mirror of
https://github.com/phishingclub/phishingclub.git
synced 2026-07-30 15:38:49 +02:00
1624 lines
57 KiB
Go
1624 lines
57 KiB
Go
package remotebrowser
|
|
|
|
import (
|
|
"bytes"
|
|
"context"
|
|
"io"
|
|
|
|
"encoding/json"
|
|
"errors"
|
|
"fmt"
|
|
"os"
|
|
"os/exec"
|
|
"path/filepath"
|
|
"regexp"
|
|
"strings"
|
|
"sync"
|
|
"sync/atomic"
|
|
"syscall"
|
|
"time"
|
|
|
|
"go.uber.org/zap"
|
|
|
|
"github.com/dop251/goja"
|
|
"github.com/go-rod/rod"
|
|
"github.com/go-rod/rod/lib/launcher"
|
|
"github.com/go-rod/rod/lib/launcher/flags"
|
|
"github.com/go-rod/rod/lib/proto"
|
|
)
|
|
|
|
// Config holds browser connection and execution settings configurable by platform admins.
|
|
type Config struct {
|
|
Mode string `json:"mode"` // "local" or "remote"
|
|
Remote string `json:"remote"` // DevTools WS URL (mode=remote)
|
|
Proxy string `json:"proxy"` // socks5:// or http:// (mode=local)
|
|
Headless bool `json:"headless"` // run Chrome in headless mode (mode=local)
|
|
Timeout int `json:"timeout"` // ms, 0 = use DefaultTimeout
|
|
Lang string `json:"lang"` // BCP 47 locale e.g. "en-US" (mode=local)
|
|
Timezone string `json:"timezone"` // IANA timezone e.g. "Europe/Copenhagen" to match the exit network
|
|
ExtraFlags []string `json:"extraFlags"` // additional Chrome CLI flags e.g. ["--use-gl=egl"] (mode=local)
|
|
}
|
|
|
|
const DefaultTimeout = 60_000 // ms
|
|
|
|
// DefaultChromiumUA is the fallback user-agent used when the real browser version
|
|
// cannot be probed. It presents a Linux Chrome identity matching the host OS so
|
|
// navigator.platform, the client hints, and every worker context stay consistent.
|
|
// A Windows disguise cannot: navigator.platform is not overridable in worker and
|
|
// service-worker contexts, so it always contradicts a Windows UA there. The major
|
|
// version tracks the bundled Chromium (rod RevisionDefault); prefer the real,
|
|
// probed version via detectChromeMajor at runtime.
|
|
const DefaultChromiumUA = "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/128.0.0.0 Safari/537.36"
|
|
|
|
// chromeVersionRe matches a full Chromium version like "128.0.6568.0".
|
|
var chromeVersionRe = regexp.MustCompile(`(\d+)\.\d+\.\d+\.\d+`)
|
|
|
|
// uaChromeMajorRe extracts the major version from a UA string ("Chrome/128...").
|
|
var uaChromeMajorRe = regexp.MustCompile(`Chrome/(\d+)`)
|
|
|
|
// detectChromeMajor runs `<bin> --version` and returns the major version, e.g.
|
|
// "128" from "Chromium 128.0.6568.0". Returns "" when the binary cannot be probed;
|
|
// the caller falls back to DefaultChromiumUA. Probing the real binary keeps the
|
|
// spoofed UA version in lockstep with the engine, so a version cross-check cannot
|
|
// catch a UA that claims a version the engine does not support.
|
|
func detectChromeMajor(ctx context.Context, bin string) string {
|
|
c, cancel := context.WithTimeout(ctx, 5*time.Second)
|
|
defer cancel()
|
|
out, err := exec.CommandContext(c, bin, "--version").Output()
|
|
if err != nil {
|
|
return ""
|
|
}
|
|
if m := chromeVersionRe.FindStringSubmatch(string(out)); m != nil {
|
|
return m[1]
|
|
}
|
|
return ""
|
|
}
|
|
|
|
// linuxChromeUA builds a reduced Linux Chrome user-agent for the given major
|
|
// version, matching Chrome's UA-reduction format (minor, build, and patch zeroed).
|
|
func linuxChromeUA(major string) string {
|
|
return fmt.Sprintf(
|
|
"Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/%s.0.0.0 Safari/537.36",
|
|
major)
|
|
}
|
|
|
|
// cleanHeadlessUA strips the "HeadlessChrome" token from a real user-agent so a
|
|
// remote headless browser reports as ordinary Chrome. Used only in remote mode,
|
|
// where a launch flag is not available.
|
|
func cleanHeadlessUA(ua string) string {
|
|
return strings.ReplaceAll(ua, "HeadlessChrome", "Chrome")
|
|
}
|
|
|
|
// uaPlatform returns the navigator.platform value that matches ua's OS token.
|
|
func uaPlatform(ua string) string {
|
|
switch {
|
|
case strings.Contains(ua, "Windows"):
|
|
return "Win32"
|
|
case strings.Contains(ua, "Macintosh"), strings.Contains(ua, "Mac OS"):
|
|
return "MacIntel"
|
|
default:
|
|
return "Linux x86_64"
|
|
}
|
|
}
|
|
|
|
// uaMetadata builds client-hint metadata (Sec-CH-UA / navigator.userAgentData)
|
|
// consistent with ua's version and OS. Setting this alongside the UA string avoids
|
|
// the empty-client-hints tell that a string-only override produces.
|
|
func uaMetadata(ua string) *proto.EmulationUserAgentMetadata {
|
|
major := "128"
|
|
if m := uaChromeMajorRe.FindStringSubmatch(ua); m != nil {
|
|
major = m[1]
|
|
}
|
|
platform := "Linux"
|
|
switch {
|
|
case strings.Contains(ua, "Windows"):
|
|
platform = "Windows"
|
|
case strings.Contains(ua, "Macintosh"), strings.Contains(ua, "Mac OS"):
|
|
platform = "macOS"
|
|
}
|
|
return &proto.EmulationUserAgentMetadata{
|
|
Brands: []*proto.EmulationUserAgentBrandVersion{
|
|
{Brand: "Not;A=Brand", Version: "24"},
|
|
{Brand: "Chromium", Version: major},
|
|
{Brand: "Google Chrome", Version: major},
|
|
},
|
|
Platform: platform,
|
|
Architecture: "x86",
|
|
Bitness: "64",
|
|
Mobile: false,
|
|
}
|
|
}
|
|
|
|
// applyRemoteIdentity aligns a remote browser's main-frame identity via CDP.
|
|
// Remote mode cannot take a launch flag, so this strips any "HeadlessChrome" token
|
|
// from the real UA and sets a matching platform and client-hint metadata. Worker
|
|
// contexts of a remote browser inherit the remote's own process UA and are out of
|
|
// our control here; operators should run a non-headless remote browser.
|
|
func applyRemoteIdentity(page *rod.Page, customUA string, emitter *channelEmitter) {
|
|
ua := customUA
|
|
if ua == "" {
|
|
if ver, err := (proto.BrowserGetVersion{}).Call(page); err == nil {
|
|
ua = cleanHeadlessUA(ver.UserAgent)
|
|
}
|
|
if ua == "" {
|
|
ua = DefaultChromiumUA
|
|
}
|
|
}
|
|
override := &proto.NetworkSetUserAgentOverride{
|
|
UserAgent: ua,
|
|
Platform: uaPlatform(ua),
|
|
UserAgentMetadata: uaMetadata(ua),
|
|
}
|
|
if err := page.SetUserAgent(override); err != nil {
|
|
emitter.log(fmt.Sprintf("[session] warning: failed to set user-agent: %v", err))
|
|
}
|
|
}
|
|
|
|
// DefaultConfig returns a default config.
|
|
func DefaultConfig() Config {
|
|
return Config{
|
|
Mode: "local",
|
|
Timeout: DefaultTimeout,
|
|
}
|
|
}
|
|
|
|
// ParseConfig parses a JSON config string, returning defaults for empty input.
|
|
func ParseConfig(raw string) (Config, error) {
|
|
cfg := DefaultConfig()
|
|
if raw == "" {
|
|
return cfg, nil
|
|
}
|
|
if err := json.Unmarshal([]byte(raw), &cfg); err != nil {
|
|
return cfg, err
|
|
}
|
|
if cfg.Timeout <= 0 {
|
|
cfg.Timeout = DefaultTimeout
|
|
}
|
|
if cfg.Mode != "remote" {
|
|
cfg.Remote = ""
|
|
}
|
|
return cfg, nil
|
|
}
|
|
|
|
// chromeEnv builds a minimal environment for the Chrome subprocess, plus any
|
|
// overrides. Passing os.Environ() to Chrome leaks secrets into the renderer process.
|
|
// Only variables Chrome actually needs are forwarded.
|
|
func chromeEnv(overrides ...string) []string {
|
|
allowed := map[string]bool{
|
|
"HOME": true, "PATH": true, "USER": true, "LOGNAME": true,
|
|
"DISPLAY": true, "XAUTHORITY": true, "DBUS_SESSION_BUS_ADDRESS": true,
|
|
"FONTCONFIG_PATH": true, "FONTCONFIG_FILE": true,
|
|
}
|
|
var env []string
|
|
for _, kv := range os.Environ() {
|
|
key := kv
|
|
if i := strings.IndexByte(kv, '='); i >= 0 {
|
|
key = kv[:i]
|
|
}
|
|
if allowed[key] {
|
|
env = append(env, kv)
|
|
}
|
|
}
|
|
return append(env, overrides...)
