char/avril

git clone https://git.t4t.associates/char/avril

Charlotte Somadd progress to module imports from registries7b011462026-10-23

use std::future::Future;
use std::io::{self, BufRead, IsTerminal, Write};
use std::sync::Arc;
use std::sync::atomic::{AtomicBool, Ordering};
use std::sync::mpsc::{self, Receiver};
use std::task::Poll;
use std::thread;
use std::time::Duration;

use avril_core::{Error, Runtime};
use futures::StreamExt;
use futures::channel::mpsc::{self as async_mpsc, UnboundedReceiver, UnboundedSender};
use futures::future::poll_fn;
use oxc::allocator::Allocator;
use oxc::ast::ast::{Expression, ImportAttributeKey, ImportDeclarationSpecifier, Statement};
use oxc::parser::Parser;
use oxc::span::{GetSpan, SourceType};
use reedline::Signal;
use serde_json::{Value, json};
use url::Url;

mod line_editor;

enum Input {
    Line(String),
    Eof,
    Cancel,
    Interrupt,
    Error(String),
}

enum Activity {
    Input(Option<Input>),
    RuntimeError(Error),
    StalledEvaluation,
    Background,
}

struct InterruptState {
    active: Arc<AtomicBool>,
    interrupted: Arc<AtomicBool>,
    handle: avril_core::v8::IsolateHandle,
}

pub(crate) fn run(
    runtime: &mut Runtime,
    referrer: &str,
    stdio: &avril_ext_stdio::Stdio,
    host: &tokio::runtime::Handle,
    reporter: Option<&crate::progress::Reporter>,
) -> Result<(), Error> {
    let source_url = Url::from_file_path(referrer)
        .map_err(|_| Error::JavaScript(format!("Invalid REPL referrer: {referrer}")))?
        .to_string();
    runtime.enable_repl()?;

    let active = Arc::new(AtomicBool::new(false));
    let interrupted = stdio.interrupt_flag();
    let handle = runtime.isolate_handle();
    let (input_sender, input_receiver) = async_mpsc::unbounded();
    let (request, requests) = mpsc::channel::<()>();
    let terminal = io::stdin().is_terminal() && io::stderr().is_terminal();
    let input_thread_sender = input_sender.clone();
    let input_thread = thread::Builder::new()
        .name("avril:repl-input".into())
        .spawn(move || read_input(terminal, requests, input_thread_sender))
        .map_err(|error| Error::JavaScript(error.to_string()));

    let signal_active = Arc::clone(&active);
    let signal_interrupted = Arc::clone(&interrupted);
    let signal_handle = handle.clone();
    let signal_sender = input_sender;
    let result = match input_thread {
        Err(error) => Err(error),
        Ok(_input_thread) => {
            // Route Ctrl-C through the shared chaining registry instead of
            // ctrlc's overwriting handler, so REPL listeners observe the
            // interrupt rather than being clobbered by it.
            let on_interrupt = move || {
                signal_interrupted.store(true, Ordering::SeqCst);
                let _ = signal_sender.unbounded_send(Input::Interrupt);
                if signal_active.load(Ordering::SeqCst) {
                    // A blocking runSync cannot observe V8 termination, so
                    // stop exactly the subprocess it is waiting on. Other
                    // session work is left alone.
                    avril_ext_process::Tree::global().stop_sync_processes();
                    signal_handle.terminate_execution();
                }
            };
            #[cfg(unix)]
            avril_ext_process::Tree::global()
                .claim_interrupt(host, Arc::new(on_interrupt))
                .map_err(|error| Error::JavaScript(error.to_string()))?;
            #[cfg(not(unix))]
            ctrlc::set_handler(on_interrupt)
                .map_err(|error| Error::JavaScript(error.to_string()))?;
            run_session(Repl {
                runtime,
                host,
                source_url: &source_url,
                request,
                input: input_receiver,
                interrupts: InterruptState { active, interrupted, handle },
                stdio,
                terminal,
                reporter,
            })
        }
    };
    let disabled = runtime.disable_repl();
    result?;
    disabled
}

