chromiumoxide/handler/
target.rs

1use std::collections::VecDeque;
2use std::pin::Pin;
3use std::sync::Arc;
4use std::time::Instant;
5
6use chromiumoxide_cdp::cdp::browser_protocol::target::DetachFromTargetParams;
7use futures::channel::oneshot::Sender;
8use futures::stream::Stream;
9use futures::task::{Context, Poll};
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                let mut cmds =  vec![
75                    (
76                        attach.identifier(),
77                        serde_json::to_value(attach).unwrap_or_default(),
78                    ),
79                    (
80                        disable_log.identifier(),
81                        serde_json::to_value(disable_log).unwrap_or_default(),
82                    )
83                ];
84
85                // enable performance on pages.
86                if cfg!(feature = "collect_metrics") {
87                    let enable_performance = chromiumoxide_cdp::cdp::browser_protocol::performance::EnableParams::default();
88                    cmds.push((
89                        enable_performance.identifier(),
90                        serde_json::to_value(enable_performance).unwrap_or_default(),
91                    ));
92                }
93
94                cmds
95            } else {
96                vec![]
97            }
98    };
99
100    /// Attach to target commands
101    static ref ATTACH_TARGET: (chromiumoxide_types::MethodId, serde_json::Value) = {
102        let runtime_cmd = RunIfWaitingForDebuggerParams::default();
103
104        (runtime_cmd.identifier(), serde_json::to_value(runtime_cmd).unwrap_or_default())
105    };
106}
107
108#[derive(Debug)]
109pub struct Target {
110    /// Info about this target as returned from the chromium instance
111    info: TargetInfo,
112    /// The type of this target
113    r#type: TargetType,
114    /// Configs for this target
115    config: TargetConfig,
116    /// The context this target is running in
117    browser_context: BrowserContext,
118    /// The frame manager that maintains the state of all frames and handles
119    /// navigations of frames
120    frame_manager: FrameManager,
121    /// Handles all the https
122    pub(crate) network_manager: NetworkManager,
123    emulation_manager: EmulationManager,
124    /// The identifier of the session this target is attached to
125    session_id: Option<SessionId>,
126    /// The handle of the browser page of this target
127    page: Option<PageHandle>,
128    /// Drives this target towards initialization
129    pub(crate) init_state: TargetInit,
130    /// Currently queued events to report to the `Handler`
131    queued_events: VecDeque<TargetEvent>,
132    /// All registered event subscriptions
133    event_listeners: EventListeners,
134    /// Senders that need to be notified once the main frame has loaded
135    wait_for_frame_navigation: Vec<Sender<ArcHttpRequest>>,
136    /// Senders that need to be notified once the main frame reaches `networkIdle`.
137    wait_for_network_idle: Vec<Sender<ArcHttpRequest>>,
138    /// (Optional) for `networkAlmostIdle` if you want it as well.
139    wait_for_network_almost_idle: Vec<Sender<ArcHttpRequest>>,
140    /// The sender who requested the page.
141    initiator: Option<Sender<Result<Page>>>,
142}
143
144impl Target {
145    /// Create a new target instance with `TargetInfo` after a
146    /// `CreateTargetParams` request.
147    pub fn new(info: TargetInfo, config: TargetConfig, browser_context: BrowserContext) -> Self {
148        let ty = TargetType::new(&info.r#type);
149        let request_timeout: Duration = config.request_timeout;
150        let mut network_manager = NetworkManager::new(config.ignore_https_errors, request_timeout);
151
152        if !config.cache_enabled {
153            network_manager.set_cache_enabled(false);
154        }
155
156        if !config.service_worker_enabled {
157            network_manager.set_service_worker_enabled(true);
158        }
159
160        network_manager.set_request_interception(config.request_intercept);
161        network_manager.max_bytes_allowed = config.max_bytes_allowed;
162
163        if let Some(ref headers) = config.extra_headers {
164            network_manager.set_extra_headers(headers.clone());
165        }
166
167        network_manager.ignore_visuals = config.ignore_visuals;
168        network_manager.block_javascript = config.ignore_javascript;
169        network_manager.block_analytics = config.ignore_analytics;
170        network_manager.block_stylesheets = config.ignore_stylesheets;
171        network_manager.only_html = config.only_html;
172        network_manager.intercept_manager = config.intercept_manager;
173
174        Self {
175            info,
176            r#type: ty,
177            config,
178            frame_manager: FrameManager::new(request_timeout),
179            network_manager,
180            emulation_manager: EmulationManager::new(request_timeout),
181            session_id: None,
182            page: None,
183            init_state: TargetInit::AttachToTarget,
184            wait_for_frame_navigation: Default::default(),
185            wait_for_network_idle: Default::default(),
186            wait_for_network_almost_idle: Default::default(),
187            queued_events: Default::default(),
188            event_listeners: Default::default(),
189            initiator: None,
190            browser_context,
191        }
192    }
193
194    /// Set the session id.
