Skip to main content

chromiumoxide/handler/
target.rs

1use std::collections::VecDeque;
2use std::sync::Arc;
3use std::time::Instant;
4
5use chromiumoxide_cdp::cdp::browser_protocol::target::DetachFromTargetParams;
6use std::task::{Context, Poll};
7use tokio::sync::oneshot::Sender;
8
9use tokio::sync::Notify;
10
11use crate::auth::Credentials;
12use crate::cdp::browser_protocol::target::CloseTargetParams;
13use crate::cmd::CommandChain;
14use crate::cmd::CommandMessage;
15use crate::error::{CdpError, Result};
16use crate::handler::browser::BrowserContext;
17use crate::handler::domworld::DOMWorldKind;
18use crate::handler::emulation::EmulationManager;
19use crate::handler::frame::FrameRequestedNavigation;
20use crate::handler::frame::{
21    FrameEvent, FrameManager, NavigationError, NavigationId, NavigationOk,
22};
23use crate::handler::network::{NetworkEvent, NetworkManager};
24use crate::handler::page::PageHandle;
25use crate::handler::viewport::Viewport;
26use crate::handler::{PageInner, REQUEST_TIMEOUT};
27use crate::listeners::{EventListenerRequest, EventListeners};
28use crate::{page::Page, ArcHttpRequest};
29use chromiumoxide_cdp::cdp::browser_protocol::{
30    browser::BrowserContextId,
31    log as cdplog,
32    page::{FrameId, GetFrameTreeParams},
33    target::{AttachToTargetParams, SessionId, SetAutoAttachParams, TargetId, TargetInfo},
34};
35use chromiumoxide_cdp::cdp::events::CdpEvent;
36use chromiumoxide_cdp::cdp::js_protocol::runtime::{
37    ExecutionContextId, RunIfWaitingForDebuggerParams,
38};
39use chromiumoxide_cdp::cdp::CdpEventMessage;
40use chromiumoxide_types::{Command, Method, Request, Response};
41use spider_network_blocker::intercept_manager::NetworkInterceptManager;
42use std::time::Duration;
43
44macro_rules! advance_state {
45    ($s:ident, $cx:ident, $now:ident, $cmds: ident, $next_state:expr ) => {{
46        if let Poll::Ready(poll) = $cmds.poll($now) {
47            return match poll {
48                None => {
49                    $s.init_state = $next_state;
50                    $s.poll($cx, $now)
51                }
52                Some(Ok((method, params))) => Some(TargetEvent::Request(Request {
53                    method,
54                    session_id: $s.session_id.clone().map(Into::into),
55                    params,
56                })),
57                Some(Err(_)) => Some($s.on_initialization_failed()),
58            };
59        } else {
60            return None;
61        }
62    }};
63}
64
65lazy_static::lazy_static! {
66    /// Initial start command params.
67    static ref INIT_COMMANDS_PARAMS: Vec<(chromiumoxide_types::MethodId, serde_json::Value)> = {
68        if let Ok(attach) = SetAutoAttachParams::builder()
69            .flatten(true)
70            .auto_attach(true)
71            .wait_for_debugger_on_start(true)
72            .build() {
73                let disable_log = cdplog::DisableParams::default();
74
75                let mut cmds =  vec![
76                    (
77                        attach.identifier(),
78                        serde_json::to_value(attach).unwrap_or_default(),
79                    ),
80                    (
81                        disable_log.identifier(),
82                        serde_json::to_value(disable_log).unwrap_or_default(),
83                    )
84                ];
85
86                // enable performance on pages.
87                if cfg!(feature = "collect_metrics") {
88                    let enable_performance = chromiumoxide_cdp::cdp::browser_protocol::performance::EnableParams::default();
89                    cmds.push((
90                        enable_performance.identifier(),
91                        serde_json::to_value(enable_performance).unwrap_or_default(),
92                    ));
93                }
94
95                cmds
96            } else {
97                vec![]
98            }
99    };
100
101    /// Attach to target commands
102    static ref ATTACH_TARGET: (chromiumoxide_types::MethodId, serde_json::Value) = {
103        let runtime_cmd = RunIfWaitingForDebuggerParams::default();
104
105        (runtime_cmd.identifier(), serde_json::to_value(runtime_cmd).unwrap_or_default())
106    };
107}
108
109/// Per-queue cap on waiter sends per `Target::poll` call.
110///
111/// Each `wait_for_*` queue can hold an unbounded number of `oneshot::Sender`s
112/// registered by concurrent callers. Firing them all in one tight `pop()`
113/// loop previously produced multi-hundred-microsecond synchronous bursts
114/// inside the handler's event loop under fan-out (e.g. 1000 tasks awaiting
115/// `wait_for_load` on one page). Capping at 64 per queue per poll keeps
116/// worst-case burst at ~5 × 64 oneshot sends (~6μs) before yielding. Any
117/// remainder is drained on subsequent polls, re-armed via `Waker::wake_by_ref`.
118const WAITER_DRAIN_BUDGET: usize = 64;
119
120/// Pop up to `budget` senders from `queue` and deliver `value` to each.
121///
122/// Returns `true` when the queue still contains senders after draining.
123/// Dropped receivers (closed senders) are silently ignored — they consume
124/// a budget slot but contribute no cost beyond the cheap `send` no-op.
125///
126/// The queue is pruned of closed senders elsewhere once per `Target::poll`
127/// (before this helper runs), so in steady state `budget` slots approximate
128/// `budget` live fan-out sends.
129#[inline]
130fn drain_waiters_bounded(
131    queue: &mut Vec<Sender<ArcHttpRequest>>,
132    http_request: Option<&Arc<crate::handler::http::HttpRequest>>,
133    budget: usize,
134) -> bool {
135    let to_fire = queue.len().min(budget);
136    for _ in 0..to_fire {
137        // `pop` cannot be `None` here: `to_fire <= queue.len()`.
138        if let Some(tx) = queue.pop() {
139            let _ = tx.send(http_request.cloned());
140        }
141    }
142    !queue.is_empty()
143}
144
145#[derive(Debug)]
146pub struct Target {
147    /// Info about this target as returned from the chromium instance
148    info: TargetInfo,
149    /// The type of this target
150    r#type: TargetType,
151    /// Configs for this target
152    config: TargetConfig,
153    /// The context this target is running in
154    browser_context: BrowserContext,
155    /// The frame manager that maintains the state of all frames and handles
156    /// navigations of frames
157    frame_manager: FrameManager,
158    /// Handles all the https
159    pub(crate) network_manager: NetworkManager,
160    emulation_manager: EmulationManager,
161    /// The identifier of the session this target is attached to
162    session_id: Option<SessionId>,
163    /// The handle of the browser page of this target
164    page: Option<PageHandle>,
165    /// Drives this target towards initialization
166    pub(crate) init_state: TargetInit,
167    /// Currently queued events to report to the `Handler`
168    queued_events: VecDeque<TargetEvent>,
169    /// All registered event subscriptions
170    event_listeners: EventListeners,
171    /// Senders that need to be notified once the main frame has loaded
172    wait_for_frame_navigation: Vec<Sender<ArcHttpRequest>>,
173    /// Senders notified once `DOMContentLoaded` fires (before `load`).
174    wait_for_dom_content_loaded: Vec<Sender<ArcHttpRequest>>,
175    /// Senders notified once the `load` event fires (all subresources done).
176    wait_for_load: Vec<Sender<ArcHttpRequest>>,
177    /// Senders that need to be notified once the main frame reaches `networkIdle`.
178    wait_for_network_idle: Vec<Sender<ArcHttpRequest>>,
179    /// (Optional) for `networkAlmostIdle` if you want it as well.
180    wait_for_network_almost_idle: Vec<Sender<ArcHttpRequest>>,
181    /// The sender who requested the page.
182    initiator: Option<Sender<Result<Page>>>,
183}
184
185impl Target {
186    /// Create a new target instance with `TargetInfo` after a
187    /// `CreateTargetParams` request.
