Add TextEncoder, TextDecoder implementations to boa_runtime (#3994)

* Add a TextEncoder implementation to boa_runtime

* Add a TextDecoder implementation as well.

Cargo.lock

@@ -535,6 +535,7 @@ version = "0.19.0"
 dependencies = [
  "boa_engine",
  "boa_gc",
+ "boa_interop",
  "indoc",
  "rustc-hash 2.0.0",
  "textwrap",

core/runtime/Cargo.toml

@@ -13,6 +13,7 @@ rust-version.workspace = true
 [dependencies]
 boa_engine.workspace = true
 boa_gc.workspace = true
+boa_interop.workspace = true
 rustc-hash = { workspace = true, features = ["std"] }
 
 [dev-dependencies]

core/runtime/src/lib.rs

@@ -56,6 +56,11 @@ mod console;
 #[doc(inline)]
 pub use console::{Console, Logger};
 
+mod text;
+
+#[doc(inline)]
+pub use text::{TextDecoder, TextEncoder};
+
 #[cfg(test)]
 pub(crate) mod test {
     use boa_engine::{builtins, Context, JsResult, JsValue, Source};

core/runtime/src/text.rs

@@ -0,0 +1,124 @@
+//! Module implementing JavaScript classes to handle text encoding and decoding.
+//!
+//! See <https://developer.mozilla.org/en-US/docs/Web/API/Encoding_API> for more information.
+
+use boa_engine::object::builtins::JsUint8Array;
+use boa_engine::string::CodePoint;
+use boa_engine::{
+    js_string, Context, Finalize, JsData, JsNativeError, JsObject, JsResult, JsString, Trace,
+};
+use boa_interop::js_class;
+
+#[cfg(test)]
+mod tests;
+
+/// The `TextDecoder`[mdn] class represents an encoder for a specific method, that is
+/// a specific character encoding, like `utf-8`.
+///
+/// [mdn]: https://developer.mozilla.org/en-US/docs/Web/API/TextDecoder
+#[derive(Debug, Clone, JsData, Trace, Finalize)]
+pub struct TextDecoder;
+
+impl TextDecoder {
+    /// Register the `TextDecoder` class into the realm.
+    ///
+    /// # Errors
+    /// This will error if the context or realm cannot register the class.
+    pub fn register(context: &mut Context) -> JsResult<()> {
+        context.register_global_class::<Self>()?;
+        Ok(())
+    }
+
+    /// The `decode()` method of the `TextDecoder` interface returns a `JsString` containing
+    /// the given `Uint8Array` decoded in the specific method. This will replace any
+    /// invalid characters with the Unicode replacement character.
+    pub fn decode(text: &JsUint8Array, context: &mut Context) -> JsString {
+        let buffer = text.iter(context).collect::<Vec<u8>>();
+        let string = String::from_utf8_lossy(&buffer);
+        JsString::from(string.as_ref())
+    }
+}
+
+js_class! {
+    class TextDecoder {
+        property encoding {
+            fn get() -> JsString {
+                js_string!("utf-8")
+            }
+        }
+
+        // Creates a new `TextEncoder` object. Encoding is optional but MUST BE
+        // "utf-8" if specified. Options is ignored.
+        constructor(encoding: Option<JsString>, _options: Option<JsObject>) {
+            if let Some(e) = encoding {
+                if e != js_string!("utf-8") {
+                    return Err(JsNativeError::typ().with_message("Only utf-8 encoding is supported").into());
+                }
+            }
+            Ok(TextDecoder)
+        }
+
+        fn decode(array: JsUint8Array, context: &mut Context) -> JsString {
+            TextDecoder::decode(&array, context)
+        }
+    }
+}
+
+/// The `TextEncoder`[mdn] class represents an encoder for a specific method, that is
+/// a specific character encoding, like `utf-8`.
+///
+/// [mdn]: https://developer.mozilla.org/en-US/docs/Web/API/TextEncoder
+#[derive(Debug, Clone, JsData, Trace, Finalize)]
+pub struct TextEncoder;
+
+impl TextEncoder {
+    /// Register the `TextEncoder` class into the realm.
+    ///
+    /// # Errors
+    /// This will error if the context or realm cannot register the class.
+    pub fn register(context: &mut Context) -> JsResult<()> {
+        context.register_global_class::<Self>()?;
+        Ok(())
+    }
+
+    /// The `encode()` method of the `TextEncoder` interface returns a `Uint8Array` containing
+    /// the given string encoded in the specific method.
+    ///
+    /// # Errors
+    /// This will error if there is an issue creating the `Uint8Array`.
+    pub fn encode(text: &JsString, context: &mut Context) -> JsResult<JsUint8Array> {
+        // TODO: move this logic to JsString.
+        JsUint8Array::from_iter(
+            text.code_points().flat_map(|s| match s {
+                CodePoint::Unicode(c) => c.to_string().as_bytes().to_vec(),
+                CodePoint::UnpairedSurrogate(_) => "\u{FFFD}".as_bytes().to_vec(),
+            }),
+            context,
+        )
+    }
+}
+
+js_class! {
+    class TextEncoder {
+        property encoding {
+            fn get() -> JsString {
+                js_string!("utf-8")
+            }
+        }
+
+        // Creates a new `TextEncoder` object. Encoding is optional but MUST BE
+        // "utf-8" if specified. Options is ignored.
+        constructor(encoding: Option<JsString>, _options: Option<JsObject>) {
+            if let Some(e) = encoding {
+                if e != js_string!("utf-8") {
+                    return Err(JsNativeError::typ().with_message("Only utf-8 encoding is supported").into());
+                }
+            }
+            Ok(TextEncoder)
+        }
+
+        fn encode(text: JsString, context: &mut Context) -> JsResult<JsUint8Array> {
+            TextEncoder::encode(&text, context)
+        }
+    }
+}