/// One interactive session's runtime, input plumbing, and terminal state.
struct Repl<'a> {
    runtime: &'a mut Runtime,
    host: &'a tokio::runtime::Handle,
    source_url: &'a str,
    request: mpsc::Sender<()>,
    input: UnboundedReceiver<Input>,
    interrupts: InterruptState,
    stdio: &'a avril_ext_stdio::Stdio,
    terminal: bool,
    reporter: Option<&'a crate::progress::Reporter>,
}

fn run_session(
    Repl {
        runtime,
        host,
        source_url,
        request,
        input,
        interrupts,
        stdio,
        terminal,
        reporter,
    }: Repl<'_>,
) -> Result<(), Error> {
    let mut input = input;
    runtime.execute_script::<()>("<repl>", "globalThis._ = undefined")?;
    let InterruptState { active, interrupted, handle } = &interrupts;
    let mut source = String::new();
    let mut input_pending = false;
    'session: loop {
        if !input_pending {
            request
                .send(())
                .map_err(|error| Error::JavaScript(error.to_string()))?;
            input_pending = true;
        }
        let line = loop {
            match host.block_on(next_activity(runtime, &mut input, active, None, terminal)) {
                Activity::Input(Some(Input::Line(line))) => {
                    input_pending = false;
                    if interrupted.swap(false, Ordering::SeqCst) {
                        handle.cancel_terminate_execution();
                        source.clear();
                        continue 'session;
                    }
                    break Some(line);
                }
                Activity::Input(Some(Input::Eof)) | Activity::Input(None) => break None,
                Activity::Input(Some(Input::Cancel)) => {
                    interrupted.store(false, Ordering::SeqCst);
                    handle.cancel_terminate_execution();
                    source.clear();
                    input_pending = false;
                    continue 'session;
                }
                Activity::Input(Some(Input::Interrupt)) => {
                    interrupted.store(false, Ordering::SeqCst);
                    handle.cancel_terminate_execution();
                    source.clear();
                    continue 'session;
                }
                Activity::Input(Some(Input::Error(error))) => {
                    return Err(Error::JavaScript(error));
                }
                Activity::RuntimeError(error) => {
                    if interrupted.swap(false, Ordering::SeqCst) {
                        handle.cancel_terminate_execution();
                        eprintln!("Evaluation interrupted");
                        source.clear();
                    } else {
                        eprintln!("{error}");
                    }
                }
                Activity::Background | Activity::StalledEvaluation => {}
            }
        };
        let Some(line) = line else {
            break;
        };
        if source.is_empty() {
            match line.trim() {
                ".exit" | ".quit" => break,
                ".help" => {
                    println!(
                        "Enter JavaScript; incomplete expressions continue on the next line.\n.help  Show this help\n.exit  Exit the REPL"
                    );
                    continue;
                }
                _ => {}
            }
        }
        if line.trim().is_empty() && source.is_empty() {
            continue;
        }
        source.push_str(&line);
        source.push('\n');
        if is_incomplete(&source) {
            continue;
        }

        let expression = prepare_imports(&prepare_expression(&source));
        source.clear();
        let shown = Session {
            runtime: &mut *runtime,
            host,
            input: &mut input,
            interrupts: &interrupts,
            stdio,
            terminal,
            reporter,
        }
        .show(&expression, source_url);
        runtime.release_repl_objects()?;
        if shown? == Waited::Eof {
            break;
        }
    }
    Ok(())
}

// Results go through avril:std's inspect so they match console.log and honor
// custom inspectors; the inspector can only reach it from the guest realm.
const INSPECT: &str = r#"async function (value, colors, remember) {
    if (remember) this._ = value;
    const { inspect } = await import("avril:std");
    return inspect(value, { colors });
}"#;

