blitz_script/

document.rs

//! [`ScriptDocument`]: a [`Document`] implementation with JavaScript support

use std::cell::RefCell;
use std::rc::Rc;
use std::sync::mpsc::{Receiver, RecvTimeoutError, Sender, channel};
use std::sync::{Arc, Mutex};
use std::task::{Context as TaskContext, Waker};

use blitz_dom::{
    BaseDocument, DEFAULT_CSS, DocGuard, DocGuardMut, Document, DocumentConfig, EventDriver, NodeId,
};
use blitz_html::stream::StreamingParser;
use blitz_html::{DocumentHtmlParser, HtmlProvider};
use blitz_traits::events::{BlitzPointerEvent, DomEvent, UiEvent};
use url::Url;
use web_time::{Duration, Instant};

use crate::event_handler::ScriptEventHandler;
use crate::fetch::{DefaultScriptFetcher, ScriptFetcher};
use crate::module::SharedFetcher;
use crate::runtime::ScriptRuntime;

type PollHook =
    Box<dyn for<'a> FnMut(&mut ScriptDocument, Option<&TaskContext<'a>>) -> bool + 'static>;

/// What a `<script>` element's `type` says it holds.
#[derive(Clone, Copy, PartialEq, Eq)]
enum ScriptKind {
    Classic,
    /// `type="module"`. Parsed in module goal and given a loader; a module run
    /// as a classic script fails on its own first `import` statement.
    Module,
    /// `type="importmap"`. Not code: the JSON that gives bare specifiers their
    /// meaning, and it has to be installed before any module resolves one.
    ImportMap,
}

/// A `<script>` element found in the document
struct PendingScript {
    node_id: NodeId,
    src: Option<String>,
    inline_text: String,
    kind: ScriptKind,
    /// Runs after every non-deferred script in the same batch.
    ///
    /// True for module scripts, which are deferred by definition, and for
    /// `<script defer src>`. `async` takes it back off both: an async script
    /// runs whenever its fetch lands, which here is immediately.
    deferred: bool,
}

/// A [`Document`] which executes the JavaScript contained in the document's
/// `<script>` tags, exposing DOM APIs backed by `blitz-dom` to the scripts.
///
/// Construct with [`ScriptDocument::from_html`], then call
/// [`execute_scripts`](ScriptDocument::execute_scripts) (this also happens
/// automatically on the first [`poll`](Document::poll)). UI events pushed via
/// [`handle_ui_event`](Document::handle_ui_event) are dispatched to JavaScript
/// event listeners before Blitz's default actions run.
pub struct ScriptDocument {
    inner: Rc<RefCell<BaseDocument>>,
    runtime: ScriptRuntime,
    base_url: Option<Url>,
    /// Shared with the module loader inside the script context.
    ///
    /// The loader is fixed at context construction, which happens in
    /// `from_html`, while an embedder installs its fetcher afterwards through
    /// [`with_fetcher`](Self::with_fetcher). The cell is what lets the later
    /// call reach the earlier object.
    fetcher: SharedFetcher,
    scripts_executed: bool,
    startup_events_pending: bool,
    load_event_pending: bool,
    /// Which `<script>` nodes have already run.
    ///
    /// By node rather than a single "done" flag, because scripts keep arriving
    /// after the first pass: a loader that appends its bundle, an analytics
    /// snippet, a lazily inserted chunk. Keyed this way, a rescan can tell a
    /// script it has run from one it has not without running anything twice.
    executed_scripts: std::collections::HashSet<NodeId>,
    poll_hook: Option<PollHook>,

    // Timer wakeups: a background thread which wakes the event loop (via the
    // `Waker` passed to `poll`) when the next JS timer is due.
    waker: Arc<Mutex<Option<Waker>>>,
    timer_thread: Option<Sender<Instant>>,
}

impl ScriptDocument {
    /// Parse HTML up to its first script boundary into a [`ScriptDocument`].
    ///
    /// Embedders install their shims on this live DOM before scripts execute.
    /// [`execute_scripts`](Self::execute_scripts), or the first `poll`, resumes
    /// the same parser without replacing any nodes.
    ///
    /// HTML without a script boundary is parsed completely before returning.
    /// XHTML keeps its non-streaming XML parser. No scripts execute here.
    pub fn from_html(html: &str, mut config: DocumentConfig) -> Self {
        if let Some(ss) = &mut config.ua_stylesheets {
            if !ss.iter().any(|s| s == DEFAULT_CSS) {
                ss.push(String::from(DEFAULT_CSS));
            }
        } else {
            config.ua_stylesheets = Some(vec![String::from(DEFAULT_CSS)]);
        }
        if config.html_parser_provider.is_none() {
            config.html_parser_provider = Some(Arc::new(HtmlProvider));
        }

