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 allow_first_party_stylesheets: self.config.allow_first_party_stylesheets,
532 allow_first_party_javascript: self.config.allow_first_party_javascript,
533 allow_first_party_visuals: self.config.allow_first_party_visuals,
534 extra_headers: self.config.extra_headers.clone(),
535 only_html: self.config.only_html && self.config.created_first_target,
536 intercept_manager: self.config.intercept_manager,
537 max_bytes_allowed: self.config.max_bytes_allowed,
538 max_redirects: self.config.max_redirects,
539 max_main_frame_navigations: self.config.max_main_frame_navigations,
540 whitelist_patterns: self.config.whitelist_patterns.clone(),
541 blacklist_patterns: self.config.blacklist_patterns.clone(),
542 #[cfg(feature = "adblock")]
543 adblock_filter_rules: self.config.adblock_filter_rules.clone(),
544 page_wake: self.page_wake.clone(),
545 page_channel_capacity: self.config.page_channel_capacity,
546 },
547 browser_ctx,
548 );
549
550 let tid = target.target_id().clone();
551 self.target_ids.push(tid.clone());
552 self.targets.insert(tid, target);
553 }
554
555 fn on_attached_to_target(&mut self, event: Box<EventAttachedToTarget>) {
557 let session = Session::new(event.session_id.clone(), event.target_info.target_id);
558 if let Some(target) = self.targets.get_mut(session.target_id()) {
559 target.set_session_id(session.session_id().clone())
560 }
561 self.sessions.insert(event.session_id, session);
562 }
563
564 fn on_detached_from_target(&mut self, event: EventDetachedFromTarget) {
568 if let Some(session) = self.sessions.remove(&event.session_id) {
570 if let Some(target) = self.targets.get_mut(session.target_id()) {
571 target.session_id_mut().take();
572 }
573 }
574 }
575
576 fn on_target_destroyed(&mut self, event: EventTargetDestroyed) {
578 self.attached_targets.remove(&event.target_id);
579
580 if let Some(target) = self.targets.remove(&event.target_id) {
581 if let Some(session) = target.session_id() {
583 self.sessions.remove(session);
584 }
585 }
586 }
587
588 fn on_target_crashed(&mut self, event: EventTargetCrashed) {
611 let crashed_id = event.target_id.clone();
612 let status = event.status.clone();
613 let error_code = event.error_code;
614
615 let to_cancel: Vec<CallId> = self
621 .pending_commands
622 .iter()
623 .filter_map(|(&call_id, (req, _, _))| match req {
624 PendingRequest::ExternalCommand {
625 target_id: Some(tid),
626 ..
627 } if *tid == crashed_id => Some(call_id),
628 PendingRequest::InternalCommand(tid) if *tid == crashed_id => Some(call_id),
629 _ => None,
630 })
631 .collect();
632
633 for call_id in to_cancel {
634 if let Some((req, _, _)) = self.pending_commands.remove(&call_id) {
635 match req {
636 PendingRequest::ExternalCommand { tx, .. } => {
637 let _ = tx.send(Err(CdpError::msg(format!(
638 "target {:?} crashed: {} (errorCode={})",
639 crashed_id, status, error_code
640 ))));
641 }
642 PendingRequest::InternalCommand(_) => {
643 }
646 _ => {}
647 }
648 }
649 }
650
651 self.attached_targets.remove(&crashed_id);
653 if let Some(target) = self.targets.remove(&crashed_id) {
654 if let Some(session) = target.session_id() {
655 self.sessions.remove(session);
656 }
657 }
658 }
659
660 fn evict_timed_out_commands(&mut self, now: Instant) {
665 let deadline = match now.checked_sub(self.config.request_timeout) {
666 Some(d) => d,
667 None => return,
668 };
669
