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 self.target_ids.push(target.target_id().clone());
526 self.targets.insert(target.target_id().clone(), target);
527 }
528
529 fn on_attached_to_target(&mut self, event: Box<EventAttachedToTarget>) {
531 let session = Session::new(event.session_id.clone(), event.target_info.target_id);
532 if let Some(target) = self.targets.get_mut(session.target_id()) {
533 target.set_session_id(session.session_id().clone())
534 }
535 self.sessions.insert(event.session_id, session);
536 }
537
538 fn on_detached_from_target(&mut self, event: EventDetachedFromTarget) {
542 if let Some(session) = self.sessions.remove(&event.session_id) {
544 if let Some(target) = self.targets.get_mut(session.target_id()) {
545 target.session_id_mut().take();
546 }
547 }
548 }
549
550 fn on_target_destroyed(&mut self, event: EventTargetDestroyed) {
552 self.attached_targets.remove(&event.target_id);
553
554 if let Some(target) = self.targets.remove(&event.target_id) {
555 if let Some(session) = target.session_id() {
557 self.sessions.remove(session);
558 }
559 }
560 }
561
562 fn evict_timed_out_commands(&mut self, now: Instant) {
567 let deadline = match now.checked_sub(self.config.request_timeout) {
568 Some(d) => d,
569 None => return,
570 };
571
572 let timed_out: Vec<_> = self
573 .pending_commands
574 .iter()
575 .filter(|(_, (_, _, timestamp))| *timestamp < deadline)
576 .map(|(k, _)| *k)
577 .collect();
578
579 for call in timed_out {
580 if let Some((req, _, _)) = self.pending_commands.remove(&call) {
581 match req {
582 PendingRequest::CreateTarget(tx) => {
583 let _ = tx.send(Err(CdpError::Timeout));
584 }
585 PendingRequest::GetTargets(tx) => {
586 let _ = tx.send(Err(CdpError::Timeout));
587 }
588 PendingRequest::Navigate(nav) => {
589 if let Some(nav) = self.navigations.remove(&nav) {
590 match nav {
591 NavigationRequest::Navigate(nav) => {
592 let _ = nav.tx.send(Err(CdpError::Timeout));
593 }
594 }
595 }
596 }
597 PendingRequest::ExternalCommand(tx) => {
598 let _ = tx.send(Err(CdpError::Timeout));
599 }
600 PendingRequest::InternalCommand(_) => {}
601 PendingRequest::CloseBrowser(tx) => {
602 let _ = tx.send(Err(CdpError::Timeout));
603 }
604 }
605 }
606 }
607 }
608
609 pub fn event_listeners_mut(&mut self) -> &mut EventListeners {
610 &mut self.event_listeners
611 }
612
613 pub async fn run(mut self) -> Result<()> {
643 use chromiumoxide_types::Message;
644 use tokio::time::MissedTickBehavior;
645 use tokio_tungstenite::tungstenite::{self, error::ProtocolError};
646
647 let page_wake = Arc::new(Notify::new());
649 self.page_wake = Some(page_wake.clone());
650
651 let conn = self
653 .conn
654 .take()
655 .ok_or_else(|| CdpError::msg("Handler::run() called with no connection"))?;
656 let async_conn = conn.into_async();
657 let mut ws_reader = async_conn.reader;
658 let ws_tx = async_conn.cmd_tx;
659 let mut writer_handle = async_conn.writer_handle;
660 let mut next_call_id = async_conn.next_id;
661
662 let mut alloc_call_id = || {
664 let id = chromiumoxide_types::CallId::new(next_call_id);
665 next_call_id = next_call_id.wrapping_add(1);
666 id
667 };
668
669 let mut evict_timer = tokio::time::interval_at(
671 tokio::time::Instant::now() + self.config.request_timeout,
672 self.config.request_timeout,
673 );
674 evict_timer.set_missed_tick_behavior(MissedTickBehavior::Delay);
675
676 macro_rules! ws_submit {
678 ($method:expr, $session_id:expr, $params:expr) => {{
679 let id = alloc_call_id();