188    pub fn new(info: TargetInfo, config: TargetConfig, browser_context: BrowserContext) -> Self {
189        let ty = TargetType::new(&info.r#type);
190        let request_timeout: Duration = config.request_timeout;
191        let mut network_manager = NetworkManager::new(config.ignore_https_errors, request_timeout);
192
193        if !config.cache_enabled {
194            network_manager.set_cache_enabled(false);
195        }
196
197        if !config.service_worker_enabled {
198            network_manager.set_service_worker_enabled(true);
199        }
200
201        network_manager.set_request_interception(config.request_intercept);
202        network_manager.max_bytes_allowed = config.max_bytes_allowed;
203        network_manager.max_redirects = config.max_redirects;
204
205        if let Some(headers) = &config.extra_headers {
206            network_manager.set_extra_headers(headers.clone());
207        }
208
209        if let Some(whitelist) = &config.whitelist_patterns {
210            network_manager.set_whitelist_patterns(whitelist.clone());
211        }
212
213        if let Some(blacklist) = &config.blacklist_patterns {
214            network_manager.set_blacklist_patterns(blacklist);
215        }
216
217        network_manager.ignore_visuals = config.ignore_visuals;
218        network_manager.block_javascript = config.ignore_javascript;
219        network_manager.block_analytics = config.ignore_analytics;
220        network_manager.block_prefetch = config.ignore_prefetch;
221
222        network_manager.block_stylesheets = config.ignore_stylesheets;
223        network_manager.allow_first_party_stylesheets = config.allow_first_party_stylesheets;
224        network_manager.allow_first_party_javascript = config.allow_first_party_javascript;
225        network_manager.allow_first_party_visuals = config.allow_first_party_visuals;
226        network_manager.only_html = config.only_html;
227        network_manager.intercept_manager = config.intercept_manager;
228        network_manager.set_remote_local_policy(config.remote_local_policy);
229
230        // Push the one-shot interception policy now that every field above is
231        // configured (must run AFTER the field setup — `set_request_interception`
232        // above only queued `Fetch.enable`). No-op unless opted in.
233        network_manager.emit_request_policy();
234
235        #[cfg(feature = "adblock")]
236        if let Some(rules) = &config.adblock_filter_rules {
237            use adblock::lists::{FilterSet, ParseOptions, RuleTypes};
238
239            let mut filter_set = FilterSet::new(false);
240            let mut opts = ParseOptions::default();
241            opts.rule_types = RuleTypes::All;
242
243            // Include built-in patterns.
244            filter_set.add_filters(&*spider_network_blocker::adblock::ADBLOCK_PATTERNS, opts);
245            // Merge user-supplied rules (e.g. EasyList / EasyPrivacy content).
246            filter_set.add_filters(rules.iter().map(|s| s.as_str()), opts);
247
248            let engine = adblock::Engine::from_filter_set(filter_set, true);
249            network_manager.set_adblock_engine(std::sync::Arc::new(engine));
250        }
251
252        let mut frame_manager = FrameManager::new(request_timeout);
253        frame_manager.set_max_main_frame_navigations(config.max_main_frame_navigations);
254
255        Self {
256            info,
257            r#type: ty,
258            config,
259            frame_manager,
260            network_manager,
261            emulation_manager: EmulationManager::new(request_timeout),
262            session_id: None,
263            page: None,
264            init_state: TargetInit::AttachToTarget,
265            wait_for_frame_navigation: Default::default(),
266            wait_for_dom_content_loaded: Default::default(),
267            wait_for_load: Default::default(),
268            wait_for_network_idle: Default::default(),
269            wait_for_network_almost_idle: Default::default(),
270            queued_events: Default::default(),
271            event_listeners: Default::default(),
272            initiator: None,
273            browser_context,
274        }
275    }
276
277    /// Set the session id.
278    pub fn set_session_id(&mut self, id: SessionId) {
279        self.session_id = Some(id)
280    }
281
282    /// Get the session id.
283    pub fn session_id(&self) -> Option<&SessionId> {
284        self.session_id.as_ref()
285    }
286
287    /// Get the session id mut.
288    pub fn session_id_mut(&mut self) -> &mut Option<SessionId> {
289        &mut self.session_id
290    }
291
292    /// Get the browser context.
293    pub fn browser_context(&self) -> &BrowserContext {
294        &self.browser_context
295    }
296
297    /// The identifier for this target
298    pub fn target_id(&self) -> &TargetId {
299        &self.info.target_id
300    }
301
302    /// The type of this target
303    pub fn r#type(&self) -> &TargetType {
304        &self.r#type
305    }
306
307    /// Whether this target is already initialized
308    pub fn is_initialized(&self) -> bool {
309        matches!(self.init_state, TargetInit::Initialized)
310    }
311
312    /// Navigate a frame
313    pub fn goto(&mut self, req: FrameRequestedNavigation) {
314        if self.network_manager.has_target_domain() {
315            self.network_manager.clear_target_domain();
316            let goto_url = req
317                .req
318                .params
319                .as_object()
320                .and_then(|o| o.get("url"))
321                .and_then(|v| v.as_str());
322
323            if let Some(url) = goto_url {
324                self.network_manager.set_page_url(url.into());
325            }
326        }
327        self.frame_manager.goto(req);
328    }
329
330    /// Create a new page from the session.
331    fn create_page(&mut self) {
332        if self.page.is_none() {
333            if let Some(session) = self.session_id.clone() {
334                let handle = PageHandle::with_capacity(
335                    self.target_id().clone(),
336                    session,
337                    self.opener_id().cloned(),
338                    self.config.request_timeout,
339                    self.config.page_wake.clone(),
340                    self.config.page_channel_capacity,
341                );
342                self.page = Some(handle);
343            }
344        }
345    }
346
347    /// Tries to create the `PageInner` if this target is already initialized
348    pub(crate) fn get_or_create_page(&mut self) -> Option<&Arc<PageInner>> {
349        self.create_page();
350        self.page.as_ref().map(|p| p.inner())
351    }
352
353    /// Mutable access to the page handle (for `try_recv` in `Handler::run()`).
354    pub(crate) fn page_mut(&mut self) -> Option<&mut PageHandle> {
355        self.page.as_mut()
356    }
357
358    /// Is the target a page?
359    pub fn is_page(&self) -> bool {
360        self.r#type().is_page()
361    }
362
363    /// The browser context ID.
364    pub fn browser_context_id(&self) -> Option<&BrowserContextId> {
365        self.info.browser_context_id.as_ref()
366    }
367
368    /// The target connection info.
369    pub fn info(&self) -> &TargetInfo {
370        &self.info
371    }
372
373    /// Get the target that opened this target. Top-level targets return `None`.
374    pub fn opener_id(&self) -> Option<&TargetId> {
375        self.info.opener_id.as_ref()
376    }
377
378    pub fn frame_manager(&self) -> &FrameManager {
379        &self.frame_manager
380    }
381
382    /// The frame manager.
383    pub fn frame_manager_mut(&mut self) -> &mut FrameManager {
384        &mut self.frame_manager
385    }
386
387    /// Get event listeners mutably.
388    pub fn event_listeners_mut(&mut self) -> &mut EventListeners {
389        &mut self.event_listeners
390    }
391
392    /// Received a response to a command issued by this target
393    pub fn on_response(&mut self, resp: Response, method: &str) {
394        if let Some(cmds) = self.init_state.commands_mut() {
395            cmds.received_response(method);
396        }
397
398        if let GetFrameTreeParams::IDENTIFIER = method {
399            if let Some(resp) = resp
400                .result
401                .and_then(|val| GetFrameTreeParams::response_from_value(val).ok())
402            {
403                self.frame_manager.on_frame_tree(resp.frame_tree);
404            }
405        }
406        // requests originated from the network manager all return an empty response, hence they
407        // can be ignored here
408    }
409
410    /// On CDP Event message.
411    pub fn on_event(&mut self, event: CdpEventMessage) {
412        let CdpEventMessage {
413            params,
414            method,
415            session_id,
416            ..
