1use crate::listeners::{EventListenerRequest, EventListeners};
2use chromiumoxide_cdp::cdp::browser_protocol::browser::*;
3use chromiumoxide_cdp::cdp::browser_protocol::target::*;
4use chromiumoxide_cdp::cdp::events::CdpEvent;
5use chromiumoxide_cdp::cdp::events::CdpEventMessage;
6use chromiumoxide_types::{CallId, Message, Method, Response};
7use chromiumoxide_types::{MethodId, Request as CdpRequest};
8use fnv::FnvHashMap;
9use futures_util::Stream;
10use hashbrown::{HashMap, HashSet};
11use spider_network_blocker::intercept_manager::NetworkInterceptManager;
12use std::pin::Pin;
13use std::task::{Context, Poll};
14use std::time::{Duration, Instant};
15use tokio::sync::mpsc::Receiver;
16use tokio::sync::oneshot::Sender as OneshotSender;
17use tokio_tungstenite::tungstenite::error::ProtocolError;
18use tokio_tungstenite::tungstenite::Error;
19
20use std::sync::Arc;
21use tokio::sync::Notify;
22
23use crate::cmd::{to_command_response, CommandMessage};
24use crate::conn::Connection;
25use crate::error::{CdpError, Result};
26use crate::handler::browser::BrowserContext;
27use crate::handler::frame::FrameRequestedNavigation;
28use crate::handler::frame::{NavigationError, NavigationId, NavigationOk};
29use crate::handler::job::PeriodicJob;
30use crate::handler::session::Session;
31use crate::handler::target::TargetEvent;
32use crate::handler::target::{Target, TargetConfig};
33use crate::handler::viewport::Viewport;
34use crate::page::Page;
35pub(crate) use page::PageInner;
36
37pub const REQUEST_TIMEOUT: u64 = 30_000;
39
40pub mod blockers;
41pub mod browser;
42pub mod commandfuture;
43pub mod domworld;
44pub mod emulation;
45pub mod frame;
46pub mod http;
47pub mod httpfuture;
48mod job;
49pub mod network;
50pub mod network_utils;
51pub mod page;
52#[cfg(feature = "parallel-handler")]
53pub mod parallel;
54pub mod sender;
55mod session;
56pub mod target;
57pub mod target_message_future;
58pub mod viewport;
59
60#[must_use = "streams do nothing unless polled"]
63#[derive(Debug)]
64pub struct Handler {
65 pub default_browser_context: BrowserContext,
66 pub browser_contexts: HashSet<BrowserContext>,
67 pending_commands: FnvHashMap<CallId, (PendingRequest, MethodId, Instant)>,
71 from_browser: Receiver<HandlerMessage>,
73 target_ids: Vec<TargetId>,
75 targets: HashMap<TargetId, Target>,
77 navigations: FnvHashMap<NavigationId, NavigationRequest>,
79 sessions: HashMap<SessionId, Session>,
83 conn: Option<Connection<CdpEventMessage>>,
86 evict_command_timeout: PeriodicJob,
88 next_navigation_id: usize,
90 config: HandlerConfig,
92 event_listeners: EventListeners,
94 closing: bool,
96 remaining_bytes: Option<u64>,
98 budget_exhausted: bool,
100 attached_targets: HashSet<TargetId>,
102 page_wake: Option<Arc<Notify>>,
105}
106
107lazy_static::lazy_static! {
108 static ref DISCOVER_ID: (std::borrow::Cow<'static, str>, serde_json::Value) = {
110 let discover = SetDiscoverTargetsParams::new(true);
111 (discover.identifier(), serde_json::to_value(discover).expect("valid discover target params"))
112 };
113 static ref TARGET_PARAMS_ID: (std::borrow::Cow<'static, str>, serde_json::Value) = {
115 let msg = GetTargetsParams { filter: None };
116 (msg.identifier(), serde_json::to_value(msg).expect("valid paramtarget"))
117 };
118 static ref CLOSE_PARAMS_ID: (std::borrow::Cow<'static, str>, serde_json::Value) = {
120 let close_msg = CloseParams::default();
121 (close_msg.identifier(), serde_json::to_value(close_msg).expect("valid close params"))
122 };
123}
124
125fn maybe_store_attach_session_id(target: &mut Target, method: &MethodId, resp: &Response) {
126 if method.as_ref() != AttachToTargetParams::IDENTIFIER {
127 return;
128 }
129
130 if let Ok(resp) = to_command_response::<AttachToTargetParams>(resp.clone(), method.clone()) {
131 target.set_session_id(resp.result.session_id);
132 }
133}
134
135impl Handler {
136 pub(crate) fn new(
139 mut conn: Connection<CdpEventMessage>,
140 rx: Receiver<HandlerMessage>,
141 config: HandlerConfig,
142 ) -> Self {
143 let discover = DISCOVER_ID.clone();
144 let _ = conn.submit_command(discover.0, None, discover.1);
145 let conn = Some(conn);
146
147 let browser_contexts = config
148 .context_ids
149 .iter()
150 .map(|id| BrowserContext::from(id.clone()))
151 .collect();
152
153 Self {
154 pending_commands: Default::default(),
155 from_browser: rx,
156 default_browser_context: Default::default(),
157 browser_contexts,
158 target_ids: Default::default(),
159 targets: Default::default(),
160 navigations: Default::default(),
161 sessions: Default::default(),
162 conn,
163 evict_command_timeout: PeriodicJob::new(config.request_timeout),
164 next_navigation_id: 0,
165 config,
166 event_listeners: Default::default(),
167 closing: false,
168 remaining_bytes: None,
169 budget_exhausted: false,
170 attached_targets: Default::default(),
171 page_wake: None,
172 }
173 }
174
175 #[inline]
178 fn conn(&mut self) -> Result<&mut Connection<CdpEventMessage>> {
179 self.conn
180 .as_mut()
181 .ok_or_else(|| CdpError::msg("connection consumed by Handler::run()"))
182 }
183
184 pub fn get_target(&self, target_id: &TargetId) -> Option<&Target> {
186 self.targets.get(target_id)
187 }
188
189 pub fn targets(&self) -> impl Iterator<Item = &Target> + '_ {
191 self.targets.values()
192 }
193
194 pub fn default_browser_context(&self) -> &BrowserContext {
196 &self.default_browser_context
197 }
198
199 pub fn browser_contexts(&self) -> impl Iterator<Item = &BrowserContext> + '_ {