        let base_url = config
            .base_url
            .as_deref()
            .and_then(|url| Url::parse(url).ok());

        let is_xml = DocumentHtmlParser::is_xhtml_document(html);
        let mut doc = BaseDocument::new(config);
        if is_xml {
            let mut mutr = doc.mutate();
            DocumentHtmlParser::parse_into_mutator(&mut mutr, html);
            drop(mutr);
        }

        let inner = Rc::new(RefCell::new(doc));
        let parser = if is_xml {
            None
        } else {
            let mut parser = StreamingParser::new(Rc::clone(&inner), html);
            if parser.next_script().is_some() {
                Some(parser)
            } else {
                parser.finish();
                None
            }
        };
        let fetcher: SharedFetcher = Rc::new(RefCell::new(Rc::new(DefaultScriptFetcher)));
        let runtime = ScriptRuntime::new(Rc::clone(&inner), base_url.as_ref(), Rc::clone(&fetcher));
        crate::docwrite::prepare(&runtime.context, html, Rc::clone(&fetcher), parser);

        Self {
            inner,
            runtime,
            base_url,
            fetcher,
            scripts_executed: false,
            startup_events_pending: false,
            load_event_pending: false,
            executed_scripts: std::collections::HashSet::new(),
            poll_hook: None,
            waker: Arc::new(Mutex::new(None)),
            timer_thread: None,
        }
    }

    /// Set the microtask budget shared by all drains in one poll.
    ///
    /// Defaults to 16 milliseconds. Unstarted jobs remain queued. Individual
    /// jobs run to completion unless the context's hard execution limit
    /// interrupts them. Panics if `duration` is zero.
    pub fn set_job_budget(&mut self, duration: Duration) {
        self.runtime.set_job_budget(duration);
    }

    /// The configured microtask budget.
    pub fn job_budget(&self) -> Duration {
        self.runtime.job_budget()
    }

    /// Whether immediate Boa jobs or observer records remain for a later host tick.
    ///
    /// Future timers and suspended module evaluations alone do not count.
    pub fn has_pending_jobs(&self) -> bool {
        self.runtime.jobs_remain()
    }

    /// Work queued for the next host tick: due timers and frame callbacks,
    /// plus one while any promise job is runnable.
    pub fn pending_task_count(&self) -> u64 {
        self.runtime.pending_task_count()
    }

    /// Drive scripts and runnable event-loop work to quiescence.
    ///
    /// Starts document scripts if necessary, then polls until parsing,
    /// deferred scripts, microtasks, observer delivery and timers due now
    /// have drained. Completion depends on pending work, not a fixed number
    /// of polls. Each poll retains the configured job budget.
    ///
    /// External scripts use the synchronous document fetcher. Imported module
    /// loads remain pending until their result futures complete, including
    /// dynamic imports from classic scripts. Pending loads are polled every
    /// five milliseconds when there is no runnable work. Suspended module
    /// evaluations also wait for their next timer or a later host poll.
    /// Unrelated future timers remain scheduled, and non-script resources
    /// are not awaited.
    ///
    /// Returns an error if module loading or evaluation remains pending after
    /// sixty seconds, or an unfinished document input stream cannot make
    /// progress. Synchronous fetchers must enforce their own I/O timeouts.
    /// A page that continuously produces runnable work without pending modules
    /// may never become idle.
    ///
    /// Call on the document's owning thread, outside poll hooks and
    /// JavaScript callbacks. This is a blocking entry point for tests and
    /// hosts that need completed script startup before inspecting the DOM.
    pub fn run_until_idle(&mut self) -> Result<(), &'static str> {
        const MODULE_TIMEOUT: Duration = Duration::from_secs(60);
        const LOAD_POLL_INTERVAL: Duration = Duration::from_millis(5);
        let module_deadline = Instant::now() + MODULE_TIMEOUT;

        loop {
            let progressed = self.poll(None);
            let now = Instant::now();
            let next_timer = self.runtime.next_timer_deadline();
            let loads_pending = self.runtime.has_pending_module_loads();
            let evaluations_pending = self.runtime.has_pending_module_evaluations();

