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;
52pub mod sender;
53mod session;
54pub mod target;
55pub mod target_message_future;
56pub mod viewport;
57
58#[must_use = "streams do nothing unless polled"]
61#[derive(Debug)]
62pub struct Handler {
63 pub default_browser_context: BrowserContext,
64 pub browser_contexts: HashSet<BrowserContext>,
65 pending_commands: FnvHashMap<CallId, (PendingRequest, MethodId, Instant)>,
69 from_browser: Receiver<HandlerMessage>,
71 target_ids: Vec<TargetId>,
73 targets: HashMap<TargetId, Target>,
75 navigations: FnvHashMap<NavigationId, NavigationRequest>,
77 sessions: HashMap<SessionId, Session>,
81 conn: Option<Connection<CdpEventMessage>>,
84 evict_command_timeout: PeriodicJob,
86 next_navigation_id: usize,
88 config: HandlerConfig,
90 event_listeners: EventListeners,
92 closing: bool,
94 remaining_bytes: Option<u64>,
96 budget_exhausted: bool,
98 attached_targets: HashSet<TargetId>,
100 page_wake: Option<Arc<Notify>>,
103}
104
105lazy_static::lazy_static! {
106 static ref DISCOVER_ID: (std::borrow::Cow<'static, str>, serde_json::Value) = {
108 let discover = SetDiscoverTargetsParams::new(true);
109 (discover.identifier(), serde_json::to_value(discover).expect("valid discover target params"))
110 };
111 static ref TARGET_PARAMS_ID: (std::borrow::Cow<'static, str>, serde_json::Value) = {
113 let msg = GetTargetsParams { filter: None };
114 (msg.identifier(), serde_json::to_value(msg).expect("valid paramtarget"))
115 };
116 static ref CLOSE_PARAMS_ID: (std::borrow::Cow<'static, str>, serde_json::Value) = {
118 let close_msg = CloseParams::default();
119 (close_msg.identifier(), serde_json::to_value(close_msg).expect("valid close params"))
120 };
121}
122
123fn maybe_store_attach_session_id(target: &mut Target, method: &MethodId, resp: &Response) {
124 if method.as_ref() != AttachToTargetParams::IDENTIFIER {
125 return;
126 }
127
128 if let Ok(resp) = to_command_response::<AttachToTargetParams>(resp.clone(), method.clone()) {
129 target.set_session_id(resp.result.session_id);
130 }
131}
132
133impl Handler {
134 pub(crate) fn new(
137 mut conn: Connection<CdpEventMessage>,
138 rx: Receiver<HandlerMessage>,
139 config: HandlerConfig,
140 ) -> Self {
141 let discover = DISCOVER_ID.clone();
142 let _ = conn.submit_command(discover.0, None, discover.1);
143 let conn = Some(conn);
144
145 let browser_contexts = config
146 .context_ids
147 .iter()
148 .map(|id| BrowserContext::from(id.clone()))
149 .collect();
150
151 Self {
152 pending_commands: Default::default(),
153 from_browser: rx,
154 default_browser_context: Default::default(),
155 browser_contexts,
156 target_ids: Default::default(),
157 targets: Default::default(),
158 navigations: Default::default(),
159 sessions: Default::default(),
160 conn,
161 evict_command_timeout: PeriodicJob::new(config.request_timeout),
162 next_navigation_id: 0,
163 config,
164 event_listeners: Default::default(),
165 closing: false,
166 remaining_bytes: None,
167 budget_exhausted: false,
168 attached_targets: Default::default(),
169 page_wake: None,
170 }
171 }
172
173 #[inline]
176 fn conn(&mut self) -> Result<&mut Connection<CdpEventMessage>> {
177 self.conn
178 .as_mut()
179 .ok_or_else(|| CdpError::msg("connection consumed by Handler::run()"))
180 }
181
182 pub fn get_target(&self, target_id: &TargetId) -> Option<&Target> {
184 self.targets.get(target_id)
185 }
186
187 pub fn targets(&self) -> impl Iterator<Item = &Target> + '_ {
189 self.targets.values()
190 }
191
192 pub fn default_browser_context(&self) -> &BrowserContext {
194 &self.default_browser_context
195 }
196
197 pub fn browser_contexts(&self) -> impl Iterator<Item = &BrowserContext> + '_ {
199 self.browser_contexts.iter()
200 }
201
202 fn on_navigation_response(&mut self, id: NavigationId, resp: Response) {
204 if let Some(nav) = self.navigations.remove(&id) {
205 match nav {
206 NavigationRequest::Navigate(mut nav) => {
207 if nav.navigated {
208 let _ = nav.tx.send(Ok(resp));
209 } else {
210 nav.set_response(resp);
211 self.navigations
212 .insert(id, NavigationRequest::Navigate(nav));
213 }
214 }
215 }
216 }
217 }
218
219 fn on_navigation_lifecycle_completed(&mut self, res: Result<NavigationOk, NavigationError>) {