680 let call = chromiumoxide_types::MethodCall {
681 id,
682 method: $method,
683 session_id: $session_id,
684 params: $params,
685 };
686 match ws_tx.try_send(call) {
687 Ok(()) => Ok::<_, CdpError>(id),
688 Err(tokio::sync::mpsc::error::TrySendError::Full(_)) => {
689 tracing::warn!("WS command channel full — dropping command");
690 Err(CdpError::msg("WS command channel full"))
691 }
692 Err(tokio::sync::mpsc::error::TrySendError::Closed(_)) => {
693 Err(CdpError::msg("WS writer closed"))
694 }
695 }
696 }};
697 }
698
699 loop {
701 let now = std::time::Instant::now();
702
703 const PER_TARGET_DRAIN_BUDGET: usize = 128;
709
710 for n in (0..self.target_ids.len()).rev() {
711 let target_id = self.target_ids.swap_remove(n);
712
713 if let Some((id, mut target)) = self.targets.remove_entry(&target_id) {
714 {
716 let mut msgs = Vec::new();
717 if let Some(handle) = target.page_mut() {
718 while msgs.len() < PER_TARGET_DRAIN_BUDGET {
719 match handle.rx.try_recv() {
720 Ok(msg) => msgs.push(msg),
721 Err(_) => break,
722 }
723 }
724 }
725 for msg in msgs {
726 target.on_page_message(msg);
727 }
728 }
729
730 while let Some(event) = target.advance(now) {
732 match event {
733 TargetEvent::Request(req) => {
734 if let Ok(call_id) = ws_submit!(
735 req.method.clone(),
736 req.session_id.map(Into::into),
737 req.params
738 ) {
739 self.pending_commands.insert(
740 call_id,
741 (
742 PendingRequest::InternalCommand(
743 target.target_id().clone(),
744 ),
745 req.method,
746 now,
747 ),
748 );
749 }
750 }
751 TargetEvent::Command(msg) => {
752 if msg.is_navigation() {
753 let (req, tx) = msg.split();
754 let nav_id = self.next_navigation_id();
755 target.goto(FrameRequestedNavigation::new(
756 nav_id,
757 req.clone(),
758 self.config.request_timeout,
759 ));
760 if let Ok(call_id) = ws_submit!(
761 req.method.clone(),
762 req.session_id.map(Into::into),
763 req.params
764 ) {
765 self.pending_commands.insert(
766 call_id,
767 (PendingRequest::Navigate(nav_id), req.method, now),
768 );
769 }
770 self.navigations.insert(
771 nav_id,
772 NavigationRequest::Navigate(NavigationInProgress::new(tx)),
773 );
774 } else {
775 if let Ok(call_id) = ws_submit!(
776 msg.method.clone(),
777 msg.session_id.map(Into::into),
778 msg.params
779 ) {
780 self.pending_commands.insert(
781 call_id,
782 (
783 PendingRequest::ExternalCommand(msg.sender),
784 msg.method,
785 now,
786 ),
787 );
788 }
789 }
790 }
791 TargetEvent::NavigationRequest(nav_id, req) => {
792 if let Ok(call_id) = ws_submit!(
793 req.method.clone(),
794 req.session_id.map(Into::into),
795 req.params
796 ) {
797 self.pending_commands.insert(
798 call_id,
799 (PendingRequest::Navigate(nav_id), req.method, now),
800 );
801 }
802 }
803 TargetEvent::NavigationResult(res) => {
804 self.on_navigation_lifecycle_completed(res);
805 }
806 TargetEvent::BytesConsumed(n) => {
807 if let Some(rem) = self.remaining_bytes.as_mut() {
808 *rem = rem.saturating_sub(n);
809 if *rem == 0 {
810 self.budget_exhausted = true;
811 }
812 }
813 }
814 }
815 }
816
817 target.event_listeners_mut().flush();
819
820 self.targets.insert(id, target);
821 self.target_ids.push(target_id);
822 }
823 }
824
825 self.event_listeners.flush();
827
828 if self.budget_exhausted {
829 for t in self.targets.values_mut() {
830 t.network_manager.set_block_all(true);
831 }
832 }
833
834 if self.closing {
835 break;
836 }
837
838 tokio::select! {
840 msg = self.from_browser.recv() => {
841 match msg {
842 Some(msg) => {
843 match msg {
844 HandlerMessage::Command(cmd) => {