#[derive(PartialEq)]
enum Waited {
    Response(String),
    Abandoned,
    Eof,
}

struct Session<'a> {
    runtime: &'a mut Runtime,
    host: &'a tokio::runtime::Handle,
    input: &'a mut UnboundedReceiver<Input>,
    interrupts: &'a InterruptState,
    stdio: &'a avril_ext_stdio::Stdio,
    terminal: bool,
    reporter: Option<&'a crate::progress::Reporter>,
}

struct ProgressScope<'a>(&'a crate::progress::Reporter);

impl Drop for ProgressScope<'_> {
    fn drop(&mut self) {
        self.0.scope(false);
    }
}

impl Session<'_> {
    fn show(&mut self, expression: &str, source_url: &str) -> Result<Waited, Error> {
        let response = match self.run(|runtime| runtime.evaluate_repl(expression, source_url))? {
            Waited::Response(response) => response,
            waited => return Ok(waited),
        };
        let (value, stderr) = match completion(&response) {
            Ok(completion) => completion,
            Err(message) => {
                eprintln!("{message}");
                return Ok(Waited::Abandoned);
            }
        };
        let colors = json!({ "value": avril_ext_console::colors(stderr) });
        let remember = json!({ "value": !stderr });
        let response = match self
            .run(|runtime| runtime.call_repl_function(INSPECT, &[value, colors, remember]))?
        {
            Waited::Response(response) => response,
            waited => return Ok(waited),
        };
        let text = inspected(&response);
        if stderr {
            eprintln!("{text}");
        } else {
            println!("{text}");
        }
        io::stdout()
            .flush()
            .map_err(|error| Error::JavaScript(error.to_string()))?;
        Ok(Waited::Abandoned)
    }

    /// Dispatch an inspector command and drive the event loop until it answers,
    /// unless the user interrupts it or input ends first.
    fn run(
        &mut self,
        dispatch: impl FnOnce(&mut Runtime) -> Result<i32, Error>,
    ) -> Result<Waited, Error> {
        let progress_scope = self.reporter.map(|reporter| {
            reporter.scope(true);
            ProgressScope(reporter)
        });
        let InterruptState { active, interrupted, handle } = self.interrupts;
        active.store(true, Ordering::SeqCst);
        if interrupted.swap(false, Ordering::SeqCst) {
            active.store(false, Ordering::SeqCst);
            handle.cancel_terminate_execution();
            return Ok(Waited::Abandoned);
        }
        let dispatched = dispatch(self.runtime);
        active.store(false, Ordering::SeqCst);
        let id = match dispatched {
            Ok(id) => id,
            Err(_) if interrupted.swap(false, Ordering::SeqCst) => {
                handle.cancel_terminate_execution();
                drop(progress_scope);
                eprintln!("Evaluation interrupted");
                return Ok(Waited::Abandoned);
            }
            Err(error) => return Err(error),
        };
        loop {
            if let Some(response) = self.runtime.take_repl_response(id) {
                self.stdio
                    .flush_interruptible(|| interrupted.load(Ordering::SeqCst))
                    .map_err(|error| Error::JavaScript(error.to_string()))?;
                if interrupted.swap(false, Ordering::SeqCst) {
                    handle.cancel_terminate_execution();
                    drop(progress_scope);
                    eprintln!("Evaluation interrupted");
                    return Ok(Waited::Abandoned);
                }
                return Ok(Waited::Response(response));
            }
            match self.host.block_on(next_activity(
                self.runtime,
                self.input,
                active,
                Some(id),
                self.terminal,
            )) {
                Activity::Input(Some(Input::Interrupt | Input::Cancel)) => {
                    if interrupted.swap(false, Ordering::SeqCst) {
                        self.runtime.terminate_repl_execution()?;
                        self.abandon(id);
                        drop(progress_scope);
                        eprintln!("Evaluation interrupted");
                        return Ok(Waited::Abandoned);
                    }
                }
                Activity::RuntimeError(error) => {
                    if interrupted.swap(false, Ordering::SeqCst) {
                        self.abandon(id);
                        drop(progress_scope);
                        eprintln!("Evaluation interrupted");
                        return Ok(Waited::Abandoned);
                    }
                    if let Some(reporter) = self.reporter {
                        reporter.flush();
                    }
                    eprintln!("{error}");
                }
                Activity::Input(Some(Input::Error(error))) => {
                    return Err(Error::JavaScript(error));
                }
                Activity::Input(Some(Input::Eof)) | Activity::Input(None) => {
                    self.runtime.terminate_repl_execution()?;
                    self.abandon(id);
                    return Ok(Waited::Eof);
                }
                Activity::StalledEvaluation => {
                    self.runtime.terminate_repl_execution()?;
                    self.abandon(id);
                    drop(progress_scope);
                    eprintln!("REPL evaluation is still pending with no runnable host work");
                    return Ok(Waited::Abandoned);
                }
                Activity::Input(Some(Input::Line(_))) | Activity::Background => {}
            }
        }
    }