670 let timed_out: Vec<_> = self
671 .pending_commands
672 .iter()
673 .filter(|(_, (_, _, timestamp))| *timestamp < deadline)
674 .map(|(k, _)| *k)
675 .collect();
676
677 for call in timed_out {
678 if let Some((req, _, _)) = self.pending_commands.remove(&call) {
679 match req {
680 PendingRequest::CreateTarget(tx) => {
681 let _ = tx.send(Err(CdpError::Timeout));
682 }
683 PendingRequest::GetTargets(tx) => {
684 let _ = tx.send(Err(CdpError::Timeout));
685 }
686 PendingRequest::Navigate(nav) => {
687 if let Some(nav) = self.navigations.remove(&nav) {
688 match nav {
689 NavigationRequest::Navigate(nav) => {
690 let _ = nav.tx.send(Err(CdpError::Timeout));
691 }
692 }
693 }
694 }
695 PendingRequest::ExternalCommand { tx, .. } => {
696 let _ = tx.send(Err(CdpError::Timeout));
697 }
698 PendingRequest::InternalCommand(_) => {}
699 PendingRequest::CloseBrowser(tx) => {
700 let _ = tx.send(Err(CdpError::Timeout));
701 }
702 }
703 }
704 }
705 }
706
707 pub fn event_listeners_mut(&mut self) -> &mut EventListeners {
708 &mut self.event_listeners
709 }
710
711 pub async fn run(mut self) -> Result<()> {
741 use chromiumoxide_types::Message;
742 use tokio::time::MissedTickBehavior;
743 use tokio_tungstenite::tungstenite::{self, error::ProtocolError};
744
745 let page_wake = Arc::new(Notify::new());
747 self.page_wake = Some(page_wake.clone());
748
749 let conn = self
751 .conn
752 .take()
753 .ok_or_else(|| CdpError::msg("Handler::run() called with no connection"))?;
754 let async_conn = conn.into_async();
755 let mut ws_reader = async_conn.reader;
756 let ws_tx = async_conn.cmd_tx;
757 let mut writer_handle = async_conn.writer_handle;
758 let reader_handle = async_conn.reader_handle;
759 let mut next_call_id = async_conn.next_id;
760
761 let mut alloc_call_id = || {
763 let id = chromiumoxide_types::CallId::new(next_call_id);
764 next_call_id = next_call_id.wrapping_add(1);
765 id
766 };
767
768 let mut evict_timer = tokio::time::interval_at(
770 tokio::time::Instant::now() + self.config.request_timeout,
771 self.config.request_timeout,
772 );
773 evict_timer.set_missed_tick_behavior(MissedTickBehavior::Delay);
774
775 macro_rules! ws_submit {
777 ($method:expr, $session_id:expr, $params:expr) => {{
778 let id = alloc_call_id();
779 let call = chromiumoxide_types::MethodCall {
780 id,
781 method: $method,
782 session_id: $session_id,
783 params: $params,
784 };
785 match ws_tx.try_send(call) {
786 Ok(()) => Ok::<_, CdpError>(id),
787 Err(tokio::sync::mpsc::error::TrySendError::Full(_)) => {
788 tracing::warn!("WS command channel full — dropping command");
789 Err(CdpError::msg("WS command channel full"))
790 }
791 Err(tokio::sync::mpsc::error::TrySendError::Closed(_)) => {
792 Err(CdpError::msg("WS writer closed"))
793 }
794 }
795 }};
796 }
797
798 let run_result: Result<()> = loop {
806 let now = std::time::Instant::now();
807
808 const PER_TARGET_DRAIN_BUDGET: usize = 128;
814
815 for n in (0..self.target_ids.len()).rev() {
816 let target_id = self.target_ids.swap_remove(n);
817
818 if let Some((id, mut target)) = self.targets.remove_entry(&target_id) {
819 {
821 let mut msgs = Vec::new();
822 if let Some(handle) = target.page_mut() {
823 while msgs.len() < PER_TARGET_DRAIN_BUDGET {
824 match handle.rx.try_recv() {
825 Ok(msg) => msgs.push(msg),
826 Err(_) => break,
827 }
828 }
829 }
830 for msg in msgs {
831 target.on_page_message(msg);
832 }
833 }
834