845 if let Ok(call_id) = ws_submit!(
846 cmd.method.clone(),
847 cmd.session_id.map(Into::into),
848 cmd.params
849 ) {
850 self.pending_commands.insert(
851 call_id,
852 (PendingRequest::ExternalCommand(cmd.sender), cmd.method, now),
853 );
854 }
855 }
856 HandlerMessage::FetchTargets(tx) => {
857 let msg = TARGET_PARAMS_ID.clone();
858 if let Ok(call_id) = ws_submit!(msg.0.clone(), None, msg.1) {
859 self.pending_commands.insert(
860 call_id,
861 (PendingRequest::GetTargets(tx), msg.0, now),
862 );
863 }
864 }
865 HandlerMessage::CloseBrowser(tx) => {
866 let close_msg = CLOSE_PARAMS_ID.clone();
867 if let Ok(call_id) = ws_submit!(close_msg.0.clone(), None, close_msg.1) {
868 self.pending_commands.insert(
869 call_id,
870 (PendingRequest::CloseBrowser(tx), close_msg.0, now),
871 );
872 }
873 }
874 HandlerMessage::CreatePage(params, tx) => {
875 if let Some(ref id) = params.browser_context_id {
876 self.browser_contexts.insert(BrowserContext::from(id.clone()));
877 }
878 self.create_page_async(params, tx, &mut alloc_call_id, &ws_tx, now);
879 }
880 HandlerMessage::GetPages(tx) => {
881 let pages: Vec<_> = self.targets.values_mut()
882 .filter(|p| p.is_page())
883 .filter_map(|target| target.get_or_create_page())
884 .map(|page| Page::from(page.clone()))
885 .collect();
886 let _ = tx.send(pages);
887 }
888 HandlerMessage::InsertContext(ctx) => {
889 if self.default_browser_context.id().is_none() {
890 self.default_browser_context = ctx.clone();
891 }
892 self.browser_contexts.insert(ctx);
893 }
894 HandlerMessage::DisposeContext(ctx) => {
895 self.browser_contexts.remove(&ctx);
896 self.attached_targets.retain(|tid| {
897 self.targets.get(tid)
898 .and_then(|t| t.browser_context_id())
899 .map(|id| Some(id) != ctx.id())
900 .unwrap_or(true)
901 });
902 self.closing = true;
903 }
904 HandlerMessage::GetPage(target_id, tx) => {
905 let page = self.targets.get_mut(&target_id)
906 .and_then(|target| target.get_or_create_page())
907 .map(|page| Page::from(page.clone()));
908 let _ = tx.send(page);
909 }
910 HandlerMessage::AddEventListener(req) => {
911 self.event_listeners.add_listener(req);
912 }
913 }
914 }
915 None => break, }
917 }
918
919 frame = ws_reader.next_message() => {
920 match frame {
921 Some(Ok(boxed_msg)) => match *boxed_msg {
922 Message::Response(resp) => {
923 self.on_response(resp);
924 }
925 Message::Event(ev) => {
926 self.on_event(ev);
927 }
928 },
929 Some(Err(err)) => {
930 tracing::error!("WS Connection error: {:?}", err);
931 if let CdpError::Ws(ref ws_error) = err {
932 match ws_error {
933 tungstenite::Error::AlreadyClosed => break,
934 tungstenite::Error::Protocol(detail)
935 if detail == &ProtocolError::ResetWithoutClosingHandshake =>
936 {
937 break;
938 }
939 _ => return Err(err),
940 }
941 } else {
942 return Err(err);
943 }
944 }
945 None => break, }
947 }
948
949 _ = page_wake.notified() => {
950 }
952
953 _ = evict_timer.tick() => {
954 self.evict_timed_out_commands(now);
955 for t in self.targets.values_mut() {
956 t.network_manager.evict_stale_entries(now);
957 t.frame_manager_mut().evict_stale_context_ids();
958 }
959 }
960
961 result = &mut writer_handle => {
962 match result {
964 Ok(Ok(())) => break,
965 Ok(Err(e)) => return Err(e),
966 Err(e) => return Err(CdpError::msg(format!("WS writer panicked: {e}"))),
967 }
968 }
969 }
970 }
971
972 writer_handle.abort();
973 Ok(())
974 }
975
976 fn create_page_async(
978 &mut self,
979 params: CreateTargetParams,
980 tx: OneshotSender<Result<Page>>,