    fn abandon(&mut self, id: i32) {
        self.interrupts.handle.cancel_terminate_execution();
        let _ = self.runtime.take_repl_response(id);
    }
}

fn next_activity<'a>(
    runtime: &'a mut Runtime,
    input: &'a mut UnboundedReceiver<Input>,
    active: &'a AtomicBool,
    evaluation: Option<i32>,
    terminal: bool,
) -> impl Future<Output = Activity> + 'a {
    let mut background_delay = None;
    poll_fn(move |context| {
        if evaluation.is_some_and(|id| runtime.has_repl_response(id)) {
            return Poll::Ready(Activity::Background);
        }
        if let Poll::Ready(input) = input.poll_next_unpin(context) {
            return Poll::Ready(Activity::Input(input));
        }
        let background_pending = runtime.has_pending_background_tasks();
        active.store(true, Ordering::SeqCst);
        let result = runtime.poll_event_loop(context);
        active.store(false, Ordering::SeqCst);
        match result {
            Err(error) => return Poll::Ready(Activity::RuntimeError(error)),
            Ok(true) => return Poll::Ready(Activity::Background),
            Ok(false) => {}
        }
        // The checkpoint can finish an inspector evaluation without a native completion.
        if evaluation.is_some_and(|id| runtime.has_repl_response(id)) {
            return Poll::Ready(Activity::Background);
        }
        // The default V8 platform does not notify the host about foreground tasks.
        if background_pending || runtime.has_pending_background_tasks() {
            let delay = background_delay
                .get_or_insert_with(|| Box::pin(tokio::time::sleep(Duration::from_millis(20))));
            if delay.as_mut().poll(context).is_ready() {
                return Poll::Ready(Activity::Background);
            }
        } else {
            background_delay = None;
            if !terminal && evaluation.is_some() && !runtime.has_pending_work() {
                return Poll::Ready(Activity::StalledEvaluation);
            }
        }
        Poll::Pending
    })
}

fn read_input(terminal: bool, requests: Receiver<()>, input: UnboundedSender<Input>) {
    if terminal {
        let mut editor = line_editor::create();
        for () in requests {
            let line = match editor.read_line(&line_editor::Prompt) {
                Ok(Signal::Success(line)) => Input::Line(line),
                Ok(Signal::CtrlC) => Input::Cancel,
                Ok(Signal::CtrlD) => Input::Eof,
                Err(error) => Input::Error(error.to_string()),
            };
            let eof = matches!(line, Input::Eof | Input::Error(_));
            if input.unbounded_send(line).is_err() || eof {
                break;
            }
        }
        return;
    }

    let stdin = io::stdin();
    let mut stdin = stdin.lock();
    for _ in requests {
        let mut line = String::new();
        let line = loop {
            match stdin.read_line(&mut line) {
                Ok(0) => break Input::Eof,
                Ok(_) => {
                    break Input::Line(line.trim_end_matches(['\r', '\n']).to_owned());
                }
                Err(error) if error.kind() == io::ErrorKind::Interrupted => continue,
                Err(error) => break Input::Error(error.to_string()),
            }
        };
        let eof = matches!(line, Input::Eof | Input::Error(_));
        if input.unbounded_send(line).is_err() || eof {
            break;
        }
    }
}