195    pub fn set_session_id(&mut self, id: SessionId) {
196        self.session_id = Some(id)
197    }
198
199    /// Get the session id.
200    pub fn session_id(&self) -> Option<&SessionId> {
201        self.session_id.as_ref()
202    }
203
204    /// Get the session id mut.
205    pub fn session_id_mut(&mut self) -> &mut Option<SessionId> {
206        &mut self.session_id
207    }
208
209    /// Get the browser context.
210    pub fn browser_context(&self) -> &BrowserContext {
211        &self.browser_context
212    }
213
214    /// The identifier for this target
215    pub fn target_id(&self) -> &TargetId {
216        &self.info.target_id
217    }
218
219    /// The type of this target
220    pub fn r#type(&self) -> &TargetType {
221        &self.r#type
222    }
223
224    /// Whether this target is already initialized
225    pub fn is_initialized(&self) -> bool {
226        matches!(self.init_state, TargetInit::Initialized)
227    }
228
229    /// Navigate a frame
230    pub fn goto(&mut self, req: FrameRequestedNavigation) {
231        self.frame_manager.goto(req)
232    }
233
234    /// Create a new page from the session.
235    fn create_page(&mut self) {
236        if self.page.is_none() {
237            if let Some(session) = self.session_id.clone() {
238                let handle =
239                    PageHandle::new(self.target_id().clone(), session, self.opener_id().cloned());
240                self.page = Some(handle);
241            }
242        }
243    }
244
245    /// Tries to create the `PageInner` if this target is already initialized
246    pub(crate) fn get_or_create_page(&mut self) -> Option<&Arc<PageInner>> {
247        self.create_page();
248        self.page.as_ref().map(|p| p.inner())
249    }
250
251    /// Is the target a page?
252    pub fn is_page(&self) -> bool {
253        self.r#type().is_page()
254    }
255
256    /// The browser context ID.
257    pub fn browser_context_id(&self) -> Option<&BrowserContextId> {
258        self.info.browser_context_id.as_ref()
259    }
260
261    /// The target connection info.
262    pub fn info(&self) -> &TargetInfo {
263        &self.info
264    }
265
266    /// Get the target that opened this target. Top-level targets return `None`.
267    pub fn opener_id(&self) -> Option<&TargetId> {
268        self.info.opener_id.as_ref()
269    }
270
271    pub fn frame_manager(&self) -> &FrameManager {
272        &self.frame_manager
273    }
274
275    /// The frame manager.
276    pub fn frame_manager_mut(&mut self) -> &mut FrameManager {
277        &mut self.frame_manager
278    }
279
280    /// Get event listeners mutably.
281    pub fn event_listeners_mut(&mut self) -> &mut EventListeners {
282        &mut self.event_listeners
283    }
284
285    /// Received a response to a command issued by this target
286    pub fn on_response(&mut self, resp: Response, method: &str) {
287        if let Some(cmds) = self.init_state.commands_mut() {
288            cmds.received_response(method);
289        }
290
291        if let GetFrameTreeParams::IDENTIFIER = method {
292            if let Some(resp) = resp
293                .result
294                .and_then(|val| GetFrameTreeParams::response_from_value(val).ok())
295            {
296                self.frame_manager.on_frame_tree(resp.frame_tree);
297            }
298        }
299        // requests originated from the network manager all return an empty response, hence they
300        // can be ignored here
301    }
302
303    /// On CDP Event message.