201 self.browser_contexts.iter()
202 }
203
204 fn on_navigation_response(&mut self, id: NavigationId, resp: Response) {
206 if let Some(nav) = self.navigations.remove(&id) {
207 match nav {
208 NavigationRequest::Navigate(mut nav) => {
209 if nav.navigated {
210 let _ = nav.tx.send(Ok(resp));
211 } else {
212 nav.set_response(resp);
213 self.navigations
214 .insert(id, NavigationRequest::Navigate(nav));
215 }
216 }
217 }
218 }
219 }
220
221 fn on_navigation_lifecycle_completed(&mut self, res: Result<NavigationOk, NavigationError>) {
223 match res {
224 Ok(ok) => {
225 let id = *ok.navigation_id();
226 if let Some(nav) = self.navigations.remove(&id) {
227 match nav {
228 NavigationRequest::Navigate(mut nav) => {
229 if let Some(resp) = nav.response.take() {
230 let _ = nav.tx.send(Ok(resp));
231 } else {
232 nav.set_navigated();
233 self.navigations
234 .insert(id, NavigationRequest::Navigate(nav));
235 }
236 }
237 }
238 }
239 }
240 Err(err) => {
241 if let Some(nav) = self.navigations.remove(err.navigation_id()) {
242 match nav {
243 NavigationRequest::Navigate(nav) => {
244 let _ = nav.tx.send(Err(err.into()));
245 }
246 }
247 }
248 }
249 }
250 }
251
252 fn on_response(&mut self, resp: Response) {
254 if let Some((req, method, _)) = self.pending_commands.remove(&resp.id) {
255 match req {
256 PendingRequest::CreateTarget(tx) => {
257 match to_command_response::<CreateTargetParams>(resp, method) {
258 Ok(resp) => {
259 if let Some(target) = self.targets.get_mut(&resp.target_id) {
260 target.set_initiator(tx);
261 } else {
262 let _ = tx.send(Err(CdpError::NotFound)).ok();
263 }
264 }
265 Err(err) => {
266 let _ = tx.send(Err(err)).ok();
267 }
268 }
269 }
270 PendingRequest::GetTargets(tx) => {
271 match to_command_response::<GetTargetsParams>(resp, method) {
272 Ok(resp) => {
273 let targets = resp.result.target_infos;
274 let results = targets.clone();
275
276 for target_info in targets {
277 let event: EventTargetCreated = EventTargetCreated { target_info };
278 self.on_target_created(event);
279 }
280
281 let _ = tx.send(Ok(results)).ok();
282 }
283 Err(err) => {
284 let _ = tx.send(Err(err)).ok();
285 }
286 }
287 }
288 PendingRequest::Navigate(id) => {
289 self.on_navigation_response(id, resp);
290 if self.config.only_html && !self.config.created_first_target {
291 self.config.created_first_target = true;
292 }
293 }
294 PendingRequest::ExternalCommand { tx, .. } => {
295 let _ = tx.send(Ok(resp)).ok();
296 }
297 PendingRequest::InternalCommand(target_id) => {
298 if let Some(target) = self.targets.get_mut(&target_id) {
299 maybe_store_attach_session_id(target, &method, &resp);
300 target.on_response(resp, method.as_ref());
301 }
302 }
303 PendingRequest::CloseBrowser(tx) => {
304 self.closing = true;
305 let _ = tx.send(Ok(CloseReturns {})).ok();
306 }
307 }
308 }
309 }
310
311 pub(crate) fn submit_external_command(
313 &mut self,
314 msg: CommandMessage,
315 now: Instant,
316 ) -> Result<()> {
317 let target_id = msg
322 .session_id
323 .as_ref()
324 .and_then(|sid| self.sessions.get(sid.as_ref()))
325 .map(|s| s.target_id().clone());
326 let call_id =
327 self.conn()?
328 .submit_command(msg.method.clone(), msg.session_id, msg.params)?;
329 self.pending_commands.insert(
330 call_id,
331 (
332 PendingRequest::ExternalCommand {
333 tx: msg.sender,
334 target_id,
335 },
336 msg.method,
337 now,
338 ),
339 );
340 Ok(())
341 }
342
343 pub(crate) fn submit_internal_command(
344 &mut self,
345 target_id: TargetId,
346 req: CdpRequest,
347 now: Instant,
348 ) -> Result<()> {
349 let call_id = self.conn()?.submit_command(
350 req.method.clone(),
351 req.session_id.map(Into::into),
352 req.params,
353 )?;
354 self.pending_commands.insert(
355 call_id,
356 (PendingRequest::InternalCommand(target_id), req.method, now),
357 );
358 Ok(())
359 }
360
361 fn submit_fetch_targets(&mut self, tx: OneshotSender<Result<Vec<TargetInfo>>>, now: Instant) {
362 let msg = TARGET_PARAMS_ID.clone();
363
364 if let Some(conn) = self.conn.as_mut() {
365 if let Ok(call_id) = conn.submit_command(msg.0.clone(), None, msg.1) {
366 self.pending_commands
367 .insert(call_id, (PendingRequest::GetTargets(tx), msg.0, now));
368 }
369 }
370 }
371
372 fn submit_navigation(&mut self, id: NavigationId, req: CdpRequest, now: Instant) {
375 if let Some(conn) = self.conn.as_mut() {
376 if let Ok(call_id) = conn.submit_command(
377 req.method.clone(),
378 req.session_id.map(Into::into),
379 req.params,
380 ) {
381 self.pending_commands
382 .insert(call_id, (PendingRequest::Navigate(id), req.method, now));
383 }
384 }
385 }
386
387 fn submit_close(&mut self, tx: OneshotSender<Result<CloseReturns>>, now: Instant) {
388 let close_msg = CLOSE_PARAMS_ID.clone();
389
390 if let Some(conn) = self.conn.as_mut() {
391 if let Ok(call_id) = conn.submit_command(close_msg.0.clone(), None, close_msg.1) {
392 self.pending_commands.insert(
393 call_id,
394 (PendingRequest::CloseBrowser(tx), close_msg.0, now),
395 );
396 }
397 }
398 }
399
400 fn on_target_message(&mut self, target: &mut Target, msg: CommandMessage, now: Instant) {
402 if msg.is_navigation() {
403 let (req, tx) = msg.split();
404 let id = self.next_navigation_id();
405
406 target.goto(FrameRequestedNavigation::new(
407 id,
408 req,
409 self.config.request_timeout,
410 ));
411
412 self.navigations.insert(
413 id,
414 NavigationRequest::Navigate(NavigationInProgress::new(tx)),
415 );
416 } else {
417 let _ = self.submit_external_command(msg, now);
418 }
419 }
420