fn is_incomplete(source: &str) -> bool {
    let allocator = Allocator::default();
    let parsed = Parser::new(&allocator, source, SourceType::mjs()).parse();
    if parsed.diagnostics.is_empty() || prepare_expression(source) != source {
        return false;
    }
    let incomplete = |candidate: &str| {
        let allocator = Allocator::default();
        let parsed = Parser::new(&allocator, candidate, SourceType::mjs()).parse();
        let eof = candidate.len();
        !parsed.diagnostics.is_empty()
            && parsed.diagnostics.iter().all(|error| {
                let message = error.message.to_ascii_lowercase();
                let at_end = error
                    .labels
                    .iter()
                    .any(|label| label.offset().saturating_add(label.len()) as usize >= eof);
                if message.contains("unterminated string") {
                    error.labels.iter().any(|label| {
                        candidate.as_bytes().get(label.offset() as usize) == Some(&b'`')
                    }) || candidate.ends_with("\\\n")
                } else {
                    at_end && !message.contains("unterminated regular expression")
                }
            })
    };
    incomplete(source) || incomplete(&format!("({source}"))
}

fn prepare_expression(source: &str) -> String {
    let wrapped = format!("({source})");
    let allocator = Allocator::default();
    let parsed = Parser::new(&allocator, &wrapped, SourceType::mjs()).parse();
    if parsed.diagnostics.is_empty()
        && matches!(
            parsed.program.body.as_slice(),
            [Statement::ExpressionStatement(statement)]
                if matches!(statement.expression.without_parentheses(), Expression::ObjectExpression(_))
        )
    {
        return wrapped;
    }
    source.to_owned()
}

fn prepare_imports(source: &str) -> String {
    let allocator = Allocator::default();
    let parsed = Parser::new(&allocator, source, SourceType::mjs()).parse();
    if !parsed.diagnostics.is_empty() {
        return source.to_owned();
    }
    let mut result = String::new();
    let mut offset = 0;
    for statement in &parsed.program.body {
        let Statement::ImportDeclaration(import) = statement else {
            continue;
        };
        if import.phase.is_some() {
            return source.to_owned();
        }
        let mut properties = Vec::new();
        let mut namespace = None;
        for specifier in import.specifiers.iter().flatten() {
            match specifier {
                ImportDeclarationSpecifier::ImportDefaultSpecifier(specifier) => {
                    properties.push(format!("\"default\": {}", specifier.local.name));
                }
                ImportDeclarationSpecifier::ImportSpecifier(specifier) => {
                    let span = specifier.imported.span();
                    let key = &source[span.start as usize..span.end as usize];
                    properties.push(format!("{key}: {}", specifier.local.name));
                }
                ImportDeclarationSpecifier::ImportNamespaceSpecifier(specifier) => {
                    namespace = Some(specifier.local.name.as_str());
                }
            }
        }
        let mut options = String::new();
        if let Some(clause) = &import.with_clause {
            let attributes = clause
                .with_entries
                .iter()
                .map(|attribute| {
                    let key = match &attribute.key {
                        ImportAttributeKey::Identifier(identifier) => {
                            serde_json::to_string(identifier.name.as_str()).unwrap()
                        }
                        ImportAttributeKey::StringLiteral(literal) => source
                            [literal.span.start as usize..literal.span.end as usize]
                            .to_owned(),
                    };
                    let value = &source
                        [attribute.value.span.start as usize..attribute.value.span.end as usize];
                    format!("[{key}]: {value}")
                })
                .collect::<Vec<_>>()
                .join(", ");
            options = format!(", {{ with: {{ {attributes} }} }}");
        }
        let specifier = &source[import.source.span.start as usize..import.source.span.end as usize];
        let initializer = format!("await import({specifier}{options})");
        let mut declarations = Vec::new();
        if let Some(namespace) = namespace {
            declarations.push(format!("{namespace} = {initializer}"));
        }
        if !properties.is_empty() || namespace.is_none() {
            declarations.push(format!(
                "{{ {} }} = {}",
                properties.join(", "),
                namespace.unwrap_or(&initializer)
            ));
        }
        let start = import.span.start as usize;
        let end = import.span.end as usize;
        result.push_str(&source[offset..start]);
        result.push_str(&format!("const {};", declarations.join(", ")));
        // Keep diagnostics for subsequent statements on their original lines.
        result.extend(
            source[start..end]
                .chars()
                .filter(|ch| matches!(ch, '\r' | '\n' | '\u{2028}' | '\u{2029}')),
        );
        offset = end;
    }
    result.push_str(&source[offset..]);
    result
}