221 match res {
222 Ok(ok) => {
223 let id = *ok.navigation_id();
224 if let Some(nav) = self.navigations.remove(&id) {
225 match nav {
226 NavigationRequest::Navigate(mut nav) => {
227 if let Some(resp) = nav.response.take() {
228 let _ = nav.tx.send(Ok(resp));
229 } else {
230 nav.set_navigated();
231 self.navigations
232 .insert(id, NavigationRequest::Navigate(nav));
233 }
234 }
235 }
236 }
237 }
238 Err(err) => {
239 if let Some(nav) = self.navigations.remove(err.navigation_id()) {
240 match nav {
241 NavigationRequest::Navigate(nav) => {
242 let _ = nav.tx.send(Err(err.into()));
243 }
244 }
245 }
246 }
247 }
248 }
249
250 fn on_response(&mut self, resp: Response) {
252 if let Some((req, method, _)) = self.pending_commands.remove(&resp.id) {
253 match req {
254 PendingRequest::CreateTarget(tx) => {
255 match to_command_response::<CreateTargetParams>(resp, method) {
256 Ok(resp) => {
257 if let Some(target) = self.targets.get_mut(&resp.target_id) {
258 target.set_initiator(tx);
259 } else {
260 let _ = tx.send(Err(CdpError::NotFound)).ok();
261 }
262 }
263 Err(err) => {
264 let _ = tx.send(Err(err)).ok();
265 }
266 }
267 }
268 PendingRequest::GetTargets(tx) => {
269 match to_command_response::<GetTargetsParams>(resp, method) {
270 Ok(resp) => {
271 let targets = resp.result.target_infos;
272 let results = targets.clone();
273
274 for target_info in targets {
275 let event: EventTargetCreated = EventTargetCreated { target_info };
276 self.on_target_created(event);
277 }
278
279 let _ = tx.send(Ok(results)).ok();
280 }
281 Err(err) => {
282 let _ = tx.send(Err(err)).ok();
283 }
284 }
285 }
286 PendingRequest::Navigate(id) => {
287 self.on_navigation_response(id, resp);
288 if self.config.only_html && !self.config.created_first_target {
289 self.config.created_first_target = true;
290 }
291 }
292 PendingRequest::ExternalCommand(tx) => {
293 let _ = tx.send(Ok(resp)).ok();
294 }
295 PendingRequest::InternalCommand(target_id) => {
296 if let Some(target) = self.targets.get_mut(&target_id) {
297 maybe_store_attach_session_id(target, &method, &resp);
298 target.on_response(resp, method.as_ref());
299 }
300 }
301 PendingRequest::CloseBrowser(tx) => {
302 self.closing = true;
303 let _ = tx.send(Ok(CloseReturns {})).ok();
304 }
305 }
306 }
307 }
308
309 pub(crate) fn submit_external_command(
311 &mut self,
312 msg: CommandMessage,
313 now: Instant,
314 ) -> Result<()> {
315 let call_id =
316 self.conn()?
317 .submit_command(msg.method.clone(), msg.session_id, msg.params)?;
318 self.pending_commands.insert(
319 call_id,
320 (PendingRequest::ExternalCommand(msg.sender), msg.method, now),
321 );
322 Ok(())
323 }
324
325 pub(crate) fn submit_internal_command(
326 &mut self,
327 target_id: TargetId,
328 req: CdpRequest,
329 now: Instant,
330 ) -> Result<()> {
331 let call_id = self.conn()?.submit_command(
332 req.method.clone(),
333 req.session_id.map(Into::into),
334 req.params,
335 )?;
336 self.pending_commands.insert(
337 call_id,
338 (PendingRequest::InternalCommand(target_id), req.method, now),
339 );
340 Ok(())
341 }
342
343 fn submit_fetch_targets(&mut self, tx: OneshotSender<Result<Vec<TargetInfo>>>, now: Instant) {
344 let msg = TARGET_PARAMS_ID.clone();
345
346 if let Some(conn) = self.conn.as_mut() {
347 if let Ok(call_id) = conn.submit_command(msg.0.clone(), None, msg.1) {
348 self.pending_commands
349 .insert(call_id, (PendingRequest::GetTargets(tx), msg.0, now));
350 }
351 }
352 }
353
354 fn submit_navigation(&mut self, id: NavigationId, req: CdpRequest, now: Instant) {
357 if let Some(conn) = self.conn.as_mut() {
358 if let Ok(call_id) = conn.submit_command(
359 req.method.clone(),
360 req.session_id.map(Into::into),
361 req.params,
362 ) {
363 self.pending_commands
364 .insert(call_id, (PendingRequest::Navigate(id), req.method, now));
365 }
366 }
367 }
368
369 fn submit_close(&mut self, tx: OneshotSender<Result<CloseReturns>>, now: Instant) {
370 let close_msg = CLOSE_PARAMS_ID.clone();
371
372 if let Some(conn) = self.conn.as_mut() {
373 if let Ok(call_id) = conn.submit_command(close_msg.0.clone(), None, close_msg.1) {
374 self.pending_commands.insert(
375 call_id,
376 (PendingRequest::CloseBrowser(tx), close_msg.0, now),
