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