fix(sourcemap): Handle multi-byte characters correctly (#5153)

crates/swc/tests/fixture/issues-4xxx/4120/1/output/index.map

@@ -1,5 +1,5 @@
 {
-  "mappings": "AAUA,cAAe,CAAEA,CAAC,CATd,wDAAsF,CAStEC,CAAC,CALjB,yGAA6J,CAK1I,AAAC",
+  "mappings": "AAUA,cAAe,CAAEA,CAAC,CATd,yCAAsF,CAStEC,CAAC,CALjB,kFAA6J,CAK1I,AAAC",
   "names": [
     "a",
     "b"

crates/swc/tests/source_map.rs

@@ -243,7 +243,7 @@ fn issue_4112() {
                 r#""use strict";
 
             export { default as Selection } from "./selection";
-            
+
             export { default as RichTextarea } from "./richTextarea";
             "#
                 .to_string(),
@@ -268,7 +268,7 @@ fn issue_4112() {
                 r#""use strict";
 
             import { React, window } from "easy";
-            
+
             window.assign({
               React
             });
@@ -311,9 +311,13 @@ fn should_work_with_emit_source_map_columns() {
         let fm = cm.new_source_file(
             swc_common::FileName::Real("./app.js".into()),
             r#"import { createElement } from "react";
-      
-  createElement('div', null, {});"#
-                .to_string(),
+
+  createElement('div', null, {});
+  /* 패딩에 대 */ /* 한 더 많은 문자. */ "테스트" + "테스트";
+  function ƒ() {
+  }
+"#
+            .to_string(),
         );
         let result = c.process_js_file(
             fm.clone(),
@@ -335,6 +339,8 @@ fn should_work_with_emit_source_map_columns() {
             Ok(result) => {
                 assert!(result.map.is_some());
                 let map = result.map.unwrap();
+
+                // lookup createElement(...) function call
                 let source_map = sourcemap::SourceMap::from_slice(map.as_bytes())
                     .expect("failed to deserialize sourcemap");
                 let token = source_map
@@ -344,6 +350,24 @@ fn should_work_with_emit_source_map_columns() {
                 assert_eq!(token.get_dst_col(), 14);
                 assert_eq!(token.get_src_line(), 2);
                 assert_eq!(token.get_src_col(), 16);
+
+                // lookup second string literal
+                let token = source_map
+                    .lookup_token(2, 37)
+                    .expect("failed to find token");
+                assert_eq!(token.get_dst_line(), 2);
+                assert_eq!(token.get_dst_col(), 37);
+                assert_eq!(token.get_src_line(), 3);
+                assert_eq!(token.get_src_col(), 39);
+
+                // lookup paren after multi-byte identifier
+                let token = source_map
+                    .lookup_token(3, 10)
+                    .expect("failed to find token");
+                assert_eq!(token.get_dst_line(), 3);
+                assert_eq!(token.get_dst_col(), 10);
+                assert_eq!(token.get_src_line(), 4);
+                assert_eq!(token.get_src_col(), 12);
             }
             Err(err) => {
                 panic!("Error: {:#?}", err);

crates/swc_common/src/source_map.rs

@@ -1157,8 +1157,6 @@ impl SourceMap {
 
         let mut cur_file: Option<Lrc<SourceFile>> = None;
 
-        let mut ch_start = 0;
-        let mut line_ch_start = 0;
         let mut prev_dst_line = u32::MAX;
 
         for (pos, lc) in mappings.iter() {
@@ -1193,8 +1191,6 @@ impl SourceMap {
                     }
 
                     cur_file = Some(f.clone());
-                    ch_start = 0;
-                    line_ch_start = 0;
                     &f
                 }
             };
@@ -1228,9 +1224,8 @@ impl SourceMap {
                 pos,
                 linebpos,
             );
-            let chpos = pos.to_u32() - self.calc_extra_bytes(f, &mut ch_start, pos);
-            let linechpos =
-                linebpos.to_u32() - self.calc_extra_bytes(f, &mut line_ch_start, linebpos);
+            let chpos = pos.to_u32() - self.calc_extra_bytes(f, &mut 0, pos);
+            let linechpos = linebpos.to_u32() - self.calc_extra_bytes(f, &mut 0, linebpos);
 
             let mut col = max(chpos, linechpos) - min(chpos, linechpos);
 

crates/swc_ecma_codegen/src/text_writer/basic_impl.rs

@@ -1,4 +1,4 @@
-use std::io::{self, Write};
+use std::io::Write;
 
 use rustc_hash::FxHashSet;
 use swc_common::{sync::Lrc, BytePos, LineCol, SourceMap, Span};
@@ -44,29 +44,26 @@ impl<'a, W: Write> JsWriter<'a, W> {
         }
     }
 