377 );
378 }
379 }
380 }
381
382 fn on_target_message(&mut self, target: &mut Target, msg: CommandMessage, now: Instant) {
384 if msg.is_navigation() {
385 let (req, tx) = msg.split();
386 let id = self.next_navigation_id();
387
388 target.goto(FrameRequestedNavigation::new(
389 id,
390 req,
391 self.config.request_timeout,
392 ));
393
394 self.navigations.insert(
395 id,
396 NavigationRequest::Navigate(NavigationInProgress::new(tx)),
397 );
398 } else {
399 let _ = self.submit_external_command(msg, now);
400 }
401 }
402
403 fn next_navigation_id(&mut self) -> NavigationId {
405 let id = NavigationId(self.next_navigation_id);
406 self.next_navigation_id = self.next_navigation_id.wrapping_add(1);
407 id
408 }
409
410 fn create_page(&mut self, params: CreateTargetParams, tx: OneshotSender<Result<Page>>) {
421 let about_blank = params.url == "about:blank";
422 let http_check =
423 !about_blank && params.url.starts_with("http") || params.url.starts_with("file://");
424
425 if about_blank || http_check {
426 let method = params.identifier();
427
428 let Some(conn) = self.conn.as_mut() else {
429 let _ = tx.send(Err(CdpError::msg("connection consumed"))).ok();
430 return;
431 };
432 match serde_json::to_value(params) {
433 Ok(params) => match conn.submit_command(method.clone(), None, params) {
434 Ok(call_id) => {
435 self.pending_commands.insert(
436 call_id,
437 (PendingRequest::CreateTarget(tx), method, Instant::now()),
438 );
439 }
440 Err(err) => {
441 let _ = tx.send(Err(err.into())).ok();
442 }
443 },
444 Err(err) => {
445 let _ = tx.send(Err(err.into())).ok();
446 }
447 }
448 } else {
449 let _ = tx.send(Err(CdpError::NotFound)).ok();
450 }
451 }
452
453 fn on_event(&mut self, event: CdpEventMessage) {
455 if let Some(session_id) = &event.session_id {
456 if let Some(session) = self.sessions.get(session_id.as_str()) {
457 if let Some(target) = self.targets.get_mut(session.target_id()) {
458 return target.on_event(event);
459 }
460 }
461 }
462 let CdpEventMessage { params, method, .. } = event;
463
464 match params {
465 CdpEvent::TargetTargetCreated(ref ev) => self.on_target_created((**ev).clone()),
466 CdpEvent::TargetAttachedToTarget(ref ev) => self.on_attached_to_target(ev.clone()),
467 CdpEvent::TargetTargetDestroyed(ref ev) => self.on_target_destroyed(ev.clone()),
468 CdpEvent::TargetDetachedFromTarget(ref ev) => self.on_detached_from_target(ev.clone()),
469 _ => {}
470 }
471
472 chromiumoxide_cdp::consume_event!(match params {
473 |ev| self.event_listeners.start_send(ev),
474 |json| { let _ = self.event_listeners.try_send_custom(&method, json);}
475 });
476 }
477
478 fn on_target_created(&mut self, event: EventTargetCreated) {
482 if !self.browser_contexts.is_empty() {
483 if let Some(ref context_id) = event.target_info.browser_context_id {
484 let bc = BrowserContext {
485 id: Some(context_id.clone()),
486 };
487 if !self.browser_contexts.contains(&bc) {
488 return;
489 }
490 }
491 }
492 let browser_ctx = event
493 .target_info
494 .browser_context_id
495 .clone()
496 .map(BrowserContext::from)
497 .unwrap_or_else(|| self.default_browser_context.clone());
498 let target = Target::new(
499 event.target_info,
500 TargetConfig {
501 ignore_https_errors: self.config.ignore_https_errors,
502 request_timeout: self.config.request_timeout,
503 viewport: self.config.viewport.clone(),
504 request_intercept: self.config.request_intercept,
505 cache_enabled: self.config.cache_enabled,
506 service_worker_enabled: self.config.service_worker_enabled,
507 ignore_visuals: self.config.ignore_visuals,
508 ignore_stylesheets: self.config.ignore_stylesheets,
509 ignore_javascript: self.config.ignore_javascript,
510 ignore_analytics: self.config.ignore_analytics,
511 ignore_prefetch: self.config.ignore_prefetch,
512 extra_headers: self.config.extra_headers.clone(),
513 only_html: self.config.only_html && self.config.created_first_target,
514 intercept_manager: self.config.intercept_manager,
515 max_bytes_allowed: self.config.max_bytes_allowed,
516 whitelist_patterns: self.config.whitelist_patterns.clone(),
517 blacklist_patterns: self.config.blacklist_patterns.clone(),
518 #[cfg(feature = "adblock")]
