From 450b1e6da9105043d403580c5fc7e5dbd7bfcab7 Mon Sep 17 00:00:00 2001
From: overlookmotel <theoverlookmotel@gmail.com>
Date: Fri, 24 Jan 2025 10:44:22 +0000
Subject: [PATCH] Fix translation

---
 crates/oxc_ast/src/utf8_to_utf16.rs | 29 ++++++++++++++++++-----------
 1 file changed, 18 insertions(+), 11 deletions(-)

diff --git a/crates/oxc_ast/src/utf8_to_utf16.rs b/crates/oxc_ast/src/utf8_to_utf16.rs
index 79ce7881f38a98..4a5911322e0621 100644
--- a/crates/oxc_ast/src/utf8_to_utf16.rs
+++ b/crates/oxc_ast/src/utf8_to_utf16.rs
@@ -5,7 +5,6 @@ use oxc_span::Span;
 use crate::{ast::Program, visit::VisitMut};
 
 /// Convert UTF8 span offsets to UTF16.
-#[derive(Default)]
 pub struct Utf8ToUtf16 {
     translations: Vec<Translation>,
 }
@@ -22,15 +21,18 @@ struct Translation {
 
 impl Utf8ToUtf16 {
     /// Create new `Utf8ToUtf16` converter.
+    #[expect(clippy::new_without_default)]
     pub fn new() -> Self {
-        Self::default()
+        let mut translations = Vec::with_capacity(16);
+        translations.push(Translation { utf8_offset: 0, utf16_difference: 0 });
+        Self { translations }
     }
 
     /// Convert all the spans in the AST to UTF16.
     pub fn convert(mut self, program: &mut Program<'_>) {
         self.build_table(program.source_text);
         // Skip if source is entirely ASCII
-        if self.translations.is_empty() {
+        if self.translations.len() == 1 {
             return;
         }
         self.visit_program(program);
@@ -46,8 +48,10 @@ impl Utf8ToUtf16 {
             if byte >= 0xC0 {
                 let difference_for_this_byte = u32::from(byte >= 0xE0) + 1;
                 utf16_difference += difference_for_this_byte;
+                // Record `utf8_offset + 1` not `utf8_offset`, because it's only offsets *after* this
+                // Unicode character that need to be shifted
                 self.translations
-                    .push(Translation { utf8_offset: utf8_offset as u32, utf16_difference });
+                    .push(Translation { utf8_offset: utf8_offset as u32 + 1, utf16_difference });
             }
         }
     }
@@ -58,14 +62,17 @@ impl Utf8ToUtf16 {
     }
 
     fn convert_offset(&self, utf8_offset: u32) -> u32 {
-        // FIXME:
-        let mut utf16_offset = utf8_offset;
+        // Find the first entry in table *after* the UTF8 offset.
+        // The difference we need to subtract is recorded in the entry prior to it.
         let index =
-            self.translations.partition_point(|&translation| translation.utf8_offset < utf8_offset);
-        if let Some(&translation) = self.translations.get(index) {
-            utf16_offset -= translation.utf16_difference;
-        }
-        utf16_offset
+            self.translations.partition_point(|translation| translation.utf8_offset <= utf8_offset);
+        // SAFETY:
+        // First entry in table is `0, 0`. `partition_point` finds the first entry where
+        // `utf8_offset < translation.utf8_offset` (or `translations.len()` if none exists).
+        // So guaranteed `index > 0`, and `index <= translations.len()`.
+        // Therefore `index - 1` cannot wrap around, and cannot be out of bounds.
+        let translation = unsafe { self.translations.get_unchecked(index - 1) };
+        utf8_offset - translation.utf16_difference
     }
 }