417        } = event;
418
419        let is_session_scoped = matches!(
420            params,
421            CdpEvent::FetchRequestPaused(_)
422                | CdpEvent::FetchAuthRequired(_)
423                | CdpEvent::NetworkRequestWillBeSent(_)
424                | CdpEvent::NetworkResponseReceived(_)
425                | CdpEvent::NetworkLoadingFinished(_)
426                | CdpEvent::NetworkLoadingFailed(_)
427                | CdpEvent::PageFrameAttached(_)
428                | CdpEvent::PageFrameDetached(_)
429                | CdpEvent::PageFrameNavigated(_)
430                | CdpEvent::PageNavigatedWithinDocument(_)
431                | CdpEvent::PageLifecycleEvent(_)
432                | CdpEvent::PageFrameStartedLoading(_)
433                | CdpEvent::PageFrameStoppedLoading(_)
434                | CdpEvent::RuntimeExecutionContextCreated(_)
435                | CdpEvent::RuntimeExecutionContextDestroyed(_)
436                | CdpEvent::RuntimeExecutionContextsCleared(_)
437                | CdpEvent::RuntimeBindingCalled(_)
438        );
439
440        if is_session_scoped {
441            let ev_sid: &str = match session_id.as_deref() {
442                Some(s) => s,
443                None => return,
444            };
445
446            let self_sid: &str = match self.session_id.as_ref() {
447                Some(sid) => sid.as_ref(),
448                None => return,
449            };
450
451            if self_sid != ev_sid {
452                return;
453            }
454        }
455
456        match &params {
457            // `FrameManager` events
458            CdpEvent::PageFrameAttached(ev) => self
459                .frame_manager
460                .on_frame_attached(ev.frame_id.clone(), Some(ev.parent_frame_id.clone())),
461            CdpEvent::PageFrameDetached(ev) => self.frame_manager.on_frame_detached(ev),
462            CdpEvent::PageFrameNavigated(ev) => {
463                self.frame_manager.on_frame_navigated(&ev.frame);
464            }
465            CdpEvent::PageNavigatedWithinDocument(ev) => {
466                self.frame_manager.on_frame_navigated_within_document(ev)
467            }
468            CdpEvent::RuntimeExecutionContextCreated(ev) => {
469                self.frame_manager.on_frame_execution_context_created(ev)
470            }
471            CdpEvent::RuntimeExecutionContextDestroyed(ev) => {
472                self.frame_manager.on_frame_execution_context_destroyed(ev)
473            }
474            CdpEvent::RuntimeExecutionContextsCleared(_) => {
475                self.frame_manager.on_execution_contexts_cleared()
476            }
477            CdpEvent::RuntimeBindingCalled(ev) => {
478                // TODO check if binding registered and payload is json
479                self.frame_manager.on_runtime_binding_called(ev)
480            }
481            CdpEvent::PageLifecycleEvent(ev) => self.frame_manager.on_page_lifecycle_event(ev),
482            CdpEvent::PageFrameStartedLoading(ev) => {
483                self.frame_manager.on_frame_started_loading(ev);
484            }
485            CdpEvent::PageFrameStoppedLoading(ev) => {
486                self.frame_manager.on_frame_stopped_loading(ev);
487            }
488            // `Target` events
489            CdpEvent::TargetAttachedToTarget(ev) => {
490                if ev.waiting_for_debugger {
491                    let runtime_cmd = ATTACH_TARGET.clone();
492
493                    self.queued_events.push_back(TargetEvent::Request(Request {
494                        method: runtime_cmd.0,
495                        session_id: Some(ev.session_id.clone().into()),
496                        params: runtime_cmd.1,
497                    }));
498                }
499
500                if "service_worker" == &ev.target_info.r#type {
501                    let detach_command = DetachFromTargetParams::builder()
502                        .session_id(ev.session_id.clone())
503                        .build();
504
505                    let method = detach_command.identifier();
506
507                    if let Ok(params) = serde_json::to_value(detach_command) {
508                        self.queued_events.push_back(TargetEvent::Request(Request {
509                            method,
510                            session_id: self.session_id.clone().map(Into::into),
511                            params,
512                        }));
513                    }
514                }
515            }
516            // `NetworkManager` events
517            CdpEvent::FetchRequestPaused(ev) => self.network_manager.on_fetch_request_paused(ev),
518            CdpEvent::FetchAuthRequired(ev) => self.network_manager.on_fetch_auth_required(ev),
519            CdpEvent::NetworkRequestWillBeSent(ev) => {
520                self.network_manager.on_request_will_be_sent(ev)
521            }
522            CdpEvent::NetworkRequestServedFromCache(ev) => {
523                self.network_manager.on_request_served_from_cache(ev)
524            }
525            CdpEvent::NetworkResponseReceived(ev) => self.network_manager.on_response_received(ev),
526            CdpEvent::NetworkLoadingFinished(ev) => {
527                self.network_manager.on_network_loading_finished(ev)
528            }
529            CdpEvent::NetworkLoadingFailed(ev) => {
530                self.network_manager.on_network_loading_failed(ev)
531            }
532            _ => (),
533        }
534        chromiumoxide_cdp::consume_event!(match params {
535           |ev| self.event_listeners.start_send(ev),
536           |json| { let _ = self.event_listeners.try_send_custom(&method, json);}
537        });
538    }
539
540    /// Called when a init command timed out
541    fn on_initialization_failed(&mut self) -> TargetEvent {
542        if let Some(initiator) = self.initiator.take() {
543            let _ = initiator.send(Err(CdpError::Timeout));
544        }
545        self.init_state = TargetInit::Closing;
546        let close_target = CloseTargetParams::new(self.info.target_id.clone());
547
548        TargetEvent::Request(Request {
549            method: close_target.identifier(),
550            session_id: self.session_id.clone().map(Into::into),
551            params: serde_json::to_value(close_target).unwrap_or_default(),
552        })
553    }
554
555    /// Advance that target's state
556    pub(crate) fn poll(&mut self, cx: &mut Context<'_>, now: Instant) -> Option<TargetEvent> {
557        if !self.is_page() {
558            // can only poll pages
559            return None;
560        }
561
562        match &mut self.init_state {
563            TargetInit::AttachToTarget => {
564                self.init_state = TargetInit::InitializingFrame(FrameManager::init_commands(
565                    self.config.request_timeout,
566                ));
567
568                if let Ok(params) = AttachToTargetParams::builder()
569                    .target_id(self.target_id().clone())
570                    .flatten(true)
571                    .build()
572                {
573                    return Some(TargetEvent::Request(Request::new(
574                        params.identifier(),
575                        serde_json::to_value(params).unwrap_or_default(),
576                    )));
577                } else {
578                    return None;
579                }
580            }
581            TargetInit::InitializingFrame(cmds) => {
582                self.session_id.as_ref()?;
583                if let Poll::Ready(poll) = cmds.poll(now) {
584                    return match poll {
585                        None => {
586                            if let Some(world_name) = self.frame_manager.get_isolated_world_name() {
587                                let world_name = world_name.clone();
588
589                                if let Some(isolated_world_cmds) =
590                                    self.frame_manager.ensure_isolated_world(&world_name)
591                                {
592                                    *cmds = isolated_world_cmds;
593                                } else {
594                                    self.init_state = TargetInit::InitializingNetwork(
595                                        self.network_manager.init_commands(),
596                                    );
597                                }
598                            } else {
599                                self.init_state = TargetInit::InitializingNetwork(
600                                    self.network_manager.init_commands(),
601                                );
602                            }
603                            self.poll(cx, now)
604                        }
605                        Some(Ok((method, params))) => Some(TargetEvent::Request(Request {
606                            method,
607                            session_id: self.session_id.clone().map(Into::into),
608                            params,
609                        })),
610                        Some(Err(_)) => Some(self.on_initialization_failed()),
611                    };
612                } else {
613                    return None;
614                }
615            }
616            TargetInit::InitializingNetwork(cmds) => {
617                advance_state!(
618                    self,
619                    cx,
620                    now,
621                    cmds,
622                    TargetInit::InitializingPage(Self::page_init_commands(
623                        self.config.request_timeout
624                    ))
625                );
626            }
627            TargetInit::InitializingPage(cmds) => {
628                advance_state!(
629                    self,
630                    cx,
631                    now,
632                    cmds,
633                    match self.config.viewport.as_ref() {
634                        Some(viewport) => TargetInit::InitializingEmulation(
635                            self.emulation_manager.init_commands(viewport)
636                        ),
637                        None => TargetInit::Initialized,
638                    }
639                );
640            }
641            TargetInit::InitializingEmulation(cmds) => {
642                advance_state!(self, cx, now, cmds, TargetInit::Initialized);
643            }
644            TargetInit::Initialized => {
645                if let Some(initiator) = self.initiator.take() {
646                    // make sure that the main frame of the page has finished loading
647                    if self
648                        .frame_manager
649                        .main_frame()
650                        .map(|frame| frame.is_loaded())
651                        .unwrap_or_default()
652                    {
653                        if let Some(page) = self.get_or_create_page() {
654                            let _ = initiator.send(Ok(page.clone().into()));
655                        } else {
656                            self.initiator = Some(initiator);
657                        }
658                    } else {
659                        self.initiator = Some(initiator);
660                    }
661                }
662            }
663            TargetInit::Closing => return None,
664        };
665
666        // Prune senders whose receivers have been dropped (caller
667        // timed out or was cancelled) so the vecs don't grow unbounded.
668        // Done once per poll() call, outside the inner loop.