519 adblock_filter_rules: self.config.adblock_filter_rules.clone(),
520 page_wake: self.page_wake.clone(),
521 },
522 browser_ctx,
523 );
524
525 let tid = target.target_id().clone();
526 self.target_ids.push(tid.clone());
527 self.targets.insert(tid, target);
528 }
529
530 fn on_attached_to_target(&mut self, event: Box<EventAttachedToTarget>) {
532 let session = Session::new(event.session_id.clone(), event.target_info.target_id);
533 if let Some(target) = self.targets.get_mut(session.target_id()) {
534 target.set_session_id(session.session_id().clone())
535 }
536 self.sessions.insert(event.session_id, session);
537 }
538
539 fn on_detached_from_target(&mut self, event: EventDetachedFromTarget) {
543 if let Some(session) = self.sessions.remove(&event.session_id) {
545 if let Some(target) = self.targets.get_mut(session.target_id()) {
546 target.session_id_mut().take();
547 }
548 }
549 }
550
551 fn on_target_destroyed(&mut self, event: EventTargetDestroyed) {
553 self.attached_targets.remove(&event.target_id);
554
555 if let Some(target) = self.targets.remove(&event.target_id) {
556 if let Some(session) = target.session_id() {
558 self.sessions.remove(session);
559 }
560 }
561 }
562
563 fn evict_timed_out_commands(&mut self, now: Instant) {
568 let deadline = match now.checked_sub(self.config.request_timeout) {
569 Some(d) => d,
570 None => return,
571 };
572
573 let timed_out: Vec<_> = self
574 .pending_commands
575 .iter()
576 .filter(|(_, (_, _, timestamp))| *timestamp < deadline)
577 .map(|(k, _)| *k)
578 .collect();
579
580 for call in timed_out {
581 if let Some((req, _, _)) = self.pending_commands.remove(&call) {
582 match req {
583 PendingRequest::CreateTarget(tx) => {
584 let _ = tx.send(Err(CdpError::Timeout));
585 }
586 PendingRequest::GetTargets(tx) => {
587 let _ = tx.send(Err(CdpError::Timeout));
588 }
589 PendingRequest::Navigate(nav) => {
590 if let Some(nav) = self.navigations.remove(&nav) {
591 match nav {
592 NavigationRequest::Navigate(nav) => {
593 let _ = nav.tx.send(Err(CdpError::Timeout));
594 }
595 }
596 }
597 }
598 PendingRequest::ExternalCommand(tx) => {
599 let _ = tx.send(Err(CdpError::Timeout));
600 }
601 PendingRequest::InternalCommand(_) => {}
602 PendingRequest::CloseBrowser(tx) => {
603 let _ = tx.send(Err(CdpError::Timeout));
604 }
605 }
606 }
607 }
608 }
609
610 pub fn event_listeners_mut(&mut self) -> &mut EventListeners {
611 &mut self.event_listeners
612 }
613
614 pub async fn run(mut self) -> Result<()> {
644 use chromiumoxide_types::Message;
645 use tokio::time::MissedTickBehavior;
646 use tokio_tungstenite::tungstenite::{self, error::ProtocolError};
647
648 let page_wake = Arc::new(Notify::new());
650 self.page_wake = Some(page_wake.clone());
651
652 let conn = self
654 .conn
655 .take()
656 .ok_or_else(|| CdpError::msg("Handler::run() called with no connection"))?;
657 let async_conn = conn.into_async();
658 let mut ws_reader = async_conn.reader;
659 let ws_tx = async_conn.cmd_tx;
660 let mut writer_handle = async_conn.writer_handle;
661 let mut next_call_id = async_conn.next_id;
662
663 let mut alloc_call_id = || {
665 let id = chromiumoxide_types::CallId::new(next_call_id);
666 next_call_id = next_call_id.wrapping_add(1);
667 id
668 };
669
670 let mut evict_timer = tokio::time::interval_at(
672 tokio::time::Instant::now() + self.config.request_timeout,
673 self.config.request_timeout,
674 );
675 evict_timer.set_missed_tick_behavior(MissedTickBehavior::Delay);
676
677 macro_rules! ws_submit {
679 ($method:expr, $session_id:expr, $params:expr) => {{
680 let id = alloc_call_id();
681 let call = chromiumoxide_types::MethodCall {
682 id,
683 method: $method,
684 session_id: $session_id,
685 params: $params,
686 };
687 match ws_tx.try_send(call) {
688 Ok(()) => Ok::<_, CdpError>(id),
689 Err(tokio::sync::mpsc::error::TrySendError::Full(_)) => {
690 tracing::warn!("WS command channel full — dropping command");
691 Err(CdpError::msg("WS command channel full"))
692 }
693 Err(tokio::sync::mpsc::error::TrySendError::Closed(_)) => {
694 Err(CdpError::msg("WS writer closed"))
695 }
696 }
697 }};
698 }
699
700 loop {
702 let now = std::time::Instant::now();
703
704 const PER_TARGET_DRAIN_BUDGET: usize = 128;
710