304    pub fn on_event(&mut self, event: CdpEventMessage) {
305        let CdpEventMessage { params, method, .. } = event;
306
307        match &params {
308            // `FrameManager` events
309            CdpEvent::PageFrameAttached(ev) => self
310                .frame_manager
311                .on_frame_attached(ev.frame_id.clone(), Some(ev.parent_frame_id.clone())),
312            CdpEvent::PageFrameDetached(ev) => self.frame_manager.on_frame_detached(ev),
313            CdpEvent::PageFrameNavigated(ev) => self.frame_manager.on_frame_navigated(&ev.frame),
314            CdpEvent::PageNavigatedWithinDocument(ev) => {
315                self.frame_manager.on_frame_navigated_within_document(ev)
316            }
317            CdpEvent::RuntimeExecutionContextCreated(ev) => {
318                self.frame_manager.on_frame_execution_context_created(ev)
319            }
320            CdpEvent::RuntimeExecutionContextDestroyed(ev) => {
321                self.frame_manager.on_frame_execution_context_destroyed(ev)
322            }
323            CdpEvent::RuntimeExecutionContextsCleared(_) => {
324                self.frame_manager.on_execution_contexts_cleared()
325            }
326            CdpEvent::RuntimeBindingCalled(ev) => {
327                // TODO check if binding registered and payload is json
328                self.frame_manager.on_runtime_binding_called(ev)
329            }
330            CdpEvent::PageLifecycleEvent(ev) => self.frame_manager.on_page_lifecycle_event(ev),
331            CdpEvent::PageFrameStartedLoading(ev) => {
332                self.frame_manager.on_frame_started_loading(ev);
333            }
334            CdpEvent::PageFrameStoppedLoading(ev) => {
335                self.frame_manager.on_frame_stopped_loading(ev);
336            }
337            // `Target` events
338            CdpEvent::TargetAttachedToTarget(ev) => {
339                if ev.waiting_for_debugger {
340                    let runtime_cmd = ATTACH_TARGET.clone();
341
342                    self.queued_events.push_back(TargetEvent::Request(Request {
343                        method: runtime_cmd.0,
344                        session_id: Some(ev.session_id.clone().into()),
345                        params: runtime_cmd.1,
346                    }));
347                }
348
349                if "service_worker" == &ev.target_info.r#type {
350                    let detach_command = DetachFromTargetParams::builder()
351                        .session_id(ev.session_id.clone())
352                        .build();
353
354                    let method = detach_command.identifier();
355
356                    if let Ok(params) = serde_json::to_value(detach_command) {
357                        self.queued_events.push_back(TargetEvent::Request(Request {
358                            method,
359                            session_id: self.session_id.clone().map(Into::into),
360                            params,
361                        }));
362                    }
363                }
364            }
365            // `NetworkManager` events
366            CdpEvent::FetchRequestPaused(ev) => self.network_manager.on_fetch_request_paused(ev),
367            CdpEvent::FetchAuthRequired(ev) => self.network_manager.on_fetch_auth_required(ev),
368            CdpEvent::NetworkRequestWillBeSent(ev) => {
369                self.network_manager.on_request_will_be_sent(ev)
370            }
371            CdpEvent::NetworkRequestServedFromCache(ev) => {
372                self.network_manager.on_request_served_from_cache(ev)
373            }
374            CdpEvent::NetworkResponseReceived(ev) => self.network_manager.on_response_received(ev),
375            CdpEvent::NetworkLoadingFinished(ev) => {
376                self.network_manager.on_network_loading_finished(ev)
377            }
378            CdpEvent::NetworkLoadingFailed(ev) => {
379                self.network_manager.on_network_loading_failed(ev)
380            }
381            _ => (),
382        }
383        chromiumoxide_cdp::consume_event!(match params {
384           |ev| self.event_listeners.start_send(ev),
385           |json| { let _ = self.event_listeners.try_send_custom(&method, json);}
386        });
387    }
388
389    /// Called when a init command timed out
390    fn on_initialization_failed(&mut self) -> TargetEvent {
391        if let Some(initiator) = self.initiator.take() {
392            let _ = initiator.send(Err(CdpError::Timeout));
393        }
394        self.init_state = TargetInit::Closing;
395        let close_target = CloseTargetParams::new(self.info.target_id.clone());
396
397        TargetEvent::Request(Request {
398            method: close_target.identifier(),
399            session_id: self.session_id.clone().map(Into::into),
400            params: serde_json::to_value(close_target).unwrap_or_default(),
401        })
402    }
403
404    /// Advance that target's state
405    pub(crate) fn poll(&mut self, cx: &mut Context<'_>, now: Instant) -> Option<TargetEvent> {
406        if !self.is_page() {
407            // can only poll pages
408            return None;
409        }
410
411        match &mut self.init_state {
412            TargetInit::AttachToTarget => {
413                self.init_state = TargetInit::InitializingFrame(FrameManager::init_commands(
414                    self.config.request_timeout,
415                ));
416
417                if let Ok(params) = AttachToTargetParams::builder()