835 while let Some(event) = target.advance(now) {
837 match event {
838 TargetEvent::Request(req) => {
839 if let Ok(call_id) =
840 ws_submit!(req.method.clone(), req.session_id, req.params)
841 {
842 self.pending_commands.insert(
843 call_id,
844 (
845 PendingRequest::InternalCommand(
846 target.target_id().clone(),
847 ),
848 req.method,
849 now,
850 ),
851 );
852 }
853 }
854 TargetEvent::Command(msg) => {
855 if msg.is_navigation() {
856 let (req, tx) = msg.split();
857 let nav_id = self.next_navigation_id();
858 target.goto(FrameRequestedNavigation::new(
859 nav_id,
860 req.clone(),
861 self.config.request_timeout,
862 ));
863 if let Ok(call_id) =
864 ws_submit!(req.method.clone(), req.session_id, req.params)
865 {
866 self.pending_commands.insert(
867 call_id,
868 (PendingRequest::Navigate(nav_id), req.method, now),
869 );
870 }
871 self.navigations.insert(
872 nav_id,
873 NavigationRequest::Navigate(NavigationInProgress::new(tx)),
874 );
875 } else if let Ok(call_id) = ws_submit!(
876 msg.method.clone(),
877 msg.session_id.map(Into::into),
878 msg.params
879 ) {
880 let target_id = Some(target.target_id().clone());
884 self.pending_commands.insert(
885 call_id,
886 (
887 PendingRequest::ExternalCommand {
888 tx: msg.sender,
889 target_id,
890 },
891 msg.method,
892 now,
893 ),
894 );
895 }
896 }
897 TargetEvent::NavigationRequest(nav_id, req) => {
898 if let Ok(call_id) =
899 ws_submit!(req.method.clone(), req.session_id, req.params)
900 {
901 self.pending_commands.insert(
902 call_id,
903 (PendingRequest::Navigate(nav_id), req.method, now),
904 );
905 }
906 }
907 TargetEvent::NavigationResult(res) => {
908 self.on_navigation_lifecycle_completed(res);
909 }
910 TargetEvent::BytesConsumed(n) => {
911 if let Some(rem) = self.remaining_bytes.as_mut() {
912 *rem = rem.saturating_sub(n);
913 if *rem == 0 {
914 self.budget_exhausted = true;
915 }
916 }
917 }
918 }
919 }
920
921 target.event_listeners_mut().flush();
923
924 self.targets.insert(id, target);
925 self.target_ids.push(target_id);
926 }
927 }
928
929 self.event_listeners.flush();
931
932 if self.budget_exhausted {
933 for t in self.targets.values_mut() {
934 t.network_manager.set_block_all(true);
935 }
936 }
937
938 if self.closing {
939 break Ok(());
940 }
941
942 tokio::select! {
944 msg = self.from_browser.recv() => {
945 match msg {
946 Some(msg) => {
947 match msg {
948 HandlerMessage::Command(cmd) => {
949 let target_id = cmd
952 .session_id
953 .as_ref()
954 .and_then(|sid| self.sessions.get(sid.as_ref()))
955 .map(|s| s.target_id().clone());
956 if let Ok(call_id) = ws_submit!(
957 cmd.method.clone(),
958 cmd.session_id.map(Into::into),
959 cmd.params
960 ) {
961 self.pending_commands.insert(
962 call_id,
963 (
964 PendingRequest::ExternalCommand {
965 tx: cmd.sender,
966 target_id,
967 },
968 cmd.method,
969 now,
970 ),
971 );
972 }
973 }
974 HandlerMessage::FetchTargets(tx) => {
975 let msg = TARGET_PARAMS_ID.clone();
976 if let Ok(call_id) = ws_submit!(msg.0.clone(), None, msg.1) {
977 self.pending_commands.insert(
978 call_id,
979 (PendingRequest::GetTargets(tx), msg.0, now),
980 );
981 }
982 }
983 HandlerMessage::CloseBrowser(tx) => {
984 let close_msg = CLOSE_PARAMS_ID.clone();
985 if let Ok(call_id) = ws_submit!(close_msg.0.clone(), None, close_msg.1) {
986 self.pending_commands.insert(
987 call_id,
988 (PendingRequest::CloseBrowser(tx), close_msg.0, now),
989 );
990 }
991 }
992 HandlerMessage::CreatePage(params, tx) => {