711 for n in (0..self.target_ids.len()).rev() {
712 let target_id = self.target_ids.swap_remove(n);
713
714 if let Some((id, mut target)) = self.targets.remove_entry(&target_id) {
715 {
717 let mut msgs = Vec::new();
718 if let Some(handle) = target.page_mut() {
719 while msgs.len() < PER_TARGET_DRAIN_BUDGET {
720 match handle.rx.try_recv() {
721 Ok(msg) => msgs.push(msg),
722 Err(_) => break,
723 }
724 }
725 }
726 for msg in msgs {
727 target.on_page_message(msg);
728 }
729 }
730
731 while let Some(event) = target.advance(now) {
733 match event {
734 TargetEvent::Request(req) => {
735 if let Ok(call_id) =
736 ws_submit!(req.method.clone(), req.session_id, req.params)
737 {
738 self.pending_commands.insert(
739 call_id,
740 (
741 PendingRequest::InternalCommand(
742 target.target_id().clone(),
743 ),
744 req.method,
745 now,
746 ),
747 );
748 }
749 }
750 TargetEvent::Command(msg) => {
751 if msg.is_navigation() {
752 let (req, tx) = msg.split();
753 let nav_id = self.next_navigation_id();
754 target.goto(FrameRequestedNavigation::new(
755 nav_id,
756 req.clone(),
757 self.config.request_timeout,
758 ));
759 if let Ok(call_id) =
760 ws_submit!(req.method.clone(), req.session_id, req.params)
761 {
762 self.pending_commands.insert(
763 call_id,
764 (PendingRequest::Navigate(nav_id), req.method, now),
765 );
766 }
767 self.navigations.insert(
768 nav_id,
769 NavigationRequest::Navigate(NavigationInProgress::new(tx)),
770 );
771 } else if let Ok(call_id) = ws_submit!(
772 msg.method.clone(),
773 msg.session_id.map(Into::into),
774 msg.params
775 ) {
776 self.pending_commands.insert(
777 call_id,
778 (
779 PendingRequest::ExternalCommand(msg.sender),
780 msg.method,
781 now,
782 ),
783 );
784 }
785 }
786 TargetEvent::NavigationRequest(nav_id, req) => {
787 if let Ok(call_id) =
788 ws_submit!(req.method.clone(), req.session_id, req.params)
789 {
790 self.pending_commands.insert(
791 call_id,
792 (PendingRequest::Navigate(nav_id), req.method, now),
793 );
794 }
795 }
796 TargetEvent::NavigationResult(res) => {
797 self.on_navigation_lifecycle_completed(res);
798 }
799 TargetEvent::BytesConsumed(n) => {
800 if let Some(rem) = self.remaining_bytes.as_mut() {
801 *rem = rem.saturating_sub(n);
802 if *rem == 0 {
803 self.budget_exhausted = true;
804 }
805 }
806 }
807 }
808 }
809
810 target.event_listeners_mut().flush();
812
813 self.targets.insert(id, target);
814 self.target_ids.push(target_id);
815 }
816 }
817
818 self.event_listeners.flush();
820
821 if self.budget_exhausted {
822 for t in self.targets.values_mut() {
823 t.network_manager.set_block_all(true);
824 }
825 }
826
827 if self.closing {
828 break;
829 }
830
831 tokio::select! {
833 msg = self.from_browser.recv() => {
834 match msg {
835 Some(msg) => {
836 match msg {
837 HandlerMessage::Command(cmd) => {
838 if let Ok(call_id) = ws_submit!(
839 cmd.method.clone(),
840 cmd.session_id.map(Into::into),
841 cmd.params
842 ) {
843 self.pending_commands.insert(
844 call_id,
845 (PendingRequest::ExternalCommand(cmd.sender), cmd.method, now),
846 );
847 }
848 }
849 HandlerMessage::FetchTargets(tx) => {
850 let msg = TARGET_PARAMS_ID.clone();
851 if let Ok(call_id) = ws_submit!(msg.0.clone(), None, msg.1) {
852 self.pending_commands.insert(
853 call_id,
854 (PendingRequest::GetTargets(tx), msg.0, now),
855 );
856 }
857 }
858 HandlerMessage::CloseBrowser(tx) => {
859 let close_msg = CLOSE_PARAMS_ID.clone();
860 if let Ok(call_id) = ws_submit!(close_msg.0.clone(), None, close_msg.1) {
861 self.pending_commands.insert(
862 call_id,
863 (PendingRequest::CloseBrowser(tx), close_msg.0, now),
864 );
865 }
866 }
867 HandlerMessage::CreatePage(params, tx) => {
868 if let Some(ref id) = params.browser_context_id {
869 self.browser_contexts.insert(BrowserContext::from(id.clone()));
870 }
871 self.create_page_async(params, tx, &mut alloc_call_id, &ws_tx, now);
872 }
873 HandlerMessage::GetPages(tx) => {
874 let pages: Vec<_> = self.targets.values_mut()
875 .filter(|p| p.is_page())
876 .filter_map(|target| target.get_or_create_page())
877 .map(|page| Page::from(page.clone()))
878 .collect();
879 let _ = tx.send(pages);
880 }
881 HandlerMessage::InsertContext(ctx) => {
882 if self.default_browser_context.id().is_none() {