418                    .target_id(self.target_id().clone())
419                    .flatten(true)
420                    .build()
421                {
422                    return Some(TargetEvent::Request(Request::new(
423                        params.identifier(),
424                        serde_json::to_value(params).unwrap_or_default(),
425                    )));
426                } else {
427                    return None;
428                }
429            }
430            TargetInit::InitializingFrame(cmds) => {
431                self.session_id.as_ref()?;
432                if let Poll::Ready(poll) = cmds.poll(now) {
433                    return match poll {
434                        None => {
435                            if let Some(world_name) = self.frame_manager.get_isolated_world_name() {
436                                let world_name = world_name.clone();
437
438                                if let Some(isolated_world_cmds) =
439                                    self.frame_manager.ensure_isolated_world(&world_name)
440                                {
441                                    *cmds = isolated_world_cmds;
442                                } else {
443                                    self.init_state = TargetInit::InitializingNetwork(
444                                        self.network_manager.init_commands(),
445                                    );
446                                }
447                            } else {
448                                self.init_state = TargetInit::InitializingNetwork(
449                                    self.network_manager.init_commands(),
450                                );
451                            }
452                            self.poll(cx, now)
453                        }
454                        Some(Ok((method, params))) => Some(TargetEvent::Request(Request {
455                            method,
456                            session_id: self.session_id.clone().map(Into::into),
457                            params,
458                        })),
459                        Some(Err(_)) => Some(self.on_initialization_failed()),
460                    };
461                } else {
462                    return None;
463                }
464            }
465            TargetInit::InitializingNetwork(cmds) => {
466                advance_state!(
467                    self,
468                    cx,
469                    now,
470                    cmds,
471                    TargetInit::InitializingPage(Self::page_init_commands(
472                        self.config.request_timeout
473                    ))
474                );
475            }
476            TargetInit::InitializingPage(cmds) => {
477                advance_state!(
478                    self,
479                    cx,
480                    now,
481                    cmds,
482                    match self.config.viewport.as_ref() {
483                        Some(viewport) => TargetInit::InitializingEmulation(
484                            self.emulation_manager.init_commands(viewport)
485                        ),
486                        None => TargetInit::Initialized,
487                    }
488                );
489            }
490            TargetInit::InitializingEmulation(cmds) => {
491                advance_state!(self, cx, now, cmds, TargetInit::Initialized);
492            }
493            TargetInit::Initialized => {
494                if let Some(initiator) = self.initiator.take() {
495                    // make sure that the main frame of the page has finished loading
496                    if self
497                        .frame_manager
498                        .main_frame()
499                        .map(|frame| frame.is_loaded())
500                        .unwrap_or_default()
501                    {
502                        if let Some(page) = self.get_or_create_page() {
503                            let _ = initiator.send(Ok(page.clone().into()));
504                        } else {
505                            self.initiator = Some(initiator);
506                        }
507                    } else {
508                        self.initiator = Some(initiator);
509                    }
510                }
511            }
512            TargetInit::Closing => return None,
513        };
514
515        loop {
516            if self.init_state == TargetInit::Closing {
517                break None;
518            }
519
520            if let Some(frame) = self.frame_manager.main_frame() {
521                if frame.is_loaded() {
522                    while let Some(tx) = self.wait_for_frame_navigation.pop() {
523                        let _ = tx.send(frame.http_request().cloned());
524                    }
525                }
526
527                if frame.is_network_idle() {
528                    while let Some(tx) = self.wait_for_network_idle.pop() {
529                        let _ = tx.send(frame.http_request().cloned());
530                    }
531                }
532
533                if frame.is_network_almost_idle() {
534                    while let Some(tx) = self.wait_for_network_almost_idle.pop() {
535                        let _ = tx.send(frame.http_request().cloned());
536                    }
537                }
538            }
539
540            // Drain queued messages first.