|
|
}
|
|
|
|
// resolveBrowserRootDir returns the directory Rod uses to cache its auto-downloaded
|
|
// Chromium. We use a path next to the running binary rather than $HOME/.cache
|
|
func resolveBrowserRootDir() (string, error) {
|
|
execPath, err := os.Executable()
|
|
if err != nil {
|
|
return "", fmt.Errorf("cannot locate browser cache dir: %w", err)
|
|
}
|
|
return filepath.Join(filepath.Dir(execPath), "data", "browser"), nil
|
|
}
|
|
|
|
// xvfbScreen is the geometry of a session's private display. The recipient's real
|
|
// viewport and screen metrics are applied over CDP after the page opens, so this
|
|
// only has to be large enough to hold the browser window.
|
|
const xvfbScreen = "1920x1080x24"
|
|
|
|
// x11SocketDir is where X servers place their per display sockets. Xvfb does not
|
|
// create it, so a bare container image where nothing else has made it would leave
|
|
// the server unable to open a socket. Creating it here keeps headful working
|
|
// without adding a step to every Dockerfile.
|
|
const x11SocketDir = "/tmp/.X11-unix"
|
|
|
|
// startPrivateDisplay launches an Xvfb dedicated to a single browser session and
|
|
// returns its display name together with a stop function.
|
|
//
|
|
// Sessions get a display each instead of sharing one for two reasons. A dedicated
|
|
// display means the browser window is the only window on it, so no two sessions
|
|
// can compete for the foreground, which is what previously froze every session but
|
|
// the most recent one. It also stops the accidental crossover a shared display
|
|
// allows, where any client can read the windows and input of every other client
|
|
// on it. This is not a boundary against a hostile escaped renderer: every browser
|
|
// runs as the same user, so such code could reach a sibling display regardless.
|
|
// That case is the already accepted remote browser RCE risk; guarding it would
|
|
// need separate users or namespaces, not an X cookie, which the same user can read.
|
|
//
|
|
// Xvfb chooses the display number itself and reports it back on the descriptor
|
|
// passed with -displayfd, so concurrent sessions never race for the same number.
|
|
func startPrivateDisplay() (string, func(), error) {
|
|
// Best effort: usually already present. If creating it was actually needed
|
|
// and failed, the read below surfaces the concrete failure.
|
|
_ = os.MkdirAll(x11SocketDir, 0o777)
|
|
|
|
pr, pw, err := os.Pipe()
|
|
if err != nil {
|
|
return "", nil, err
|
|
}
|
|
cmd := exec.Command("Xvfb", "-displayfd", "3", "-screen", "0", xvfbScreen, "-nolisten", "tcp", "-noreset")
|
|
cmd.ExtraFiles = []*os.File{pw} // the child sees this as descriptor 3
|
|
if err := cmd.Start(); err != nil {
|
|
pr.Close() //nolint:errcheck
|
|
pw.Close() //nolint:errcheck
|
|
return "", nil, fmt.Errorf("xvfb start failed: %w", err)
|
|
}
|
|
// Xvfb owns the write end now. Dropping ours means a read returns instead of
|
|
// blocking if the server dies before reporting a display.
|
|
pw.Close() //nolint:errcheck
|
|
|
|
exited := make(chan error, 1)
|
|
go func() { exited <- cmd.Wait() }()
|
|
|
|
var stopOnce sync.Once
|
|
stop := func() {
|
|
stopOnce.Do(func() {
|
|
if cmd.Process != nil {
|
|
// SIGTERM first so Xvfb removes its own lock file and socket;
|
|
// SIGKILL would strand both and slowly litter /tmp.
|
|
cmd.Process.Signal(syscall.SIGTERM) //nolint:errcheck
|
|
select {
|
|
case <-exited:
|
|
case <-time.After(3 * time.Second):
|
|
cmd.Process.Kill() //nolint:errcheck
|
|
<-exited
|
|
}
|
|
}
|
|
pr.Close() //nolint:errcheck
|
|
})
|
|
}
|
|
|
|
pr.SetReadDeadline(time.Now().Add(10 * time.Second)) //nolint:errcheck
|
|
var out []byte
|
|
chunk := make([]byte, 16)
|
|
for !bytes.ContainsRune(out, '\n') && len(out) < 32 {
|
|
n, readErr := pr.Read(chunk)
|
|
if n > 0 {
|
|
out = append(out, chunk[:n]...)
|
|
}
|
|
if readErr != nil {
|
|
stop()
|
|
return "", nil, fmt.Errorf("xvfb did not report a display: %w", readErr)
|
|
}
|
|
}
|
|
num := strings.TrimSpace(string(out))
|
|
if num == "" {
|
|
stop()
|
|
return "", nil, errors.New("xvfb reported an empty display")
|
|
}
|
|
return ":" + num, stop, nil
|
|
}
|
|
|
|
// chromeSterrWriter is an io.Writer that fans Chrome stdout/stderr lines out to
|
|
// the session emitter (when emitter != nil) and/or the app logger (when logger != nil).
|
|
type chromeSterrWriter struct {
|
|
emitter *channelEmitter // non-nil → forward to session event stream
|
|
logger *zap.SugaredLogger // non-nil → forward to app debug log
|
|
buf []byte
|
|
}
|
|
|
|
func (w *chromeSterrWriter) Write(p []byte) (int, error) {
|
|
w.buf = append(w.buf, p...)
|
|
for {
|
|
idx := bytes.IndexByte(w.buf, '\n')
|
|
if idx < 0 {
|
|
break
|
|
}
|
|
line := strings.TrimSpace(string(w.buf[:idx]))
|
|
w.buf = w.buf[idx+1:]
|
|
if line != "" {
|
|
if w.emitter != nil {
|
|
w.emitter.log("[chrome] " + line)
|
|
}
|
|
if w.logger != nil {
|
|
w.logger.Debugw(line, "source", "chrome")
|
|
}
|
|
}
|
|
}
|
|
return len(p), nil
|
|
}
|
|
|
|
// scriptStopError is thrown by stop() for a clean script exit with no error emitted.
|
|
type scriptStopError struct{}
|
|
|
|
func (scriptStopError) Error() string { return "script stopped" }
|
|
|
|
// knownGoErrors maps substrings in a GoError message to a friendlier single line description.
|
|
// Only Rod/CDP/network-specific errors belong here. context.Canceled and
|
|
// context.DeadlineExceeded are handled earlier via errors.Is and never reach cleanGoError.
|
|
var knownGoErrors = []struct {
|
|
substr string
|
|
message string
|
|
}{
|
|
{"connection refused", "browser connection refused — is Chrome running?"},
|
|
{"use of closed network connection", "browser connection closed unexpectedly"},
|
|
{"i/o timeout", "browser CDP connection timed out"},
|
|
{"EOF", "browser disconnected unexpectedly"},
|
|
{"Target closed", "browser tab was closed"},
|
|
{"page not found", "browser tab was closed"},
|
|
}
|
|
|
|
// cleanGoError returns a friendly single line message for common GoError strings,
|
|
// stripping the "GoError: " prefix and JS stack trace. Unknown errors are returned
|
|
// with the prefix stripped but the stack trace removed.
|
|
func cleanGoError(s string) string {
|
|
// Strip leading "GoError: " prefix if present.
|
|
msg := strings.TrimPrefix(s, "GoError: ")
|
|
// Remove everything from the first newline (stack trace).
|
|
if idx := strings.IndexByte(msg, '\n'); idx >= 0 {
|
|
msg = msg[:idx]
|
|
}
|
|
// Map messages to friendly descriptions.
|
|
lower := strings.ToLower(msg)
|
|
for _, e := range knownGoErrors {
|
|
if strings.Contains(lower, e.substr) {
|
|
return e.message
|
|
}
|
|
}
|
|
return msg
|
|
}
|
|
|
|
// Runner executes a JS script against a Chrome instance and streams events
|
|
// to the Events channel. The caller must close or drain Events after Run returns.
|
|
type Runner struct {
|
|
Script string
|
|
Config Config
|
|
// ExecPath is the server-configured Chrome binary path (from config.json,
|
|
// not user-supplied). Empty = Rod auto-download.
|
|
ExecPath string
|
|
// Logger, when set, receives Chrome process stdout/stderr at debug level.
|
|
Logger *zap.SugaredLogger
|
|
Events chan RunEvent // server → client
|
|
Incoming chan IncomingMsg // client → script (victim events / test injections)
|
|
// BrowserCh receives the *rod.Page as soon as newSession() spawns the browser.
|
|
// The caller reads this once to obtain the page for streaming.
|
|
BrowserCh chan *rod.Page
|
|
// LiveCh receives the *rod.Page when s.keepAlive() is called (kept for compat).
|
|
LiveCh chan *rod.Page
|
|
// StreamCh receives commands from s.stream(selector, name) / stop().
|
|
StreamCh chan StreamCmd
|
|
// keepAliveActive is set by s.keepAlive() so Run() parks after the script
|
|
// finishes, waiting for the operator to explicitly end the session.