883 self.default_browser_context = ctx.clone();
884 }
885 self.browser_contexts.insert(ctx);
886 }
887 HandlerMessage::DisposeContext(ctx) => {
888 self.browser_contexts.remove(&ctx);
889 self.attached_targets.retain(|tid| {
890 self.targets.get(tid)
891 .and_then(|t| t.browser_context_id())
892 .map(|id| Some(id) != ctx.id())
893 .unwrap_or(true)
894 });
895 self.closing = true;
896 }
897 HandlerMessage::GetPage(target_id, tx) => {
898 let page = self.targets.get_mut(&target_id)
899 .and_then(|target| target.get_or_create_page())
900 .map(|page| Page::from(page.clone()));
901 let _ = tx.send(page);
902 }
903 HandlerMessage::AddEventListener(req) => {
904 self.event_listeners.add_listener(req);
905 }
906 }
907 }
908 None => break, }
910 }
911
912 frame = ws_reader.next_message() => {
913 match frame {
914 Some(Ok(boxed_msg)) => match *boxed_msg {
915 Message::Response(resp) => {
916 self.on_response(resp);
917 }
918 Message::Event(ev) => {
919 self.on_event(ev);
920 }
921 },
922 Some(Err(err)) => {
923 tracing::error!("WS Connection error: {:?}", err);
924 if let CdpError::Ws(ref ws_error) = err {
925 match ws_error {
926 tungstenite::Error::AlreadyClosed => break,
927 tungstenite::Error::Protocol(detail)
928 if detail == &ProtocolError::ResetWithoutClosingHandshake =>
929 {
930 break;
931 }
932 _ => return Err(err),
933 }
934 } else {
935 return Err(err);
936 }
937 }
938 None => break, }
940 }
941
942 _ = page_wake.notified() => {
943 }
945
946 _ = evict_timer.tick() => {
947 self.evict_timed_out_commands(now);
948 for t in self.targets.values_mut() {
949 t.network_manager.evict_stale_entries(now);
950 t.frame_manager_mut().evict_stale_context_ids();
951 }
952 }
953
954 result = &mut writer_handle => {
955 match result {
957 Ok(Ok(())) => break,
958 Ok(Err(e)) => return Err(e),
959 Err(e) => return Err(CdpError::msg(format!("WS writer panicked: {e}"))),
960 }
961 }
962 }
963 }
964
965 writer_handle.abort();
966 Ok(())
967 }
968
969 fn create_page_async(
971 &mut self,
972 params: CreateTargetParams,
973 tx: OneshotSender<Result<Page>>,
974 alloc_call_id: &mut impl FnMut() -> chromiumoxide_types::CallId,
975 ws_tx: &tokio::sync::mpsc::Sender<chromiumoxide_types::MethodCall>,
976 now: std::time::Instant,
977 ) {
978 let about_blank = params.url == "about:blank";
979 let http_check =
980 !about_blank && params.url.starts_with("http") || params.url.starts_with("file://");
981
982 if about_blank || http_check {
983 let method = params.identifier();
984 match serde_json::to_value(params) {
985 Ok(params) => {
986 let id = alloc_call_id();
987 let call = chromiumoxide_types::MethodCall {
988 id,
989 method: method.clone(),
990 session_id: None,
991 params,
992 };
993 match ws_tx.try_send(call) {
994 Ok(()) => {
995 self.pending_commands
996 .insert(id, (PendingRequest::CreateTarget(tx), method, now));
997 }
998 Err(_) => {
999 let _ = tx
1000 .send(Err(CdpError::msg("WS command channel full or closed")))
1001 .ok();
1002 }
1003 }
1004 }
1005 Err(err) => {
1006 let _ = tx.send(Err(err.into())).ok();
1007 }
1008 }
1009 } else {
1010 let _ = tx.send(Err(CdpError::NotFound)).ok();
1011 }
1012 }
1013}
1014
1015impl Stream for Handler {
1016 type Item = Result<()>;
1017
1018 fn poll_next(self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Option<Self::Item>> {
1019 let pin = self.get_mut();
1020
1021 let mut dispose = false;
1022
1023 let now = Instant::now();
1024
1025 loop {
1026 while let Poll::Ready(Some(msg)) = pin.from_browser.poll_recv(cx) {
1030 match msg {
1031 HandlerMessage::Command(cmd) => {
1032 pin.submit_external_command(cmd, now)?;
1033 }
1034 HandlerMessage::FetchTargets(tx) => {
1035 pin.submit_fetch_targets(tx, now);
1036 }
1037 HandlerMessage::CloseBrowser(tx) => {
1038 pin.submit_close(tx, now);
1039 }
1040 HandlerMessage::CreatePage(params, tx) => {
1041 if let Some(ref id) = params.browser_context_id {
1042 pin.browser_contexts
1043 .insert(BrowserContext::from(id.clone()));
1044 }
1045 pin.create_page(params, tx);
1046 }
1047 HandlerMessage::GetPages(tx) => {
1048 let pages: Vec<_> = pin
1049 .targets
1050 .values_mut()
1051 .filter(|p: &&mut Target| p.is_page())
1052 .filter_map(|target| target.get_or_create_page())