541            if let Some(ev) = self.queued_events.pop_front() {
542                return Some(ev);
543            }
544
545            if let Some(handle) = self.page.as_mut() {
546                while let Poll::Ready(Some(msg)) = Pin::new(&mut handle.rx).poll_next(cx) {
547                    if self.init_state == TargetInit::Closing {
548                        break;
549                    }
550
551                    match msg {
552                        TargetMessage::Command(cmd) => {
553                            if cmd.method == "Network.setBlockedURLs" {
554                                if let Some(arr) = cmd.params.get("urls").and_then(|v| v.as_array())
555                                {
556                                    let mut unblock_all = false;
557                                    let mut block_all = false;
558
559                                    for s in arr.iter().filter_map(|v| v.as_str()) {
560                                        if s == "!*" {
561                                            unblock_all = true;
562                                            break; // "!*" overrides any block rules
563                                        }
564                                        if s.contains('*') {
565                                            block_all = true;
566                                        }
567                                    }
568
569                                    if unblock_all {
570                                        self.network_manager.set_block_all(false);
571                                    } else if block_all {
572                                        self.network_manager.set_block_all(true);
573                                    }
574                                }
575                            }
576                            self.queued_events.push_back(TargetEvent::Command(cmd));
577                        }
578                        TargetMessage::MainFrame(tx) => {
579                            let _ =
580                                tx.send(self.frame_manager.main_frame().map(|f| f.id().clone()));
581                        }
582                        TargetMessage::AllFrames(tx) => {
583                            let _ = tx.send(
584                                self.frame_manager
585                                    .frames()
586                                    .map(|f| f.id().clone())
587                                    .collect(),
588                            );
589                        }
590                        TargetMessage::Url(req) => {
591                            let GetUrl { frame_id, tx } = req;
592                            let frame = if let Some(frame_id) = frame_id {
593                                self.frame_manager.frame(&frame_id)
594                            } else {
595                                self.frame_manager.main_frame()
596                            };
597                            let _ = tx.send(frame.and_then(|f| f.url().map(str::to_string)));
598                        }
599                        TargetMessage::Name(req) => {
600                            let GetName { frame_id, tx } = req;
601                            let frame = if let Some(frame_id) = frame_id {
602                                self.frame_manager.frame(&frame_id)
603                            } else {
604                                self.frame_manager.main_frame()
605                            };
606                            let _ = tx.send(frame.and_then(|f| f.name().map(str::to_string)));
607                        }
608                        TargetMessage::Parent(req) => {
609                            let GetParent { frame_id, tx } = req;
610                            let frame = self.frame_manager.frame(&frame_id);
611                            let _ = tx.send(frame.and_then(|f| f.parent_id().cloned()));
612                        }
613                        TargetMessage::WaitForNavigation(tx) => {
614                            if let Some(frame) = self.frame_manager.main_frame() {
615                                // TODO submit a navigation watcher: waitForFrameNavigation
616
617                                // TODO return the watchers navigationResponse
618                                if frame.is_loaded() {
619                                    let _ = tx.send(frame.http_request().cloned());
620                                } else {
621                                    self.wait_for_frame_navigation.push(tx);
622                                }
623                            } else {
624                                self.wait_for_frame_navigation.push(tx);
625                            }
626                        }
627                        TargetMessage::WaitForNetworkIdle(tx) => {
628                            if let Some(frame) = self.frame_manager.main_frame() {
629                                if frame.is_network_idle() {
630                                    let _ = tx.send(frame.http_request().cloned());
631                                } else {
632                                    self.wait_for_network_idle.push(tx);
633                                }
634                            } else {
635                                self.wait_for_network_idle.push(tx);
636                            }
637                        }
638                        TargetMessage::WaitForNetworkAlmostIdle(tx) => {
639                            if let Some(frame) = self.frame_manager.main_frame() {
640                                if frame.is_network_almost_idle() {