669        if !self.wait_for_frame_navigation.is_empty() {
670            self.wait_for_frame_navigation.retain(|tx| !tx.is_closed());
671        }
672        if !self.wait_for_dom_content_loaded.is_empty() {
673            self.wait_for_dom_content_loaded
674                .retain(|tx| !tx.is_closed());
675        }
676        if !self.wait_for_load.is_empty() {
677            self.wait_for_load.retain(|tx| !tx.is_closed());
678        }
679        if !self.wait_for_network_idle.is_empty() {
680            self.wait_for_network_idle.retain(|tx| !tx.is_closed());
681        }
682        if !self.wait_for_network_almost_idle.is_empty() {
683            self.wait_for_network_almost_idle
684                .retain(|tx| !tx.is_closed());
685        }
686
687        loop {
688            if self.init_state == TargetInit::Closing {
689                break None;
690            }
691
692            if let Some(frame) = self.frame_manager.main_frame() {
693                let req = frame.http_request();
694                let mut waiters_remaining = false;
695
696                if frame.is_dom_content_loaded() {
697                    waiters_remaining |= drain_waiters_bounded(
698                        &mut self.wait_for_dom_content_loaded,
699                        req,
700                        WAITER_DRAIN_BUDGET,
701                    );
702                    waiters_remaining |= drain_waiters_bounded(
703                        &mut self.wait_for_frame_navigation,
704                        req,
705                        WAITER_DRAIN_BUDGET,
706                    );
707                }
708
709                if frame.is_loaded() {
710                    waiters_remaining |=
711                        drain_waiters_bounded(&mut self.wait_for_load, req, WAITER_DRAIN_BUDGET);
712                }
713
714                if frame.is_network_idle() {
715                    waiters_remaining |= drain_waiters_bounded(
716                        &mut self.wait_for_network_idle,
717                        req,
718                        WAITER_DRAIN_BUDGET,
719                    );
720                }
721
722                if frame.is_network_almost_idle() {
723                    waiters_remaining |= drain_waiters_bounded(
724                        &mut self.wait_for_network_almost_idle,
725                        req,
726                        WAITER_DRAIN_BUDGET,
727                    );
728                }
729
730                if waiters_remaining {
731                    // More waiters queued than the per-poll budget.
732                    // Self-wake so the handler re-enters and drains the
733                    // remainder on the next tick instead of stalling.
734                    cx.waker().wake_by_ref();
735                }
736            }
737
738            // Drain queued messages first.
739            if let Some(ev) = self.queued_events.pop_front() {
740                return Some(ev);
741            }
742
743            if let Some(handle) = self.page.as_mut() {
744                while let Poll::Ready(Some(msg)) = handle.rx.poll_recv(cx) {
745                    if self.init_state == TargetInit::Closing {
746                        break;
747                    }
748
749                    match msg {
750                        TargetMessage::Command(cmd) => {
751                            if cmd.method == "Network.setBlockedURLs" {
752                                if let Some(arr) = cmd.params.get("urls").and_then(|v| v.as_array())
753                                {
754                                    let mut unblock_all = false;
755                                    let mut block_all = false;
756
757                                    for s in arr.iter().filter_map(|v| v.as_str()) {
758                                        if s == "!*" {
759                                            unblock_all = true;
760                                            break; // "!*" overrides any block rules
761                                        }
762                                        if s.contains('*') {
763                                            block_all = true;
764                                        }
765                                    }
766
767                                    if unblock_all {
768                                        self.network_manager.set_block_all(false);
769                                    } else if block_all {
770                                        self.network_manager.set_block_all(true);
771                                    }
772                                }
773                            }
774                            self.queued_events.push_back(TargetEvent::Command(cmd));
775                        }
776                        TargetMessage::MainFrame(tx) => {
777                            let _ =
778                                tx.send(self.frame_manager.main_frame().map(|f| f.id().clone()));
779                        }
780                        TargetMessage::AllFrames(tx) => {
781                            let _ = tx.send(
782                                self.frame_manager
783                                    .frames()
784                                    .map(|f| f.id().clone())
785                                    .collect(),
786                            );
787                        }
788                        #[cfg(feature = "_cache")]
789                        TargetMessage::CacheKey((cache_key, cache_policy)) => {
790                            self.network_manager.set_cache_site_key(cache_key);
791                            self.network_manager.set_cache_policy(cache_policy);
792                        }
793                        TargetMessage::Url(req) => {
794                            let GetUrl { frame_id, tx } = req;
795                            let frame = if let Some(frame_id) = frame_id {
796                                self.frame_manager.frame(&frame_id)
797                            } else {
798                                self.frame_manager.main_frame()
799                            };
800                            let _ = tx.send(frame.and_then(|f| f.url().map(str::to_string)));
801                        }
802                        TargetMessage::Name(req) => {
803                            let GetName { frame_id, tx } = req;
804                            let frame = if let Some(frame_id) = frame_id {
805                                self.frame_manager.frame(&frame_id)
806                            } else {
807                                self.frame_manager.main_frame()
808                            };
809                            let _ = tx.send(frame.and_then(|f| f.name().map(str::to_string)));
810                        }
811                        TargetMessage::Parent(req) => {
812                            let GetParent { frame_id, tx } = req;
813                            let frame = self.frame_manager.frame(&frame_id);
814                            let _ = tx.send(frame.and_then(|f| f.parent_id().cloned()));
815                        }
816                        TargetMessage::WaitForNavigation(tx) => {
817                            if let Some(frame) = self.frame_manager.main_frame() {
818                                if frame.is_dom_content_loaded() {
819                                    let _ = tx.send(frame.http_request().cloned());
820                                } else {
821                                    self.wait_for_frame_navigation.push(tx);
822                                }
823                            } else {
824                                self.wait_for_frame_navigation.push(tx);
825                            }
826                        }
827                        TargetMessage::WaitForDomContentLoaded(tx) => {
828                            if let Some(frame) = self.frame_manager.main_frame() {
829                                if frame.is_dom_content_loaded() {
830                                    let _ = tx.send(frame.http_request().cloned());
831                                } else {
832                                    self.wait_for_dom_content_loaded.push(tx);
833                                }
834                            } else {
835                                self.wait_for_dom_content_loaded.push(tx);
836                            }
837                        }
838                        TargetMessage::WaitForLoad(tx) => {
839                            if let Some(frame) = self.frame_manager.main_frame() {
840                                if frame.is_loaded() {
841                                    let _ = tx.send(frame.http_request().cloned());
842                                } else {
843                                    self.wait_for_load.push(tx);
844                                }
845                            } else {
846                                self.wait_for_load.push(tx);
847                            }
848                        }
849                        TargetMessage::WaitForNetworkIdle(tx) => {
850                            if let Some(frame) = self.frame_manager.main_frame() {
851                                if frame.is_network_idle() {
852                                    let _ = tx.send(frame.http_request().cloned());
853                                } else {
854                                    self.wait_for_network_idle.push(tx);
855                                }
856                            } else {
857                                self.wait_for_network_idle.push(tx);
858                            }
859                        }
860                        TargetMessage::WaitForNetworkAlmostIdle(tx) => {
861                            if let Some(frame) = self.frame_manager.main_frame() {
862                                if frame.is_network_almost_idle() {
863                                    let _ = tx.send(frame.http_request().cloned());
864                                } else {
865                                    self.wait_for_network_almost_idle.push(tx);
866                                }
867                            } else {
868                                self.wait_for_network_almost_idle.push(tx);
869                            }
870                        }
871                        TargetMessage::AddEventListener(req) => {
872                            if req.method == "Fetch.requestPaused" {
873                                self.network_manager.enable_request_intercept();
874                            }
875                            // register a new listener
876                            self.event_listeners.add_listener(req);
877                        }
878                        TargetMessage::GetExecutionContext(ctx) => {
879                            let GetExecutionContext {
880                                dom_world,
881                                frame_id,
882                                tx,
883                            } = ctx;
884                            let frame = if let Some(frame_id) = frame_id {
885                                self.frame_manager.frame(&frame_id)
886                            } else {
887                                self.frame_manager.main_frame()
888                            };
889
890                            if let Some(frame) = frame {
891                                match dom_world {
892                                    DOMWorldKind::Main => {
893                                        let _ = tx.send(frame.main_world().execution_context());
894                                    }
895                                    DOMWorldKind::Secondary => {
896                                        let _ =
897                                            tx.send(frame.secondary_world().execution_context());
898                                    }
899                                }
900                            } else {
901                                let _ = tx.send(None);
902                            }
903                        }
904                        TargetMessage::Authenticate(credentials) => {
905                            self.network_manager.authenticate(credentials);
906                        }
907                        TargetMessage::BlockNetwork(blocked) => {
908                            self.network_manager.set_block_all(blocked);
909                        }
910                        TargetMessage::EnableInterception(enabled) => {
911                            // if interception is enabled disable the user facing handling.
912                            self.network_manager.user_request_interception_enabled = !enabled;
913                        }
914                    }
915                }
916            }
917
918            while let Some(event) = self.network_manager.poll() {
919                if self.init_state == TargetInit::Closing {
920                    break;
921                }
922                match event {
923                    NetworkEvent::SendCdpRequest((method, params)) => {
924                        // send a message to the browser
925                        self.queued_events.push_back(TargetEvent::Request(Request {
926                            method,
927                            session_id: self.session_id.clone().map(Into::into),
928                            params,
929                        }))
930                    }
931                    NetworkEvent::Request(_) => {}
932                    NetworkEvent::Response(_) => {}
933                    NetworkEvent::RequestFailed(request) => {
934                        self.frame_manager.on_http_request_finished(request);
935                    }
936                    NetworkEvent::RequestFinished(request) => {
937                        self.frame_manager.on_http_request_finished(request);
938                    }
939                    NetworkEvent::BytesConsumed(n) => {
940                        self.queued_events.push_back(TargetEvent::BytesConsumed(n));
941                    }
942                }
943            }
944
945            while let Some(event) = self.frame_manager.poll(now) {
946                if self.init_state == TargetInit::Closing {
947                    break;
948                }
949                match event {
950                    FrameEvent::NavigationResult(res) => {
951                        self.queued_events
952                            .push_back(TargetEvent::NavigationResult(res));
953                    }
954                    FrameEvent::NavigationRequest(id, req) => {
955                        self.queued_events
956                            .push_back(TargetEvent::NavigationRequest(id, req));
957                    }
958                }
959            }
960
961            if self.queued_events.is_empty() {
962                return None;
963            }
964        }
965    }
966
967    /// Process a single message from the page channel.
