
	
	// 检测常见框架
	if strings.Contains(bodyLower, "wordpress") {
		webInfo.Technologies = append(webInfo.Technologies, "WordPress")
	}
	
	if strings.Contains(bodyLower, "joomla") {
		webInfo.Technologies = append(webInfo.Technologies, "Joomla")
	}
	
	if strings.Contains(bodyLower, "drupal") {
		webInfo.Technologies = append(webInfo.Technologies, "Drupal")
	}

	// 检测JavaScript框架
	if strings.Contains(bodyLower, "jquery") {
		webInfo.Technologies = append(webInfo.Technologies, "jQuery")
	}
	
	if strings.Contains(bodyLower, "angular") {
		webInfo.Technologies = append(webInfo.Technologies, "AngularJS")
	}
	
	if strings.Contains(bodyLower, "react") {
		webInfo.Technologies = append(webInfo.Technologies, "React")
	}

	// 检测Web服务器相关
	if strings.Contains(bodyLower, "apache tomcat") || strings.Contains(bodyLower, "tomcat") {
		webInfo.Technologies = append(webInfo.Technologies, "Apache Tomcat")
	}
	
	if strings.Contains(bodyLower, "jetty") {
		webInfo.Technologies = append(webInfo.Technologies, "Eclipse Jetty")
	}
}

// init 自动注册Web插件
func init() {
	RegisterWebPlugin("webtitle", func() WebPlugin {
		return NewWebTitlePlugin()
	})
}