641                                    let _ = tx.send(frame.http_request().cloned());
642                                } else {
643                                    self.wait_for_network_almost_idle.push(tx);
644                                }
645                            } else {
646                                self.wait_for_network_almost_idle.push(tx);
647                            }
648                        }
649
650                        TargetMessage::AddEventListener(req) => {
651                            if req.method == "Fetch.requestPaused" {
652                                self.network_manager.disable_request_intercept();
653                            }
654                            // register a new listener
655                            self.event_listeners.add_listener(req);
656                        }
657                        TargetMessage::GetExecutionContext(ctx) => {
658                            let GetExecutionContext {
659                                dom_world,
660                                frame_id,
661                                tx,
662                            } = ctx;
663                            let frame = if let Some(frame_id) = frame_id {
664                                self.frame_manager.frame(&frame_id)
665                            } else {
666                                self.frame_manager.main_frame()
667                            };
668
669                            if let Some(frame) = frame {
670                                match dom_world {
671                                    DOMWorldKind::Main => {
672                                        let _ = tx.send(frame.main_world().execution_context());
673                                    }
674                                    DOMWorldKind::Secondary => {
675                                        let _ =
676                                            tx.send(frame.secondary_world().execution_context());
677                                    }
678                                }
679                            } else {
680                                let _ = tx.send(None);
681                            }
682                        }
683                        TargetMessage::Authenticate(credentials) => {
684                            self.network_manager.authenticate(credentials);
685                        }
686                    }
687                }
688            }
689
690            while let Some(event) = self.network_manager.poll() {
691                if self.init_state == TargetInit::Closing {
692                    break;
693                }
694                match event {
695                    NetworkEvent::SendCdpRequest((method, params)) => {
696                        // send a message to the browser
697                        self.queued_events.push_back(TargetEvent::Request(Request {
698                            method,
699                            session_id: self.session_id.clone().map(Into::into),
700                            params,
701                        }))
702                    }
703                    NetworkEvent::Request(_) => {}
704                    NetworkEvent::Response(_) => {}
705                    NetworkEvent::RequestFailed(request) => {
706                        self.frame_manager.on_http_request_finished(request);
707                    }
708                    NetworkEvent::RequestFinished(request) => {
709                        self.frame_manager.on_http_request_finished(request);
710                    }
711                    NetworkEvent::BytesConsumed(n) => {
712                        self.queued_events.push_back(TargetEvent::BytesConsumed(n));
713                    }
714                }
715            }
716
717            while let Some(event) = self.frame_manager.poll(now) {
718                if self.init_state == TargetInit::Closing {
719                    break;
720                }
721                match event {
722                    FrameEvent::NavigationResult(res) => {
723                        self.queued_events
724                            .push_back(TargetEvent::NavigationResult(res));
725                    }
726                    FrameEvent::NavigationRequest(id, req) => {
727                        self.queued_events
728                            .push_back(TargetEvent::NavigationRequest(id, req));
729                    }
730                }
731            }
732
733            if self.queued_events.is_empty() {
734                return None;
735            }
736        }
737    }
738
739    /// Set the sender half of the channel who requested the creation of this
740    /// target
741    pub fn set_initiator(&mut self, tx: Sender<Result<Page>>) {
742        self.initiator = Some(tx);
743    }
744
745    pub(crate) fn page_init_commands(timeout: Duration) -> CommandChain {
746        CommandChain::new(INIT_COMMANDS_PARAMS.clone(), timeout)
747    }
748}
749
750/// Configuration for how a single target/page should be fetched and processed.
751#[derive(Debug, Clone)]
752pub struct TargetConfig {
753    /// Whether to ignore TLS/HTTPS certificate errors (e.g. self-signed or expired certs).
754    /// When `true`, connections will proceed even if certificate validation fails.
755    pub ignore_https_errors: bool,
756    /// Request timeout to use for the main navigation / resource fetch.
757    /// This is the total time allowed before a request is considered failed.
758    pub request_timeout: Duration,
759    /// Optional browser viewport to use for this target.