1053 .map(|page| Page::from(page.clone()))
1054 .collect();
1055 let _ = tx.send(pages);
1056 }
1057 HandlerMessage::InsertContext(ctx) => {
1058 if pin.default_browser_context.id().is_none() {
1059 pin.default_browser_context = ctx.clone();
1060 }
1061 pin.browser_contexts.insert(ctx);
1062 }
1063 HandlerMessage::DisposeContext(ctx) => {
1064 pin.browser_contexts.remove(&ctx);
1065 pin.attached_targets.retain(|tid| {
1066 pin.targets
1067 .get(tid)
1068 .and_then(|t| t.browser_context_id()) .map(|id| Some(id) != ctx.id())
1070 .unwrap_or(true)
1071 });
1072 pin.closing = true;
1073 dispose = true;
1074 }
1075 HandlerMessage::GetPage(target_id, tx) => {
1076 let page = pin
1077 .targets
1078 .get_mut(&target_id)
1079 .and_then(|target| target.get_or_create_page())
1080 .map(|page| Page::from(page.clone()));
1081 let _ = tx.send(page);
1082 }
1083 HandlerMessage::AddEventListener(req) => {
1084 pin.event_listeners.add_listener(req);
1085 }
1086 }
1087 }
1088
1089 for n in (0..pin.target_ids.len()).rev() {
1090 let target_id = pin.target_ids.swap_remove(n);
1091
1092 if let Some((id, mut target)) = pin.targets.remove_entry(&target_id) {
1093 while let Some(event) = target.poll(cx, now) {
1094 match event {
1095 TargetEvent::Request(req) => {
1096 let _ = pin.submit_internal_command(
1097 target.target_id().clone(),
1098 req,
1099 now,
1100 );
1101 }
1102 TargetEvent::Command(msg) => {
1103 pin.on_target_message(&mut target, msg, now);
1104 }
1105 TargetEvent::NavigationRequest(id, req) => {
1106 pin.submit_navigation(id, req, now);
1107 }
1108 TargetEvent::NavigationResult(res) => {
1109 pin.on_navigation_lifecycle_completed(res)
1110 }
1111 TargetEvent::BytesConsumed(n) => {
1112 if let Some(rem) = pin.remaining_bytes.as_mut() {
1113 *rem = rem.saturating_sub(n);
1114 if *rem == 0 {
1115 pin.budget_exhausted = true;
1116 }
1117 }
1118 }
1119 }
1120 }
1121
1122 target.event_listeners_mut().poll(cx);
1124
1125 pin.targets.insert(id, target);
1126 pin.target_ids.push(target_id);
1127 }
1128 }
1129
1130 pin.event_listeners_mut().poll(cx);
1133
1134 let mut done = true;
1135
1136 let mut ws_msgs = Vec::new();
1139 let mut ws_err = None;
1140 {
1141 let Some(conn) = pin.conn.as_mut() else {
1142 return Poll::Ready(Some(Err(CdpError::msg(
1143 "connection consumed by Handler::run()",
1144 ))));
1145 };
1146 while let Poll::Ready(Some(ev)) = Pin::new(&mut *conn).poll_next(cx) {
1147 match ev {
1148 Ok(msg) => ws_msgs.push(msg),
1149 Err(err) => {
1150 ws_err = Some(err);
1151 break;
1152 }
1153 }
1154 }
1155 }
1156
1157 for boxed_msg in ws_msgs {
1158 match *boxed_msg {
1159 Message::Response(resp) => {
1160 pin.on_response(resp);
1161 if pin.closing {
1162 return Poll::Ready(None);
1163 }
1164 }
1165 Message::Event(ev) => {
1166 pin.on_event(ev);
1167 }
1168 }
1169 done = false;
1170 }
1171
1172 if let Some(err) = ws_err {
1173 tracing::error!("WS Connection error: {:?}", err);
1174 if let CdpError::Ws(ref ws_error) = err {
1175 match ws_error {
1176 Error::AlreadyClosed => {
1177 pin.closing = true;
1178 dispose = true;
1179 }
1180 Error::Protocol(detail)
1181 if detail == &ProtocolError::ResetWithoutClosingHandshake =>
1182 {
1183 pin.closing = true;
1184 dispose = true;
1185 }
1186 _ => return Poll::Ready(Some(Err(err))),
1187 }
1188 } else {
1189 return Poll::Ready(Some(Err(err)));
1190 }
1191 }
1192
1193 if pin.evict_command_timeout.poll_ready(cx) {
1194 pin.evict_timed_out_commands(now);
1196 for t in pin.targets.values_mut() {
1199 t.network_manager.evict_stale_entries(now);
1200 t.frame_manager_mut().evict_stale_context_ids();
1201 }
1202 }
1203
1204 if pin.budget_exhausted {
1205 for t in pin.targets.values_mut() {
1206 t.network_manager.set_block_all(true);
1207 }
1208 }
1209
1210 if dispose {
1211 return Poll::Ready(None);
1212 }
1213
1214 if done {
1215 return Poll::Pending;
1217 }
1218 }
1219 }
1220}
1221
1222#[derive(Debug, Clone)]
1224pub struct HandlerConfig {
1225 pub ignore_https_errors: bool,
1227 pub viewport: Option<Viewport>,
1229 pub context_ids: Vec<BrowserContextId>,
1231 pub request_timeout: Duration,
1233 pub request_intercept: bool,
1235 pub cache_enabled: bool,
1237 pub service_worker_enabled: bool,