|
|
keepAliveActive atomic.Bool
|
|
}
|
|
|
|
// IncomingMsg is an event sent from the client into the running script.
|
|
// Wire format: {"event": "credentials", "data": {"username": "...", "password": "..."}}
|
|
type IncomingMsg struct {
|
|
Event string `json:"event"`
|
|
Data any `json:"data"`
|
|
}
|
|
|
|
// StreamCmd is sent on Runner.StreamCh when the script calls s.stream() or the returned stop().
|
|
type StreamCmd struct {
|
|
Op string // "start" | "stop"
|
|
Selector string // CSS selector (Op=start)
|
|
Name string // stream name
|
|
Page *rod.Page // rod page (Op=start)
|
|
MaxFps int // 0 = unlimited
|
|
Quality int // JPEG re-encode quality 1-100; 0 = default (92)
|
|
}
|
|
|
|
// NewRunner creates a Runner with buffered event channels.
|
|
func NewRunner(script string, cfg Config) *Runner {
|
|
return &Runner{
|
|
Script: script,
|
|
Config: cfg,
|
|
Events: make(chan RunEvent, 256),
|
|
Incoming: make(chan IncomingMsg, 256),
|
|
BrowserCh: make(chan *rod.Page, 1),
|
|
LiveCh: make(chan *rod.Page, 1),
|
|
StreamCh: make(chan StreamCmd, 16),
|
|
}
|
|
}
|
|
|
|
// Run executes the script. It blocks until the script finishes, the context is
|
|
// cancelled, or the global timeout fires. The Events channel is closed when Run
|
|
// returns.
|
|
func (r *Runner) Run(ctx context.Context) error {
|
|
defer close(r.Events)
|
|
defer close(r.StreamCh)
|
|
|
|
emitter := newChannelEmitter(r.Events)
|
|
|
|
// Last resort recovery: rod can panic with nil-pointer dereferences inside goja
|
|
// native callbacks. Goja panics non-Value panics, which would crash the process.
|
|
// Catch anything that escapes vm.RunString so a broken script never takes down the server.
|
|
defer func() {
|
|
if rec := recover(); rec != nil {
|
|
emitter.errorf(fmt.Sprintf("internal error: %v", rec))
|
|
}
|
|
}()
|
|
|
|
timeout := time.Duration(r.Config.Timeout) * time.Millisecond
|
|
outerCtx := ctx // operator-level context; only cancelled explicitly
|
|
ctx, timeoutCancel := context.WithTimeout(outerCtx, timeout)
|
|
defer timeoutCancel()
|
|
|
|
vm := goja.New()
|
|
|
|
// Interrupt the JS VM on termination. Two cases:
|
|
// DeadlineExceeded — real timeout: interrupt immediately.
|
|
// Canceled — either keepAlive() cancelled the script timeout to
|
|
// park the session, or the operator cancelled. Either
|
|
// way wait for the outer context so we only interrupt
|
|
// when the operator actually ends the session.
|
|
go func() {
|
|
<-ctx.Done()
|
|
if errors.Is(ctx.Err(), context.DeadlineExceeded) {
|
|
vm.Interrupt(ctx.Err())
|
|
return
|
|
}
|
|
<-outerCtx.Done()
|
|
vm.Interrupt(outerCtx.Err())
|
|
}()
|
|
|
|
// vmArgStr returns "" for undefined/null instead of the string "undefined".
|
|
vmArgStr := func(v goja.Value) string {
|
|
if goja.IsUndefined(v) || goja.IsNull(v) {
|
|
return ""
|
|
}
|
|
return v.String()
|
|
}
|
|
|
|
vm.Set("stop", func(call goja.FunctionCall) goja.Value {
|
|
panic(vm.NewGoError(scriptStopError{}))
|
|
})
|
|
|
|
vm.Set("emit", func(call goja.FunctionCall) goja.Value {
|
|
key := vmArgStr(call.Argument(0))
|
|
value := call.Argument(1).Export()
|
|
emitter.emit(key, value)
|
|
return goja.Undefined()
|
|
})
|
|
|
|
vm.Set("log", func(call goja.FunctionCall) goja.Value {
|
|
msg := vmArgStr(call.Argument(0))
|
|
if len(call.Arguments) > 1 && !goja.IsUndefined(call.Argument(1)) && !goja.IsNull(call.Argument(1)) {
|
|
emitter.log(msg, call.Argument(1).Export())
|
|
} else {
|
|
emitter.log(msg)
|
|
}
|
|
return goja.Undefined()
|
|
})
|
|
|
|
vm.Set("info", func(call goja.FunctionCall) goja.Value {
|
|
msg := vmArgStr(call.Argument(0))
|
|
emitter.info(msg)
|
|
return goja.Undefined()
|
|
})
|
|
|
|
vm.Set("submitData", func(call goja.FunctionCall) goja.Value {
|
|
data := call.Argument(0).Export()
|
|
emitter.submitData(data)
|
|
return goja.Undefined()
|
|
})
|
|
|
|
// eventQueue buffers events received by race() that didn't match any race condition,
|
|
// so they remain available for subsequent waitForEvent / waitForAny calls.
|
|
var eventQueue []IncomingMsg
|
|
enqueue := func(msg IncomingMsg) {
|
|
if len(eventQueue) < 512 {
|
|
eventQueue = append(eventQueue, msg)
|
|
}
|
|
}
|
|
|
|
// nextMatchingEvent returns the first event in eventQueue whose name is in keySet,
|
|
// removing it from the queue. Returns ok=false if no match is buffered.
|
|
nextMatchingEvent := func(keySet map[string]bool) (IncomingMsg, bool) {
|
|
for i, msg := range eventQueue {
|
|
if keySet[msg.Event] {
|
|
eventQueue = append(eventQueue[:i], eventQueue[i+1:]...)
|
|
return msg, true
|
|
}
|
|
}
|
|
return IncomingMsg{}, false
|
|
}
|
|
|
|
vm.Set("waitForEvent", func(call goja.FunctionCall) goja.Value {
|
|
key := vmArgStr(call.Argument(0))
|
|
emitter.log(fmt.Sprintf("[waitForEvent] waiting for %q", key))
|
|
ks := map[string]bool{key: true}
|
|
if msg, ok := nextMatchingEvent(ks); ok {
|
|
emitter.log(fmt.Sprintf("[waitForEvent] received %q (queued)", key))
|
|
return vm.ToValue(msg.Data)
|
|
}
|
|
for {
|
|
select {
|
|
case <-ctx.Done():
|
|
panic(vm.NewGoError(ctx.Err()))
|
|
case msg := <-r.Incoming:
|
|
if msg.Event == key {
|
|
emitter.log(fmt.Sprintf("[waitForEvent] received %q", key))
|
|
return vm.ToValue(msg.Data)
|
|
}
|
|
enqueue(msg)
|
|
}
|
|
}
|
|
})
|
|
|
|
vm.Set("waitForAny", func(call goja.FunctionCall) goja.Value {
|
|
// waitForAny(["password", "username"]) or waitForAny("password", "username")
|
|
// Returns {event: "...", data: ...} for whichever arrives first.
|
|
keySet := make(map[string]bool)
|
|
if len(call.Arguments) == 1 {
|
|
if arr, ok := call.Argument(0).Export().([]interface{}); ok {
|
|
for _, v := range arr {
|
|
keySet[fmt.Sprintf("%v", v)] = true
|
|
}
|
|
} else {
|
|
keySet[vmArgStr(call.Argument(0))] = true
|
|
}
|
|
} else {
|
|
for _, a := range call.Arguments {
|
|
keySet[vmArgStr(a)] = true
|
|
}
|
|
}
|
|
if msg, ok := nextMatchingEvent(keySet); ok {
|
|
result := vm.NewObject()
|
|
result.Set("event", msg.Event)
|
|
result.Set("data", vm.ToValue(msg.Data))
|
|
return result
|
|
}
|
|
for {
|
|
select {
|
|
case <-ctx.Done():
|
|
panic(vm.NewGoError(ctx.Err()))
|
|
case msg := <-r.Incoming:
|
|
if keySet[msg.Event] {
|
|
result := vm.NewObject()
|
|
result.Set("event", msg.Event)
|
|
result.Set("data", vm.ToValue(msg.Data))
|
|
return result
|
|
}
|
|
enqueue(msg)
|
|
}
|
|
}
|
|
})
|
|
|
|
// retry(max, fn) or retry({max, wait}, fn)
|
|
// fn receives a RetryContext and returns truthy to break (the value is returned),
|
|
// or false/undefined to retry. Returns null when all attempts are exhausted.