            // Check before the progress branch so continuing module work
            // cannot bypass the deadline by keeping each poll busy.
            if now >= module_deadline {
                if loads_pending {
                    return Err("module loading did not complete within sixty seconds");
                }
                if evaluations_pending {
                    return Err("module evaluation did not complete within sixty seconds");
                }
            }

            if self.runtime.jobs_remain()
                || next_timer.is_some_and(|deadline| deadline <= now)
                || progressed
            {
                continue;
            }

            if loads_pending || evaluations_pending {
                // A timer is not required for a loader completion. While a
                // load is outstanding, keep polling even if the next timer
                // belongs to unrelated page work far in the future.
                let wait = if loads_pending {
                    LOAD_POLL_INTERVAL
                } else {
                    next_timer.map_or(LOAD_POLL_INTERVAL, |deadline| {
                        deadline.saturating_duration_since(now)
                    })
                };
                let wait = next_timer.map_or(wait, |deadline| {
                    wait.min(deadline.saturating_duration_since(now))
                });
                std::thread::sleep(wait.min(module_deadline.saturating_duration_since(now)));
                continue;
            }

            if crate::docwrite::active(&self.runtime.context)
                || self.startup_events_pending
                || self.collect_scripts().iter().any(|script| {
                    !self.executed_scripts.contains(&script.node_id)
                        && !crate::docwrite::was_executed(&self.runtime.context, script.node_id)
                })
            {
                return Err("script startup cannot make progress");
            }

            return Ok(());
        }
    }

    /// When the soonest page timer is due, so a host can sleep exactly until
    /// then instead of on a fixed tick (a `setTimeout(0)` continuation should
    /// not wait for the host's next frame).
    pub fn next_timer_deadline(&self) -> Option<std::time::Instant> {
        self.runtime.next_timer_deadline()
    }

    /// Retain the context's hard execution deadline across a host operation.
    ///
    /// Input, polling and evaluation enter scopes automatically. An embedder
    /// can retain an outer scope across several input events in one action.
    pub fn enter_execution_time_limit(&self) -> boa_engine::vm::ExecutionScope {
        self.runtime.context.enter_execution_time_limit()
    }
    /// Override the [`ScriptFetcher`] used to load external (`src="..."`) scripts.
    /// The default fetcher supports `file:` and `data:` URLs.
    pub fn with_fetcher(self, fetcher: impl ScriptFetcher) -> Self {
        *self.fetcher.borrow_mut() = Rc::new(fetcher);
        self
    }

    /// Install the host callback invoked by `window.ipc.postMessage(body)`.
    ///
    /// The callback may forward work to another thread, but JavaScript and DOM state remain on
    /// the document's owning thread. Replacing the callback is supported before script startup.
    pub fn set_ipc_handler(&mut self, handler: impl Fn(String) + 'static) {
        self.runtime.ctx.state.borrow_mut().ipc_handler = Some(Rc::new(handler));
    }

    /// Install work that an embedder needs to run on the document thread during polling.
    ///
    /// The hook runs after document scripts and due timers. Returning `true` requests a redraw.
    pub fn set_poll_hook(
        &mut self,
        hook: impl for<'a> FnMut(&mut ScriptDocument, Option<&TaskContext<'a>>) -> bool + 'static,
    ) {
        self.poll_hook = Some(Box::new(hook));
    }

    /// Append document-thread work without discarding an embedder's existing
    /// poll lifecycle.
    pub fn add_poll_hook(
        &mut self,
        mut hook: impl for<'a> FnMut(&mut ScriptDocument, Option<&TaskContext<'a>>) -> bool + 'static,
    ) {
        let Some(mut existing) = self.poll_hook.take() else {
            self.poll_hook = Some(Box::new(hook));
            return;
        };
        self.poll_hook = Some(Box::new(move |document, task_context| {
            let ran_existing = existing(document, task_context);
            let ran_added = hook(document, task_context);
            ran_existing | ran_added
        }));
    }

    /// Execute the document's `<script>` elements in document order, then fire
    /// the `DOMContentLoaded` and `load` events.
    ///
    /// Does nothing if scripts have already been executed.
    pub fn execute_scripts(&mut self) {
        let _execution = self.enter_execution_time_limit();
        let _profiling = self.runtime.ctx.enter_profiling_boundary();
        let _slice = self.runtime.begin_job_slice();
        if self.scripts_executed {
            return;
        }
        self.scripts_executed = true;
        self.startup_events_pending = true;

        crate::docwrite::begin(&mut self.runtime.context);
        self.resume_parser_and_scripts();
        self.finish_startup_events();

        self.request_redraw();
        self.arm_timer_thread();
    }

    /// The resolved URLs of the document's external (`<script src="...">`) scripts,
    /// in document order.
    ///
    /// The [`ScriptFetcher`] API is synchronous, so embedders with asynchronous
    /// networking can use this to prefetch script sources before calling
    /// [`execute_scripts`](Self::execute_scripts), and then serve them from memory
    /// via a custom fetcher (see [`with_fetcher`](Self::with_fetcher)).
    pub fn external_script_urls(&self) -> Vec<Url> {
        if let Some(original) = crate::docwrite::original(&self.runtime.context) {
            // The navigation DOM stops at its first boundary. Discover later
            // source scripts without creating native nodes or loading resources.
            return blitz_html::scan_script_tags(&original)
                .into_iter()
                .filter(|script| script.is_javascript())
                .filter_map(|script| script.src)
                .filter_map(|src| self.resolve_script_url(src.as_ref()))
                .collect();
        }
        self.collect_scripts()
            .iter()
            // An import map is inline-only per the HTML spec, so a `src` on one
            // names nothing an embedder should prefetch.
            .filter(|script| script.kind != ScriptKind::ImportMap)
            .filter_map(|script| script.src.as_deref())
            .filter_map(|src| self.resolve_script_url(src))
            .collect()
    }