421 fn next_navigation_id(&mut self) -> NavigationId {
423 let id = NavigationId(self.next_navigation_id);
424 self.next_navigation_id = self.next_navigation_id.wrapping_add(1);
425 id
426 }
427
428 fn create_page(&mut self, params: CreateTargetParams, tx: OneshotSender<Result<Page>>) {
439 let about_blank = params.url == "about:blank";
440 let http_check =
441 !about_blank && params.url.starts_with("http") || params.url.starts_with("file://");
442
443 if about_blank || http_check {
444 let method = params.identifier();
445
446 let Some(conn) = self.conn.as_mut() else {
447 let _ = tx.send(Err(CdpError::msg("connection consumed"))).ok();
448 return;
449 };
450 match serde_json::to_value(params) {
451 Ok(params) => match conn.submit_command(method.clone(), None, params) {
452 Ok(call_id) => {
453 self.pending_commands.insert(
454 call_id,
455 (PendingRequest::CreateTarget(tx), method, Instant::now()),
456 );
457 }
458 Err(err) => {
459 let _ = tx.send(Err(err.into())).ok();
460 }
461 },
462 Err(err) => {
463 let _ = tx.send(Err(err.into())).ok();
464 }
465 }
466 } else {
467 let _ = tx.send(Err(CdpError::NotFound)).ok();
468 }
469 }
470
471 fn on_event(&mut self, event: CdpEventMessage) {
473 if let Some(session_id) = &event.session_id {
474 if let Some(session) = self.sessions.get(session_id.as_str()) {
475 if let Some(target) = self.targets.get_mut(session.target_id()) {
476 return target.on_event(event);
477 }
478 }
479 }
480 let CdpEventMessage { params, method, .. } = event;
481
482 match params {
483 CdpEvent::TargetTargetCreated(ref ev) => self.on_target_created((**ev).clone()),
484 CdpEvent::TargetAttachedToTarget(ref ev) => self.on_attached_to_target(ev.clone()),
485 CdpEvent::TargetTargetDestroyed(ref ev) => self.on_target_destroyed(ev.clone()),
486 CdpEvent::TargetTargetCrashed(ref ev) => self.on_target_crashed(ev.clone()),
487 CdpEvent::TargetDetachedFromTarget(ref ev) => self.on_detached_from_target(ev.clone()),
488 _ => {}
489 }
490
491 chromiumoxide_cdp::consume_event!(match params {
492 |ev| self.event_listeners.start_send(ev),
493 |json| { let _ = self.event_listeners.try_send_custom(&method, json);}
494 });
495 }
496
497 fn on_target_created(&mut self, event: EventTargetCreated) {
501 if !self.browser_contexts.is_empty() {
502 if let Some(ref context_id) = event.target_info.browser_context_id {
503 let bc = BrowserContext {
504 id: Some(context_id.clone()),
505 };
506 if !self.browser_contexts.contains(&bc) {
507 return;
508 }
509 }
510 }
511 let browser_ctx = event
512 .target_info
513 .browser_context_id
514 .clone()
515 .map(BrowserContext::from)
516 .unwrap_or_else(|| self.default_browser_context.clone());
517 let target = Target::new(
518 event.target_info,
519 TargetConfig {
520 ignore_https_errors: self.config.ignore_https_errors,
521 request_timeout: self.config.request_timeout,
522 viewport: self.config.viewport.clone(),
523 request_intercept: self.config.request_intercept,
524 cache_enabled: self.config.cache_enabled,
525 service_worker_enabled: self.config.service_worker_enabled,
526 ignore_visuals: self.config.ignore_visuals,
527 ignore_stylesheets: self.config.ignore_stylesheets,
528 ignore_javascript: self.config.ignore_javascript,
529 ignore_analytics: self.config.ignore_analytics,
530 ignore_prefetch: self.config.ignore_prefetch,
531 extra_headers: self.config.extra_headers.clone(),
532 only_html: self.config.only_html && self.config.created_first_target,
533 intercept_manager: self.config.intercept_manager,
534 max_bytes_allowed: self.config.max_bytes_allowed,
535 max_redirects: self.config.max_redirects,
536 max_main_frame_navigations: self.config.max_main_frame_navigations,
537 whitelist_patterns: self.config.whitelist_patterns.clone(),
538 blacklist_patterns: self.config.blacklist_patterns.clone(),
539 #[cfg(feature = "adblock")]
540 adblock_filter_rules: self.config.adblock_filter_rules.clone(),
541 page_wake: self.page_wake.clone(),
542 page_channel_capacity: self.config.page_channel_capacity,
543 },
544 browser_ctx,
545 );
546
547 let tid = target.target_id().clone();
548 self.target_ids.push(tid.clone());
549 self.targets.insert(tid, target);
550 }
551
552 fn on_attached_to_target(&mut self, event: Box<EventAttachedToTarget>) {
554 let session = Session::new(event.session_id.clone(), event.target_info.target_id);
555 if let Some(target) = self.targets.get_mut(session.target_id()) {
556 target.set_session_id(session.session_id().clone())
557 }
558 self.sessions.insert(event.session_id, session);
559 }
560
561 fn on_detached_from_target(&mut self, event: EventDetachedFromTarget) {
565 if let Some(session) = self.sessions.remove(&event.session_id) {
567 if let Some(target) = self.targets.get_mut(session.target_id()) {
568 target.session_id_mut().take();
569 }
570 }
571 }
572
573 fn on_target_destroyed(&mut self, event: EventTargetDestroyed) {
575 self.attached_targets.remove(&event.target_id);
576
577 if let Some(target) = self.targets.remove(&event.target_id) {
578 if let Some(session) = target.session_id() {
580 self.sessions.remove(session);
581 }
582 }
583 }
584
585 fn on_target_crashed(&mut self, event: EventTargetCrashed) {
608 let crashed_id = event.target_id.clone();
609 let status = event.status.clone();
610 let error_code = event.error_code;
611
612 let to_cancel: Vec<CallId> = self
618 .pending_commands
619 .iter()
620 .filter_map(|(&call_id, (req, _, _))| match req {
621 PendingRequest::ExternalCommand {
622 target_id: Some(tid),
623 ..