968    ///
969    /// Used by `Handler::run()` after `try_recv()` drains the page channel.
970    pub(crate) fn on_page_message(&mut self, msg: TargetMessage) {
971        if self.init_state == TargetInit::Closing {
972            return;
973        }
974        match msg {
975            TargetMessage::Command(cmd) => {
976                if cmd.method == "Network.setBlockedURLs" {
977                    if let Some(arr) = cmd.params.get("urls").and_then(|v| v.as_array()) {
978                        let mut unblock_all = false;
979                        let mut block_all = false;
980                        for s in arr.iter().filter_map(|v| v.as_str()) {
981                            if s == "!*" {
982                                unblock_all = true;
983                                break;
984                            }
985                            if s.contains('*') {
986                                block_all = true;
987                            }
988                        }
989                        if unblock_all {
990                            self.network_manager.set_block_all(false);
991                        } else if block_all {
992                            self.network_manager.set_block_all(true);
993                        }
994                    }
995                }
996                self.queued_events.push_back(TargetEvent::Command(cmd));
997            }
998            TargetMessage::MainFrame(tx) => {
999                let _ = tx.send(self.frame_manager.main_frame().map(|f| f.id().clone()));
1000            }
1001            TargetMessage::AllFrames(tx) => {
1002                let _ = tx.send(
1003                    self.frame_manager
1004                        .frames()
1005                        .map(|f| f.id().clone())
1006                        .collect(),
1007                );
1008            }
1009            #[cfg(feature = "_cache")]
1010            TargetMessage::CacheKey((cache_key, cache_policy)) => {
1011                self.network_manager.set_cache_site_key(cache_key);
1012                self.network_manager.set_cache_policy(cache_policy);
1013            }
1014            TargetMessage::Url(req) => {
1015                let GetUrl { frame_id, tx } = req;
1016                let frame = if let Some(frame_id) = frame_id {
1017                    self.frame_manager.frame(&frame_id)
1018                } else {
1019                    self.frame_manager.main_frame()
1020                };
1021                let _ = tx.send(frame.and_then(|f| f.url().map(str::to_string)));
1022            }
1023            TargetMessage::Name(req) => {
1024                let GetName { frame_id, tx } = req;
1025                let frame = if let Some(frame_id) = frame_id {
1026                    self.frame_manager.frame(&frame_id)
1027                } else {
1028                    self.frame_manager.main_frame()
1029                };
1030                let _ = tx.send(frame.and_then(|f| f.name().map(str::to_string)));
1031            }
1032            TargetMessage::Parent(req) => {
1033                let GetParent { frame_id, tx } = req;
1034                let frame = self.frame_manager.frame(&frame_id);
1035                let _ = tx.send(frame.and_then(|f| f.parent_id().cloned()));
1036            }
1037            TargetMessage::WaitForNavigation(tx) => {
1038                if let Some(frame) = self.frame_manager.main_frame() {
1039                    if frame.is_dom_content_loaded() {
1040                        let _ = tx.send(frame.http_request().cloned());
1041                    } else {
1042                        self.wait_for_frame_navigation.push(tx);
1043                    }
1044                } else {
1045                    self.wait_for_frame_navigation.push(tx);
1046                }
1047            }
1048            TargetMessage::WaitForDomContentLoaded(tx) => {
1049                if let Some(frame) = self.frame_manager.main_frame() {
1050                    if frame.is_dom_content_loaded() {
1051                        let _ = tx.send(frame.http_request().cloned());
1052                    } else {
1053                        self.wait_for_dom_content_loaded.push(tx);
1054                    }
1055                } else {
1056                    self.wait_for_dom_content_loaded.push(tx);
1057                }
1058            }
1059            TargetMessage::WaitForLoad(tx) => {
1060                if let Some(frame) = self.frame_manager.main_frame() {
1061                    if frame.is_loaded() {
1062                        let _ = tx.send(frame.http_request().cloned());
1063                    } else {
1064                        self.wait_for_load.push(tx);
1065                    }
1066                } else {
1067                    self.wait_for_load.push(tx);
1068                }
1069            }
1070            TargetMessage::WaitForNetworkIdle(tx) => {
1071                if let Some(frame) = self.frame_manager.main_frame() {
1072                    if frame.is_network_idle() {
1073                        let _ = tx.send(frame.http_request().cloned());
1074                    } else {
1075                        self.wait_for_network_idle.push(tx);
1076                    }
1077                } else {
1078                    self.wait_for_network_idle.push(tx);
1079                }
1080            }
1081            TargetMessage::WaitForNetworkAlmostIdle(tx) => {
1082                if let Some(frame) = self.frame_manager.main_frame() {
1083                    if frame.is_network_almost_idle() {
1084                        let _ = tx.send(frame.http_request().cloned());
1085                    } else {
1086                        self.wait_for_network_almost_idle.push(tx);
1087                    }
1088                } else {
1089                    self.wait_for_network_almost_idle.push(tx);
1090                }
1091            }
1092            TargetMessage::AddEventListener(req) => {
1093                if req.method == "Fetch.requestPaused" {
1094                    self.network_manager.enable_request_intercept();
1095                }
1096                self.event_listeners.add_listener(req);
1097            }
1098            TargetMessage::GetExecutionContext(ctx) => {
1099                let GetExecutionContext {
1100                    dom_world,
1101                    frame_id,
1102                    tx,
1103                } = ctx;
1104                let frame = if let Some(frame_id) = frame_id {
1105                    self.frame_manager.frame(&frame_id)
1106                } else {
1107                    self.frame_manager.main_frame()
1108                };
1109                if let Some(frame) = frame {
1110                    match dom_world {
1111                        DOMWorldKind::Main => {
1112                            let _ = tx.send(frame.main_world().execution_context());
1113                        }
1114                        DOMWorldKind::Secondary => {
1115                            let _ = tx.send(frame.secondary_world().execution_context());
1116                        }
1117                    }
1118                } else {
1119                    let _ = tx.send(None);
1120                }
1121            }
1122            TargetMessage::Authenticate(credentials) => {
1123                self.network_manager.authenticate(credentials);
1124            }
1125            TargetMessage::BlockNetwork(blocked) => {
1126                self.network_manager.set_block_all(blocked);
1127            }
1128            TargetMessage::EnableInterception(enabled) => {
1129                self.network_manager.user_request_interception_enabled = !enabled;
1130            }
1131        }
1132    }
1133
1134    /// Advance the target's state machine and drain queued events.
1135    ///
1136    /// Like [`poll`](Self::poll) but does **not** read from the page channel
1137    /// (that is handled externally by `Handler::run()` via `try_recv`).