core/runtime/src/text/tests.rs

@@ -0,0 +1,152 @@
+use crate::test::{run_test_actions_with, TestAction};
+use crate::{TextDecoder, TextEncoder};
+use boa_engine::object::builtins::JsUint8Array;
+use boa_engine::property::Attribute;
+use boa_engine::{js_str, js_string, Context, JsString};
+use indoc::indoc;
+
+#[test]
+fn encoder_js() {
+    let context = &mut Context::default();
+    TextEncoder::register(context).unwrap();
+
+    run_test_actions_with(
+        [
+            TestAction::run(indoc! {r#"
+                const encoder = new TextEncoder();
+                encoded = encoder.encode("Hello, World!");
+            "#}),
+            TestAction::inspect_context(|context| {
+                let encoded = context
+                    .global_object()
+                    .get(js_str!("encoded"), context)
+                    .unwrap();
+                let array =
+                    JsUint8Array::from_object(encoded.as_object().unwrap().clone()).unwrap();
+                let buffer = array.iter(context).collect::<Vec<_>>();
+
+                assert_eq!(buffer, b"Hello, World!");
+            }),
+        ],
+        context,
+    );
+}
+
+#[test]
+fn encoder_js_unpaired() {
+    use crate::test::{run_test_actions_with, TestAction};
+    use indoc::indoc;
+
+    let context = &mut Context::default();
+    TextEncoder::register(context).unwrap();
+
+    let unpaired_surrogates: [u16; 3] = [0xDC58, 0xD83C, 0x0015];
+    let text = JsString::from(&unpaired_surrogates);
+    context
+        .register_global_property(js_str!("text"), text, Attribute::default())
+        .unwrap();
+
+    run_test_actions_with(
+        [
+            TestAction::run(indoc! {r#"
+                const encoder = new TextEncoder();
+                encoded = encoder.encode(text);
+            "#}),
+            TestAction::inspect_context(|context| {
+                let encoded = context
+                    .global_object()
+                    .get(js_str!("encoded"), context)
+                    .unwrap();
+                let array =
+                    JsUint8Array::from_object(encoded.as_object().unwrap().clone()).unwrap();
+                let buffer = array.iter(context).collect::<Vec<_>>();
+
+                assert_eq!(buffer, "\u{FFFD}\u{FFFD}\u{15}".as_bytes());
+            }),
+        ],
+        context,
+    );
+}
+
+#[test]
+fn decoder_js() {
+    let context = &mut Context::default();
+    TextDecoder::register(context).unwrap();
+
+    run_test_actions_with(
+        [
+            TestAction::run(indoc! {r#"
+                const d = new TextDecoder();
+                decoded = d.decode(
+                    Uint8Array.from([ 72, 101, 108, 108, 111, 44, 32, 87, 111, 114, 108, 100, 33 ])
+                );
+            "#}),
+            TestAction::inspect_context(|context| {
+                let decoded = context
+                    .global_object()
+                    .get(js_str!("decoded"), context)
+                    .unwrap();
+                assert_eq!(decoded.as_string(), Some(&js_string!("Hello, World!")));
+            }),
+        ],
+        context,
+    );
+}
+
+#[test]
+fn decoder_js_invalid() {
+    use crate::test::{run_test_actions_with, TestAction};
+    use indoc::indoc;
+
+    let context = &mut Context::default();
+    TextDecoder::register(context).unwrap();
+
+    run_test_actions_with(
+        [
+            TestAction::run(indoc! {r#"
+                const d = new TextDecoder();
+                decoded = d.decode(
+                    Uint8Array.from([ 72, 101, 108, 108, 111, 160, 87, 111, 114, 108, 100, 161 ])
+                );
+            "#}),
+            TestAction::inspect_context(|context| {
+                let decoded = context
+                    .global_object()
+                    .get(js_str!("decoded"), context)
+                    .unwrap();
+                assert_eq!(
+                    decoded.as_string(),
+                    Some(&js_string!("Hello\u{FFFD}World\u{FFFD}"))
+                );
+            }),
+        ],
+        context,
+    );
+}
+
+#[test]
+fn roundtrip() {
+    let context = &mut Context::default();
+    TextEncoder::register(context).unwrap();
+    TextDecoder::register(context).unwrap();
+
+    run_test_actions_with(
+        [
+            TestAction::run(indoc! {r#"
+                const encoder = new TextEncoder();
+                const decoder = new TextDecoder();
+                const text = "Hello, World!";
+                const encoded = encoder.encode(text);
+                decoded = decoder.decode(encoded);
+            "#}),
+            TestAction::inspect_context(|context| {
+                let decoded = context
+                    .global_object()
+                    .get(js_str!("decoded"), context)
+                    .unwrap();
+                assert_eq!(decoded.as_string(), Some(&js_string!("Hello, World!")));
+            }),
+        ],
+        context,
+    );
+}