624 } if *tid == crashed_id => Some(call_id),
625 PendingRequest::InternalCommand(tid) if *tid == crashed_id => Some(call_id),
626 _ => None,
627 })
628 .collect();
629
630 for call_id in to_cancel {
631 if let Some((req, _, _)) = self.pending_commands.remove(&call_id) {
632 match req {
633 PendingRequest::ExternalCommand { tx, .. } => {
634 let _ = tx.send(Err(CdpError::msg(format!(
635 "target {:?} crashed: {} (errorCode={})",
636 crashed_id, status, error_code
637 ))));
638 }
639 PendingRequest::InternalCommand(_) => {
640 }
643 _ => {}
644 }
645 }
646 }
647
648 self.attached_targets.remove(&crashed_id);
650 if let Some(target) = self.targets.remove(&crashed_id) {
651 if let Some(session) = target.session_id() {
652 self.sessions.remove(session);
653 }
654 }
655 }
656
657 fn evict_timed_out_commands(&mut self, now: Instant) {
662 let deadline = match now.checked_sub(self.config.request_timeout) {
663 Some(d) => d,
664 None => return,
665 };
666
667 let timed_out: Vec<_> = self
668 .pending_commands
669 .iter()
670 .filter(|(_, (_, _, timestamp))| *timestamp < deadline)
671 .map(|(k, _)| *k)
672 .collect();
673
674 for call in timed_out {
675 if let Some((req, _, _)) = self.pending_commands.remove(&call) {
676 match req {
677 PendingRequest::CreateTarget(tx) => {
678 let _ = tx.send(Err(CdpError::Timeout));
679 }
680 PendingRequest::GetTargets(tx) => {
681 let _ = tx.send(Err(CdpError::Timeout));
682 }
683 PendingRequest::Navigate(nav) => {
684 if let Some(nav) = self.navigations.remove(&nav) {
685 match nav {
686 NavigationRequest::Navigate(nav) => {
687 let _ = nav.tx.send(Err(CdpError::Timeout));
688 }
689 }
690 }
691 }
692 PendingRequest::ExternalCommand { tx, .. } => {
693 let _ = tx.send(Err(CdpError::Timeout));
694 }
695 PendingRequest::InternalCommand(_) => {}
696 PendingRequest::CloseBrowser(tx) => {
697 let _ = tx.send(Err(CdpError::Timeout));
698 }
699 }
700 }
701 }
702 }
703
704 pub fn event_listeners_mut(&mut self) -> &mut EventListeners {
705 &mut self.event_listeners
706 }
707
708 pub async fn run(mut self) -> Result<()> {
738 use chromiumoxide_types::Message;
739 use tokio::time::MissedTickBehavior;
740 use tokio_tungstenite::tungstenite::{self, error::ProtocolError};
741
742 let page_wake = Arc::new(Notify::new());
744 self.page_wake = Some(page_wake.clone());
745
746 let conn = self
748 .conn
749 .take()
750 .ok_or_else(|| CdpError::msg("Handler::run() called with no connection"))?;
751 let async_conn = conn.into_async();
752 let mut ws_reader = async_conn.reader;
753 let ws_tx = async_conn.cmd_tx;
754 let mut writer_handle = async_conn.writer_handle;
755 let reader_handle = async_conn.reader_handle;
756 let mut next_call_id = async_conn.next_id;
757
758 let mut alloc_call_id = || {
760 let id = chromiumoxide_types::CallId::new(next_call_id);
761 next_call_id = next_call_id.wrapping_add(1);
762 id
763 };
764
765 let mut evict_timer = tokio::time::interval_at(
767 tokio::time::Instant::now() + self.config.request_timeout,
768 self.config.request_timeout,
769 );
770 evict_timer.set_missed_tick_behavior(MissedTickBehavior::Delay);
771
772 macro_rules! ws_submit {
774 ($method:expr, $session_id:expr, $params:expr) => {{
775 let id = alloc_call_id();
776 let call = chromiumoxide_types::MethodCall {
777 id,
778 method: $method,
779 session_id: $session_id,
780 params: $params,
781 };
782 match ws_tx.try_send(call) {
783 Ok(()) => Ok::<_, CdpError>(id),
784 Err(tokio::sync::mpsc::error::TrySendError::Full(_)) => {
785 tracing::warn!("WS command channel full — dropping command");
786 Err(CdpError::msg("WS command channel full"))
787 }
788 Err(tokio::sync::mpsc::error::TrySendError::Closed(_)) => {
789 Err(CdpError::msg("WS writer closed"))
790 }
791 }
792 }};
793 }
794
795 let run_result: Result<()> = loop {
803 let now = std::time::Instant::now();
804
805 const PER_TARGET_DRAIN_BUDGET: usize = 128;
811
812 for n in (0..self.target_ids.len()).rev() {
813 let target_id = self.target_ids.swap_remove(n);
814
815 if let Some((id, mut target)) = self.targets.remove_entry(&target_id) {
816 {
818 let mut msgs = Vec::new();
819 if let Some(handle) = target.page_mut() {
820 while msgs.len() < PER_TARGET_DRAIN_BUDGET {
821 match handle.rx.try_recv() {
822 Ok(msg) => msgs.push(msg),
823 Err(_) => break,
824 }
825 }
826 }
827 for msg in msgs {
828 target.on_page_message(msg);
829 }
830 }
831
832 while let Some(event) = target.advance(now) {
834 match event {
835 TargetEvent::Request(req) => {
836 if let Ok(call_id) =
837 ws_submit!(req.method.clone(), req.session_id, req.params)
838 {
839 self.pending_commands.insert(
840 call_id,
841 (
842 PendingRequest::InternalCommand(
843 target.target_id().clone(),
844 ),
845 req.method,
846 now,
847 ),
848 );
849 }
850 }
851 TargetEvent::Command(msg) => {
852 if msg.is_navigation() {
853 let (req, tx) = msg.split();
854 let nav_id = self.next_navigation_id();
855 target.goto(FrameRequestedNavigation::new(
