[Merged by Bors] - Add URI encoding and decoding functions

boa_engine/src/builtins/mod.rs

@@ -32,6 +32,7 @@ pub mod string;
 pub mod symbol;
 pub mod typed_array;
 pub mod undefined;
+pub mod uri;
 
 #[cfg(feature = "console")]
 pub mod console;
@@ -81,7 +82,7 @@ use crate::{
     builtins::{
         array_buffer::ArrayBuffer, async_generator::AsyncGenerator,
         async_generator_function::AsyncGeneratorFunction, generator::Generator,
-        generator_function::GeneratorFunction, typed_array::TypedArray,
+        generator_function::GeneratorFunction, typed_array::TypedArray, uri::Uri,
     },
     property::{Attribute, PropertyDescriptor},
     Context, JsValue,
@@ -193,7 +194,8 @@ pub fn init(context: &mut Context) {
         Promise,
         AsyncFunction,
         AsyncGenerator,
-        AsyncGeneratorFunction
+        AsyncGeneratorFunction,
+        Uri
     };
 
     #[cfg(feature = "intl")]

boa_engine/src/builtins/regexp/mod.rs

@@ -1745,7 +1745,8 @@ fn advance_string_index(s: &JsString, index: u64, unicode: bool) -> u64 {
     }
 
     // 5. Let cp be ! CodePointAt(S, index).
-    let (_, offset, _) = crate::builtins::string::code_point_at(s, index);
+    let cp = crate::builtins::string::code_point_at(s, index);
 
-    index + u64::from(offset)
+    // 6. Return index + cp.[[CodeUnitCount]].
+    index + u64::from(cp.code_unit_count)
 }

boa_engine/src/builtins/string/mod.rs

@@ -40,29 +40,53 @@ pub(crate) enum Placement {
     End,
 }
 
-pub(crate) fn code_point_at(string: &JsString, position: u64) -> (u32, u8, bool) {
+/// Code point information for the `code_point_at()` function.
+#[derive(Debug, Clone, Copy)]
+pub(crate) struct CodePointInfo {
+    pub(crate) code_point: u32,
+    pub(crate) code_unit_count: u8,
+    pub(crate) is_unpaired_surrogate: bool,
+}
+
+pub(crate) fn code_point_at(string: &JsString, position: u64) -> CodePointInfo {
     let mut encoded = string.encode_utf16();
     let size = encoded.clone().count() as u64;
 
     let first = encoded
         .nth(position as usize)
         .expect("The callers of this function must've already checked bounds.");
     if !is_leading_surrogate(first) && !is_trailing_surrogate(first) {
-        return (u32::from(first), 1, false);
+        return CodePointInfo {
+            code_point: u32::from(first),
+            code_unit_count: 1,
+            is_unpaired_surrogate: false,
+        };
     }
 
     if is_trailing_surrogate(first) || position + 1 == size {
-        return (u32::from(first), 1, true);
+        return CodePointInfo {
+            code_point: u32::from(first),
+            code_unit_count: 1,
+            is_unpaired_surrogate: true,
+        };
     }
 
     let second = encoded
         .next()
         .expect("The callers of this function must've already checked bounds.");
     if !is_trailing_surrogate(second) {
-        return (u32::from(first), 1, true);
+        return CodePointInfo {
+            code_point: u32::from(first),
+            code_unit_count: 1,
+            is_unpaired_surrogate: true,
+        };
     }
     let cp = (u32::from(first) - 0xD800) * 0x400 + (u32::from(second) - 0xDC00) + 0x10000;
-    (cp, 2, false)
+    CodePointInfo {
+        code_point: cp,
+        code_unit_count: 2,
+        is_unpaired_surrogate: false,
+    }
 }
 
 /// Helper function to check if a `char` is trimmable.
@@ -544,7 +568,7 @@ impl String {
             IntegerOrInfinity::Integer(position) if (0..size).contains(&position) => {
                 // 6. Let cp be ! CodePointAt(S, position).
                 // 7. Return 𝔽(cp.[[CodePoint]]).
-                Ok(code_point_at(&string, position as u64).0.into())
+                Ok(code_point_at(&string, position as u64).code_point.into())
             }
             // 5. If position < 0 or position ≥ size, return undefined.
             _ => Ok(JsValue::undefined()),

boa_engine/src/builtins/string/string_iterator.rs

@@ -10,6 +10,8 @@ use crate::{
 use boa_gc::{Finalize, Trace};
 use boa_profiler::Profiler;
 
+use super::CodePointInfo;
+
 #[derive(Debug, Clone, Finalize, Trace)]
 pub struct StringIterator {
     string: JsValue,
@@ -61,7 +63,11 @@ impl StringIterator {
                 context,
             ));
         }
-        let (_, code_unit_count, _) = code_point_at(&native_string, position as u64);
+        let CodePointInfo {
+            code_point: _,
+            code_unit_count,
+            is_unpaired_surrogate: _,
+        } = code_point_at(&native_string, position as u64);
         string_iterator.next_index += i32::from(code_unit_count);
         let result_string = crate::builtins::string::String::substring(
             &string_iterator.string,

