vhosts support

pkg/readers/file.go

@@ -3,6 +3,7 @@ package readers
 import (
 	"bufio"
 	"fmt"
+	"net"
 	"net/url"
 	"os"
 	"strconv"
@@ -120,7 +121,7 @@ func (fr *FileReader) urlsFor(candidate string, ports []int) []string {
 
 	if hasScheme && hasPort {
 		// return the candidate as is
-		urls = append(urls, parsedURL.String())
+		urls = append(urls, decodeLastHashtag(parsedURL.String()))
 		return urls
 	}
 
@@ -163,12 +164,13 @@ func (fr *FileReader) urlsFor(candidate string, ports []int) []string {
 			}
 
 			fullURL := url.URL{
-				Scheme: scheme,
-				Host:   host,
-				Path:   parsedURL.Path,
+				Scheme:   scheme,
+				Host:     host,
+				Path:     parsedURL.Path,
+				Fragment: parsedURL.Fragment,
 			}
 
-			urls = append(urls, fullURL.String())
+			urls = append(urls, decodeLastHashtag(fullURL.String()))
 		}
 	}
 
@@ -197,3 +199,25 @@ func (fr *FileReader) ports() []int {
 func isIPv6(hostname string) bool {
 	return len(hostname) > 0 && hostname[0] == '[' && hostname[len(hostname)-1] == ']'
 }
+
+// it's for vhosts screenshooting. ## may be encoded
+func decodeLastHashtag(url string) string {
+	i := strings.LastIndex(url, "%23%23")
+	pad := 0
+	if i == -1 {
+		i = strings.LastIndex(url, "#%23")
+		pad = 4
+	} else {
+		pad = 6
+	}
+	if i != -1 {
+		_url := url[:i] + "##" + url[i+pad:]
+		ipaddrIndex := strings.LastIndex(_url, "##") + 2
+		ipaddr := net.ParseIP(_url[ipaddrIndex:])
+		if ipaddr != nil {
+			return _url
+		}
+	}
+
+	return url
+}

pkg/runner/drivers/chromedp.go

@@ -8,6 +8,8 @@ import (
 	"fmt"
 	"image"
 	"log/slog"
+	"net"
+	"net/url"
 	"os"
 	"os/exec"
 	"path/filepath"
@@ -57,7 +59,7 @@ func (b *browserInstance) Close() {
 // see Witness for more information on why we're explicitly not using tabs
 // (to do that we would alloc in the NewChromedp function and make sure that
 // we have the browser started with chromedp.Run(browserCtx)).
-func getChromedpAllocator(opts runner.Options) (*browserInstance, error) {
+func getChromedpAllocator(opts runner.Options, target string) (*browserInstance, error) {
 	var (
 		allocCtx    context.Context
 		allocCancel context.CancelFunc
@@ -96,6 +98,22 @@ func getChromedpAllocator(opts runner.Options) (*browserInstance, error) {
 			allocOpts = append(allocOpts, chromedp.ExecPath(opts.Chrome.Path))
 		}
 
+		// Use specific ip address for domain if it's included and addres is hostname
+		if strings.Contains(target, "##") {
+			targetUri, _ := url.ParseRequestURI(target)
+
+			ipaddrIndex := strings.LastIndex(target, "##") + 2
+			ipaddr := net.ParseIP(target[ipaddrIndex:])
+
+			// ignore ipaddr from hashtag if hostname is already an ip address
+			if ipaddr != nil && net.ParseIP(targetUri.Host) == nil {
+				allocOpts = append(allocOpts,
+					chromedp.Flag("host-resolver-rules", fmt.Sprintf("MAP %s %s", targetUri.Hostname(), ipaddr)),
+					chromedp.Flag("host-rules", fmt.Sprintf("MAP %s %s", targetUri.Hostname(), ipaddr)),
+				)
+			}
+		}
+
 		allocCtx, allocCancel = chromedp.NewExecAllocator(context.Background(), allocOpts...)
 
 	} else {
@@ -128,7 +146,7 @@ func (run *Chromedp) Witness(target string, thisRunner *runner.Runner) (*models.
 	// a resources thing I guess with a parent browser process? so, using this
 	// driver now means the resource usage will be higher, but, your accuracy
 	// will also be amazing.
-	allocator, err := getChromedpAllocator(run.options)
+	allocator, err := getChromedpAllocator(run.options, target)
 	if err != nil {
 		return nil, err
 	}
@@ -339,7 +357,7 @@ func (run *Chromedp) Witness(target string, thisRunner *runner.Runner) (*models.
 
 	// navigate to the target
 	if err := chromedp.Run(
-		navigationCtx, chromedp.Navigate(target),
+		navigationCtx, chromedp.Navigate(stripHashtagIp(target)),
 	); err != nil && err != context.DeadlineExceeded {
 		return nil, fmt.Errorf("could not navigate to target: %w", err)
 	}
@@ -474,3 +492,17 @@ func (run *Chromedp) Witness(target string, thisRunner *runner.Runner) (*models.
 func (run *Chromedp) Close() {
 	run.log.Debug("closing browser allocation context")
 }
+
+// removes last ## with ipaddress
+func stripHashtagIp(target string) string {
+	if strings.Contains(target, "##") {
+		i := strings.LastIndex(target, "##")
+		ipaddr := net.ParseIP(target[i+2:])
+
+		// ignore ipaddr from hashtag if hostname is already an ip address
+		if ipaddr != nil {
+			return target[:i]
+		}
+	}
+	return target
+}