856 nav_id,
857 req.clone(),
858 self.config.request_timeout,
859 ));
860 if let Ok(call_id) =
861 ws_submit!(req.method.clone(), req.session_id, req.params)
862 {
863 self.pending_commands.insert(
864 call_id,
865 (PendingRequest::Navigate(nav_id), req.method, now),
866 );
867 }
868 self.navigations.insert(
869 nav_id,
870 NavigationRequest::Navigate(NavigationInProgress::new(tx)),
871 );
872 } else if let Ok(call_id) = ws_submit!(
873 msg.method.clone(),
874 msg.session_id.map(Into::into),
875 msg.params
876 ) {
877 let target_id = Some(target.target_id().clone());
881 self.pending_commands.insert(
882 call_id,
883 (
884 PendingRequest::ExternalCommand {
885 tx: msg.sender,
886 target_id,
887 },
888 msg.method,
889 now,
890 ),
891 );
892 }
893 }
894 TargetEvent::NavigationRequest(nav_id, req) => {
895 if let Ok(call_id) =
896 ws_submit!(req.method.clone(), req.session_id, req.params)
897 {
898 self.pending_commands.insert(
899 call_id,
900 (PendingRequest::Navigate(nav_id), req.method, now),
901 );
902 }
903 }
904 TargetEvent::NavigationResult(res) => {
905 self.on_navigation_lifecycle_completed(res);
906 }
907 TargetEvent::BytesConsumed(n) => {
908 if let Some(rem) = self.remaining_bytes.as_mut() {
909 *rem = rem.saturating_sub(n);
910 if *rem == 0 {
911 self.budget_exhausted = true;
912 }
913 }
914 }
915 }
916 }
917
918 target.event_listeners_mut().flush();
920
921 self.targets.insert(id, target);
922 self.target_ids.push(target_id);
923 }
924 }
925
926 self.event_listeners.flush();
928
929 if self.budget_exhausted {
930 for t in self.targets.values_mut() {
931 t.network_manager.set_block_all(true);
932 }
933 }
934
935 if self.closing {
936 break Ok(());
937 }
938
939 tokio::select! {
941 msg = self.from_browser.recv() => {
942 match msg {
943 Some(msg) => {
944 match msg {
945 HandlerMessage::Command(cmd) => {
946 let target_id = cmd
949 .session_id
950 .as_ref()
951 .and_then(|sid| self.sessions.get(sid.as_ref()))
952 .map(|s| s.target_id().clone());
953 if let Ok(call_id) = ws_submit!(
954 cmd.method.clone(),
955 cmd.session_id.map(Into::into),
956 cmd.params
957 ) {
958 self.pending_commands.insert(
959 call_id,
960 (
961 PendingRequest::ExternalCommand {
962 tx: cmd.sender,
963 target_id,
964 },
965 cmd.method,
966 now,
967 ),
968 );
969 }
970 }
971 HandlerMessage::FetchTargets(tx) => {
972 let msg = TARGET_PARAMS_ID.clone();
973 if let Ok(call_id) = ws_submit!(msg.0.clone(), None, msg.1) {
974 self.pending_commands.insert(
975 call_id,
976 (PendingRequest::GetTargets(tx), msg.0, now),
977 );
978 }
979 }
980 HandlerMessage::CloseBrowser(tx) => {
981 let close_msg = CLOSE_PARAMS_ID.clone();
982 if let Ok(call_id) = ws_submit!(close_msg.0.clone(), None, close_msg.1) {
983 self.pending_commands.insert(
984 call_id,
985 (PendingRequest::CloseBrowser(tx), close_msg.0, now),
986 );
987 }
988 }
989 HandlerMessage::CreatePage(params, tx) => {
990 if let Some(ref id) = params.browser_context_id {
991 self.browser_contexts.insert(BrowserContext::from(id.clone()));
992 }
993 self.create_page_async(params, tx, &mut alloc_call_id, &ws_tx, now);
994 }
995 HandlerMessage::GetPages(tx) => {
996 let pages: Vec<_> = self.targets.values_mut()
997 .filter(|p| p.is_page())
998 .filter_map(|target| target.get_or_create_page())
999 .map(|page| Page::from(page.clone()))
1000 .collect();
1001 let _ = tx.send(pages);
1002 }
1003 HandlerMessage::InsertContext(ctx) => {
1004 if self.default_browser_context.id().is_none() {
1005 self.default_browser_context = ctx.clone();
1006 }
1007 self.browser_contexts.insert(ctx);
1008 }
1009 HandlerMessage::DisposeContext(ctx) => {
1010 self.browser_contexts.remove(&ctx);
1011 self.attached_targets.retain(|tid| {
1012 self.targets.get(tid)
1013 .and_then(|t| t.browser_context_id())
1014 .map(|id| Some(id) != ctx.id())
1015 .unwrap_or(true)
1016 });
1017 self.closing = true;
1018 }
1019 HandlerMessage::GetPage(target_id, tx) => {
1020 let page = self.targets.get_mut(&target_id)
1021 .and_then(|target| target.get_or_create_page())
1022 .map(|page| Page::from(page.clone()));
1023 let _ = tx.send(page);
1024 }
1025 HandlerMessage::AddEventListener(req) => {
1026 self.event_listeners.add_listener(req);
1027 }
1028 }
1029 }
1030 None => break Ok(()), }
1032 }
1033
1034 frame = ws_reader.next_message() => {
1035 match frame {
1036 Some(Ok(boxed_msg)) => match *boxed_msg {
1037 Message::Response(resp) => {
1038 self.on_response(resp);
1039 }
1040 Message::Event(ev) => {
1041 self.on_event(ev);
1042 }
1043 },
1044 Some(Err(err)) => {
1045 tracing::error!("WS Connection error: {:?}", err);
1046 if let CdpError::Ws(ref ws_error) = err {
1047 match ws_error {
1048 tungstenite::Error::AlreadyClosed => break Ok(()),
1049 tungstenite::Error::Protocol(detail)
1050 if detail == &ProtocolError::ResetWithoutClosingHandshake =>
1051 {
1052 break Ok(());
1053 }
1054 _ => break Err(err),
1055 }
1056 } else {
1057 break Err(err);