981 alloc_call_id: &mut impl FnMut() -> chromiumoxide_types::CallId,
982 ws_tx: &tokio::sync::mpsc::Sender<chromiumoxide_types::MethodCall>,
983 now: std::time::Instant,
984 ) {
985 let about_blank = params.url == "about:blank";
986 let http_check =
987 !about_blank && params.url.starts_with("http") || params.url.starts_with("file://");
988
989 if about_blank || http_check {
990 let method = params.identifier();
991 match serde_json::to_value(params) {
992 Ok(params) => {
993 let id = alloc_call_id();
994 let call = chromiumoxide_types::MethodCall {
995 id,
996 method: method.clone(),
997 session_id: None,
998 params,
999 };
1000 match ws_tx.try_send(call) {
1001 Ok(()) => {
1002 self.pending_commands
1003 .insert(id, (PendingRequest::CreateTarget(tx), method, now));
1004 }
1005 Err(_) => {
1006 let _ = tx
1007 .send(Err(CdpError::msg("WS command channel full or closed")))
1008 .ok();
1009 }
1010 }
1011 }
1012 Err(err) => {
1013 let _ = tx.send(Err(err.into())).ok();
1014 }
1015 }
1016 } else {
1017 let _ = tx.send(Err(CdpError::NotFound)).ok();
1018 }
1019 }
1020}
1021
1022impl Stream for Handler {
1023 type Item = Result<()>;
1024
1025 fn poll_next(self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Option<Self::Item>> {
1026 let pin = self.get_mut();
1027
1028 let mut dispose = false;
1029
1030 let now = Instant::now();
1031
1032 loop {
1033 while let Poll::Ready(Some(msg)) = pin.from_browser.poll_recv(cx) {
1037 match msg {
1038 HandlerMessage::Command(cmd) => {
1039 pin.submit_external_command(cmd, now)?;
1040 }
1041 HandlerMessage::FetchTargets(tx) => {
1042 pin.submit_fetch_targets(tx, now);
1043 }
1044 HandlerMessage::CloseBrowser(tx) => {
1045 pin.submit_close(tx, now);
1046 }
1047 HandlerMessage::CreatePage(params, tx) => {
1048 if let Some(ref id) = params.browser_context_id {
1049 pin.browser_contexts
1050 .insert(BrowserContext::from(id.clone()));
1051 }
1052 pin.create_page(params, tx);
1053 }
1054 HandlerMessage::GetPages(tx) => {
1055 let pages: Vec<_> = pin
1056 .targets
1057 .values_mut()
1058 .filter(|p: &&mut Target| p.is_page())
1059 .filter_map(|target| target.get_or_create_page())
1060 .map(|page| Page::from(page.clone()))
1061 .collect();
1062 let _ = tx.send(pages);
1063 }
1064 HandlerMessage::InsertContext(ctx) => {
1065 if pin.default_browser_context.id().is_none() {
1066 pin.default_browser_context = ctx.clone();
1067 }
1068 pin.browser_contexts.insert(ctx);
1069 }
1070 HandlerMessage::DisposeContext(ctx) => {
1071 pin.browser_contexts.remove(&ctx);
1072 pin.attached_targets.retain(|tid| {
1073 pin.targets
1074 .get(tid)
1075 .and_then(|t| t.browser_context_id()) .map(|id| Some(id) != ctx.id())
1077 .unwrap_or(true)
1078 });
1079 pin.closing = true;
1080 dispose = true;
1081 }
1082 HandlerMessage::GetPage(target_id, tx) => {
1083 let page = pin
1084 .targets
1085 .get_mut(&target_id)
1086 .and_then(|target| target.get_or_create_page())
1087 .map(|page| Page::from(page.clone()));
1088 let _ = tx.send(page);
1089 }
1090 HandlerMessage::AddEventListener(req) => {
1091 pin.event_listeners.add_listener(req);
1092 }
1093 }
1094 }
1095
1096 for n in (0..pin.target_ids.len()).rev() {
1097 let target_id = pin.target_ids.swap_remove(n);
1098
1099 if let Some((id, mut target)) = pin.targets.remove_entry(&target_id) {
1100 while let Some(event) = target.poll(cx, now) {
1101 match event {
1102 TargetEvent::Request(req) => {
1103 let _ = pin.submit_internal_command(
1104 target.target_id().clone(),
1105 req,
1106 now,
1107 );
1108 }