993 if let Some(ref id) = params.browser_context_id {
994 self.browser_contexts.insert(BrowserContext::from(id.clone()));
995 }
996 self.create_page_async(params, tx, &mut alloc_call_id, &ws_tx, now);
997 }
998 HandlerMessage::GetPages(tx) => {
999 let pages: Vec<_> = self.targets.values_mut()
1000 .filter(|p| p.is_page())
1001 .filter_map(|target| target.get_or_create_page())
1002 .map(|page| Page::from(page.clone()))
1003 .collect();
1004 let _ = tx.send(pages);
1005 }
1006 HandlerMessage::InsertContext(ctx) => {
1007 if self.default_browser_context.id().is_none() {
1008 self.default_browser_context = ctx.clone();
1009 }
1010 self.browser_contexts.insert(ctx);
1011 }
1012 HandlerMessage::DisposeContext(ctx) => {
1013 self.browser_contexts.remove(&ctx);
1014 self.attached_targets.retain(|tid| {
1015 self.targets.get(tid)
1016 .and_then(|t| t.browser_context_id())
1017 .map(|id| Some(id) != ctx.id())
1018 .unwrap_or(true)
1019 });
1020 self.closing = true;
1021 }
1022 HandlerMessage::GetPage(target_id, tx) => {
1023 let page = self.targets.get_mut(&target_id)
1024 .and_then(|target| target.get_or_create_page())
1025 .map(|page| Page::from(page.clone()));
1026 let _ = tx.send(page);
1027 }
1028 HandlerMessage::AddEventListener(req) => {
1029 self.event_listeners.add_listener(req);
1030 }
1031 }
1032 }
1033 None => break Ok(()), }
1035 }
1036
1037 frame = ws_reader.next_message() => {
1038 match frame {
1039 Some(Ok(boxed_msg)) => match *boxed_msg {
1040 Message::Response(resp) => {
1041 self.on_response(resp);
1042 }
1043 Message::Event(ev) => {
1044 self.on_event(ev);
1045 }
1046 },
1047 Some(Err(err)) => {
1048 tracing::error!("WS Connection error: {:?}", err);
1049 if let CdpError::Ws(ref ws_error) = err {
1050 match ws_error {
1051 tungstenite::Error::AlreadyClosed => break Ok(()),
1052 tungstenite::Error::Protocol(detail)
1053 if detail == &ProtocolError::ResetWithoutClosingHandshake =>
1054 {
1055 break Ok(());
1056 }
1057 _ => break Err(err),
1058 }
1059 } else {
1060 break Err(err);
1061 }
1062 }
1063 None => break Ok(()), }
1065 }
1066
1067 _ = page_wake.notified() => {
1068 }
1070
1071 _ = evict_timer.tick() => {
1072 self.evict_timed_out_commands(now);
1073 for t in self.targets.values_mut() {
1074 t.network_manager.evict_stale_entries(now);
1075 t.frame_manager_mut().evict_stale_context_ids();
1076 }
1077 }
1078
1079 result = &mut writer_handle => {
1080 match result {
1082 Ok(Ok(())) => break Ok(()),
1083 Ok(Err(e)) => break Err(e),
1084 Err(e) => break Err(CdpError::msg(format!("WS writer panicked: {e}"))),
1085 }
1086 }
1087 }
1088 };
1089
1090 drop(ws_tx);
1098
1099 if !writer_handle.is_finished() {
1104 let _ = tokio::time::timeout(std::time::Duration::from_millis(500), &mut writer_handle)
1105 .await;
1106 if !writer_handle.is_finished() {
1107 writer_handle.abort();
1108 }
1109 }
1110
1111 reader_handle.abort();
1117
1118 run_result
1119 }
1120
1121 fn create_page_async(
1123 &mut self,
1124 params: CreateTargetParams,
1125 tx: OneshotSender<Result<Page>>,
1126 alloc_call_id: &mut impl FnMut() -> chromiumoxide_types::CallId,
1127 ws_tx: &tokio::sync::mpsc::Sender<chromiumoxide_types::MethodCall>,
1128 now: std::time::Instant,
1129 ) {
1130 let about_blank = params.url == "about:blank";
1131 let http_check =
1132 !about_blank && params.url.starts_with("http") || params.url.starts_with("file://");
1133