|
|
vm.Set("retry", func(call goja.FunctionCall) goja.Value {
|
|
if len(call.Arguments) < 2 {
|
|
panic(vm.NewTypeError("retry: expected (max, fn) or ({max, wait}, fn)"))
|
|
}
|
|
|
|
maxAttempts := 10
|
|
waitMs := 0
|
|
|
|
firstArg := call.Argument(0)
|
|
switch firstArg.Export().(type) {
|
|
case int64, float64, int, int32, int16, int8, uint, uint64, uint32, uint16, uint8:
|
|
maxAttempts = int(firstArg.ToInteger())
|
|
default:
|
|
obj := firstArg.ToObject(vm)
|
|
if mv := obj.Get("max"); mv != nil && !goja.IsUndefined(mv) && !goja.IsNull(mv) {
|
|
maxAttempts = int(mv.ToInteger())
|
|
}
|
|
if wv := obj.Get("wait"); wv != nil && !goja.IsUndefined(wv) && !goja.IsNull(wv) {
|
|
waitMs = int(wv.ToInteger())
|
|
}
|
|
}
|
|
|
|
fn, ok := goja.AssertFunction(call.Argument(1))
|
|
if !ok {
|
|
panic(vm.NewTypeError("retry: second argument must be a function"))
|
|
}
|
|
|
|
for attempt := 1; attempt <= maxAttempts; attempt++ {
|
|
loopCtx := vm.NewObject()
|
|
loopCtx.Set("attempt", attempt)
|
|
loopCtx.Set("max", maxAttempts)
|
|
loopCtx.Set("isFirst", attempt == 1)
|
|
loopCtx.Set("isLast", attempt == maxAttempts)
|
|
|
|
result, err := fn(goja.Undefined(), loopCtx)
|
|
if err != nil {
|
|
panic(err)
|
|
}
|
|
|
|
// explicit false → retry; any other truthy value (including true) → done
|
|
if !goja.IsNull(result) && !goja.IsUndefined(result) {
|
|
if b, isBool := result.Export().(bool); !isBool || b {
|
|
return result
|
|
}
|
|
}
|
|
|
|
if waitMs > 0 && attempt < maxAttempts {
|
|
timer := time.NewTimer(time.Duration(waitMs) * time.Millisecond)
|
|
select {
|
|
case <-ctx.Done():
|
|
timer.Stop()
|
|
panic(vm.NewGoError(ctx.Err()))
|
|
case <-timer.C:
|
|
}
|
|
}
|
|
}
|
|
|
|
return goja.Null()
|
|
})
|
|
|
|
// Note: withTimeout is intentionally NOT available at the top level because
|
|
// there is no page context to thread through. Use session.withTimeout(ms, fn) instead.
|
|
|
|
vm.Set("newSession", func(call goja.FunctionCall) goja.Value {
|
|
type sessionOpts struct {
|
|
Proxy string
|
|
Remote string
|
|
Headless bool
|
|
IdleTimeout int // ms; 0 = disabled
|
|
Debug bool
|
|
ChromeDebug bool
|
|
QueryTimeout int // ms; 0 = no timeout on read ops
|
|
UserAgent string
|
|
Lang string // BCP 47 locale, e.g. "en-US" - sets --lang flag (local mode only)
|
|
Timezone string // IANA timezone, e.g. "Europe/Copenhagen" - overrides browser timezone
|
|
ExtraFlags []string // additional Chrome CLI flags e.g. ["--use-gl=egl"] (local mode only)
|
|
}
|
|
opts := sessionOpts{
|
|
Headless: r.Config.Headless,
|
|
Lang: r.Config.Lang,
|
|
Timezone: r.Config.Timezone,
|
|
ExtraFlags: r.Config.ExtraFlags,
|
|
}
|
|
if r.Config.Proxy != "" {
|
|
opts.Proxy = r.Config.Proxy
|
|
}
|
|
if r.Config.Mode == "remote" && r.Config.Remote != "" {
|
|
opts.Remote = r.Config.Remote
|
|
}
|
|
|
|
if len(call.Arguments) > 0 {
|
|
obj := call.Argument(0).ToObject(vm)
|
|
if obj != nil {
|
|
if v := obj.Get("proxy"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
opts.Proxy = v.String()
|
|
}
|
|
if v := obj.Get("remote"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
opts.Remote = v.String()
|
|
}
|
|
if v := obj.Get("headless"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
opts.Headless = v.ToBoolean()
|
|
}
|
|
if v := obj.Get("idleTimeout"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
opts.IdleTimeout = int(v.ToInteger())
|
|
}
|
|
if v := obj.Get("debug"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
opts.Debug = v.ToBoolean()
|
|
}
|
|
if v := obj.Get("chromeDebug"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
opts.ChromeDebug = v.ToBoolean()
|
|
}
|
|
if v := obj.Get("queryTimeout"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
opts.QueryTimeout = int(v.ToInteger())
|
|
}
|
|
if v := obj.Get("userAgent"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
opts.UserAgent = v.String()
|
|
}
|
|
if v := obj.Get("lang"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
opts.Lang = v.String()
|
|
}
|
|
if v := obj.Get("timezone"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
opts.Timezone = v.String()
|
|
}
|
|
if v := obj.Get("extraFlags"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
if arr, ok := v.Export().([]interface{}); ok {
|
|
for _, item := range arr {
|
|
if s, ok := item.(string); ok && s != "" {
|
|
opts.ExtraFlags = append(opts.ExtraFlags, s)
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
var browser *rod.Browser
|
|
var page *rod.Page
|
|
// launchUA is the identity applied in local mode via the --user-agent launch
|
|
// flag; it also drives the OS-appropriate WebGL renderer choice below.
|
|
var launchUA string
|
|
|
|
if opts.Remote != "" {
|
|
if opts.Proxy != "" {
|
|
emitter.log(fmt.Sprintf("[session] warning: proxy %q ignored for remote browser", opts.Proxy))
|
|
}
|
|
// ResolveURL normalises any of: bare port "9222", "http://host:9222",
|
|
// or a full "ws://..." URL to the actual DevTools WebSocket URL by
|
|
// fetching /json/version. Users don't need to copy the UUID manually.
|
|
wsURL, resolveErr := launcher.ResolveURL(opts.Remote)
|
|
if resolveErr != nil {
|
|
emitter.errorf(fmt.Sprintf("browser connect failed: cannot resolve remote URL %q: %v", opts.Remote, resolveErr))
|
|
return goja.Undefined()
|
|
}
|
|
emitter.log(fmt.Sprintf("[session] connecting to remote browser at %s", wsURL))
|
|
// NoDefaultDevice disables rod's default LaptopWithMDPIScreen emulation,
|
|
// which otherwise overrides every new page with a macOS Chrome 114 UA, an
|
|
// "en" Accept-Language, and a 1280x800 device-metrics override. Those
|
|
// contradict the real Linux platform, blank the client hints, and make the
|
|
// viewport exceed the screen. We manage identity ourselves instead.
|
|
browser = rod.New().ControlURL(wsURL).Context(outerCtx).NoDefaultDevice()
|
|
if connectErr := browser.Connect(); connectErr != nil {
|
|
emitter.errorf(fmt.Sprintf("browser connect failed: %v", connectErr))
|
|
return goja.Undefined()
|
|
}
|
|
// Every session runs in its own browser context. All sessions share one
|
|
// remote browser, so on the default context they would also share one
|
|
// cookie jar, one localStorage and one cache: a session captured for one
|
|
// recipient would be readable from the page of any other recipient
|
|
// running at the same time. A browser context keeps that storage separate
|
|
// and discards all of it when the context is disposed.
|
|
//
|
|
// DisposeOnDetach makes the browser drop the context when this control
|
|
// connection goes away, which covers the paths where the session ends
|
|
// without running its own cleanup: victim disconnect, script timeout,
|
|
// operator cancel, and a server crash.
|
|
ctxRes, ctxErr := (proto.TargetCreateBrowserContext{DisposeOnDetach: true}).Call(browser)
|
|
if ctxErr != nil {
|
|
// Do not close the browser here. It is shared with every other
|
|
// session and closing it would end all of them.
|
|
emitter.errorf(fmt.Sprintf("browser context create failed: %v", ctxErr))
|
|
return goja.Undefined()
|
|
}
|
|
// Rebind to the new context. rod carries BrowserContextID into every
|
|
// target it creates, and its Close disposes the context instead of the
|
|
// whole browser once the field is set, so from here on the shared remote
|
|
// browser is never closed by this session.
|
|
isolated := *browser
|
|
isolated.BrowserContextID = ctxRes.BrowserContextID
|
|
browser = &isolated
|
|
emitter.log(fmt.Sprintf("[session] isolated browser context %s", ctxRes.BrowserContextID))
|
|
|
|
var pageErr error
|
|
page, pageErr = browser.Page(proto.TargetCreateTarget{URL: "about:blank"})
|
|
if pageErr != nil {
|
|
emitter.errorf(fmt.Sprintf("page create failed: %v", pageErr))
|
|
browser.Close() //nolint:errcheck
|
|
return goja.Undefined()
|
|
}
|
|
} else {
|
|
rootDir, err := resolveBrowserRootDir()
|
|
if err != nil {
|
|
emitter.errorf(err.Error())
|
|
return goja.Undefined()
|
|
}
|
|
// Newer Chromium requires writable XDG dirs and a crash-dumps-dir at
|
|
// startup (see go-rod#1126). Use subdirs of the browser root so all
|
|
// Chrome-related data lives in one place.