-    fn write_indent_string(&mut self) -> io::Result<usize> {
-        const INDENT: &[u8] = b"    ";
+    fn write_indent_string(&mut self) -> Result {
+        const INDENT: &str = "    ";
 
-        let mut cnt = 0;
         for _ in 0..self.indent {
-            cnt += self.raw_write(INDENT)?;
+            self.raw_write(INDENT)?;
         }
 
-        Ok(cnt)
+        Ok(())
     }
 
-    fn raw_write(&mut self, data: &[u8]) -> io::Result<usize> {
-        let written = self.wr.write(data)?;
-        self.line_pos += written;
-        Ok(written)
+    fn raw_write(&mut self, data: &str) -> Result {
+        self.wr.write_all(data.as_bytes())?;
+        self.line_pos += data.chars().count();
+        Ok(())
     }
 
-    fn write(&mut self, span: Option<Span>, data: &str) -> io::Result<usize> {
-        let mut cnt = 0;
-
+    fn write(&mut self, span: Option<Span>, data: &str) -> Result {
         if !data.is_empty() {
             if self.line_start {
-                cnt += self.write_indent_string()?;
+                self.write_indent_string()?;
                 self.line_start = false;
 
                 if let Some(pending) = self.pending_srcmap.take() {
@@ -80,7 +77,7 @@ impl<'a, W: Write> JsWriter<'a, W> {
                 }
             }
 
-            cnt += self.raw_write(data.as_bytes())?;
+            self.raw_write(data)?;
 
             if let Some(span) = span {
                 if !span.is_dummy() {
@@ -89,7 +86,7 @@ impl<'a, W: Write> JsWriter<'a, W> {
             }
         }
 
-        Ok(cnt)
+        Ok(())
     }
 
     fn srcmap(&mut self, byte_pos: BytePos) {
@@ -156,7 +153,7 @@ impl<'a, W: Write> WriteJs for JsWriter<'a, W> {
     fn write_line(&mut self) -> Result {
         let pending = self.pending_srcmap.take();
         if !self.line_start {
-            self.raw_write(self.new_line.as_bytes())?;
+            self.raw_write(self.new_line)?;
             self.line_count += 1;
             self.line_pos = 0;
             self.line_start = true;
@@ -180,7 +177,8 @@ impl<'a, W: Write> WriteJs for JsWriter<'a, W> {
             let line_start_of_s = compute_line_starts(s);
             if line_start_of_s.len() > 1 {
                 self.line_count = self.line_count + line_start_of_s.len() - 1;
-                self.line_pos = s.len() - line_start_of_s.last().cloned().unwrap_or(0);
+                let last_line_byte_index = line_start_of_s.last().cloned().unwrap_or(0);
+                self.line_pos = s[last_line_byte_index..].chars().count();
             }
 
             if !span.is_dummy() {
@@ -197,7 +195,8 @@ impl<'a, W: Write> WriteJs for JsWriter<'a, W> {
             let line_start_of_s = compute_line_starts(s);
             if line_start_of_s.len() > 1 {
                 self.line_count = self.line_count + line_start_of_s.len() - 1;
-                self.line_pos = s.len() - line_start_of_s.last().cloned().unwrap_or(0);
+                let last_line_byte_index = line_start_of_s.last().cloned().unwrap_or(0);
+                self.line_pos = s[last_line_byte_index..].chars().count();
             }
         }
         Ok(())
@@ -214,7 +213,8 @@ impl<'a, W: Write> WriteJs for JsWriter<'a, W> {
             let line_start_of_s = compute_line_starts(s);
             if line_start_of_s.len() > 1 {
                 self.line_count = self.line_count + line_start_of_s.len() - 1;
-                self.line_pos = s.len() - line_start_of_s.last().cloned().unwrap_or(0);
+                let last_line_byte_index = line_start_of_s.last().cloned().unwrap_or(0);
+                self.line_pos = s[last_line_byte_index..].chars().count();
             }
 
             if !span.is_dummy() {