1134 if about_blank || http_check {
1135 let method = params.identifier();
1136 match serde_json::to_value(params) {
1137 Ok(params) => {
1138 let id = alloc_call_id();
1139 let call = chromiumoxide_types::MethodCall {
1140 id,
1141 method: method.clone(),
1142 session_id: None,
1143 params,
1144 };
1145 match ws_tx.try_send(call) {
1146 Ok(()) => {
1147 self.pending_commands
1148 .insert(id, (PendingRequest::CreateTarget(tx), method, now));
1149 }
1150 Err(_) => {
1151 let _ = tx
1152 .send(Err(CdpError::msg("WS command channel full or closed")))
1153 .ok();
1154 }
1155 }
1156 }
1157 Err(err) => {
1158 let _ = tx.send(Err(err.into())).ok();
1159 }
1160 }
1161 } else {
1162 let _ = tx.send(Err(CdpError::NotFound)).ok();
1163 }
1164 }
1165
1166 #[cfg(feature = "parallel-handler")]
1172 pub async fn run_parallel(mut self) -> Result<()> {
1173 let conn = self
1177 .conn
1178 .take()
1179 .ok_or_else(|| CdpError::msg("Handler::run_parallel() called with no connection"))?;
1180 let async_conn = conn.into_async();
1181
1182 let next_id = async_conn.next_id;
1185 let boot_call_id = chromiumoxide_types::CallId::new(next_id.saturating_sub(1));
1186 let boot_method = DISCOVER_ID.0.clone();
1187
1188 let router = parallel::Router::new(
1189 self.config,
1190 self.default_browser_context,
1191 self.from_browser,
1192 async_conn.reader,
1193 async_conn.cmd_tx,
1194 boot_call_id,
1195 boot_method,
1196 next_id,
1197 );
1198 let result = router.run().await;
1199
1200 async_conn.writer_handle.abort();
1202 async_conn.reader_handle.abort();
1203
1204 result
1205 }
1206}
1207
1208impl Stream for Handler {
1209 type Item = Result<()>;
1210
1211 fn poll_next(self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Option<Self::Item>> {
1212 const BROWSER_MSG_BUDGET: usize = 128;
1218 const PER_TARGET_DRAIN_BUDGET: usize = 128;
1219 const WS_MSG_BUDGET: usize = 512;
1220
1221 let pin = self.get_mut();
1222
1223 let mut dispose = false;
1224 let mut budget_hit = false;
1225
1226 let now = Instant::now();
1227
1228 loop {
1229 let mut browser_msgs = 0usize;
1233 while let Poll::Ready(Some(msg)) = pin.from_browser.poll_recv(cx) {
1234 match msg {
1235 HandlerMessage::Command(cmd) => {
1236 pin.submit_external_command(cmd, now)?;
1237 }
1238 HandlerMessage::FetchTargets(tx) => {
1239 pin.submit_fetch_targets(tx, now);
1240 }
1241 HandlerMessage::CloseBrowser(tx) => {
1242 pin.submit_close(tx, now);
1243 }
1244 HandlerMessage::CreatePage(params, tx) => {
1245 if let Some(ref id) = params.browser_context_id {
1246 pin.browser_contexts
1247 .insert(BrowserContext::from(id.clone()));
1248 }
1249 pin.create_page(params, tx);
1250 }
1251 HandlerMessage::GetPages(tx) => {
1252 let pages: Vec<_> = pin
1253 .targets
1254 .values_mut()
1255 .filter(|p: &&mut Target| p.is_page())
1256 .filter_map(|target| target.get_or_create_page())
1257 .map(|page| Page::from(page.clone()))
1258 .collect();
1259 let _ = tx.send(pages);
1260 }
1261 HandlerMessage::InsertContext(ctx) => {
1262 if pin.default_browser_context.id().is_none() {
1263 pin.default_browser_context = ctx.clone();
1264 }
1265 pin.browser_contexts.insert(ctx);
1266 }
1267 HandlerMessage::DisposeContext(ctx) => {
1268 pin.browser_contexts.remove(&ctx);
1269 pin.attached_targets.retain(|tid| {
1270 pin.targets
1271 .get(tid)
1272 .and_then(|t| t.browser_context_id()) .map(|id| Some(id) != ctx.id())
1274 .unwrap_or(true)
1275 });
1276 pin.closing = true;