1239 pub ignore_visuals: bool,
1241 pub ignore_stylesheets: bool,
1243 pub ignore_javascript: bool,
1245 pub ignore_analytics: bool,
1247 pub ignore_prefetch: bool,
1249 pub ignore_ads: bool,
1251 pub extra_headers: Option<std::collections::HashMap<String, String>>,
1253 pub only_html: bool,
1255 pub created_first_target: bool,
1257 pub intercept_manager: NetworkInterceptManager,
1259 pub max_bytes_allowed: Option<u64>,
1261 pub whitelist_patterns: Option<Vec<String>>,
1263 pub blacklist_patterns: Option<Vec<String>>,
1265 #[cfg(feature = "adblock")]
1267 pub adblock_filter_rules: Option<Vec<String>>,
1268 pub channel_capacity: usize,
1271 pub connection_retries: u32,
1274}
1275
1276impl Default for HandlerConfig {
1277 fn default() -> Self {
1278 Self {
1279 ignore_https_errors: true,
1280 viewport: Default::default(),
1281 context_ids: Vec::new(),
1282 request_timeout: Duration::from_millis(REQUEST_TIMEOUT),
1283 request_intercept: false,
1284 cache_enabled: true,
1285 service_worker_enabled: true,
1286 ignore_visuals: false,
1287 ignore_stylesheets: false,
1288 ignore_ads: false,
1289 ignore_javascript: false,
1290 ignore_analytics: true,
1291 ignore_prefetch: true,
1292 only_html: false,
1293 extra_headers: Default::default(),
1294 created_first_target: false,
1295 intercept_manager: NetworkInterceptManager::Unknown,
1296 max_bytes_allowed: None,
1297 whitelist_patterns: None,
1298 blacklist_patterns: None,
1299 #[cfg(feature = "adblock")]
1300 adblock_filter_rules: None,
1301 channel_capacity: 4096,
1302 connection_retries: crate::conn::DEFAULT_CONNECTION_RETRIES,
1303 }
1304 }
1305}
1306
1307#[derive(Debug)]
1309pub struct NavigationInProgress<T> {
1310 navigated: bool,
1312 response: Option<Response>,
1314 tx: OneshotSender<T>,
1316}
1317
1318impl<T> NavigationInProgress<T> {
1319 fn new(tx: OneshotSender<T>) -> Self {
1320 Self {
1321 navigated: false,
1322 response: None,
1323 tx,
1324 }
1325 }
1326
1327 fn set_response(&mut self, resp: Response) {
1329 self.response = Some(resp);
1330 }
1331
1332 fn set_navigated(&mut self) {
1334 self.navigated = true;
1335 }
1336}
1337
1338#[derive(Debug)]
1340enum NavigationRequest {
1341 Navigate(NavigationInProgress<Result<Response>>),
1343 }
1345
1346#[derive(Debug)]
1349enum PendingRequest {
1350 CreateTarget(OneshotSender<Result<Page>>),
1353 GetTargets(OneshotSender<Result<Vec<TargetInfo>>>),
1355 Navigate(NavigationId),
1362 ExternalCommand(OneshotSender<Result<Response>>),
1364 InternalCommand(TargetId),
1367 CloseBrowser(OneshotSender<Result<CloseReturns>>),
1369}
1370
1371#[derive(Debug)]
1375pub(crate) enum HandlerMessage {
1376 CreatePage(CreateTargetParams, OneshotSender<Result<Page>>),
1377 FetchTargets(OneshotSender<Result<Vec<TargetInfo>>>),
1378 InsertContext(BrowserContext),
1379 DisposeContext(BrowserContext),
1380 GetPages(OneshotSender<Vec<Page>>),
1381 Command(CommandMessage),
1382 GetPage(TargetId, OneshotSender<Option<Page>>),
1383 AddEventListener(EventListenerRequest),
1384 CloseBrowser(OneshotSender<Result<CloseReturns>>),
1385}
1386
1387#[cfg(test)]
1388mod tests {
1389 use super::*;
1390 use chromiumoxide_cdp::cdp::browser_protocol::target::{AttachToTargetReturns, TargetInfo};
1391
1392 #[test]
1393 fn attach_to_target_response_sets_session_id_before_event_arrives() {
1394 let info = TargetInfo::builder()
1395 .target_id("target-1".to_string())
1396 .r#type("page")
1397 .title("")
1398 .url("about:blank")
1399 .attached(false)
1400 .can_access_opener(false)
1401 .build()
1402 .expect("target info");
1403 let mut target = Target::new(info, TargetConfig::default(), BrowserContext::default());
1404 let method: MethodId = AttachToTargetParams::IDENTIFIER.into();
1405 let result = serde_json::to_value(AttachToTargetReturns::new("session-1".to_string()))
1406 .expect("attach result");
1407 let resp = Response {
1408 id: CallId::new(1),
1409 result: Some(result),
1410 error: None,
1411 };
1412
1413 maybe_store_attach_session_id(&mut target, &method, &resp);
1414
1415 assert_eq!(
1416 target.session_id().map(AsRef::as_ref),
1417 Some("session-1"),
1418 "attach response should seed the flat session id even before Target.attachedToTarget"
1419 );
1420 }
1421}