1109 TargetEvent::Command(msg) => {
1110 pin.on_target_message(&mut target, msg, now);
1111 }
1112 TargetEvent::NavigationRequest(id, req) => {
1113 pin.submit_navigation(id, req, now);
1114 }
1115 TargetEvent::NavigationResult(res) => {
1116 pin.on_navigation_lifecycle_completed(res)
1117 }
1118 TargetEvent::BytesConsumed(n) => {
1119 if let Some(rem) = pin.remaining_bytes.as_mut() {
1120 *rem = rem.saturating_sub(n);
1121 if *rem == 0 {
1122 pin.budget_exhausted = true;
1123 }
1124 }
1125 }
1126 }
1127 }
1128
1129 target.event_listeners_mut().poll(cx);
1131
1132 pin.targets.insert(id, target);
1133 pin.target_ids.push(target_id);
1134 }
1135 }
1136
1137 pin.event_listeners_mut().poll(cx);
1140
1141 let mut done = true;
1142
1143 let mut ws_msgs = Vec::new();
1146 let mut ws_err = None;
1147 {
1148 let Some(conn) = pin.conn.as_mut() else {
1149 return Poll::Ready(Some(Err(CdpError::msg(
1150 "connection consumed by Handler::run()",
1151 ))));
1152 };
1153 while let Poll::Ready(Some(ev)) = Pin::new(&mut *conn).poll_next(cx) {
1154 match ev {
1155 Ok(msg) => ws_msgs.push(msg),
1156 Err(err) => {
1157 ws_err = Some(err);
1158 break;
1159 }
1160 }
1161 }
1162 }
1163
1164 for boxed_msg in ws_msgs {
1165 match *boxed_msg {
1166 Message::Response(resp) => {
1167 pin.on_response(resp);
1168 if pin.closing {
1169 return Poll::Ready(None);
1170 }
1171 }
1172 Message::Event(ev) => {
1173 pin.on_event(ev);
1174 }
1175 }
1176 done = false;
1177 }
1178
1179 if let Some(err) = ws_err {
1180 tracing::error!("WS Connection error: {:?}", err);
1181 if let CdpError::Ws(ref ws_error) = err {
1182 match ws_error {
1183 Error::AlreadyClosed => {
1184 pin.closing = true;
1185 dispose = true;
1186 }
1187 Error::Protocol(detail)
1188 if detail == &ProtocolError::ResetWithoutClosingHandshake =>
1189 {
1190 pin.closing = true;
1191 dispose = true;
1192 }
1193 _ => return Poll::Ready(Some(Err(err))),
1194 }
1195 } else {
1196 return Poll::Ready(Some(Err(err)));
1197 }
1198 }
1199
1200 if pin.evict_command_timeout.poll_ready(cx) {
1201 pin.evict_timed_out_commands(now);
1203 for t in pin.targets.values_mut() {
1206 t.network_manager.evict_stale_entries(now);
1207 t.frame_manager_mut().evict_stale_context_ids();
1208 }
1209 }
1210
1211 if pin.budget_exhausted {
1212 for t in pin.targets.values_mut() {
1213 t.network_manager.set_block_all(true);
1214 }
1215 }
1216
1217 if dispose {
1218 return Poll::Ready(None);
1219 }
1220
1221 if done {
1222 return Poll::Pending;
1224 }
1225 }
1226 }
1227}
1228
1229#[derive(Debug, Clone)]
1231pub struct HandlerConfig {
1232 pub ignore_https_errors: bool,
1234 pub viewport: Option<Viewport>,
1236 pub context_ids: Vec<BrowserContextId>,
1238 pub request_timeout: Duration,
1240 pub request_intercept: bool,
1242 pub cache_enabled: bool,
1244 pub service_worker_enabled: bool,
1246 pub ignore_visuals: bool,
1248 pub ignore_stylesheets: bool,
1250 pub ignore_javascript: bool,
1252 pub ignore_analytics: bool,
1254 pub ignore_prefetch: bool,
1256 pub ignore_ads: bool,
1258 pub extra_headers: Option<std::collections::HashMap<String, String>>,
1260 pub only_html: bool,
1262 pub created_first_target: bool,
1264 pub intercept_manager: NetworkInterceptManager,
1266 pub max_bytes_allowed: Option<u64>,
1268 pub whitelist_patterns: Option<Vec<String>>,
1270 pub blacklist_patterns: Option<Vec<String>>,
1272 #[cfg(feature = "adblock")]
1274 pub adblock_filter_rules: Option<Vec<String>>,
1275 pub channel_capacity: usize,
1278 pub connection_retries: u32,
1281}
1282