    /// Resolve a script `src` attribute against the document's base URL
    fn resolve_script_url(&self, src: &str) -> Option<Url> {
        match &self.base_url {
            Some(base) => base.join(src).ok(),
            None => Url::parse(src).ok(),
        }
    }

    /// Extend the native JavaScript surface on the document's owning thread.
    ///
    /// The callback does not execute a microtask checkpoint or request a
    /// redraw. Poll hooks can therefore deliver native values and report work
    /// only when an event actually reaches the page.
    pub fn with_js_context<R>(
        &mut self,
        callback: impl FnOnce(&mut boa_engine::Context) -> R,
    ) -> R {
        let _execution = self.enter_execution_time_limit();
        callback(&mut self.runtime.context)
    }

    /// Evaluate arbitrary JavaScript code in the document's script context
    pub fn eval(&mut self, code: &str) {
        let _execution = self.enter_execution_time_limit();
        let _profiling = self.runtime.ctx.enter_profiling_boundary();
        self.runtime.eval(code, "<eval>");
        self.request_redraw();
        self.arm_timer_thread();
    }

    /// Evaluate an embedder prelude under a name, which becomes its source path
    /// in stack traces and profiles (for example `<chuzz-webgpu>`).
    pub fn eval_named(&mut self, code: &str, name: &str) {
        let _execution = self.enter_execution_time_limit();
        let _profiling = self.runtime.ctx.enter_profiling_boundary();
        self.runtime.eval_internal(code, name);
        self.runtime.run_jobs(name);
        self.request_redraw();
        self.arm_timer_thread();
    }

    /// Evaluate JavaScript and convert its result to JSON.
    ///
    /// Embedders use this for APIs that return an evaluation result, such as Tauri's
    /// `eval_script_with_callback`. JavaScript exceptions are recorded in the runtime diagnostics
    /// and returned as an error.
    pub fn eval_json(&mut self, code: &str) -> Result<serde_json::Value, String> {
        let _execution = self.enter_execution_time_limit();
        let _profiling = self.runtime.ctx.enter_profiling_boundary();
        let result = self.runtime.eval_json(code, "<eval with result>");
        self.request_redraw();
        self.arm_timer_thread();
        result
    }

    #[cfg(feature = "debug-control")]
    pub fn diagnostic_batch(
        &self,
        console_after: u64,
        errors_after: u64,
        limit: usize,
    ) -> crate::DiagnosticBatch {
        self.runtime
            .diagnostic_batch(console_after, errors_after, limit)
    }

    #[cfg(feature = "debug-control")]
    pub(crate) fn console_entries_after(
        &self,
        sequence: u64,
    ) -> Vec<crate::runtime::DiagnosticEntry> {
        self.runtime.console_entries_after(sequence)
    }

    #[cfg(feature = "debug-control")]
    pub(crate) fn runtime_errors_after(
        &self,
        sequence: u64,
    ) -> Vec<crate::runtime::DiagnosticEntry> {
        self.runtime.runtime_errors_after(sequence)
    }

    /// Current document URL for automation and diagnostics.
    pub fn current_url(&self) -> Option<&Url> {
        self.base_url.as_ref()
    }

    /// Serialize the current document tree as HTML.
    pub fn page_source(&self) -> String {
        self.inner.borrow().root_element().outer_html()
    }

    /// Dispatch a synthetic DOM event (e.g. a click created with
    /// [`Node::synthetic_click_event`](blitz_dom::Node::synthetic_click_event))
    /// through the document's event driver. The event is exposed to JavaScript
    /// event listeners, and Blitz's default actions run unless prevented.
    pub fn dispatch_dom_event(&mut self, event: DomEvent) {
        let _execution = self.enter_execution_time_limit();
        let profiling_boundary = self.runtime.ctx.enter_profiling_boundary();
        let profiling = profiling_boundary.enabled();
        let handler = ScriptEventHandler {
            runtime: &mut self.runtime,
            profiling,
        };
        let mut driver = EventDriver::new(&mut self.inner, handler);
        driver.handle_dom_event(event);

        self.request_redraw();
        self.arm_timer_thread();
    }

    /// Resume navigation input through completed script checkpoints.
    ///
    /// Startup and subsequent polls must use the same continuation loop.
    /// After a yielded checkpoint completes, keep feeding the retained input
    /// through each script boundary until EOF or another checkpoint yields.
    fn resume_parser_and_scripts(&mut self) -> bool {
        let _slice = self.runtime.begin_job_slice();
        let mut ran = false;
        loop {
            let progressed = self.run_pending_scripts();
            ran |= progressed;
            if !progressed
                || self.runtime.jobs_remain()
                || !crate::docwrite::active(&self.runtime.context)
            {
                break;
            }
        }
        ran
    }