|
|
crashDir := filepath.Join(rootDir, "crashes")
|
|
if err := os.MkdirAll(crashDir, 0755); err != nil {
|
|
emitter.log(fmt.Sprintf("[session] warning: could not create crash dir: %v", err))
|
|
}
|
|
if err := os.MkdirAll(filepath.Join(rootDir, "config"), 0755); err != nil {
|
|
emitter.log(fmt.Sprintf("[session] warning: could not create config dir: %v", err))
|
|
}
|
|
if err := os.MkdirAll(filepath.Join(rootDir, "cache"), 0755); err != nil {
|
|
emitter.log(fmt.Sprintf("[session] warning: could not create cache dir: %v", err))
|
|
}
|
|
|
|
var chromeLogger io.Writer = io.Discard
|
|
if opts.ChromeDebug || r.Logger != nil {
|
|
cw := &chromeSterrWriter{logger: r.Logger}
|
|
if opts.ChromeDebug {
|
|
cw.emitter = emitter
|
|
}
|
|
chromeLogger = cw
|
|
}
|
|
chromeVars := []string{
|
|
"XDG_CONFIG_HOME=" + filepath.Join(rootDir, "config"),
|
|
"XDG_CACHE_HOME=" + filepath.Join(rootDir, "cache"),
|
|
}
|
|
// Headful needs a display, and it has to be one this session alone owns.
|
|
if !opts.Headless {
|
|
display, stopDisplay, dispErr := startPrivateDisplay()
|
|
switch {
|
|
case dispErr == nil:
|
|
emitter.log(fmt.Sprintf("[session] private display %s", display))
|
|
go func() {
|
|
<-outerCtx.Done()
|
|
stopDisplay()
|
|
}()
|
|
// Appended last so it beats any DISPLAY inherited from the
|
|
// server process: the last value wins once the environment is
|
|
// deduplicated on the way to the browser.
|
|
chromeVars = append(chromeVars, "DISPLAY="+display)
|
|
case os.Getenv("DISPLAY") != "":
|
|
// A desktop session already has a display. Use it rather than
|
|
// failing, and say so, because it is shared with everything
|
|
// else drawing on that desktop.
|
|
emitter.log(fmt.Sprintf("[session] warning: no private display (%v), falling back to the shared %s", dispErr, os.Getenv("DISPLAY")))
|
|
default:
|
|
emitter.errorf(fmt.Sprintf("headful browser needs a display: %v", dispErr))
|
|
return goja.Undefined()
|
|
}
|
|
}
|
|
l := launcher.New().
|
|
Headless(opts.Headless).
|
|
Logger(chromeLogger).
|
|
Set("disable-crash-reporter").
|
|
Set("crash-dumps-dir", crashDir).
|
|
Set("disable-blink-features", "AutomationControlled").
|
|
Delete("enable-automation").
|
|
Env(chromeEnv(chromeVars...)...)
|
|
|
|
var binPath string
|
|
if r.ExecPath != "" {
|
|
binPath = r.ExecPath
|
|
emitter.log(fmt.Sprintf("[session] using browser: %s", binPath))
|
|
l = l.Bin(binPath)
|
|
} else {
|
|
b := launcher.NewBrowser()
|
|
b.RootDir = rootDir
|
|
binPath = b.BinPath()
|
|
if _, err := os.Stat(binPath); os.IsNotExist(err) {
|
|
if err := b.Download(); err != nil {
|
|
emitter.errorf(fmt.Sprintf("browser download failed: %v", err))
|
|
return goja.Undefined()
|
|
}
|
|
}
|
|
emitter.log(fmt.Sprintf("[session] using browser: %s", binPath))
|
|
l = l.Bin(binPath)
|
|
}
|
|
|
|
// Set the user-agent at the process level via the launch flag so every
|
|
// context (main frame, web workers, and service workers) reports the same
|
|
// identity. A page-level CDP override does not reach service workers, which
|
|
// then leak the real "HeadlessChrome" UA and the true OS and version. When
|
|
// no custom UA is given we present a Linux Chrome identity matching the host
|
|
// OS and the real engine version, so navigator.platform, the client hints,
|
|
// and the version stay consistent across every context.
|
|
launchUA = opts.UserAgent
|
|
if launchUA == "" {
|
|
if major := detectChromeMajor(outerCtx, binPath); major != "" {
|
|
launchUA = linuxChromeUA(major)
|
|
} else {
|
|
launchUA = DefaultChromiumUA
|
|
}
|
|
}
|
|
emitter.log(fmt.Sprintf("[session] using user-agent: %s", launchUA))
|
|
l = l.Set("user-agent", launchUA)
|
|
|
|
if opts.Proxy != "" {
|
|
emitter.log(fmt.Sprintf("[session] using proxy: %s", opts.Proxy))
|
|
l = l.Proxy(opts.Proxy).Set("proxy-bypass-list", "")
|
|
}
|
|
if opts.Lang != "" {
|
|
// Sets navigator.language, navigator.languages, and the Accept-Language
|
|
// HTTP header at the process level - consistent across main frame and
|
|
// Web Workers (unlike Page.addScriptToEvaluateOnNewDocument which only
|
|
// runs in the main frame).
|
|
emitter.log(fmt.Sprintf("[session] using lang: %s", opts.Lang))
|
|
l = l.Set("lang", opts.Lang).Set("accept-lang", opts.Lang)
|
|
}
|
|
for _, rawFlag := range opts.ExtraFlags {
|
|
rawFlag = strings.TrimSpace(rawFlag)
|
|
// "!--flag-name" removes a flag from the launcher (e.g. to strip a rod default).
|
|
if strings.HasPrefix(rawFlag, "!--") {
|
|
name := flags.Flag(strings.TrimPrefix(rawFlag, "!--"))
|
|
l = l.Delete(name)
|
|
emitter.log(fmt.Sprintf("[session] removed flag: --%s", string(name)))
|
|
continue
|
|
}
|
|
if !strings.HasPrefix(rawFlag, "--") {
|
|
continue
|
|
}
|
|
rawFlag = strings.TrimPrefix(rawFlag, "--")
|
|
parts := strings.SplitN(rawFlag, "=", 2)
|
|
key := flags.Flag(parts[0])
|
|
if len(parts) == 2 {
|
|
l = l.Set(key, parts[1])
|
|
} else {
|
|
l = l.Set(key)
|
|
}
|
|
emitter.log(fmt.Sprintf("[session] extra flag: --%s", parts[0]))
|
|
}
|
|
u, err := l.Launch()
|
|
if err != nil {
|
|
emitter.errorf(fmt.Sprintf("browser launch failed: %v", err))
|
|
return goja.Undefined()
|
|
}
|
|
// Kill and clean up Chrome when the session ends. Without this, stopping
|
|
// a run before s.close() is called leaves a zombie Chrome process (~200-300 MB).
|
|
go func() {
|
|
<-outerCtx.Done()
|
|
l.Kill()
|
|
l.Cleanup()
|
|
}()
|
|
// NoDefaultDevice disables rod's default LaptopWithMDPIScreen emulation
|
|
// (macOS Chrome 114 UA, "en" Accept-Language, 1280x800 device metrics),
|
|
// which would override the launch-flag identity on the window and make the
|
|
// viewport exceed the screen. We manage identity ourselves instead.
|
|
browser = rod.New().ControlURL(u).Context(outerCtx).NoDefaultDevice()
|
|
if err := browser.Connect(); err != nil {
|
|
emitter.errorf(fmt.Sprintf("browser connect failed: %v", err))
|
|
return goja.Undefined()
|
|
}
|
|
page, err = browser.Page(proto.TargetCreateTarget{URL: "about:blank"})
|
|
if err != nil {
|
|
emitter.errorf(fmt.Sprintf("page create failed: %v", err))
|
|
browser.Close() //nolint:errcheck
|
|
return goja.Undefined()
|
|
}
|
|
// ignore "debugger;" code lines explicitly
|
|
err = proto.DebuggerSetSkipAllPauses{Skip: true}.Call(page)
|
|
if err != nil {
|
|
emitter.errorf(fmt.Sprintf("skip debug failed: %v", err))
|
|
browser.Close() //nolint:errcheck
|
|
return goja.Undefined()
|
|
}
|
|
// Note: console.* are deliberately left native. Replacing them with noop
|
|
// functions (to suppress page logging) makes console.log.toString() non-native,
|
|
// which trips TamperedFunctions. The console is not captured anyway — we never
|
|
// enable the Runtime domain (see the opsec note in browser.go), so no page
|
|
// console output reaches the server regardless.
|
|
}
|
|
|
|
// Identity handling differs by mode:
|
|
// local — the UA is already set via the --user-agent launch flag above,
|
|
// which keeps every context (including service workers) consistent
|
|
// and preserves Chrome's native client hints. No CDP override here.
|
|
// remote — no launch flag is possible, so align the main frame via CDP:
|
|
// fetch the real UA, strip any "HeadlessChrome" token, and set a
|
|
// matching platform and client-hint metadata.
|
|
if opts.Remote != "" {
|
|
applyRemoteIdentity(page, opts.UserAgent, emitter)
|
|
}
|
|
|
|
// Headful Chrome ties a page's focus and render pipeline to the state of its
|
|
// operating system window, and one browser or one display can only ever have
|
|
// a single foreground window. Concurrent sessions would then fight over it:
|
|
// whichever page was raised last keeps rendering while the rest are treated
|
|
// as background, stop producing screencast frames, and report
|
|
// document.hasFocus() false. Focus emulation detaches the page from that
|
|
// window state, so every session renders and reads as focused at the same
|
|
// time. It also removes the unfocused window tell in headful mode.