boa_engine/src/builtins/uri/consts.rs

@@ -0,0 +1,203 @@
+/// Constant with all the unescaped URI characters.
+///
+/// Contains `uriAlpha`, `DecimalDigit` and `uriMark`.
+///
+/// More information:
+///  - [ECMAScript reference][spec]
+///
+/// [spec]: https://tc39.es/ecma262/#prod-uriUnescaped
+pub(super) const URI_UNESCAPED: [u16; 71] = [
+    // uriAlpha
+    b'a' as u16,
+    b'b' as u16,
+    b'c' as u16,
+    b'd' as u16,
+    b'e' as u16,
+    b'f' as u16,
+    b'g' as u16,
+    b'h' as u16,
+    b'i' as u16,
+    b'j' as u16,
+    b'k' as u16,
+    b'l' as u16,
+    b'm' as u16,
+    b'n' as u16,
+    b'o' as u16,
+    b'p' as u16,
+    b'q' as u16,
+    b'r' as u16,
+    b's' as u16,
+    b't' as u16,
+    b'u' as u16,
+    b'v' as u16,
+    b'w' as u16,
+    b'x' as u16,
+    b'y' as u16,
+    b'z' as u16,
+    b'A' as u16,
+    b'B' as u16,
+    b'C' as u16,
+    b'D' as u16,
+    b'E' as u16,
+    b'F' as u16,
+    b'G' as u16,
+    b'H' as u16,
+    b'I' as u16,
+    b'J' as u16,
+    b'K' as u16,
+    b'L' as u16,
+    b'M' as u16,
+    b'N' as u16,
+    b'O' as u16,
+    b'P' as u16,
+    b'Q' as u16,
+    b'R' as u16,
+    b'S' as u16,
+    b'T' as u16,
+    b'U' as u16,
+    b'V' as u16,
+    b'W' as u16,
+    b'X' as u16,
+    b'Y' as u16,
+    b'Z' as u16,
+    // DecimalDigit
+    b'0' as u16,
+    b'1' as u16,
+    b'2' as u16,
+    b'3' as u16,
+    b'4' as u16,
+    b'5' as u16,
+    b'6' as u16,
+    b'7' as u16,
+    b'8' as u16,
+    b'9' as u16,
+    // uriMark
+    b'-' as u16,
+    b'_' as u16,
+    b'.' as u16,
+    b'!' as u16,
+    b'~' as u16,
+    b'*' as u16,
+    b'\'' as u16,
+    b'(' as u16,
+    b')' as u16,
+];
+
+/// Constant with all the reserved URI characters, plus the hashtag symbol (`#`).
+///
+/// More information:
+///  - [ECMAScript reference][spec]
+///
+/// [spec]: https://tc39.es/ecma262/#prod-uriReserved
+pub(super) const URI_RESERVED_HASH: [u16; 11] = [
+    b';' as u16,
+    b'/' as u16,
+    b'?' as u16,
+    b':' as u16,
+    b'@' as u16,
+    b'&' as u16,
+    b'=' as u16,
+    b'+' as u16,
+    b'$' as u16,
+    b',' as u16,
+    // Extra: # symbol
+    b'#' as u16,
+];
+
+/// Constant with all the reserved and unescaped URI characters, plus the hashtag symbol (`#`).
+///
+/// More information:
+///  - [`uriReserved` in ECMAScript spec][uri_reserved]
+///  - [`uriUnescaped` in ECMAScript spec][uri_unescaped]
+///
+/// [uri_reserved]: https://tc39.es/ecma262/#prod-uriReserved
+/// [uri_unescaped]: https://tc39.es/ecma262/#prod-uriUnescaped
+pub(super) const URI_RESERVED_UNESCAPED_HASH: [u16; 82] = [
+    // uriAlpha
+    b'a' as u16,
+    b'b' as u16,
+    b'c' as u16,
+    b'd' as u16,
+    b'e' as u16,
+    b'f' as u16,
+    b'g' as u16,
+    b'h' as u16,
+    b'i' as u16,
+    b'j' as u16,
+    b'k' as u16,
+    b'l' as u16,
+    b'm' as u16,
+    b'n' as u16,
+    b'o' as u16,
+    b'p' as u16,
+    b'q' as u16,
+    b'r' as u16,
+    b's' as u16,
+    b't' as u16,
+    b'u' as u16,
+    b'v' as u16,
+    b'w' as u16,
+    b'x' as u16,
+    b'y' as u16,
+    b'z' as u16,
+    b'A' as u16,
+    b'B' as u16,
+    b'C' as u16,
+    b'D' as u16,
+    b'E' as u16,
+    b'F' as u16,
+    b'G' as u16,
+    b'H' as u16,
+    b'I' as u16,
+    b'J' as u16,
+    b'K' as u16,
+    b'L' as u16,
+    b'M' as u16,
+    b'N' as u16,
+    b'O' as u16,
+    b'P' as u16,
+    b'Q' as u16,
+    b'R' as u16,
+    b'S' as u16,
+    b'T' as u16,
+    b'U' as u16,
+    b'V' as u16,
+    b'W' as u16,
+    b'X' as u16,
+    b'Y' as u16,
+    b'Z' as u16,
+    // DecimalDigit
+    b'0' as u16,
+    b'1' as u16,
+    b'2' as u16,
+    b'3' as u16,
+    b'4' as u16,
+    b'5' as u16,
+    b'6' as u16,
+    b'7' as u16,
+    b'8' as u16,
+    b'9' as u16,
+    // uriMark
+    b'-' as u16,
+    b'_' as u16,
+    b'.' as u16,
+    b'!' as u16,
+    b'~' as u16,
+    b'*' as u16,
+    b'\'' as u16,
+    b'(' as u16,
+    b')' as u16,
+    // uriReserved
+    b';' as u16,
+    b'/' as u16,
+    b'?' as u16,
+    b':' as u16,
+    b'@' as u16,
+    b'&' as u16,
+    b'=' as u16,
+    b'+' as u16,
+    b'$' as u16,
+    b',' as u16,
+    // Extra: # symbol
+    b'#' as u16,
+];