760    /// When `None`, the default viewport (or headless browser default) is used.
761    pub viewport: Option<Viewport>,
762    /// Enable request interception for this target.
763    /// When `true`, all network requests will pass through the intercept manager.
764    pub request_intercept: bool,
765    /// Enable caching for this target.
766    /// When `true`, responses may be read from and written to the cache layer.
767    pub cache_enabled: bool,
768    /// If `true`, skip visual/asset resources that are not required for HTML content
769    /// (e.g. images, fonts, media). Useful for performance-oriented crawls.
770    pub ignore_visuals: bool,
771    /// If `true`, block JavaScript execution (or avoid loading JS resources)
772    /// for this target. This is useful for purely static HTML crawls.
773    pub ignore_javascript: bool,
774    /// If `true`, block analytics / tracking requests (e.g. Google Analytics,
775    /// common tracker domains, etc.).
776    pub ignore_analytics: bool,
777    /// If `true`, block stylesheets and related CSS resources for this target.
778    /// This can reduce bandwidth when only raw HTML is needed.
779    pub ignore_stylesheets: bool,
780    /// If `true`, only HTML documents will be fetched/kept.
781    /// Non-HTML subresources may be skipped entirely.
782    pub only_html: bool,
783    /// Whether service workers are allowed for this target.
784    /// When `true`, service workers may register and intercept requests.
785    pub service_worker_enabled: bool,
786    /// Extra HTTP headers to send with each request for this target.
787    /// Keys should be header names, values their corresponding header values.
788    pub extra_headers: Option<std::collections::HashMap<String, String>>,
789    /// Network intercept manager used to make allow/deny/modify decisions
790    /// for requests when `request_intercept` is enabled.
791    pub intercept_manager: NetworkInterceptManager,
792    /// The maximum number of response bytes allowed for this target.
793    /// When set, responses larger than this limit may be truncated or aborted.
794    pub max_bytes_allowed: Option<u64>,
795}
796
797impl Default for TargetConfig {
798    fn default() -> Self {
799        Self {
800            ignore_https_errors: true,
801            request_timeout: Duration::from_secs(REQUEST_TIMEOUT),
802            viewport: Default::default(),
803            request_intercept: false,
804            cache_enabled: true,
805            service_worker_enabled: true,
806            ignore_javascript: false,
807            ignore_visuals: false,
808            ignore_stylesheets: false,
809            ignore_analytics: true,
810            only_html: false,
811            extra_headers: Default::default(),
812            intercept_manager: NetworkInterceptManager::Unknown,
813            max_bytes_allowed: None,
814        }
815    }
816}
817
818#[derive(Debug, Clone, Eq, PartialEq)]
819pub enum TargetType {
820    Page,
821    BackgroundPage,
822    ServiceWorker,
823    SharedWorker,
824    Other,
825    Browser,
826    Webview,
827    Unknown(String),
828}
829
830impl TargetType {
831    pub fn new(ty: &str) -> Self {
832        match ty {
833            "page" => TargetType::Page,
834            "background_page" => TargetType::BackgroundPage,
835            "service_worker" => TargetType::ServiceWorker,
836            "shared_worker" => TargetType::SharedWorker,
837            "other" => TargetType::Other,
838            "browser" => TargetType::Browser,
839            "webview" => TargetType::Webview,
840            s => TargetType::Unknown(s.to_string()),
841        }
842    }
843
844    pub fn is_page(&self) -> bool {
845        matches!(self, TargetType::Page)
846    }
847
848    pub fn is_background_page(&self) -> bool {
849        matches!(self, TargetType::BackgroundPage)
850    }
851
852    pub fn is_service_worker(&self) -> bool {
853        matches!(self, TargetType::ServiceWorker)
854    }
855
856    pub fn is_shared_worker(&self) -> bool {
857        matches!(self, TargetType::SharedWorker)
858    }
859
860    pub fn is_other(&self) -> bool {
861        matches!(self, TargetType::Other)
862    }
863
864    pub fn is_browser(&self) -> bool {
865        matches!(self, TargetType::Browser)
866    }
867
868    pub fn is_webview(&self) -> bool {
869        matches!(self, TargetType::Webview)
870    }
871}
872
873#[derive(Debug)]
874pub(crate) enum TargetEvent {
875    /// An internal request
876    Request(Request),
877    /// An internal navigation request
878    NavigationRequest(NavigationId, Request),
879    /// Indicates that a previous requested navigation has finished
880    NavigationResult(Result<NavigationOk, NavigationError>),
881    /// A new command arrived via a channel
882    Command(CommandMessage),
883    /// The bytes consumed by the network.