1058 }
1059 }
1060 None => break Ok(()), }
1062 }
1063
1064 _ = page_wake.notified() => {
1065 }
1067
1068 _ = evict_timer.tick() => {
1069 self.evict_timed_out_commands(now);
1070 for t in self.targets.values_mut() {
1071 t.network_manager.evict_stale_entries(now);
1072 t.frame_manager_mut().evict_stale_context_ids();
1073 }
1074 }
1075
1076 result = &mut writer_handle => {
1077 match result {
1079 Ok(Ok(())) => break Ok(()),
1080 Ok(Err(e)) => break Err(e),
1081 Err(e) => break Err(CdpError::msg(format!("WS writer panicked: {e}"))),
1082 }
1083 }
1084 }
1085 };
1086
1087 drop(ws_tx);
1095
1096 if !writer_handle.is_finished() {
1101 let _ = tokio::time::timeout(std::time::Duration::from_millis(500), &mut writer_handle)
1102 .await;
1103 if !writer_handle.is_finished() {
1104 writer_handle.abort();
1105 }
1106 }
1107
1108 reader_handle.abort();
1114
1115 run_result
1116 }
1117
1118 fn create_page_async(
1120 &mut self,
1121 params: CreateTargetParams,
1122 tx: OneshotSender<Result<Page>>,
1123 alloc_call_id: &mut impl FnMut() -> chromiumoxide_types::CallId,
1124 ws_tx: &tokio::sync::mpsc::Sender<chromiumoxide_types::MethodCall>,
1125 now: std::time::Instant,
1126 ) {
1127 let about_blank = params.url == "about:blank";
1128 let http_check =
1129 !about_blank && params.url.starts_with("http") || params.url.starts_with("file://");
1130
1131 if about_blank || http_check {
1132 let method = params.identifier();
1133 match serde_json::to_value(params) {
1134 Ok(params) => {
1135 let id = alloc_call_id();
1136 let call = chromiumoxide_types::MethodCall {
1137 id,
1138 method: method.clone(),
1139 session_id: None,
1140 params,
1141 };
1142 match ws_tx.try_send(call) {
1143 Ok(()) => {
1144 self.pending_commands
1145 .insert(id, (PendingRequest::CreateTarget(tx), method, now));
1146 }
1147 Err(_) => {
1148 let _ = tx
1149 .send(Err(CdpError::msg("WS command channel full or closed")))
1150 .ok();
1151 }
1152 }
1153 }
1154 Err(err) => {
1155 let _ = tx.send(Err(err.into())).ok();
1156 }
1157 }
1158 } else {
1159 let _ = tx.send(Err(CdpError::NotFound)).ok();
1160 }
1161 }
1162
1163 #[cfg(feature = "parallel-handler")]
1169 pub async fn run_parallel(mut self) -> Result<()> {
1170 let conn = self
1174 .conn
1175 .take()
1176 .ok_or_else(|| CdpError::msg("Handler::run_parallel() called with no connection"))?;
1177 let async_conn = conn.into_async();
1178
1179 let next_id = async_conn.next_id;
1182 let boot_call_id = chromiumoxide_types::CallId::new(next_id.saturating_sub(1));
1183 let boot_method = DISCOVER_ID.0.clone();
1184
1185 let router = parallel::Router::new(
1186 self.config,
1187 self.default_browser_context,
1188 self.from_browser,
1189 async_conn.reader,
1190 async_conn.cmd_tx,
1191 boot_call_id,
1192 boot_method,
1193 next_id,
1194 );
1195 let result = router.run().await;
1196
1197 async_conn.writer_handle.abort();
1199 async_conn.reader_handle.abort();
1200
1201 result
1202 }
1203}
1204
1205impl Stream for Handler {
1206 type Item = Result<()>;
1207
1208 fn poll_next(self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Option<Self::Item>> {
1209 const BROWSER_MSG_BUDGET: usize = 128;
1215 const PER_TARGET_DRAIN_BUDGET: usize = 128;
1216 const WS_MSG_BUDGET: usize = 512;
1217
1218 let pin = self.get_mut();
1219
1220 let mut dispose = false;
1221 let mut budget_hit = false;
1222
1223 let now = Instant::now();
1224
1225 loop {
1226 let mut browser_msgs = 0usize;
1230 while let Poll::Ready(Some(msg)) = pin.from_browser.poll_recv(cx) {
1231 match msg {
1232 HandlerMessage::Command(cmd) => {
1233 pin.submit_external_command(cmd, now)?;
1234 }
1235 HandlerMessage::FetchTargets(tx) => {
1236 pin.submit_fetch_targets(tx, now);
1237 }
1238 HandlerMessage::CloseBrowser(tx) => {
1239 pin.submit_close(tx, now);
1240 }
1241 HandlerMessage::CreatePage(params, tx) => {
1242 if let Some(ref id) = params.browser_context_id {
1243 pin.browser_contexts
1244 .insert(BrowserContext::from(id.clone()));
1245 }
1246 pin.create_page(params, tx);
1247 }
1248 HandlerMessage::GetPages(tx) => {
1249 let pages: Vec<_> = pin
1250 .targets
1251 .values_mut()
1252 .filter(|p: &&mut Target| p.is_page())
1253 .filter_map(|target| target.get_or_create_page())
1254 .map(|page| Page::from(page.clone()))
1255 .collect();
1256 let _ = tx.send(pages);
1257 }
1258 HandlerMessage::InsertContext(ctx) => {
1259 if pin.default_browser_context.id().is_none() {
1260 pin.default_browser_context = ctx.clone();
1261 }
1262 pin.browser_contexts.insert(ctx);
1263 }
1264 HandlerMessage::DisposeContext(ctx) => {
1265 pin.browser_contexts.remove(&ctx);
1266 pin.attached_targets.retain(|tid| {
1267 pin.targets
1268 .get(tid)
1269 .and_then(|t| t.browser_context_id()) .map(|id| Some(id) != ctx.id())
1271 .unwrap_or(true)
1272 });
1273 pin.closing = true;
1274 dispose = true;
1275 }
1276 HandlerMessage::GetPage(target_id, tx) => {
1277 let page = pin
1278 .targets
1279 .get_mut(&target_id)