|
|
if err := (proto.EmulationSetFocusEmulationEnabled{Enabled: true}).Call(page); err != nil {
|
|
emitter.log(fmt.Sprintf("[session] warning: focus emulation failed: %v", err))
|
|
}
|
|
|
|
// Timezone override keeps the navigator/Intl timezone aligned with the exit
|
|
// network's geolocation. Left to the operator because only they know the
|
|
// proxy/exit location; an unset value keeps the host timezone.
|
|
if opts.Timezone != "" {
|
|
if err := (proto.EmulationSetTimezoneOverride{TimezoneID: opts.Timezone}).Call(page); err != nil {
|
|
emitter.log(fmt.Sprintf("[session] warning: failed to set timezone %q: %v", opts.Timezone, err))
|
|
} else {
|
|
emitter.log(fmt.Sprintf("[session] using timezone: %s", opts.Timezone))
|
|
}
|
|
}
|
|
|
|
// Note: the software WebGL renderer (SwiftShader, from headless Chrome without a
|
|
// GPU) is intentionally NOT masked in JS. Overriding getParameter plus
|
|
// Function.prototype.toString to hide it trips TamperedFunctions, and a
|
|
// document-script override does not reach worker OffscreenCanvas contexts, so the
|
|
// window and workers report different renderers (hasInconsistentWorkerValues).
|
|
// Both are far stronger bot signals than SwiftShader itself. The correct fix is a
|
|
// real GPU at the browser level (headful + GPU, or an ANGLE GL backend), not JS.
|
|
|
|
// Signal the controller that a page is ready for streaming.
|
|
select {
|
|
case r.BrowserCh <- page:
|
|
default:
|
|
}
|
|
|
|
session := vm.NewObject()
|
|
RegisterBrowserBindings(vm, session, page, emitter, opts.Debug, opts.QueryTimeout)
|
|
|
|
session.Set("withTimeout", func(call goja.FunctionCall) goja.Value {
|
|
ms := call.Argument(0).ToInteger()
|
|
fn, ok := goja.AssertFunction(call.Argument(1))
|
|
if !ok {
|
|
panic(vm.NewTypeError("withTimeout: second argument must be a function"))
|
|
}
|
|
tCtx, tCancel := context.WithTimeout(page.GetContext(), time.Duration(ms)*time.Millisecond)
|
|
defer tCancel()
|
|
tPage := page.Context(tCtx)
|
|
tmpSession := vm.NewObject()
|
|
RegisterBrowserBindings(vm, tmpSession, tPage, nil, opts.Debug, opts.QueryTimeout)
|
|
_, err := fn(goja.Undefined(), tmpSession)
|
|
if err != nil {
|
|
// If our own timeout context expired, return false instead of
|
|
// propagating — lets callers branch without try/catch.
|
|
if tCtx.Err() != nil {
|
|
return vm.ToValue(false)
|
|
}
|
|
panic(err)
|
|
}
|
|
return vm.ToValue(true)
|
|
})
|
|
|
|
session.Set("close", func(call goja.FunctionCall) goja.Value {
|
|
emitter.log("[session] closing")
|
|
// Both branches call Close on the browser handle, and in remote mode
|
|
// that handle is bound to this session's browser context, so it
|
|
// disposes the context and its storage rather than the shared browser.
|
|
browser.Close() //nolint:errcheck
|
|
return goja.Undefined()
|
|
})
|
|
|
|
// keepAlive signals the caller that the page is available for operator
|
|
// takeover and cancels the script timeout so the parked session isn't
|
|
// killed after 60 s. It is intentionally non-blocking: the script
|
|
// continues after the call (so emit() calls after keepAlive() reach
|
|
// the victim). Run() parks itself after RunString returns.
|
|
session.Set("keepAlive", func(call goja.FunctionCall) goja.Value {
|
|
emitter.log("[session] keeping alive for remote takeover")
|
|
// Re-bind page to outerCtx before cancelling the timeout context.
|
|
// The browser was created with the timed ctx; if we cancel it first,
|
|
// page.GetContext() closes and StreamLiveSession fires "Session ended"
|
|
// before the operator even connects.
|
|
livePage := page.Context(outerCtx)
|
|
select {
|
|
case r.LiveCh <- livePage:
|
|
default:
|
|
}
|
|
emitter.sendMust(outerCtx, RunEvent{
|
|
Type: "keep_alive",
|
|
Time: time.Now().UTC().Format(time.RFC3339Nano),
|
|
})
|
|
r.keepAliveActive.Store(true)
|
|
timeoutCancel() // release script timeout — operator controls lifetime now
|
|
return goja.Undefined()
|
|
})
|
|
|
|
// Event-driven API: s.on(event, fn) + s.listen() + s.done()
|
|
// Built-in lifecycle events emitted by the server:
|
|
// "disconnect" — victim WebSocket connection dropped
|
|
// "navigate" — main frame navigated; data: { url: string }
|
|
handlers := map[string]goja.Callable{}
|
|
listenDone := make(chan struct{}, 1)
|
|
|
|
session.Set("on", func(call goja.FunctionCall) goja.Value {
|
|
event := call.Argument(0).String()
|
|
fn, ok := goja.AssertFunction(call.Argument(1))
|
|
if !ok {
|
|
panic(vm.NewTypeError("on: second argument must be a function"))
|
|
}
|
|
handlers[event] = fn
|
|
return goja.Undefined()
|
|
})
|
|
|
|
session.Set("done", func(call goja.FunctionCall) goja.Value {
|
|
select {
|
|
case listenDone <- struct{}{}:
|
|
default:
|
|
}
|
|
return goja.Undefined()
|
|
})
|
|
|
|
session.Set("listen", func(call goja.FunctionCall) goja.Value {
|
|
emitter.log("[session] listening for events")
|
|
|
|
var idleCh <-chan time.Time
|
|
var idleTimer *time.Timer
|
|
resetIdle := func() {
|
|
if opts.IdleTimeout > 0 {
|
|
if idleTimer != nil {
|
|
idleTimer.Stop()
|
|
}
|
|
idleTimer = time.NewTimer(time.Duration(opts.IdleTimeout) * time.Millisecond)
|
|
idleCh = idleTimer.C
|
|
}
|
|
}
|
|
defer func() {
|
|
if idleTimer != nil {
|
|
idleTimer.Stop()
|
|
}
|
|
}()
|
|
resetIdle()
|
|
|
|
for {
|
|
select {
|
|
case <-ctx.Done():
|
|
return goja.Undefined()
|
|
case <-listenDone:
|
|
return goja.Undefined()
|
|
case <-idleCh:
|
|
emitter.log(fmt.Sprintf("[session] idle timeout (%dms), closing", opts.IdleTimeout))
|
|
browser.Close() //nolint:errcheck
|
|
return goja.Undefined()
|
|
case msg := <-r.Incoming:
|
|
resetIdle()
|
|
fn, exists := handlers[msg.Event]
|
|
if !exists {
|
|
emitter.log(fmt.Sprintf("[session] no handler for %q, ignoring", msg.Event))
|
|
continue
|
|
}
|
|
if _, err := fn(goja.Undefined(), vm.ToValue(msg.Data)); err != nil {
|
|
panic(err)
|
|
}
|
|
}
|
|
}
|
|
})
|
|
|
|
// s.stream(selector, name) — non-blocking; returns {stop()} to end the stream.
|
|
// The caller (controller) watches StreamCh to start/stop cropped frame forwarding.
|
|
streamDebug := opts.Debug // capture bool, not struct field, to match RegisterBrowserBindings pattern
|
|
session.Set("stream", func(call goja.FunctionCall) goja.Value {
|
|
selector := vmArgStr(call.Argument(0))
|
|
name := vmArgStr(call.Argument(1))
|
|
if selector == "" || name == "" {
|
|
panic(vm.NewTypeError("stream: selector and name are required"))
|
|
}
|
|
maxFps := 0
|
|
quality := 0
|
|
if len(call.Arguments) > 2 {
|
|
if obj := call.Argument(2).ToObject(vm); obj != nil {
|
|
if v := obj.Get("maxFps"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
maxFps = int(v.ToInteger())
|
|
}
|
|
if v := obj.Get("quality"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
quality = int(v.ToInteger())
|
|
}
|
|
}
|
|
}
|
|
if streamDebug {
|
|
emitter.log(fmt.Sprintf("[dbg] → stream %s as %q (maxFps=%d, quality=%d)", selector, name, maxFps, quality))
|
|
}
|
|
select {
|
|
case r.StreamCh <- StreamCmd{Op: "start", Selector: selector, Name: name, Page: page, MaxFps: maxFps, Quality: quality}:
|
|
default:
|
|
emitter.log(fmt.Sprintf("[stream] warning: StreamCh full, dropping start for %q", name))
|
|
}
|
|
if streamDebug {
|
|
emitter.log(fmt.Sprintf("[dbg] ✓ stream %s as %q started", selector, name))
|
|
}
|
|
stopObj := vm.NewObject()
|
|
stopped := false
|
|
stopObj.Set("stop", func(call goja.FunctionCall) goja.Value {
|
|
if !stopped {
|
|
stopped = true
|
|
if streamDebug {
|
|
emitter.log(fmt.Sprintf("[dbg] → stream stop %q", name))
|
|
}
|
|
select {
|
|
case r.StreamCh <- StreamCmd{Op: "stop", Name: name}:
|
|
default:
|
|
emitter.log(fmt.Sprintf("[stream] warning: StreamCh full, dropping stop for %q", name))
|
|
}
|
|
if streamDebug {
|
|
emitter.log(fmt.Sprintf("[dbg] ✓ stream stop %q", name))
|
|
}
|
|
}
|
|
return goja.Undefined()
|
|
})
|
|
return stopObj
|
|
})
|
|
|
|
session.Set("capture", func(call goja.FunctionCall) goja.Value {
|
|
// Options:
|
|
// domains []string - filter cookies to these domains via CDP (e.g. ["google.com"])
|
|
// cookieNames []string - only keep cookies with these names
|
|
// localStorage bool - include localStorage (default true when no domains given)
|
|
// sessionStorage bool - include sessionStorage (default true when no domains given)
|
|
var domains []string
|
|
var cookieNames []string
|
|
lsExplicit, ssExplicit := false, false
|
|
capLS := true
|
|
capSS := true
|
|
|
|
if len(call.Arguments) > 0 {
|
|
obj := call.Argument(0).ToObject(vm)
|
|
if obj != nil {
|
|
if v := obj.Get("domains"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
if arr, ok := v.Export().([]any); ok {
|
|
for _, d := range arr {
|
|
domains = append(domains, fmt.Sprintf("%v", d))
|
|
}
|
|
}
|
|
}
|
|
if v := obj.Get("cookieNames"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
if arr, ok := v.Export().([]any); ok {
|
|
for _, n := range arr {
|
|
cookieNames = append(cookieNames, fmt.Sprintf("%v", n))
|
|
}
|
|
}
|
|
}
|
|
if v := obj.Get("localStorage"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
capLS = v.ToBoolean()
|
|
lsExplicit = true
|
|
}
|
|
if v := obj.Get("sessionStorage"); v != nil && !goja.IsUndefined(v) && !goja.IsNull(v) {
|
|
capSS = v.ToBoolean()
|
|
ssExplicit = true
|
|
}
|
|
}
|
|
}
|
|
|
|
// When domains are specified, skip storage by default unless the caller explicitly opted in.