/// An evaluation's completion or thrown value as an inspector `CallArgument`,
/// and whether it was thrown.
fn completion(response: &str) -> Result<(Value, bool), String> {
    let response = serde_json::from_str::<Value>(response)
        .map_err(|_| "Invalid V8 inspector response".to_owned())?;
    if let Some(error) = response.get("error") {
        return Err(error["message"]
            .as_str()
            .unwrap_or("Inspector error")
            .to_owned());
    }
    let result = &response["result"];
    if let Some(exception) = result.get("exceptionDetails") {
        return match exception.get("exception") {
            Some(thrown) => Ok((call_argument(thrown), true)),
            None => Err(exception["text"]
                .as_str()
                .unwrap_or("JavaScript exception")
                .to_owned()),
        };
    }
    Ok((call_argument(&result["result"]), false))
}

fn call_argument(remote: &Value) -> Value {
    for key in ["objectId", "unserializableValue", "value"] {
        if let Some(value) = remote.get(key) {
            return json!({ key: value });
        }
    }
    json!({})
}

fn inspected(response: &str) -> String {
    let response = serde_json::from_str::<Value>(response).unwrap_or_default();
    let result = &response["result"];
    result["result"]["value"]
        .as_str()
        .or_else(|| result["exceptionDetails"]["exception"]["description"].as_str())
        .or_else(|| response["error"]["message"].as_str())
        .unwrap_or("Inspection failed")
        .to_owned()
}

#[cfg(test)]
mod tests {
    use super::{is_incomplete, prepare_expression, prepare_imports};

    #[test]
    fn multiline_completeness_uses_the_javascript_parser() {
        assert!(is_incomplete("function greet() {\n"));
        assert!(is_incomplete("const value = {\n  answer: 42,\n"));
        assert!(!is_incomplete("const value = 42;\n"));
        assert!(!is_incomplete("const = 42;\n"));
        assert!(is_incomplete("const value =\n"));
        assert!(is_incomplete("{\n  answer: 42,\n"));
        assert!(is_incomplete("`brace {\n"));
        assert!(is_incomplete("/* comment\n"));
        assert!(!is_incomplete("'broken\n"));
        assert!(!is_incomplete("/broken\n"));
        assert!(is_incomplete("import {\n  inspect as show,\n"));
        assert!(is_incomplete("import { inspect } from\n"));
        assert!(!is_incomplete("import { inspect } from 'avril:std'\n"));
    }

    #[test]
    fn imports_preserve_all_javascript_line_terminators() {
        for newline in ["\n", "\r", "\r\n", "\u{2028}", "\u{2029}"] {
            let source = format!(
                "import {{{newline}inspect{newline}}} from 'avril:std';{newline}throw Error('original line');"
            );
            let prepared = prepare_imports(&source);
            let original_line = source
                .split(newline)
                .position(|line| line.contains("throw Error"))
                .unwrap();
            let prepared_line = prepared
                .split(newline)
                .position(|line| line.contains("throw Error"))
                .unwrap();
            assert_eq!(prepared_line, original_line, "{newline:?}");
        }
    }

    #[test]
    fn object_literals_are_evaluated_as_expressions() {
        assert_eq!(prepare_expression("{ answer: 42 }\n"), "({ answer: 42 }\n)");
        assert_eq!(
            prepare_expression("/* comment */ { answer: 42 }\n"),
            "(/* comment */ { answer: 42 }\n)"
        );
        assert_eq!(prepare_expression("if (true) {}\n"), "if (true) {}\n");
    }
}