1280 .and_then(|target| target.get_or_create_page())
1281 .map(|page| Page::from(page.clone()));
1282 let _ = tx.send(page);
1283 }
1284 HandlerMessage::AddEventListener(req) => {
1285 pin.event_listeners.add_listener(req);
1286 }
1287 }
1288 browser_msgs += 1;
1289 if browser_msgs >= BROWSER_MSG_BUDGET {
1290 budget_hit = true;
1291 break;
1292 }
1293 }
1294
1295 for n in (0..pin.target_ids.len()).rev() {
1296 let target_id = pin.target_ids.swap_remove(n);
1297
1298 if let Some((id, mut target)) = pin.targets.remove_entry(&target_id) {
1299 let mut drained = 0usize;
1300 while let Some(event) = target.poll(cx, now) {
1301 match event {
1302 TargetEvent::Request(req) => {
1303 let _ = pin.submit_internal_command(
1304 target.target_id().clone(),
1305 req,
1306 now,
1307 );
1308 }
1309 TargetEvent::Command(msg) => {
1310 pin.on_target_message(&mut target, msg, now);
1311 }
1312 TargetEvent::NavigationRequest(id, req) => {
1313 pin.submit_navigation(id, req, now);
1314 }
1315 TargetEvent::NavigationResult(res) => {
1316 pin.on_navigation_lifecycle_completed(res)
1317 }
1318 TargetEvent::BytesConsumed(n) => {
1319 if let Some(rem) = pin.remaining_bytes.as_mut() {
1320 *rem = rem.saturating_sub(n);
1321 if *rem == 0 {
1322 pin.budget_exhausted = true;
1323 }
1324 }
1325 }
1326 }
1327 drained += 1;
1328 if drained >= PER_TARGET_DRAIN_BUDGET {
1329 budget_hit = true;
1330 break;
1331 }
1332 }
1333
1334 target.event_listeners_mut().poll(cx);
1336
1337 pin.targets.insert(id, target);
1338 pin.target_ids.push(target_id);
1339 }
1340 }
1341
1342 pin.event_listeners_mut().poll(cx);
1345
1346 let mut done = true;
1347
1348 let mut ws_msgs = Vec::new();
1351 let mut ws_err = None;
1352 {
1353 let Some(conn) = pin.conn.as_mut() else {
1354 return Poll::Ready(Some(Err(CdpError::msg(
1355 "connection consumed by Handler::run()",
1356 ))));
1357 };
1358 while let Poll::Ready(Some(ev)) = Pin::new(&mut *conn).poll_next(cx) {
1359 match ev {
1360 Ok(msg) => ws_msgs.push(msg),
1361 Err(err) => {
1362 ws_err = Some(err);
1363 break;
1364 }
1365 }
1366 if ws_msgs.len() >= WS_MSG_BUDGET {
1367 budget_hit = true;
1368 break;
1369 }
1370 }
1371 }
1372
1373 for boxed_msg in ws_msgs {
1374 match *boxed_msg {
1375 Message::Response(resp) => {
1376 pin.on_response(resp);
1377 if pin.closing {
1378 return Poll::Ready(None);
1379 }
1380 }
1381 Message::Event(ev) => {
1382 pin.on_event(ev);
1383 }
1384 }
1385 done = false;
1386 }
1387
1388 if let Some(err) = ws_err {
1389 tracing::error!("WS Connection error: {:?}", err);
1390 if let CdpError::Ws(ref ws_error) = err {
1391 match ws_error {
1392 Error::AlreadyClosed => {
1393 pin.closing = true;
1394 dispose = true;
1395 }
1396 Error::Protocol(detail)
1397 if detail == &ProtocolError::ResetWithoutClosingHandshake =>
1398 {
1399 pin.closing = true;
1400 dispose = true;
1401 }
1402 _ => return Poll::Ready(Some(Err(err))),
1403 }
1404 } else {
1405 return Poll::Ready(Some(Err(err)));
1406 }
1407 }
1408
1409 if pin.evict_command_timeout.poll_ready(cx) {
1410 pin.evict_timed_out_commands(now);
1412 for t in pin.targets.values_mut() {
1415 t.network_manager.evict_stale_entries(now);
1416 t.frame_manager_mut().evict_stale_context_ids();
1417 }
1418 }
1419
1420 if pin.budget_exhausted {
1421 for t in pin.targets.values_mut() {
1422 t.network_manager.set_block_all(true);
1423 }
1424 }
1425
1426 if dispose {
1427 return Poll::Ready(None);
1428 }
1429
1430 if budget_hit {
1431 cx.waker().wake_by_ref();
1435 return Poll::Pending;
1436 }
1437
1438 if done {
1439 return Poll::Pending;
1441 }
1442 }
1443 }
1444}
1445
1446#[derive(Debug, Clone)]
1448pub struct HandlerConfig {
1449 pub ignore_https_errors: bool,
1451 pub viewport: Option<Viewport>,
1453 pub context_ids: Vec<BrowserContextId>,
1455 pub request_timeout: Duration,
1457 pub request_intercept: bool,
1459 pub cache_enabled: bool,
1461 pub service_worker_enabled: bool,
1463 pub ignore_visuals: bool,
1465 pub ignore_stylesheets: bool,
1467 pub ignore_javascript: bool,
1469 pub ignore_analytics: bool,
1471 pub ignore_prefetch: bool,
1473 pub ignore_ads: bool,
1475 pub extra_headers: Option<std::collections::HashMap<String, String>>,
1477 pub only_html: bool,
1479 pub created_first_target: bool,
1481 pub intercept_manager: NetworkInterceptManager,
1483 pub max_bytes_allowed: Option<u64>,
1485 pub max_redirects: Option<usize>,
1492 pub max_main_frame_navigations: Option<u32>,
1496 pub whitelist_patterns: Option<Vec<String>>,
1498 pub blacklist_patterns: Option<Vec<String>>,
1500 #[cfg(feature = "adblock")]
1502 pub adblock_filter_rules: Option<Vec<String>>,
1503 pub channel_capacity: usize,
1506 pub page_channel_capacity: usize,
1517 pub connection_retries: u32,
1520}
1521
1522impl Default for HandlerConfig {
1523 fn default() -> Self {
1524 Self {
1525 ignore_https_errors: true,
1526 viewport: Default::default(),
1527 context_ids: Vec::new(),
1528 request_timeout: Duration::from_millis(REQUEST_TIMEOUT),