1138    pub(crate) fn advance(&mut self, now: Instant) -> Option<TargetEvent> {
1139        if !self.is_page() {
1140            return None;
1141        }
1142
1143        // Init state machine
1144        match &mut self.init_state {
1145            TargetInit::AttachToTarget => {
1146                self.init_state = TargetInit::InitializingFrame(FrameManager::init_commands(
1147                    self.config.request_timeout,
1148                ));
1149                if let Ok(params) = AttachToTargetParams::builder()
1150                    .target_id(self.target_id().clone())
1151                    .flatten(true)
1152                    .build()
1153                {
1154                    return Some(TargetEvent::Request(Request::new(
1155                        params.identifier(),
1156                        serde_json::to_value(params).unwrap_or_default(),
1157                    )));
1158                } else {
1159                    return None;
1160                }
1161            }
1162            TargetInit::InitializingFrame(cmds) => {
1163                self.session_id.as_ref()?;
1164                if let Poll::Ready(poll) = cmds.poll(now) {
1165                    return match poll {
1166                        None => {
1167                            if let Some(world_name) = self.frame_manager.get_isolated_world_name() {
1168                                let world_name = world_name.clone();
1169                                if let Some(isolated_world_cmds) =
1170                                    self.frame_manager.ensure_isolated_world(&world_name)
1171                                {
1172                                    *cmds = isolated_world_cmds;
1173                                } else {
1174                                    self.init_state = TargetInit::InitializingNetwork(
1175                                        self.network_manager.init_commands(),
1176                                    );
1177                                }
1178                            } else {
1179                                self.init_state = TargetInit::InitializingNetwork(
1180                                    self.network_manager.init_commands(),
1181                                );
1182                            }
1183                            self.advance(now)
1184                        }
1185                        Some(Ok((method, params))) => Some(TargetEvent::Request(Request {
1186                            method,
1187                            session_id: self.session_id.clone().map(Into::into),
1188                            params,
1189                        })),
1190                        Some(Err(_)) => Some(self.on_initialization_failed()),
1191                    };
1192                } else {
1193                    return None;
1194                }
1195            }
1196            TargetInit::InitializingNetwork(cmds) => {
1197                if let Poll::Ready(poll) = cmds.poll(now) {
1198                    return match poll {
1199                        None => {
1200                            self.init_state = TargetInit::InitializingPage(
1201                                Self::page_init_commands(self.config.request_timeout),
1202                            );
1203                            self.advance(now)
1204                        }
1205                        Some(Ok((method, params))) => Some(TargetEvent::Request(Request {
1206                            method,
1207                            session_id: self.session_id.clone().map(Into::into),
1208                            params,
1209                        })),
1210                        Some(Err(_)) => Some(self.on_initialization_failed()),
1211                    };
1212                } else {
1213                    return None;
1214                }
1215            }
1216            TargetInit::InitializingPage(cmds) => {
1217                if let Poll::Ready(poll) = cmds.poll(now) {
1218                    return match poll {
1219                        None => {
1220                            self.init_state = match self.config.viewport.as_ref() {
1221                                Some(viewport) => TargetInit::InitializingEmulation(
1222                                    self.emulation_manager.init_commands(viewport),
1223                                ),
1224                                None => TargetInit::Initialized,
1225                            };
1226                            self.advance(now)
1227                        }
1228                        Some(Ok((method, params))) => Some(TargetEvent::Request(Request {
1229                            method,
1230                            session_id: self.session_id.clone().map(Into::into),
1231                            params,
1232                        })),
1233                        Some(Err(_)) => Some(self.on_initialization_failed()),
1234                    };
1235                } else {
1236                    return None;
1237                }
1238            }
1239            TargetInit::InitializingEmulation(cmds) => {
1240                if let Poll::Ready(poll) = cmds.poll(now) {
1241                    return match poll {
1242                        None => {
1243                            self.init_state = TargetInit::Initialized;
1244                            self.advance(now)
1245                        }
1246                        Some(Ok((method, params))) => Some(TargetEvent::Request(Request {
1247                            method,
1248                            session_id: self.session_id.clone().map(Into::into),
1249                            params,
1250                        })),
1251                        Some(Err(_)) => Some(self.on_initialization_failed()),
1252                    };
1253                } else {
1254                    return None;
1255                }
1256            }
1257            TargetInit::Initialized => {
1258                if let Some(initiator) = self.initiator.take() {
1259                    if self
1260                        .frame_manager
1261                        .main_frame()
1262                        .map(|frame| frame.is_loaded())
1263                        .unwrap_or_default()
1264                    {
1265                        if let Some(page) = self.get_or_create_page() {
1266                            let _ = initiator.send(Ok(page.clone().into()));
1267                        } else {
1268                            self.initiator = Some(initiator);
1269                        }
1270                    } else {
1271                        self.initiator = Some(initiator);
1272                    }
1273                }
1274            }
1275            TargetInit::Closing => return None,
1276        };
1277
1278        // Prune dead waiters
1279        if !self.wait_for_frame_navigation.is_empty() {
1280            self.wait_for_frame_navigation.retain(|tx| !tx.is_closed());
1281        }
1282        if !self.wait_for_dom_content_loaded.is_empty() {
1283            self.wait_for_dom_content_loaded
1284                .retain(|tx| !tx.is_closed());
1285        }
1286        if !self.wait_for_load.is_empty() {
1287            self.wait_for_load.retain(|tx| !tx.is_closed());
1288        }
1289        if !self.wait_for_network_idle.is_empty() {
1290            self.wait_for_network_idle.retain(|tx| !tx.is_closed());
1291        }
1292        if !self.wait_for_network_almost_idle.is_empty() {
1293            self.wait_for_network_almost_idle
1294                .retain(|tx| !tx.is_closed());
1295        }
1296
1297        // Drain events loop (same as poll's inner loop, minus page channel reading)
1298        loop {
1299            if self.init_state == TargetInit::Closing {
1300                break None;
1301            }
1302
1303            if let Some(frame) = self.frame_manager.main_frame() {
1304                if frame.is_dom_content_loaded() {
1305                    while let Some(tx) = self.wait_for_dom_content_loaded.pop() {
1306                        let _ = tx.send(frame.http_request().cloned());
1307                    }
1308                    while let Some(tx) = self.wait_for_frame_navigation.pop() {
1309                        let _ = tx.send(frame.http_request().cloned());
1310                    }
1311                }
1312                if frame.is_loaded() {
1313                    while let Some(tx) = self.wait_for_load.pop() {
1314                        let _ = tx.send(frame.http_request().cloned());
1315                    }
1316                }
1317                if frame.is_network_idle() {
1318                    while let Some(tx) = self.wait_for_network_idle.pop() {
1319                        let _ = tx.send(frame.http_request().cloned());
1320                    }
1321                }
1322                if frame.is_network_almost_idle() {
1323                    while let Some(tx) = self.wait_for_network_almost_idle.pop() {
1324                        let _ = tx.send(frame.http_request().cloned());
1325                    }
1326                }
1327            }
1328
1329            if let Some(ev) = self.queued_events.pop_front() {
1330                return Some(ev);
1331            }
1332
1333            while let Some(event) = self.network_manager.poll() {
1334                if self.init_state == TargetInit::Closing {
1335                    break;
1336                }
1337                match event {
1338                    NetworkEvent::SendCdpRequest((method, params)) => {
1339                        self.queued_events.push_back(TargetEvent::Request(Request {
1340                            method,
1341                            session_id: self.session_id.clone().map(Into::into),
1342                            params,
1343                        }));
1344                    }
1345                    NetworkEvent::Request(_) => {}
1346                    NetworkEvent::Response(_) => {}
1347                    NetworkEvent::RequestFailed(request) => {
1348                        self.frame_manager.on_http_request_finished(request);
1349                    }
1350                    NetworkEvent::RequestFinished(request) => {
1351                        self.frame_manager.on_http_request_finished(request);
1352                    }
1353                    NetworkEvent::BytesConsumed(n) => {
1354                        self.queued_events.push_back(TargetEvent::BytesConsumed(n));
1355                    }
1356                }
1357            }
1358
1359            while let Some(event) = self.frame_manager.poll(now) {
1360                if self.init_state == TargetInit::Closing {
1361                    break;
1362                }
1363                match event {
1364                    FrameEvent::NavigationResult(res) => {
1365                        self.queued_events
1366                            .push_back(TargetEvent::NavigationResult(res));
1367                    }
1368                    FrameEvent::NavigationRequest(id, req) => {
1369                        self.queued_events
1370                            .push_back(TargetEvent::NavigationRequest(id, req));
1371                    }
1372                }
1373            }
1374
1375            if self.queued_events.is_empty() {
1376                return None;
1377            }
1378        }
1379    }
1380
1381    /// Set the sender half of the channel who requested the creation of this
1382    /// target
1383    pub fn set_initiator(&mut self, tx: Sender<Result<Page>>) {
1384        self.initiator = Some(tx);
1385    }
1386
1387    pub(crate) fn page_init_commands(timeout: Duration) -> CommandChain {
1388        CommandChain::new(INIT_COMMANDS_PARAMS.clone(), timeout)
1389    }
1390}
1391
1392/// Configuration for how a single target/page should be fetched and processed.
1393#[derive(Debug, Clone)]
1394pub struct TargetConfig {
1395    /// Whether to ignore TLS/HTTPS certificate errors (e.g. self-signed or expired certs).
1396    /// When `true`, connections will proceed even if certificate validation fails.
1397    pub ignore_https_errors: bool,
1398    /// Request timeout to use for the main navigation / resource fetch.
1399    /// This is the total time allowed before a request is considered failed.
1400    pub request_timeout: Duration,
1401    /// Optional browser viewport to use for this target.
1402    /// When `None`, the default viewport (or headless browser default) is used.
1403    pub viewport: Option<Viewport>,
1404    /// Enable request interception for this target.
1405    /// When `true`, all network requests will pass through the intercept manager.
1406    pub request_intercept: bool,
1407    /// Enable caching for this target.
1408    /// When `true`, responses may be read from and written to the cache layer.
1409    pub cache_enabled: bool,
1410    /// If `true`, skip visual/asset resources that are not required for HTML content
1411    /// (e.g. images, fonts, media). Useful for performance-oriented crawls.