    /// Run every `<script>` that has appeared since the last pass.
    ///
    /// Called on each poll as well as at startup, because script elements are
    /// not only a parser product: a page can build one and append it, and that
    /// is how most real loaders bring in their bundle. Running only the markup
    /// the parser produced meant those were fetched by nobody and the page
    /// simply never started.
    ///
    /// A `src` script reports `load` when it runs and `error` when it does not.
    /// Loaders wait on those before continuing — nofilter.io keeps the body
    /// hidden until the bundle's `load` arrives — so a script that ran silently
    /// would still leave the page blank.
    fn run_pending_scripts(&mut self) -> bool {
        let _slice = self.runtime.begin_job_slice();
        if self.runtime.jobs_remain() || self.runtime.job_slice_exhausted() {
            return false;
        }
        if crate::docwrite::take_restart(&self.runtime.context) {
            self.executed_scripts.clear();
            self.startup_events_pending = true;
            self.load_event_pending = false;
        }
        let parsed = crate::docwrite::advance(&mut self.runtime.context);
        let generation = crate::docwrite::generation(&self.runtime.context);
        if parsed {
            self.runtime.ctx.mark_layout_dirty();
        }
        // Collect first, then run: executing a script can append more, and the
        // borrow on the document has to be released before any of them runs.
        let pending: Vec<PendingScript> = self
            .collect_scripts()
            .into_iter()
            .filter(|script| !self.executed_scripts.contains(&script.node_id))
            .filter(|script| {
                !crate::docwrite::was_executed(&self.runtime.context, script.node_id)
                    && (!script.deferred || !crate::docwrite::active(&self.runtime.context))
            })
            .collect();
        let ran = parsed || !pending.is_empty();

        // Import maps first, across the whole batch. A page is free to write
        // its map after the module that needs it, and a map installed too late
        // is the same as no map at all: the module has already failed to
        // resolve its first bare specifier.
        for script in &pending {
            if script.kind == ScriptKind::ImportMap && !script.inline_text.trim().is_empty() {
                self.runtime
                    .set_import_map(&script.inline_text, self.base_url.as_ref());
            }
        }

        // Two passes, because a module is a deferred script and a browser runs
        // every non-deferred classic script before any of them. Measured
        // against Chromium on a page mixing all five forms: the parser-blocking
        // classics run first in document order, then modules and `defer`
        // scripts together, also in document order.
        //
        // Document order alone was right until modules existed. It is now
        // wrong in a way that bites: an inline classic script writing
        // `window.__CONFIG__` after a module tag runs *before* that module in
        // a browser, and the module reads the config it expects.
        let (deferred, immediate): (Vec<_>, Vec<_>) =
            pending.into_iter().partition(|script| script.deferred);

        for script in immediate.into_iter().chain(deferred) {
            if self.runtime.job_slice_exhausted() {
                break;
            }
            if generation != crate::docwrite::generation(&self.runtime.context) {
                break;
            }
            if self.startup_events_pending && script.deferred {
                self.runtime.set_document_ready_state("interactive");
                if self.runtime.jobs_remain() {
                    break;
                }
            }
            // Marked before running, not after: a script that appends another
            // copy of itself, or that throws, must not be retried on every
            // poll for the life of the page.
            self.executed_scripts.insert(script.node_id);
            if script.kind == ScriptKind::ImportMap {
                continue;
            }
            match script.src {
                Some(src) => {
                    let Some(url) = self.resolve_script_url(&src) else {
                        eprintln!("blitz-script: could not resolve script URL {src:?}");
                        self.runtime.dispatch_node_event(script.node_id, "error");
                        if self.runtime.jobs_remain() {
                            break;
                        }
                        continue;
                    };
                    let fetcher = Rc::clone(&self.fetcher.borrow());
                    match fetcher.fetch(&url) {
                        Ok(code) => {
                            if script.kind == ScriptKind::Module {
                                self.runtime.eval_module(&code, Some(&url), url.as_str());
                            } else {
                                crate::domc::set_current_script(
                                    &self.runtime.context,
                                    Some(script.node_id),
                                );
                                self.runtime.eval_at(&code, Some(&url), url.as_str());
                            }
                            self.runtime.dispatch_node_event(script.node_id, "load");
                        }
                        Err(error) => {
                            eprintln!("blitz-script: failed to fetch script {url}: {error}");
                            self.runtime.dispatch_node_event(script.node_id, "error");
                        }
                    }
                }
                None => {
                    if !script.inline_text.trim().is_empty() {
                        // An inline script has no URL of its own, so its
                        // relative imports and `import.meta.url` resolve
                        // against the document, exactly as in a browser.
                        let base_url = self.base_url.clone();
                        if script.kind == ScriptKind::Module {
                            self.runtime.eval_module(
                                &script.inline_text,
                                base_url.as_ref(),
                                "<inline module>",
                            );
                        } else {
                            crate::domc::set_current_script(
                                &self.runtime.context,
                                Some(script.node_id),
                            );
                            self.runtime.eval_at(
                                &script.inline_text,
                                base_url.as_ref(),
                                "<inline script>",
                            );
                        }
                    }
                }
            }
            // Preserve the checkpoint before the next script, even when it
            // takes several host ticks to finish the queued promise jobs.
            if self.runtime.jobs_remain() {
                break;
            }
        }
        ran
    }