|
|
if len(domains) > 0 {
|
|
if !lsExplicit {
|
|
capLS = false
|
|
}
|
|
if !ssExplicit {
|
|
capSS = false
|
|
}
|
|
}
|
|
|
|
result := map[string]any{}
|
|
|
|
if len(domains) > 0 {
|
|
urls := make([]string, len(domains))
|
|
for i, d := range domains {
|
|
d = strings.TrimPrefix(d, ".")
|
|
if !strings.HasPrefix(d, "http") {
|
|
d = "https://" + d
|
|
}
|
|
urls[i] = d
|
|
}
|
|
res, err := proto.NetworkGetCookies{Urls: urls}.Call(page)
|
|
if err == nil {
|
|
cookies := res.Cookies
|
|
// Filter by name if requested.
|
|
if len(cookieNames) > 0 {
|
|
nameSet := make(map[string]bool, len(cookieNames))
|
|
for _, n := range cookieNames {
|
|
nameSet[n] = true
|
|
}
|
|
filtered := cookies[:0]
|
|
for _, c := range cookies {
|
|
if nameSet[c.Name] {
|
|
filtered = append(filtered, c)
|
|
}
|
|
}
|
|
cookies = filtered
|
|
}
|
|
result["cookies"] = cookies
|
|
} else {
|
|
emitter.log(fmt.Sprintf("[capture] cookies: %s", err))
|
|
}
|
|
} else {
|
|
res, err := proto.StorageGetCookies{}.Call(page)
|
|
if err == nil {
|
|
cookies := res.Cookies
|
|
// Filter by name if requested.
|
|
if len(cookieNames) > 0 {
|
|
nameSet := make(map[string]bool, len(cookieNames))
|
|
for _, n := range cookieNames {
|
|
nameSet[n] = true
|
|
}
|
|
filtered := cookies[:0]
|
|
for _, c := range cookies {
|
|
if nameSet[c.Name] {
|
|
filtered = append(filtered, c)
|
|
}
|
|
}
|
|
cookies = filtered
|
|
}
|
|
result["cookies"] = cookies
|
|
} else {
|
|
emitter.log(fmt.Sprintf("[capture] cookies: %s", err))
|
|
}
|
|
}
|
|
|
|
evalStorage := func(key string, storeName string) {
|
|
script := fmt.Sprintf(`() => (function(){try{var s=window[%q],o={};for(var i=0;i<s.length;i++){var k=s.key(i);o[k]=s.getItem(k);}return JSON.stringify(o);}catch(e){return "{}"}})()`, storeName)
|
|
res, err := page.Eval(script)
|
|
if err != nil {
|
|
emitter.log(fmt.Sprintf("[capture] %s: %s", key, err))
|
|
return
|
|
}
|
|
var data interface{}
|
|
if json.Unmarshal([]byte(res.Value.Str()), &data) == nil {
|
|
result[key] = data
|
|
}
|
|
}
|
|
|
|
if capLS {
|
|
evalStorage("localStorage", "localStorage")
|
|
}
|
|
if capSS {
|
|
evalStorage("sessionStorage", "sessionStorage")
|
|
}
|
|
|
|
emitter.capture(result)
|
|
return vm.ToValue(result)
|
|
})
|
|
|
|
// makeFrameSession builds a DOM-only sub-session bound to framePage.
|
|
// Exposes RegisterBrowserBindings + withTimeout + frame (for nested iframes).
|
|
var makeFrameSession func(framePage *rod.Page) goja.Value
|
|
makeFrameSession = func(framePage *rod.Page) goja.Value {
|
|
fs := vm.NewObject()
|
|
RegisterBrowserBindings(vm, fs, framePage, emitter, opts.Debug, opts.QueryTimeout)
|
|
fs.Set("withTimeout", func(call goja.FunctionCall) goja.Value {
|
|
ms := call.Argument(0).ToInteger()
|
|
fn, ok := goja.AssertFunction(call.Argument(1))
|
|
if !ok {
|
|
panic(vm.NewTypeError("withTimeout: second argument must be a function"))
|
|
}
|
|
tCtx, tCancel := context.WithTimeout(framePage.GetContext(), time.Duration(ms)*time.Millisecond)
|
|
defer tCancel()
|
|
tFramePage := framePage.Context(tCtx)
|
|
tmpFrame := vm.NewObject()
|
|
RegisterBrowserBindings(vm, tmpFrame, tFramePage, nil, opts.Debug, opts.QueryTimeout)
|
|
_, err := fn(goja.Undefined(), tmpFrame)
|
|
if err != nil {
|
|
if tCtx.Err() != nil {
|
|
return vm.ToValue(false)
|
|
}
|
|
panic(err)
|
|
}
|
|
return vm.ToValue(true)
|
|
})
|
|
fs.Set("frame", func(call goja.FunctionCall) goja.Value {
|
|
sel := vmArgStr(call.Argument(0))
|
|
if sel == "" {
|
|
panic(vm.NewTypeError("frame: selector is required"))
|
|
}
|
|
if opts.Debug {
|
|
emitter.log("[dbg] → frame " + sel)
|
|
}
|
|
el, err := framePage.Element(sel)
|
|
if err != nil {
|
|
emitter.log(fmt.Sprintf("[frame] %q not found: %v", sel, err))
|
|
return goja.Null()
|
|
}
|
|
nestedPage, err := el.Frame()
|
|
if err != nil {
|
|
emitter.log(fmt.Sprintf("[frame] %q frame resolve failed: %v", sel, err))
|
|
return goja.Null()
|
|
}
|
|
if opts.Debug {
|
|
emitter.log("[dbg] ✓ frame " + sel)
|
|
}
|
|
return makeFrameSession(nestedPage.Context(framePage.GetContext()))
|
|
})
|
|
return fs
|
|
}
|
|
|
|
session.Set("frame", func(call goja.FunctionCall) goja.Value {
|
|
sel := vmArgStr(call.Argument(0))
|
|
if sel == "" {
|
|
panic(vm.NewTypeError("frame: selector is required"))
|
|
}
|
|
if opts.Debug {
|
|
emitter.log("[dbg] → frame " + sel)
|
|
}
|
|
el, err := page.Element(sel)
|
|
if err != nil {
|
|
emitter.log(fmt.Sprintf("[frame] %q not found: %v", sel, err))
|
|
return goja.Null()
|
|
}
|
|
framePage, err := el.Frame()
|
|
if err != nil {
|
|
emitter.log(fmt.Sprintf("[frame] %q frame resolve failed: %v", sel, err))
|
|
return goja.Null()
|
|
}
|
|
if opts.Debug {
|
|
emitter.log("[dbg] ✓ frame " + sel)
|
|
}
|
|
return makeFrameSession(framePage.Context(page.GetContext()))
|
|
})
|
|
|
|
// raceDOMCondJS returns a JS function body that evaluates the given condition
|
|
// type against sel, returning the selector string on match or null otherwise.