1412    pub ignore_visuals: bool,
1413    /// If `true`, block JavaScript execution (or avoid loading JS resources)
1414    /// for this target. This is useful for purely static HTML crawls.
1415    pub ignore_javascript: bool,
1416    /// If `true`, block analytics / tracking requests (e.g. Google Analytics,
1417    /// common tracker domains, etc.).
1418    pub ignore_analytics: bool,
1419    /// Ignore prefetching.
1420    pub ignore_prefetch: bool,
1421    /// If `true`, block stylesheets and related CSS resources for this target.
1422    /// This can reduce bandwidth when only raw HTML is needed.
1423    pub ignore_stylesheets: bool,
1424    /// When `ignore_stylesheets` would skip a stylesheet, allow it through if
1425    /// the request URL is first-party (registrable domain matches the page's
1426    /// primary frame). Default `true`. Set `false` for strict block-all.
1427    pub allow_first_party_stylesheets: bool,
1428    /// When a downstream blocker would skip a script, allow it through if
1429    /// first-party. Default `true`.
1430    pub allow_first_party_javascript: bool,
1431    /// When `ignore_visuals` would skip an image/media/font, allow it through
1432    /// if first-party. Default `true`.
1433    pub allow_first_party_visuals: bool,
1434    /// If `true`, only HTML documents will be fetched/kept.
1435    /// Non-HTML subresources may be skipped entirely.
1436    pub only_html: bool,
1437    /// Whether service workers are allowed for this target.
1438    /// When `true`, service workers may register and intercept requests.
1439    pub service_worker_enabled: bool,
1440    /// Extra HTTP headers to send with each request for this target.
1441    /// Keys should be header names, values their corresponding header values.
1442    pub extra_headers: Option<std::collections::HashMap<String, String>>,
1443    /// Network intercept manager used to make allow/deny/modify decisions
1444    /// for requests when `request_intercept` is enabled.
1445    pub intercept_manager: NetworkInterceptManager,
1446    /// Push the interception policy to a capable remote engine once per
1447    /// navigation (`Interception.setPolicy`) so it resolves block/allow
1448    /// locally instead of round-tripping each `Fetch.requestPaused`. Default
1449    /// `false`; safe against any target.
1450    pub remote_local_policy: bool,
1451    /// The maximum number of response bytes allowed for this target.
1452    /// When set, responses larger than this limit may be truncated or aborted.
1453    pub max_bytes_allowed: Option<u64>,
1454    /// Cap on Document-type redirect hops before the navigation is aborted.
1455    /// `None` disables enforcement; `Some(n)` mirrors `reqwest::redirect::Policy::limited(n)`.
1456    pub max_redirects: Option<usize>,
1457    /// Cap on main-frame cross-document navigations per `goto`. Defends against
1458    /// JS / meta-refresh loops that bypass the HTTP redirect guard. `None`
1459    /// disables the guard.
1460    pub max_main_frame_navigations: Option<u32>,
1461    /// Whitelist patterns to allow through the network.
1462    pub whitelist_patterns: Option<Vec<String>>,
1463    /// Blacklist patterns to black through the network.
1464    pub blacklist_patterns: Option<Vec<String>>,
1465    /// Extra ABP/uBO filter rules for the adblock engine.
1466    #[cfg(feature = "adblock")]
1467    pub adblock_filter_rules: Option<Vec<String>>,
1468    /// Optional notify handle for waking `Handler::run()`'s select loop.
1469    /// `None` when using the `impl Stream for Handler` path (no overhead).
1470    pub page_wake: Option<Arc<Notify>>,
1471    /// Capacity of the per-page mpsc channel carrying `TargetMessage`s
1472    /// from the page handle to the handler. Defaults to
1473    /// `crate::handler::page::DEFAULT_PAGE_CHANNEL_CAPACITY` (2048);
1474    /// override via `HandlerConfig::page_channel_capacity`. Clamped to
1475    /// a minimum of 1 at channel creation time.
1476    pub page_channel_capacity: usize,
1477}
1478
1479impl Default for TargetConfig {
1480    fn default() -> Self {
1481        Self {
1482            ignore_https_errors: true,
1483            request_timeout: Duration::from_millis(REQUEST_TIMEOUT),
1484            viewport: Default::default(),
1485            request_intercept: false,
1486            cache_enabled: true,
1487            service_worker_enabled: true,
1488            ignore_javascript: false,
1489            ignore_visuals: false,
1490            ignore_stylesheets: false,
1491            allow_first_party_stylesheets: true,
1492            allow_first_party_javascript: true,
1493            allow_first_party_visuals: true,
1494            ignore_analytics: true,
1495            ignore_prefetch: true,
1496            only_html: false,
1497            extra_headers: Default::default(),
1498            intercept_manager: NetworkInterceptManager::Unknown,
1499            remote_local_policy: false,
1500            max_bytes_allowed: None,
1501            max_redirects: None,
1502            max_main_frame_navigations: None,
1503            whitelist_patterns: None,
1504            blacklist_patterns: None,
1505            #[cfg(feature = "adblock")]
1506            adblock_filter_rules: None,
1507            page_wake: None,
1508            page_channel_capacity: crate::handler::page::DEFAULT_PAGE_CHANNEL_CAPACITY,
1509        }
1510    }
1511}
1512
1513#[derive(Debug, Clone, Eq, PartialEq)]
1514pub enum TargetType {
1515    Page,
1516    BackgroundPage,
1517    ServiceWorker,
1518    SharedWorker,
1519    Other,
1520    Browser,
1521    Webview,
1522    Unknown(String),
1523}
1524
1525impl TargetType {
1526    pub fn new(ty: &str) -> Self {
1527        match ty {
1528            "page" => TargetType::Page,
1529            "background_page" => TargetType::BackgroundPage,
1530            "service_worker" => TargetType::ServiceWorker,
1531            "shared_worker" => TargetType::SharedWorker,
1532            "other" => TargetType::Other,
1533            "browser" => TargetType::Browser,
1534            "webview" => TargetType::Webview,
1535            s => TargetType::Unknown(s.to_string()),
1536        }
1537    }
1538
1539    pub fn is_page(&self) -> bool {
1540        matches!(self, TargetType::Page)
1541    }
1542
1543    pub fn is_background_page(&self) -> bool {
1544        matches!(self, TargetType::BackgroundPage)
1545    }
1546
1547    pub fn is_service_worker(&self) -> bool {
1548        matches!(self, TargetType::ServiceWorker)
1549    }
1550
1551    pub fn is_shared_worker(&self) -> bool {
1552        matches!(self, TargetType::SharedWorker)
1553    }
1554
1555    pub fn is_other(&self) -> bool {
1556        matches!(self, TargetType::Other)
1557    }
1558
1559    pub fn is_browser(&self) -> bool {
1560        matches!(self, TargetType::Browser)
1561    }
1562
1563    pub fn is_webview(&self) -> bool {
1564        matches!(self, TargetType::Webview)
1565    }
1566}
1567
1568#[derive(Debug)]
1569pub(crate) enum TargetEvent {
1570    /// An internal request
1571    Request(Request),
1572    /// An internal navigation request
1573    NavigationRequest(NavigationId, Request),
1574    /// Indicates that a previous requested navigation has finished
1575    NavigationResult(Result<NavigationOk, NavigationError>),
1576    /// A new command arrived via a channel
1577    Command(CommandMessage),
1578    /// The bytes consumed by the network.
1579    BytesConsumed(u64),
1580}
1581
1582// TODO this can be moved into the classes?