    fn finish_startup_events(&mut self) -> bool {
        if !(self.startup_events_pending || self.load_event_pending)
            || self.runtime.jobs_remain()
            || crate::docwrite::active(&self.runtime.context)
            || self.runtime.job_slice_exhausted()
        {
            return false;
        }
        if self.startup_events_pending
            && (self.runtime.has_pending_module_evaluations()
                || self.collect_scripts().iter().any(|script| {
                    !self.executed_scripts.contains(&script.node_id)
                        && !crate::docwrite::was_executed(&self.runtime.context, script.node_id)
                }))
        {
            return false;
        }

        let mut ran = false;
        if self.startup_events_pending {
            ran |= self.runtime.set_document_ready_state("interactive");
            if self.runtime.jobs_remain() {
                return ran;
            }
            self.startup_events_pending = false;
            self.load_event_pending = true;
            self.runtime.dispatch_document_event("DOMContentLoaded");
            ran = true;
        }
        let resources_pending = {
            let doc = self.inner.borrow();
            doc.has_pending_critical_resources() || doc.pending_image_count() != 0
        };
        if self.load_event_pending
            && !self.runtime.job_slice_exhausted()
            && !self.runtime.jobs_remain()
            && !resources_pending
            && !self.runtime.has_pending_module_evaluations()
        {
            ran |= self.runtime.set_document_ready_state("complete");
            if !self.runtime.jobs_remain() {
                self.load_event_pending = false;
                self.runtime.dispatch_window_event("load");
                ran = true;
            }
        }
        ran
    }

    /// Find `<script>` elements in document order
    fn collect_scripts(&self) -> Vec<PendingScript> {
        let doc = self.inner.borrow();
        let mut scripts = Vec::new();
        let mut stack = vec![doc.root_node().id];

        while let Some(node_id) = stack.pop() {
            let Some(node) = doc.get_node(node_id) else {
                continue;
            };

            if let Some(element) = node.element_data() {
                if element.name.local == blitz_dom::local_name!("script") {
                    if node
                        .flags
                        .contains(blitz_dom::node::NodeFlags::IS_PARSER_INERT_SCRIPT)
                    {
                        continue;
                    }
                    // Skip non-JavaScript script types (e.g. JSON data blocks).
                    let script_type = element
                        .attr(blitz_dom::local_name!("type"))
                        .unwrap_or("")
                        .trim()
                        .to_ascii_lowercase();
                    let kind = match script_type.as_str() {
                        "module" => Some(ScriptKind::Module),
                        "importmap" => Some(ScriptKind::ImportMap),
                        "" | "text/javascript" | "application/javascript" => {
                            // `nomodule` marks the classic fallback a page
                            // ships for engines without module support. Now
                            // that modules run, taking the fallback as well
                            // would mount the same application twice.
                            if element.attr(blitz_dom::local_name!("nomodule")).is_some() {
                                None
                            } else {
                                Some(ScriptKind::Classic)
                            }
                        }
                        _ => None,
                    };

                    if let Some(kind) = kind {
                        let src = element
                            .attr(blitz_dom::local_name!("src"))
                            .map(str::to_string);
                        let is_async = element.attr(blitz_dom::local_name!("async")).is_some();

                        // `defer` is only meaningful on an external classic
                        // script; the spec ignores it on an inline one. A
                        // module is deferred whether or not it says so.
                        let deferred = !is_async
                            && match kind {
                                ScriptKind::Module => true,
                                ScriptKind::Classic => {
                                    src.is_some()
                                        && element.attr(blitz_dom::local_name!("defer")).is_some()
                                }
                                ScriptKind::ImportMap => false,
                            };