1277 dispose = true;
1278 }
1279 HandlerMessage::GetPage(target_id, tx) => {
1280 let page = pin
1281 .targets
1282 .get_mut(&target_id)
1283 .and_then(|target| target.get_or_create_page())
1284 .map(|page| Page::from(page.clone()));
1285 let _ = tx.send(page);
1286 }
1287 HandlerMessage::AddEventListener(req) => {
1288 pin.event_listeners.add_listener(req);
1289 }
1290 }
1291 browser_msgs += 1;
1292 if browser_msgs >= BROWSER_MSG_BUDGET {
1293 budget_hit = true;
1294 break;
1295 }
1296 }
1297
1298 for n in (0..pin.target_ids.len()).rev() {
1299 let target_id = pin.target_ids.swap_remove(n);
1300
1301 if let Some((id, mut target)) = pin.targets.remove_entry(&target_id) {
1302 let mut drained = 0usize;
1303 while let Some(event) = target.poll(cx, now) {
1304 match event {
1305 TargetEvent::Request(req) => {
1306 let _ = pin.submit_internal_command(
1307 target.target_id().clone(),
1308 req,
1309 now,
1310 );
1311 }
1312 TargetEvent::Command(msg) => {
1313 pin.on_target_message(&mut target, msg, now);
1314 }
1315 TargetEvent::NavigationRequest(id, req) => {
1316 pin.submit_navigation(id, req, now);
1317 }
1318 TargetEvent::NavigationResult(res) => {
1319 pin.on_navigation_lifecycle_completed(res)
1320 }
1321 TargetEvent::BytesConsumed(n) => {
1322 if let Some(rem) = pin.remaining_bytes.as_mut() {
1323 *rem = rem.saturating_sub(n);
1324 if *rem == 0 {
1325 pin.budget_exhausted = true;
1326 }
1327 }
1328 }
1329 }
1330 drained += 1;
1331 if drained >= PER_TARGET_DRAIN_BUDGET {
1332 budget_hit = true;
1333 break;
1334 }
1335 }
1336
1337 target.event_listeners_mut().poll(cx);
1339
1340 pin.targets.insert(id, target);
1341 pin.target_ids.push(target_id);
1342 }
1343 }
1344
1345 pin.event_listeners_mut().poll(cx);
1348
1349 let mut done = true;
1350
1351 let mut ws_msgs = Vec::new();
1354 let mut ws_err = None;
1355 {
1356 let Some(conn) = pin.conn.as_mut() else {
1357 return Poll::Ready(Some(Err(CdpError::msg(
1358 "connection consumed by Handler::run()",
1359 ))));
1360 };
1361 while let Poll::Ready(Some(ev)) = Pin::new(&mut *conn).poll_next(cx) {
1362 match ev {
1363 Ok(msg) => ws_msgs.push(msg),
1364 Err(err) => {
1365 ws_err = Some(err);
1366 break;
1367 }
1368 }
1369 if ws_msgs.len() >= WS_MSG_BUDGET {
1370 budget_hit = true;
1371 break;
1372 }
1373 }
1374 }
1375
1376 for boxed_msg in ws_msgs {
1377 match *boxed_msg {
1378 Message::Response(resp) => {
1379 pin.on_response(resp);
1380 if pin.closing {
1381 return Poll::Ready(None);
1382 }
1383 }
1384 Message::Event(ev) => {
1385 pin.on_event(ev);
1386 }
1387 }
1388 done = false;
1389 }
1390
1391 if let Some(err) = ws_err {
1392 tracing::error!("WS Connection error: {:?}", err);
1393 if let CdpError::Ws(ref ws_error) = err {
1394 match ws_error {
1395 Error::AlreadyClosed => {
1396 pin.closing = true;
1397 dispose = true;
1398 }
1399 Error::Protocol(detail)
1400 if detail == &ProtocolError::ResetWithoutClosingHandshake =>
1401 {
1402 pin.closing = true;
1403 dispose = true;
1404 }
1405 _ => return Poll::Ready(Some(Err(err))),
1406 }
1407 } else {
1408 return Poll::Ready(Some(Err(err)));
1409 }
1410 }
1411
1412 if pin.evict_command_timeout.poll_ready(cx) {
1413 pin.evict_timed_out_commands(now);
1415 for t in pin.targets.values_mut() {
1418 t.network_manager.evict_stale_entries(now);
1419 t.frame_manager_mut().evict_stale_context_ids();
1420 }
1421 }
1422
1423 if pin.budget_exhausted {
1424 for t in pin.targets.values_mut() {