1583#[derive(Debug, PartialEq)]
1584pub enum TargetInit {
1585    InitializingFrame(CommandChain),
1586    InitializingNetwork(CommandChain),
1587    InitializingPage(CommandChain),
1588    InitializingEmulation(CommandChain),
1589    AttachToTarget,
1590    Initialized,
1591    Closing,
1592}
1593
1594impl TargetInit {
1595    fn commands_mut(&mut self) -> Option<&mut CommandChain> {
1596        match self {
1597            TargetInit::InitializingFrame(cmd) => Some(cmd),
1598            TargetInit::InitializingNetwork(cmd) => Some(cmd),
1599            TargetInit::InitializingPage(cmd) => Some(cmd),
1600            TargetInit::InitializingEmulation(cmd) => Some(cmd),
1601            TargetInit::AttachToTarget => None,
1602            TargetInit::Initialized => None,
1603            TargetInit::Closing => None,
1604        }
1605    }
1606}
1607
1608#[derive(Debug)]
1609pub struct GetExecutionContext {
1610    /// For which world the execution context was requested
1611    pub dom_world: DOMWorldKind,
1612    /// The if of the frame to get the `ExecutionContext` for
1613    pub frame_id: Option<FrameId>,
1614    /// Sender half of the channel to send the response back
1615    pub tx: Sender<Option<ExecutionContextId>>,
1616}
1617
1618impl GetExecutionContext {
1619    pub fn new(tx: Sender<Option<ExecutionContextId>>) -> Self {
1620        Self {
1621            dom_world: DOMWorldKind::Main,
1622            frame_id: None,
1623            tx,
1624        }
1625    }
1626}
1627
1628#[derive(Debug)]
1629pub struct GetUrl {
1630    /// The id of the frame to get the url for (None = main frame)
1631    pub frame_id: Option<FrameId>,
1632    /// Sender half of the channel to send the response back
1633    pub tx: Sender<Option<String>>,
1634}
1635
1636impl GetUrl {
1637    pub fn new(tx: Sender<Option<String>>) -> Self {
1638        Self { frame_id: None, tx }
1639    }
1640}
1641
1642#[derive(Debug)]
1643pub struct GetName {
1644    /// The id of the frame to get the name for (None = main frame)
1645    pub frame_id: Option<FrameId>,
1646    /// Sender half of the channel to send the response back
1647    pub tx: Sender<Option<String>>,
1648}
1649
1650#[derive(Debug)]
1651pub struct GetParent {
1652    /// The id of the frame to get the parent for (None = main frame)
1653    pub frame_id: FrameId,
1654    /// Sender half of the channel to send the response back
1655    pub tx: Sender<Option<FrameId>>,
1656}
1657
1658#[derive(Debug)]
1659pub enum TargetMessage {
1660    /// Execute a command within the session of this target
1661    Command(CommandMessage),
1662    /// Return the main frame of this target's page
1663    MainFrame(Sender<Option<FrameId>>),
1664    /// Return all the frames of this target's page
1665    AllFrames(Sender<Vec<FrameId>>),
1666    #[cfg(feature = "_cache")]
1667    /// Set the cache key and policy for the target page.
1668    CacheKey((Option<String>, Option<crate::cache::BasicCachePolicy>)),
1669    /// Return the url if available
1670    Url(GetUrl),
1671    /// Return the name if available
1672    Name(GetName),
1673    /// Return the parent id of a frame
1674    Parent(GetParent),
1675    /// A Message that resolves when the frame finished loading a new url
1676    WaitForNavigation(Sender<ArcHttpRequest>),
1677    /// Resolves when `DOMContentLoaded` fires (HTML parsed, sync scripts
1678    /// executed) — before `load`, so subresources may still be in-flight.
1679    WaitForDomContentLoaded(Sender<ArcHttpRequest>),
1680    /// Resolves when the `load` event fires — all subresources (images,
1681    /// fonts, XHRs) are done. Slower than `WaitForNavigation` through proxies.
1682    WaitForLoad(Sender<ArcHttpRequest>),
1683    /// A Message that resolves when the frame network is idle
1684    WaitForNetworkIdle(Sender<ArcHttpRequest>),
1685    /// A Message that resolves when the frame network is almost idle
1686    WaitForNetworkAlmostIdle(Sender<ArcHttpRequest>),
1687    /// A request to submit a new listener that gets notified with every
1688    /// received event
1689    AddEventListener(EventListenerRequest),
1690    /// Get the `ExecutionContext` if available
1691    GetExecutionContext(GetExecutionContext),
1692    Authenticate(Credentials),
1693    /// Set block/unblocked networking
1694    BlockNetwork(bool),
1695    /// Enable/Disable internal request paused interception
1696    EnableInterception(bool),
1697}
1698
1699#[cfg(test)]
1700mod waiter_drain_tests {
1701    //! Unit tests for `drain_waiters_bounded`.
1702    //!
1703    //! These cover the isolated drain helper — they do not spin up a real
1704    //! `Target` or browser, so they run in microseconds and exhaustively
1705    //! exercise the budget / re-arm contract:
1706    //!
1707    //! - drain with no waiters is a no-op and reports `remaining = false`
1708    //! - drain with fewer waiters than budget fires all and reports `false`
1709    //! - drain with exactly `budget` waiters fires all and reports `false`
1710    //! - drain with more waiters than `budget` fires `budget` and reports `true`
1711    //! - senders whose receivers were dropped don't panic or consume extra work
1712    //! - repeated draining eventually empties any queue (no deadlock)
1713    //!
1714    //! The last test is the key "no deadlock" property: if re-arm were broken
1715    //! (say, we forgot to wake), the handler could stall with waiters pending
1716    //! forever. Here we prove the helper itself always makes forward progress.
1717    use super::{drain_waiters_bounded, WAITER_DRAIN_BUDGET};
1718    use crate::ArcHttpRequest;
1719    use tokio::sync::oneshot::{self, Sender};
1720
1721    fn make_waiters(
1722        n: usize,
1723    ) -> (
1724        Vec<Sender<ArcHttpRequest>>,
1725        Vec<oneshot::Receiver<ArcHttpRequest>>,
1726    ) {
1727        let mut txs = Vec::with_capacity(n);
1728        let mut rxs = Vec::with_capacity(n);
1729        for _ in 0..n {
1730            let (tx, rx) = oneshot::channel();
1731            txs.push(tx);
1732            rxs.push(rx);
1733        }
1734        (txs, rxs)
1735    }
1736
1737    #[test]
1738    fn empty_queue_is_noop() {
1739        let mut queue: Vec<Sender<ArcHttpRequest>> = Vec::new();
1740        let remaining = drain_waiters_bounded(&mut queue, None, WAITER_DRAIN_BUDGET);
1741        assert!(!remaining, "empty queue should not mark 'remaining'");
1742        assert!(queue.is_empty());
1743    }
1744
1745    #[test]
1746    fn drains_fewer_than_budget() {
1747        let (mut queue, mut rxs) = make_waiters(10);
1748        let remaining = drain_waiters_bounded(&mut queue, None, WAITER_DRAIN_BUDGET);
1749        assert!(!remaining);
1750        assert!(queue.is_empty());
1751        // All receivers got a value.
1752        for rx in rxs.iter_mut() {
1753            assert!(rx.try_recv().is_ok(), "every waiter must receive a value");
1754        }
1755    }
1756
1757    #[test]
1758    fn drains_exactly_budget() {
1759        let (mut queue, mut rxs) = make_waiters(WAITER_DRAIN_BUDGET);
1760        let remaining = drain_waiters_bounded(&mut queue, None, WAITER_DRAIN_BUDGET);
1761        assert!(!remaining, "exactly-budget drain should empty the queue");
1762        assert!(queue.is_empty());
1763        for rx in rxs.iter_mut() {
1764            assert!(rx.try_recv().is_ok());
1765        }
1766    }
1767
1768    #[test]
1769    fn drains_budget_when_over_capacity() {
1770        let n = WAITER_DRAIN_BUDGET * 3 + 7; // 199 waiters at the default 64
1771        let (mut queue, _rxs) = make_waiters(n);
1772        let remaining = drain_waiters_bounded(&mut queue, None, WAITER_DRAIN_BUDGET);
1773        assert!(remaining, "over-budget drain must mark 'remaining = true'");
1774        assert_eq!(
1775            queue.len(),
1776            n - WAITER_DRAIN_BUDGET,
1777            "exactly `budget` waiters should be popped per call"
1778        );
1779    }
1780
1781    #[test]
1782    fn dropped_receiver_does_not_panic() {
1783        let (mut queue, mut rxs) = make_waiters(4);
1784        // Drop half the receivers — their senders become closed.
1785        rxs.truncate(2);
1786        let remaining = drain_waiters_bounded(&mut queue, None, WAITER_DRAIN_BUDGET);
1787        assert!(!remaining);
1788        assert!(queue.is_empty());
1789        // The remaining receivers either got a value or were the popped ones;
1790        // at minimum, no panic occurred.
1791    }
1792
1793    #[test]
1794    fn repeated_draining_empties_any_queue() {
1795        // "No deadlock" property: repeatedly calling the helper always makes
1796        // forward progress and eventually empties the queue. If this loop
1797        // ever ran forever, the re-arm contract would be unreachable.
1798        let n = 10_000;
1799        let (mut queue, _rxs) = make_waiters(n);
1800        let mut rounds = 0;
1801        loop {
1802            let remaining = drain_waiters_bounded(&mut queue, None, WAITER_DRAIN_BUDGET);
1803            rounds += 1;
1804            if !remaining {
1805                break;
1806            }
1807            assert!(rounds < n, "drain must make forward progress on every call");
1808        }
1809        assert!(queue.is_empty());
1810        // 10_000 / 64 = 156.25 → 157 full rounds + final clean-up = 157
1811        assert_eq!(
1812            rounds,
1813            n.div_ceil(WAITER_DRAIN_BUDGET),
1814            "each round should pop exactly `budget` waiters until the tail"
1815        );
1816    }
1817}