                        scripts.push(PendingScript {
                            node_id,
                            src,
                            inline_text: node.text_content(),
                            kind,
                            deferred,
                        });
                    }
                    continue;
                }
            }

            stack.extend(node.children.iter().rev().copied());
        }

        scripts
    }

    fn request_redraw(&self) {
        self.inner.borrow().shell_provider.request_redraw();
    }

    /// Ensure the timer thread is armed to wake the event loop when the next
    /// JS timer is due.
    fn arm_timer_thread(&mut self) {
        let Some(deadline) = self.runtime.next_timer_deadline() else {
            return;
        };

        let sender = self.timer_thread.get_or_insert_with(|| {
            let (tx, rx) = channel::<Instant>();
            let waker = Arc::clone(&self.waker);
            std::thread::Builder::new()
                .name("blitz-script-timers".to_string())
                .spawn(move || timer_thread_main(rx, waker))
                .expect("failed to spawn timer thread");
            tx
        });

        // If the thread has exited (channel disconnected) drop the sender so a
        // new thread is spawned next time.
        if sender.send(deadline).is_err() {
            self.timer_thread = None;
        }
    }
}

/// Background thread which wakes the event loop when JS timers are due
fn timer_thread_main(rx: Receiver<Instant>, waker: Arc<Mutex<Option<Waker>>>) {
    let mut deadline: Option<Instant> = None;

    loop {
        match deadline {
            None => match rx.recv() {
                Ok(new_deadline) => deadline = Some(new_deadline),
                Err(_) => return,
            },
            Some(current) => {
                let now = Instant::now();
                if current <= now {
                    if let Some(waker) = waker.lock().unwrap().as_ref() {
                        waker.wake_by_ref();
                    }
                    deadline = None;
                    continue;
                }
                match rx.recv_timeout(current - now) {
                    Ok(new_deadline) => deadline = Some(new_deadline.min(current)),
                    Err(RecvTimeoutError::Timeout) => {
                        if let Some(waker) = waker.lock().unwrap().as_ref() {
                            waker.wake_by_ref();
                        }
                        deadline = None;
                    }
                    Err(RecvTimeoutError::Disconnected) => return,
                }
            }
        }
    }
}

impl Document for ScriptDocument {
    fn inner(&self) -> DocGuard<'_> {
        DocGuard::RefCell(self.inner.borrow())
    }

    fn inner_mut(&mut self) -> DocGuardMut<'_> {
        DocGuardMut::RefCell(self.inner.borrow_mut())
    }

    fn handle_ui_event(&mut self, event: UiEvent) {
        let _execution = self.enter_execution_time_limit();
        let profiling_boundary = self.runtime.ctx.enter_profiling_boundary();
        let profiling = profiling_boundary.enabled();
        let handler = ScriptEventHandler {
            runtime: &mut self.runtime,
            profiling,
        };
        let mut driver = EventDriver::new(&mut self.inner, handler);
        driver.handle_ui_event(event);

        // JS may have mutated the DOM or scheduled timers
        self.request_redraw();
        self.arm_timer_thread();
    }

    fn poll(&mut self, task_context: Option<TaskContext>) -> bool {
        self.poll_with_timer_policy(task_context, None)
    }

    fn poll_for_settle(&mut self, task_context: Option<TaskContext>, run_timers: bool) -> bool {
        self.poll_with_timer_policy(task_context, Some(run_timers))
    }
}

impl ScriptDocument {
    /// `None` preserves normal polling; `Some` selects a settle pass's policy.
    fn poll_with_timer_policy(
        &mut self,
        task_context: Option<TaskContext>,
        settle_timers: Option<bool>,
    ) -> bool {
        let _execution = self.enter_execution_time_limit();
        let profiling_boundary = self.runtime.ctx.enter_profiling_boundary();
        let profiling = profiling_boundary.enabled();
        let poll_started = profiling.then(std::time::Instant::now);
        let _slice = self.runtime.begin_job_slice();
        let ran = self.poll_inner(task_context, profiling, settle_timers);
        if let Some(started) = poll_started {
            crate::script_stats::record_poll(started.elapsed(), ran);
        }
        ran
    }