1425 t.network_manager.set_block_all(true);
1426 }
1427 }
1428
1429 if dispose {
1430 return Poll::Ready(None);
1431 }
1432
1433 if budget_hit {
1434 cx.waker().wake_by_ref();
1438 return Poll::Pending;
1439 }
1440
1441 if done {
1442 return Poll::Pending;
1444 }
1445 }
1446 }
1447}
1448
1449#[derive(Debug, Clone)]
1451pub struct HandlerConfig {
1452 pub ignore_https_errors: bool,
1454 pub viewport: Option<Viewport>,
1456 pub context_ids: Vec<BrowserContextId>,
1458 pub request_timeout: Duration,
1460 pub request_intercept: bool,
1462 pub cache_enabled: bool,
1464 pub service_worker_enabled: bool,
1466 pub ignore_visuals: bool,
1468 pub ignore_stylesheets: bool,
1470 pub ignore_javascript: bool,
1472 pub allow_first_party_stylesheets: bool,
1477 pub allow_first_party_javascript: bool,
1481 pub allow_first_party_visuals: bool,
1485 pub ignore_analytics: bool,
1487 pub ignore_prefetch: bool,
1489 pub ignore_ads: bool,
1491 pub extra_headers: Option<std::collections::HashMap<String, String>>,
1493 pub only_html: bool,
1495 pub created_first_target: bool,
1497 pub intercept_manager: NetworkInterceptManager,
1499 pub max_bytes_allowed: Option<u64>,
1501 pub max_redirects: Option<usize>,
1508 pub max_main_frame_navigations: Option<u32>,
1512 pub whitelist_patterns: Option<Vec<String>>,
1514 pub blacklist_patterns: Option<Vec<String>>,
1516 #[cfg(feature = "adblock")]
1518 pub adblock_filter_rules: Option<Vec<String>>,
1519 pub channel_capacity: usize,
1522 pub page_channel_capacity: usize,
1533 pub connection_retries: u32,
1536}
1537
1538impl Default for HandlerConfig {
1539 fn default() -> Self {
1540 Self {
1541 ignore_https_errors: true,
1542 viewport: Default::default(),
1543 context_ids: Vec::new(),
1544 request_timeout: Duration::from_millis(REQUEST_TIMEOUT),
1545 request_intercept: false,
1546 cache_enabled: true,
1547 service_worker_enabled: true,
1548 ignore_visuals: false,
1549 ignore_stylesheets: false,
1550 ignore_ads: false,
1551 ignore_javascript: false,
1552 allow_first_party_stylesheets: true,
1553 allow_first_party_javascript: true,
1554 allow_first_party_visuals: true,
1555 ignore_analytics: true,
1556 ignore_prefetch: true,
1557 only_html: false,
1558 extra_headers: Default::default(),
1559 created_first_target: false,
1560 intercept_manager: NetworkInterceptManager::Unknown,
1561 max_bytes_allowed: None,
1562 max_redirects: None,
1563 max_main_frame_navigations: None,
1564 whitelist_patterns: None,
1565 blacklist_patterns: None,
1566 #[cfg(feature = "adblock")]
1567 adblock_filter_rules: None,
1568 channel_capacity: 4096,
1569 page_channel_capacity: crate::handler::page::DEFAULT_PAGE_CHANNEL_CAPACITY,
1570 connection_retries: crate::conn::DEFAULT_CONNECTION_RETRIES,
1571 }
1572 }
1573}
1574
1575#[derive(Debug)]
1577pub struct NavigationInProgress<T> {
1578 navigated: bool,
1580 response: Option<Response>,
1582 tx: OneshotSender<T>,
1584}
1585
1586impl<T> NavigationInProgress<T> {
1587 pub(crate) fn new(tx: OneshotSender<T>) -> Self {
1588 Self {
1589 navigated: false,
1590 response: None,
1591 tx,
1592 }
1593 }
1594
1595 pub(crate) fn set_response(&mut self, resp: Response) {
1597 self.response = Some(resp);
1598 }
1599
1600 pub(crate) fn set_navigated(&mut self) {
1602 self.navigated = true;
1603 }
1604
1605 #[cfg_attr(not(feature = "parallel-handler"), allow(dead_code))]
1609 pub(crate) fn is_navigated(&self) -> bool {
