Fix non‑BMP code point handling in quick‑fixes and markers (#20526)

Co-authored-by: Micha Reiser <micha@reiser.io>

crates/ruff_wasm/README.md

@@ -19,7 +19,7 @@ There are multiple versions for the different wasm-pack targets. See [here](http
 This example uses the wasm-pack web target and is known to work with Vite.
 
 ```ts
-import init, { Workspace, type Diagnostic } from '@astral-sh/ruff-wasm-web';
+import init, { Workspace, type Diagnostic, PositionEncoding } from '@astral-sh/ruff-wasm-web';
 
 const exampleDocument = `print('hello'); print("world")`
 
@@ -42,7 +42,7 @@ const workspace = new Workspace({
       'F'
     ],
   },
-});
+}, PositionEncoding.UTF16);
 
 // Will contain 1 diagnostic code for E702: Multiple statements on one line
 const diagnostics: Diagnostic[] = workspace.check(exampleDocument);

crates/ruff_wasm/src/lib.rs

@@ -19,7 +19,7 @@ use ruff_python_formatter::{PyFormatContext, QuoteStyle, format_module_ast, pret
 use ruff_python_index::Indexer;
 use ruff_python_parser::{Mode, ParseOptions, Parsed, parse, parse_unchecked};
 use ruff_python_trivia::CommentRanges;
-use ruff_source_file::{LineColumn, OneIndexed};
+use ruff_source_file::{OneIndexed, PositionEncoding as SourcePositionEncoding, SourceLocation};
 use ruff_text_size::Ranged;
 use ruff_workspace::Settings;
 use ruff_workspace::configuration::Configuration;
@@ -117,6 +117,7 @@ pub fn run() {
 #[wasm_bindgen]
 pub struct Workspace {
     settings: Settings,
+    position_encoding: SourcePositionEncoding,
 }
 
 #[wasm_bindgen]
@@ -126,7 +127,7 @@ impl Workspace {
     }
 
     #[wasm_bindgen(constructor)]
-    pub fn new(options: JsValue) -> Result<Workspace, Error> {
+    pub fn new(options: JsValue, position_encoding: PositionEncoding) -> Result<Workspace, Error> {
         let options: Options = serde_wasm_bindgen::from_value(options).map_err(into_error)?;
         let configuration =
             Configuration::from_options(options, Some(Path::new(".")), Path::new("."))
@@ -135,7 +136,10 @@ impl Workspace {
             .into_settings(Path::new("."))
             .map_err(into_error)?;
 
-        Ok(Workspace { settings })
+        Ok(Workspace {
+            settings,
+            position_encoding: position_encoding.into(),
+        })
     }
 
     #[wasm_bindgen(js_name = defaultSettings)]
@@ -228,14 +232,16 @@ impl Workspace {
 
         let messages: Vec<ExpandedMessage> = diagnostics
             .into_iter()
-            .map(|msg| ExpandedMessage {
+            .map(|msg| {
+                let range = msg.range().unwrap_or_default();
+                ExpandedMessage {
                     code: msg.secondary_code_or_id().to_string(),
                     message: msg.body().to_string(),
                     start_location: source_code
-                    .line_column(msg.range().unwrap_or_default().start())
+                        .source_location(range.start(), self.position_encoding)
                         .into(),
                     end_location: source_code
-                    .line_column(msg.range().unwrap_or_default().end())
+                        .source_location(range.end(), self.position_encoding)
                         .into(),
                     fix: msg.fix().map(|fix| ExpandedFix {
                         message: msg.first_help_text().map(ToString::to_string),
@@ -243,12 +249,17 @@ impl Workspace {
                             .edits()
                             .iter()
                             .map(|edit| ExpandedEdit {
-                            location: source_code.line_column(edit.start()).into(),
-                            end_location: source_code.line_column(edit.end()).into(),
+                                location: source_code
+                                    .source_location(edit.start(), self.position_encoding)
+                                    .into(),
+                                end_location: source_code
+                                    .source_location(edit.end(), self.position_encoding)
+                                    .into(),
                                 content: edit.content().map(ToString::to_string),
                             })
                             .collect(),
                     }),
+                }
             })
             .collect();
 
@@ -331,14 +342,37 @@ impl<'a> ParsedModule<'a> {
 #[derive(Serialize, Deserialize, Eq, PartialEq, Debug)]
 pub struct Location {
     pub row: OneIndexed,
+    /// The character offset from the start of the line.
+    ///
+    /// The semantic of the offset depends on the [`PositionEncoding`] used when creating
+    /// the [`Workspace`].
     pub column: OneIndexed,
 }
 
-impl From<LineColumn> for Location {
-    fn from(value: LineColumn) -> Self {
+impl From<SourceLocation> for Location {
+    fn from(value: SourceLocation) -> Self {
         Self {
             row: value.line,
-            column: value.column,
+            column: value.character_offset,
+        }
+    }
+}
+
+#[derive(Default, Copy, Clone)]
+#[wasm_bindgen]
+pub enum PositionEncoding {
+    #[default]
+    Utf8,
+    Utf16,
+    Utf32,
+}
+
+impl From<PositionEncoding> for SourcePositionEncoding {
+    fn from(value: PositionEncoding) -> Self {
+        match value {
+            PositionEncoding::Utf8 => Self::Utf8,
+            PositionEncoding::Utf16 => Self::Utf16,
+            PositionEncoding::Utf32 => Self::Utf32,
         }
     }
 }

crates/ruff_wasm/tests/api.rs

@@ -4,12 +4,15 @@ use wasm_bindgen_test::wasm_bindgen_test;
 
 use ruff_linter::registry::Rule;
 use ruff_source_file::OneIndexed;
-use ruff_wasm::{ExpandedMessage, Location, Workspace};
+use ruff_wasm::{ExpandedMessage, Location, PositionEncoding, Workspace};
 
 macro_rules! check {
     ($source:expr, $config:expr, $expected:expr) => {{
         let config = js_sys::JSON::parse($config).unwrap();
-        match Workspace::new(config).unwrap().check($source) {
+        match Workspace::new(config, PositionEncoding::Utf8)
+            .unwrap()
+            .check($source)
+        {
             Ok(output) => {
                 let result: Vec<ExpandedMessage> = serde_wasm_bindgen::from_value(output).unwrap();
                 assert_eq!(result, $expected);

playground/ruff/src/Editor/Editor.tsx

@@ -6,7 +6,7 @@ import {
   useState,
 } from "react";
 import { Panel, PanelGroup } from "react-resizable-panels";
-import { Diagnostic, Workspace } from "ruff_wasm";
+import { Diagnostic, Workspace, PositionEncoding } from "ruff_wasm";
 import {
   ErrorMessage,
   Theme,
@@ -173,7 +173,7 @@ export default function Editor({
 
     try {
       const config = JSON.parse(settingsSource);
-      const workspace = new Workspace(config);
+      const workspace = new Workspace(config, PositionEncoding.Utf16);
       const diagnostics = workspace.check(pythonSource);
 
       let secondary: SecondaryPanelResult = null;