1529 request_intercept: false,
1530 cache_enabled: true,
1531 service_worker_enabled: true,
1532 ignore_visuals: false,
1533 ignore_stylesheets: false,
1534 ignore_ads: false,
1535 ignore_javascript: false,
1536 ignore_analytics: true,
1537 ignore_prefetch: true,
1538 only_html: false,
1539 extra_headers: Default::default(),
1540 created_first_target: false,
1541 intercept_manager: NetworkInterceptManager::Unknown,
1542 max_bytes_allowed: None,
1543 max_redirects: None,
1544 max_main_frame_navigations: None,
1545 whitelist_patterns: None,
1546 blacklist_patterns: None,
1547 #[cfg(feature = "adblock")]
1548 adblock_filter_rules: None,
1549 channel_capacity: 4096,
1550 page_channel_capacity: crate::handler::page::DEFAULT_PAGE_CHANNEL_CAPACITY,
1551 connection_retries: crate::conn::DEFAULT_CONNECTION_RETRIES,
1552 }
1553 }
1554}
1555
1556#[derive(Debug)]
1558pub struct NavigationInProgress<T> {
1559 navigated: bool,
1561 response: Option<Response>,
1563 tx: OneshotSender<T>,
1565}
1566
1567impl<T> NavigationInProgress<T> {
1568 pub(crate) fn new(tx: OneshotSender<T>) -> Self {
1569 Self {
1570 navigated: false,
1571 response: None,
1572 tx,
1573 }
1574 }
1575
1576 pub(crate) fn set_response(&mut self, resp: Response) {
1578 self.response = Some(resp);
1579 }
1580
1581 pub(crate) fn set_navigated(&mut self) {
1583 self.navigated = true;
1584 }
1585
1586 #[cfg_attr(not(feature = "parallel-handler"), allow(dead_code))]
1590 pub(crate) fn is_navigated(&self) -> bool {
1591 self.navigated
1592 }
1593
1594 #[cfg_attr(not(feature = "parallel-handler"), allow(dead_code))]
1595 pub(crate) fn take_response(&mut self) -> Option<Response> {
1596 self.response.take()
1597 }
1598
1599 #[cfg_attr(not(feature = "parallel-handler"), allow(dead_code))]
1600 pub(crate) fn into_tx(self) -> OneshotSender<T> {
1601 self.tx
1602 }
1603}
1604
1605#[derive(Debug)]
1607enum NavigationRequest {
1608 Navigate(NavigationInProgress<Result<Response>>),
1610 }
1612
1613#[derive(Debug)]
1616enum PendingRequest {
1617 CreateTarget(OneshotSender<Result<Page>>),
1620 GetTargets(OneshotSender<Result<Vec<TargetInfo>>>),
1622 Navigate(NavigationId),
1629 ExternalCommand {
1638 tx: OneshotSender<Result<Response>>,
1639 target_id: Option<TargetId>,
1640 },
1641 InternalCommand(TargetId),
1644 CloseBrowser(OneshotSender<Result<CloseReturns>>),
1646}
1647
1648#[derive(Debug)]
1652pub(crate) enum HandlerMessage {
1653 CreatePage(CreateTargetParams, OneshotSender<Result<Page>>),
1654 FetchTargets(OneshotSender<Result<Vec<TargetInfo>>>),
1655 InsertContext(BrowserContext),
1656 DisposeContext(BrowserContext),
1657 GetPages(OneshotSender<Vec<Page>>),
1658 Command(CommandMessage),
1659 GetPage(TargetId, OneshotSender<Option<Page>>),
1660 AddEventListener(EventListenerRequest),
1661 CloseBrowser(OneshotSender<Result<CloseReturns>>),
1662}
1663
1664#[cfg(test)]
1665mod tests {
1666 use super::*;
1667 use chromiumoxide_cdp::cdp::browser_protocol::target::{AttachToTargetReturns, TargetInfo};
1668
1669 #[test]
1670 fn attach_to_target_response_sets_session_id_before_event_arrives() {
1671 let info = TargetInfo::builder()
1672 .target_id("target-1".to_string())
1673 .r#type("page")
1674 .title("")
1675 .url("about:blank")
1676 .attached(false)
1677 .can_access_opener(false)
1678 .build()
1679 .expect("target info");
1680 let mut target = Target::new(info, TargetConfig::default(), BrowserContext::default());
1681 let method: MethodId = AttachToTargetParams::IDENTIFIER.into();
1682 let result = serde_json::to_value(AttachToTargetReturns::new("session-1".to_string()))
1683 .expect("attach result");
1684 let resp = Response {
1685 id: CallId::new(1),
1686 result: Some(result),
1687 error: None,
1688 };
1689
1690 maybe_store_attach_session_id(&mut target, &method, &resp);
1691
1692 assert_eq!(
1693 target.session_id().map(AsRef::as_ref),
1694 Some("session-1"),
1695 "attach response should seed the flat session id even before Target.attachedToTarget"
1696 );
1697 }
1698
1699 #[test]
1704 fn page_channel_capacity_defaults_to_2048_across_configs() {
1705 use crate::browser::BrowserConfigBuilder;
1706 use crate::handler::page::DEFAULT_PAGE_CHANNEL_CAPACITY;
1707 use crate::handler::target::TargetConfig;
1708
1709 assert_eq!(DEFAULT_PAGE_CHANNEL_CAPACITY, 2048);
1710 assert_eq!(
1711 HandlerConfig::default().page_channel_capacity,
1712 DEFAULT_PAGE_CHANNEL_CAPACITY,
1713 "HandlerConfig default must match the historical 2048 slot count"
1714 );
1715 assert_eq!(
1716 TargetConfig::default().page_channel_capacity,
1717 DEFAULT_PAGE_CHANNEL_CAPACITY,
1718 "TargetConfig default must match the historical 2048 slot count"
1719 );
1720 let builder = BrowserConfigBuilder::default();
1723 let bc = format!("{:?}", builder);
1724 assert!(
1725 bc.contains("page_channel_capacity: 2048"),
1726 "BrowserConfigBuilder must default page_channel_capacity to 2048, got: {bc}",
1727 );
1728 }
1729}