1610 self.navigated
1611 }
1612
1613 #[cfg_attr(not(feature = "parallel-handler"), allow(dead_code))]
1614 pub(crate) fn take_response(&mut self) -> Option<Response> {
1615 self.response.take()
1616 }
1617
1618 #[cfg_attr(not(feature = "parallel-handler"), allow(dead_code))]
1619 pub(crate) fn into_tx(self) -> OneshotSender<T> {
1620 self.tx
1621 }
1622}
1623
1624#[derive(Debug)]
1626enum NavigationRequest {
1627 Navigate(NavigationInProgress<Result<Response>>),
1629 }
1631
1632#[derive(Debug)]
1635enum PendingRequest {
1636 CreateTarget(OneshotSender<Result<Page>>),
1639 GetTargets(OneshotSender<Result<Vec<TargetInfo>>>),
1641 Navigate(NavigationId),
1648 ExternalCommand {
1657 tx: OneshotSender<Result<Response>>,
1658 target_id: Option<TargetId>,
1659 },
1660 InternalCommand(TargetId),
1663 CloseBrowser(OneshotSender<Result<CloseReturns>>),
1665}
1666
1667#[derive(Debug)]
1671pub(crate) enum HandlerMessage {
1672 CreatePage(CreateTargetParams, OneshotSender<Result<Page>>),
1673 FetchTargets(OneshotSender<Result<Vec<TargetInfo>>>),
1674 InsertContext(BrowserContext),
1675 DisposeContext(BrowserContext),
1676 GetPages(OneshotSender<Vec<Page>>),
1677 Command(CommandMessage),
1678 GetPage(TargetId, OneshotSender<Option<Page>>),
1679 AddEventListener(EventListenerRequest),
1680 CloseBrowser(OneshotSender<Result<CloseReturns>>),
1681}
1682
1683#[cfg(test)]
1684mod tests {
1685 use super::*;
1686 use chromiumoxide_cdp::cdp::browser_protocol::target::{AttachToTargetReturns, TargetInfo};
1687
1688 #[test]
1689 fn attach_to_target_response_sets_session_id_before_event_arrives() {
1690 let info = TargetInfo::builder()
1691 .target_id("target-1".to_string())
1692 .r#type("page")
1693 .title("")
1694 .url("about:blank")
1695 .attached(false)
1696 .can_access_opener(false)
1697 .build()
1698 .expect("target info");
1699 let mut target = Target::new(info, TargetConfig::default(), BrowserContext::default());
1700 let method: MethodId = AttachToTargetParams::IDENTIFIER.into();
1701 let result = serde_json::to_value(AttachToTargetReturns::new("session-1".to_string()))
1702 .expect("attach result");
1703 let resp = Response {
1704 id: CallId::new(1),
1705 result: Some(result),
1706 error: None,
1707 };
1708
1709 maybe_store_attach_session_id(&mut target, &method, &resp);
1710
1711 assert_eq!(
1712 target.session_id().map(AsRef::as_ref),
1713 Some("session-1"),
1714 "attach response should seed the flat session id even before Target.attachedToTarget"
1715 );
1716 }
1717
1718 #[test]
1723 fn page_channel_capacity_defaults_to_2048_across_configs() {
1724 use crate::browser::BrowserConfigBuilder;
1725 use crate::handler::page::DEFAULT_PAGE_CHANNEL_CAPACITY;
1726 use crate::handler::target::TargetConfig;
1727
1728 assert_eq!(DEFAULT_PAGE_CHANNEL_CAPACITY, 2048);
1729 assert_eq!(
1730 HandlerConfig::default().page_channel_capacity,
1731 DEFAULT_PAGE_CHANNEL_CAPACITY,
1732 "HandlerConfig default must match the historical 2048 slot count"
1733 );
1734 assert_eq!(
1735 TargetConfig::default().page_channel_capacity,
1736 DEFAULT_PAGE_CHANNEL_CAPACITY,
1737 "TargetConfig default must match the historical 2048 slot count"
1738 );
1739 let builder = BrowserConfigBuilder::default();
1742 let bc = format!("{:?}", builder);
1743 assert!(
1744 bc.contains("page_channel_capacity: 2048"),
1745 "BrowserConfigBuilder must default page_channel_capacity to 2048, got: {bc}",
1746 );
1747 }
1748}