char/avril

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

Charlotte Somport encoding extension to typescriptf2e5e0e2026-10-19

use std::cell::RefCell;
use std::collections::HashMap;
use std::convert::Infallible;
use std::io;
use std::rc::Rc;

use avril_core::serde_v8::{JsBuffer, ToJsBuffer};
use avril_core::{Extension, Op, v8};
use encoding_rs::{CoderResult, Decoder, DecoderResult, Encoding, REPLACEMENT, UTF_8};

type DecodeArgs = (Option<u64>, String, bool, bool, JsBuffer, bool);

struct State {
    decoder: Decoder,
    fatal: bool,
}

impl State {
    fn decode(&mut self, bytes: &[u8], stream: bool) -> io::Result<(String, bool)> {
        let capacity = self
            .decoder
            .max_utf8_buffer_length(bytes.len())
            .ok_or_else(|| io::Error::other("Decoded string is too large"))?;
        let mut output = String::new();
        output.try_reserve(capacity).map_err(io::Error::other)?;
        let invalid = if self.fatal {
            let (result, _) =
                self.decoder
                    .decode_to_string_without_replacement(bytes, &mut output, !stream);
            match result {
                DecoderResult::InputEmpty => false,
                // Browsers and WPT discard the bad chunk's remainder but retain codec state.
                DecoderResult::Malformed(_, _) => true,
                DecoderResult::OutputFull => {
                    return Err(io::Error::other("Decoder output exceeded its size bound"));
                }
            }
        } else {
            let (result, _, _) = self.decoder.decode_to_string(bytes, &mut output, !stream);
            if result == CoderResult::OutputFull {
                return Err(io::Error::other("Decoder output exceeded its size bound"));
            }
            false
        };
        Ok((output, invalid))
    }
}

#[derive(Default)]
struct Decoders {
    next: u64,
    states: HashMap<u64, State>,
}

pub fn init() -> Extension {
    let decoders = Rc::new(RefCell::new(Decoders::default()));
    let decode = decoders.clone();
    Extension {
        name: "ext:encoding",
        dependencies: &["ext:web", "ext:webstreams"],
        ops: vec![
            Op::new("op_encoding_label", |label: String| {
                Ok::<_, Infallible>(
                    Encoding::for_label(label.as_bytes())
                        .filter(|encoding| *encoding != REPLACEMENT)
                        .map(|encoding| encoding.name().to_ascii_lowercase()),
                )
            }),
            Op::new(
                "op_encoding_decode",
                move |(id, label, fatal, ignore_bom, bytes, stream): DecodeArgs| {
                    let mut decoders = decode.borrow_mut();
                    let mut state = if let Some(id) = id {
                        decoders
                            .states
                            .remove(&id)
                            .ok_or_else(|| io::Error::other("Decoder is closed"))?
                    } else {
                        let encoding = Encoding::for_label(label.as_bytes())
                            .filter(|encoding| *encoding != REPLACEMENT)
                            .ok_or_else(|| io::Error::other("Invalid encoding label"))?;
                        let decoder = if ignore_bom {
                            encoding.new_decoder_without_bom_handling()
                        } else {
                            encoding.new_decoder_with_bom_removal()
                        };
                        State { decoder, fatal }
                    };
                    let (text, invalid) = state.decode(&bytes, stream)?;
                    let id = if stream {
                        let id = match id {
                            Some(id) => id,
                            None => {
                                let id = decoders
                                    .next
                                    .checked_add(1)
                                    .filter(|id| *id <= 9_007_199_254_740_991)
                                    .ok_or_else(|| {
                                        io::Error::other("Encoding resource IDs exhausted")
                                    })?;
                                decoders.next = id;
                                id
                            }
                        };
                        decoders.states.insert(id, state);
                        Some(id)
                    } else {
                        None
                    };
                    Ok::<_, io::Error>((text, invalid, id))
                },
            ),
            Op::new("op_encoding_close", move |id: u64| {
                decoders.borrow_mut().states.remove(&id);
                Ok::<_, Infallible>(())
            }),
            Op::try_raw("op_encoding_decode_utf8", |scope, args| {
                let bytes: JsBuffer =
                    avril_core::serde_v8::from_v8(scope, args.get(0)).map_err(io::Error::other)?;
                // Body collection already owns fixed, unshared bytes; no snapshot or decoder state.
                let text = if bytes.is_ascii() {
                    v8::String::new_from_one_byte(scope, &bytes, v8::NewStringType::Normal)
                } else {
                    let (text, _) = UTF_8.decode_with_bom_removal(&bytes);
                    v8::String::new(scope, &text)
                }
                .ok_or_else(|| io::Error::other("Decoded string is too large"))?;
                Ok::<_, io::Error>(text.into())
            }),
            Op::new("op_encoding_encode", |text: String| {
                Ok::<_, Infallible>(ToJsBuffer::from(text.into_bytes()))
            }),
            Op::try_raw("op_encoding_encode_into", |scope, args| {
                let text = v8::Local::<v8::String>::try_from(args.get(0))
                    .map_err(|_| io::Error::other("Expected an encoding source string"))?;
                let mut destination: JsBuffer =
                    avril_core::serde_v8::from_v8(scope, args.get(1)).map_err(io::Error::other)?;
                let mut read = 0;
                let written = text.write_utf8_v2(
                    scope,
                    &mut destination,
                    v8::WriteFlags::kReplaceInvalidUtf8,
                    Some(&mut read),
                );
                #[derive(serde::Serialize)]
                struct Counts {
                    read: usize,
                    written: usize,
                }
                avril_core::serde_v8::to_v8(scope, Counts { read, written })
                    .map_err(io::Error::other)
            }),
        ],
        modules: &[(
            "ext:encoding/main.js",
            include_str!(concat!(env!("OUT_DIR"), "/encoding.js")),
        )],
        entry_point: Some("ext:encoding/main.js"),
    }
}