|
|
raceDOMCondJS := func(condType, sel string) string {
|
|
q := fmt.Sprintf("document.querySelector(%q)", sel)
|
|
r := fmt.Sprintf("%q", sel)
|
|
switch condType {
|
|
case "visible":
|
|
return fmt.Sprintf(`()=>{var el=%s;if(!el)return null;var b=el.getBoundingClientRect();return(b.width!==0||b.height!==0)?%s:null}`, q, r)
|
|
case "ready":
|
|
return fmt.Sprintf(`()=>{var el=%s;if(!el)return null;var b=el.getBoundingClientRect();if(b.width===0&&b.height===0)return null;return el.disabled?null:%s}`, q, r)
|
|
case "enabled":
|
|
return fmt.Sprintf(`()=>{var el=%s;return(el&&!el.disabled)?%s:null}`, q, r)
|
|
case "notVisible":
|
|
return fmt.Sprintf(`()=>{var el=%s;if(!el)return %s;var b=el.getBoundingClientRect();return(b.width===0&&b.height===0)?%s:null}`, q, r, r)
|
|
case "notPresent":
|
|
return fmt.Sprintf(`()=>{return !%s?%s:null}`, q, r)
|
|
default: // "present"
|
|
return fmt.Sprintf(`()=>{return %s?%s:null}`, q, r)
|
|
}
|
|
}
|
|
|
|
// s.race(conditions) races DOM conditions, URL substrings, and incoming events
|
|
// simultaneously. condition keys: visible, ready, enabled, notVisible, notPresent,
|
|
// present (DOM), url (substring), event (name).
|
|
// Returns { key, value } for whichever condition fires first.
|
|
session.Set("race", func(call goja.FunctionCall) goja.Value {
|
|
if len(call.Arguments) == 0 {
|
|
panic(vm.NewTypeError("race: expected an object with named conditions"))
|
|
}
|
|
condObj := call.Argument(0).ToObject(vm)
|
|
|
|
type domCond struct{ key, selector, condJS string }
|
|
type urlCond struct {
|
|
key string
|
|
match func(string) bool
|
|
display string // for logging
|
|
}
|
|
type evtCond struct{ key, event string }
|
|
|
|
var doms []domCond
|
|
var urls []urlCond
|
|
var evts []evtCond
|
|
evtSet := map[string]string{}
|
|
var timeoutCh <-chan time.Time
|
|
var timeoutKey string
|
|
|
|
domTypes := []string{"visible", "ready", "enabled", "notVisible", "notPresent", "present"}
|
|
|
|
for _, k := range condObj.Keys() {
|
|
v := condObj.Get(k).ToObject(vm)
|
|
if v == nil {
|
|
continue
|
|
}
|
|
matched := false
|
|
for _, ct := range domTypes {
|
|
if sel := v.Get(ct); sel != nil && !goja.IsUndefined(sel) && !goja.IsNull(sel) {
|
|
doms = append(doms, domCond{k, sel.String(), raceDOMCondJS(ct, sel.String())})
|
|
matched = true
|
|
break
|
|
}
|
|
}
|
|
if matched {
|
|
continue
|
|
}
|
|
if u := v.Get("urlContains"); u != nil && !goja.IsUndefined(u) && !goja.IsNull(u) {
|
|
pat := u.String()
|
|
urls = append(urls, urlCond{k, func(s string) bool { return strings.Contains(s, pat) }, pat})
|
|
} else if u := v.Get("urlMatch"); u != nil && !goja.IsUndefined(u) && !goja.IsNull(u) {
|
|
re, ok := u.Export().(*regexp.Regexp)
|
|
if !ok {
|
|
panic(vm.NewTypeError("race: urlMatch value must be a RegExp"))
|
|
}
|
|
urls = append(urls, urlCond{k, re.MatchString, re.String()})
|
|
} else if e := v.Get("event"); e != nil && !goja.IsUndefined(e) && !goja.IsNull(e) {
|
|
evts = append(evts, evtCond{k, e.String()})
|
|
evtSet[e.String()] = k
|
|
} else if a := v.Get("after"); a != nil && !goja.IsUndefined(a) && !goja.IsNull(a) {
|
|
timeoutCh = time.After(time.Duration(a.ToInteger()) * time.Millisecond)
|
|
timeoutKey = k
|
|
}
|
|
}
|
|
|
|
makeResult := func(key string, value interface{}) goja.Value {
|
|
emitter.log(fmt.Sprintf("[race] matched %q = %v", key, value))
|
|
res := vm.NewObject()
|
|
res.Set("key", key)
|
|
res.Set("value", vm.ToValue(value))
|
|
return res
|
|
}
|
|
|
|
// Drain event queue before blocking
|
|
for i, msg := range eventQueue {
|
|
if key, ok := evtSet[msg.Event]; ok {
|
|
eventQueue = append(eventQueue[:i], eventQueue[i+1:]...)
|
|
return makeResult(key, msg.Data)
|
|
}
|
|
}
|
|
|
|
urlDisplays := make([]string, len(urls))
|
|
for i, u := range urls {
|
|
urlDisplays[i] = u.display
|
|
}
|
|
emitter.log(fmt.Sprintf("[race] waiting: doms=%d urls=%v events=%d", len(doms), urlDisplays, len(evts)))
|
|
|
|
ticker := time.NewTicker(100 * time.Millisecond)
|
|
defer ticker.Stop()
|
|
|
|
for {
|
|
select {
|
|
case <-ctx.Done():
|
|
panic(vm.NewGoError(ctx.Err()))
|
|
|
|
case <-timeoutCh:
|
|
return makeResult(timeoutKey, "timeout")
|
|
|
|
case msg := <-r.Incoming:
|
|
if key, matched := evtSet[msg.Event]; matched {
|
|
return makeResult(key, msg.Data)
|
|
}
|
|
enqueue(msg)
|
|
|
|
case <-ticker.C:
|
|
for _, d := range doms {
|
|
res, err := page.Eval(d.condJS)
|
|
if err != nil {
|
|
continue
|
|
}
|
|
if res.Value.Nil() {
|
|
continue
|
|
}
|
|
if s := res.Value.Str(); s != "" && s != "null" && s != "undefined" {
|
|
return makeResult(d.key, d.selector)
|
|
}
|
|
}
|
|
if len(urls) > 0 {
|
|
info, err := page.Info()
|
|
if err == nil {
|
|
for _, u := range urls {
|
|
if u.match(info.URL) {
|
|
return makeResult(u.key, info.URL)
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
})
|
|
|
|
return session
|
|
})
|
|
|
|
_, err := vm.RunString("(function(){\n" + r.Script + "\n})()")
|
|
if err != nil {
|
|
// errors.Is/As traverse goja.Exception.Unwrap(), which extracts the Go error
|
|
// stored in the "value" property of a GoError object. Do NOT use
|
|
// ex.Value().Export().(error) — that returns map[string]interface{} for JS
|
|
// objects and always fails the type assertion.
|
|
var stopErr scriptStopError
|
|
if errors.As(err, &stopErr) {
|
|
emitter.done()
|
|
return nil
|
|
}
|
|
if errors.Is(err, context.DeadlineExceeded) {
|
|
// Check whether the *global* script timeout fired. If ctx.Err() is
|
|
// DeadlineExceeded, the outer context expired — surface a clear timeout
|
|
// message and send the session_timeout lifecycle event to the victim page.
|
|
// Otherwise this is a per-operation timeout (withTimeout, queryTimeout).
|
|
if errors.Is(ctx.Err(), context.DeadlineExceeded) {
|
|
emitter.errorf(fmt.Sprintf("script timed out after %ds", r.Config.Timeout/1000))
|
|
emitter.emit("session_timeout", nil)
|
|
} else {
|
|
emitter.errorf("operation timed out")
|
|
}
|
|
emitter.done()
|
|
return nil
|
|
}
|
|
if errors.Is(err, context.Canceled) {
|
|
// Three distinct cancellation sources share this error value:
|
|
// 1. keepAlive() called timeoutCancel() → ctx canceled, outerCtx still live.
|
|
// Any subsequent ctx.Done() select in race/waitFor/listen fires immediately
|
|
// with context.Canceled, causing RunString to return here before the park
|
|
// block below is ever reached. Park explicitly so the browser stays alive.
|
|
// 2. Operator or visitor ended the session → outerCtx canceled.
|
|
// Emit session_closed so the victim page can react (e.g. redirect).
|
|
if r.keepAliveActive.Load() {
|
|
emitter.log("[session] script complete (after keepAlive), parked for operator takeover")
|
|
<-outerCtx.Done()
|
|
emitter.log("[session] keep-alive ended")
|
|
emitter.done()
|
|
return nil
|
|
}
|
|
emitter.emit("session_closed", nil)
|
|
emitter.done()
|
|
return nil
|
|
}
|
|
// Script-level error: surface the formatted exception.
|
|
var ex *goja.Exception
|
|
if errors.As(err, &ex) {
|
|
emitter.errorf(cleanGoError(ex.String()))
|
|
return fmt.Errorf("script error: %s", ex.String())
|
|
}
|
|
emitter.errorf(err.Error())
|
|
return err
|
|
}
|
|
|
|
// keepAlive() was called: script has finished but the browser must stay
|
|
// alive for operator takeover. Block here until the operator explicitly
|
|
// cancels the session (outerCtx), then emit done so the Events channel
|
|
// drains cleanly.
|
|
if r.keepAliveActive.Load() {
|
|
emitter.log("[session] script complete, parked for operator takeover")
|
|
<-outerCtx.Done()
|
|
emitter.log("[session] keep-alive ended")
|
|
}
|
|
|
|
emitter.done()
|
|
return nil
|
|
}
|