1283impl Default for HandlerConfig {
1284 fn default() -> Self {
1285 Self {
1286 ignore_https_errors: true,
1287 viewport: Default::default(),
1288 context_ids: Vec::new(),
1289 request_timeout: Duration::from_millis(REQUEST_TIMEOUT),
1290 request_intercept: false,
1291 cache_enabled: true,
1292 service_worker_enabled: true,
1293 ignore_visuals: false,
1294 ignore_stylesheets: false,
1295 ignore_ads: false,
1296 ignore_javascript: false,
1297 ignore_analytics: true,
1298 ignore_prefetch: true,
1299 only_html: false,
1300 extra_headers: Default::default(),
1301 created_first_target: false,
1302 intercept_manager: NetworkInterceptManager::Unknown,
1303 max_bytes_allowed: None,
1304 whitelist_patterns: None,
1305 blacklist_patterns: None,
1306 #[cfg(feature = "adblock")]
1307 adblock_filter_rules: None,
1308 channel_capacity: 4096,
1309 connection_retries: crate::conn::DEFAULT_CONNECTION_RETRIES,
1310 }
1311 }
1312}
1313
1314#[derive(Debug)]
1316pub struct NavigationInProgress<T> {
1317 navigated: bool,
1319 response: Option<Response>,
1321 tx: OneshotSender<T>,
1323}
1324
1325impl<T> NavigationInProgress<T> {
1326 fn new(tx: OneshotSender<T>) -> Self {
1327 Self {
1328 navigated: false,
1329 response: None,
1330 tx,
1331 }
1332 }
1333
1334 fn set_response(&mut self, resp: Response) {
1336 self.response = Some(resp);
1337 }
1338
1339 fn set_navigated(&mut self) {
1341 self.navigated = true;
1342 }
1343}
1344
1345#[derive(Debug)]
1347enum NavigationRequest {
1348 Navigate(NavigationInProgress<Result<Response>>),
1350 }
1352
1353#[derive(Debug)]
1356enum PendingRequest {
1357 CreateTarget(OneshotSender<Result<Page>>),
1360 GetTargets(OneshotSender<Result<Vec<TargetInfo>>>),
1362 Navigate(NavigationId),
1369 ExternalCommand(OneshotSender<Result<Response>>),
1371 InternalCommand(TargetId),
1374 CloseBrowser(OneshotSender<Result<CloseReturns>>),
1376}
1377
1378#[derive(Debug)]
1382pub(crate) enum HandlerMessage {
1383 CreatePage(CreateTargetParams, OneshotSender<Result<Page>>),
1384 FetchTargets(OneshotSender<Result<Vec<TargetInfo>>>),
1385 InsertContext(BrowserContext),
1386 DisposeContext(BrowserContext),
1387 GetPages(OneshotSender<Vec<Page>>),
1388 Command(CommandMessage),
1389 GetPage(TargetId, OneshotSender<Option<Page>>),
1390 AddEventListener(EventListenerRequest),
1391 CloseBrowser(OneshotSender<Result<CloseReturns>>),
1392}
1393
1394#[cfg(test)]
1395mod tests {
1396 use super::*;
1397 use chromiumoxide_cdp::cdp::browser_protocol::target::{AttachToTargetReturns, TargetInfo};
1398
1399 #[test]
1400 fn attach_to_target_response_sets_session_id_before_event_arrives() {
1401 let info = TargetInfo::builder()
1402 .target_id("target-1".to_string())
1403 .r#type("page")
1404 .title("")
1405 .url("about:blank")
1406 .attached(false)
1407 .can_access_opener(false)
1408 .build()
1409 .expect("target info");
1410 let mut target = Target::new(info, TargetConfig::default(), BrowserContext::default());
1411 let method: MethodId = AttachToTargetParams::IDENTIFIER.into();
1412 let result = serde_json::to_value(AttachToTargetReturns::new("session-1".to_string()))
1413 .expect("attach result");
1414 let resp = Response {
1415 id: CallId::new(1),
1416 result: Some(result),
1417 error: None,
1418 };
1419
1420 maybe_store_attach_session_id(&mut target, &method, &resp);
1421
1422 assert_eq!(
1423 target.session_id().map(AsRef::as_ref),
1424 Some("session-1"),
1425 "attach response should seed the flat session id even before Target.attachedToTarget"
1426 );
1427 }
1428}