884    BytesConsumed(u64),
885}
886
887// TODO this can be moved into the classes?
888#[derive(Debug, PartialEq)]
889pub enum TargetInit {
890    InitializingFrame(CommandChain),
891    InitializingNetwork(CommandChain),
892    InitializingPage(CommandChain),
893    InitializingEmulation(CommandChain),
894    AttachToTarget,
895    Initialized,
896    Closing,
897}
898
899impl TargetInit {
900    fn commands_mut(&mut self) -> Option<&mut CommandChain> {
901        match self {
902            TargetInit::InitializingFrame(cmd) => Some(cmd),
903            TargetInit::InitializingNetwork(cmd) => Some(cmd),
904            TargetInit::InitializingPage(cmd) => Some(cmd),
905            TargetInit::InitializingEmulation(cmd) => Some(cmd),
906            TargetInit::AttachToTarget => None,
907            TargetInit::Initialized => None,
908            TargetInit::Closing => None,
909        }
910    }
911}
912
913#[derive(Debug)]
914pub struct GetExecutionContext {
915    /// For which world the execution context was requested
916    pub dom_world: DOMWorldKind,
917    /// The if of the frame to get the `ExecutionContext` for
918    pub frame_id: Option<FrameId>,
919    /// Sender half of the channel to send the response back
920    pub tx: Sender<Option<ExecutionContextId>>,
921}
922
923impl GetExecutionContext {
924    pub fn new(tx: Sender<Option<ExecutionContextId>>) -> Self {
925        Self {
926            dom_world: DOMWorldKind::Main,
927            frame_id: None,
928            tx,
929        }
930    }
931}
932
933#[derive(Debug)]
934pub struct GetUrl {
935    /// The id of the frame to get the url for (None = main frame)
936    pub frame_id: Option<FrameId>,
937    /// Sender half of the channel to send the response back
938    pub tx: Sender<Option<String>>,
939}
940
941impl GetUrl {
942    pub fn new(tx: Sender<Option<String>>) -> Self {
943        Self { frame_id: None, tx }
944    }
945}
946
947#[derive(Debug)]
948pub struct GetName {
949    /// The id of the frame to get the name for (None = main frame)
950    pub frame_id: Option<FrameId>,
951    /// Sender half of the channel to send the response back
952    pub tx: Sender<Option<String>>,
953}
954
955#[derive(Debug)]
956pub struct GetParent {
957    /// The id of the frame to get the parent for (None = main frame)
958    pub frame_id: FrameId,
959    /// Sender half of the channel to send the response back
960    pub tx: Sender<Option<FrameId>>,
961}
962
963#[derive(Debug)]
964pub enum TargetMessage {
965    /// Execute a command within the session of this target
966    Command(CommandMessage),
967    /// Return the main frame of this target's page
968    MainFrame(Sender<Option<FrameId>>),
969    /// Return all the frames of this target's page
970    AllFrames(Sender<Vec<FrameId>>),
971    /// Return the url if available
972    Url(GetUrl),
973    /// Return the name if available
974    Name(GetName),
975    /// Return the parent id of a frame
976    Parent(GetParent),
977    /// A Message that resolves when the frame finished loading a new url
978    WaitForNavigation(Sender<ArcHttpRequest>),
979    /// A Message that resolves when the frame network is idle
980    WaitForNetworkIdle(Sender<ArcHttpRequest>),
981    /// A Message that resolves when the frame network is almost idle
982    WaitForNetworkAlmostIdle(Sender<ArcHttpRequest>),
983    /// A request to submit a new listener that gets notified with every
984    /// received event
985    AddEventListener(EventListenerRequest),
986    /// Get the `ExecutionContext` if available
987    GetExecutionContext(GetExecutionContext),
988    Authenticate(Credentials),
989}