    /// Move a semantic automation pointer to the exact resolved DOM node.
    ///
    /// Normal window input remains coordinate hit-tested through
    /// [`Document::handle_ui_event`]. Debug-control callers already selected a
    /// node by id, so routing those coordinates through hit testing again can
    /// silently select an overlapping descendant or retained overlay.
    pub fn handle_pointer_move_to_node(&mut self, event: BlitzPointerEvent, node_id: NodeId) {
        let _execution = self.enter_execution_time_limit();
        let profiling_boundary = self.runtime.ctx.enter_profiling_boundary();
        let profiling = profiling_boundary.enabled();
        let handler = ScriptEventHandler {
            runtime: &mut self.runtime,
            profiling,
        };
        let mut driver = EventDriver::new(&mut self.inner, handler);
        driver.handle_pointer_move_to_node(&event, node_id);

        self.request_redraw();
        self.arm_timer_thread();
    }

    /// Deliver one automation input event to the exact semantic node selected
    /// by the caller, while retaining the normal pointer event ordering,
    /// compatibility mouse events, activation state and click synthesis.
    pub fn handle_ui_event_to_node(&mut self, event: UiEvent, node_id: NodeId) {
        let _execution = self.enter_execution_time_limit();
        let profiling_boundary = self.runtime.ctx.enter_profiling_boundary();
        let profiling = profiling_boundary.enabled();
        let handler = ScriptEventHandler {
            runtime: &mut self.runtime,
            profiling,
        };
        let mut driver = EventDriver::new(&mut self.inner, handler);
        driver.handle_ui_event_to_node(event, node_id);

        self.request_redraw();
        self.arm_timer_thread();
    }

    /// The real poll. Split out so every exit path is timed by the wrapper
    /// above rather than by a stopwatch threaded through each early return.
    fn poll_inner(
        &mut self,
        task_context: Option<TaskContext>,
        profiling: bool,
        settle_timers: Option<bool>,
    ) -> bool {
        // Store the waker so the timer thread can wake the event loop
        if let Some(cx) = &task_context {
            let mut waker = self.waker.lock().unwrap();
            let stale = waker
                .as_ref()
                .map(|old| !old.will_wake(cx.waker()))
                .unwrap_or(true);
            if stale {
                *waker = Some(cx.waker().clone());
            }
        }

        // A scripted document may itself be an embedder. Chuzz's Solid chrome
        // is one: its `<web-view>` elements own the page documents. Poll those
        // children at the same outer boundary so their timers, resource
        // completions, and script work continue to make progress.
        let subdocument_changes = {
            let mut inner = self.inner.borrow_mut();
            let waker = task_context.as_ref().map(TaskContext::waker);
            match settle_timers {
                Some(run_timers) => inner.poll_subdocuments_for_settle(waker, run_timers),
                None => inner.poll_subdocuments(waker),
            }
        };

        let mut ran = subdocument_changes;
        ran |= self.runtime.poll_domc();
        // Every poll resumes a yielded checkpoint, including a timer-free
        // settle pass. Jobs do not need a timer or a dummy eval to advance.
        ran |= self.runtime.run_jobs_with_progress("poll microtasks");

        // Execute scripts on first poll if they haven't been run explicitly
        if !self.scripts_executed {
            // One-time: parsing and running the application bundle. Separated
            // because it is startup cost, and folding it into the steady-state
            // numbers made every per-poll average meaningless.
            let started = profiling.then(std::time::Instant::now);
            self.execute_scripts();
            if let Some(started) = started {
                crate::script_stats::record_work("startup:execute_scripts", started.elapsed());
            }
            ran = true;
        } else {
            // Resume navigation after a yielded checkpoint using the same
            // continuation loop as startup. A completed script boundary must
            // not leave the remaining input waiting for another host poll.
            // After EOF, this still picks up scripts appended by the page.
            ran |= self.resume_parser_and_scripts();
        }
        ran |= self.finish_startup_events();

        if settle_timers.unwrap_or(true) {
            ran |= self.runtime.run_due_timers(profiling);
        }

        // A module that opened with a top-level `await` may have settled since
        // the last turn. Checking here is what turns a silent half-mounted page
        // into a reported rejection.
        self.runtime.poll_module_evaluations();

        if let Some(mut hook) = self.poll_hook.take() {
            // The embedder's per-poll work. For a Solid application this is
            // where reactive updates and DOM mutation actually happen, so it is
            // the bucket that matters once startup is excluded.
            let started = profiling.then(std::time::Instant::now);
            ran |= hook(self, task_context.as_ref());
            if let Some(started) = started {
                crate::script_stats::record_work("poll_hook", started.elapsed());
            }
            self.poll_hook = Some(hook);
        }

        // Reclaim what removal could not judge at the time. A node is detached
        // while script may still hold its wrapper, and the answer only becomes
        // knowable once the collector has run; this is the point where asking
        // is cheap and the answer is current.
        self.runtime.sweep_detached_nodes();

        if self.runtime.jobs_remain() {
            ran = true;
            self.request_redraw();
            if let Some(cx) = &task_context {
                cx.waker().wake_by_ref();
            }
        }
        self.arm_timer_thread();
        ran
    }
}