fix(es/codegen): Fix LineCol calculation of printed files (#6763)

**Description:**

There were several issues with the way we updated the current `LineCol` position during the printing of the generated file:

- We used `chars` and `char_indices` (UTF-32) instead of `encode_utf16` (UTF-16) chars.
- JS uses UCS-2 (basically UTF-16) for its strings, and source maps default to that implicitly.
- `\r` was incorrectly handled
  - it didn't add a `line_start`; only `\n` did
- `\r\n` was incorrectly handled
- It was trying to let the `\n` path handle the `line_start`, but it called
`chars.next()` which ate the `\n` char.

I also took the opportunity to avoid the `Vec` allocations and reduced some code duplication.

See the
[before](https://evanw.github.io/source-map-visualization/#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)
and
[after](https://evanw.github.io/source-map-visualization/#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)


**Related issue:**

 - Closes https://github.com/swc-project/swc/issues/6694.

crates/swc/tests/source_map.rs

@@ -16,6 +16,7 @@ use swc::{
     },
     Compiler,
 };
+use swc_ecma_parser::Syntax;
 use testing::{assert_eq, NormalizedOutput, StdErr, Tester};
 use walkdir::WalkDir;
 
@@ -426,3 +427,87 @@ fn issue_4578() {
     )
     .unwrap();
 }
+
+#[test]
+fn issue_6694() {
+    Tester::new().print_errors(|cm, handler| {
+        let c = Compiler::new(cm.clone());
+        let fm = cm.new_source_file(
+            swc_common::FileName::Real("./app.js".into()),
+            r#"/**
+ * foo
+ * @param data foo
+ * @returns foo
+ */
+export const fixupRiskConfigData = (data: any): types.RiskConfigType => {
+  if (x) {
+    return 123;
+  } else {
+    return 456;
+  }
+};"#
+            .replace('\n', "\r\n"),
+        );
+        let result = c.process_js_file(
+            fm,
+            &handler,
+            &Options {
+                swcrc: false,
+                source_maps: Some(SourceMapsConfig::Bool(true)),
+                config: Config {
+                    jsc: JscConfig {
+                        target: Some(swc_ecma_ast::EsVersion::Es5),
+                        syntax: Some(Syntax::Typescript(Default::default())),
+                        ..Default::default()
+                    },
+                    is_module: IsModule::Bool(true),
+                    inline_sources_content: true.into(),
+                    emit_source_map_columns: true.into(),
+                    ..Default::default()
+                },
+                ..Default::default()
+            },
+        );
+
+        fn line_col(needle: &str, haystack: &str) -> Option<(u32, u32)> {
+            let lines = haystack.lines().enumerate();
+            for (i, line) in lines {
+                if let Some(c) = line.find(needle) {
+                    return Some((i as _, c as _));
+                }
+            }
+
+            None
+        }
+
+        match result {
+            Ok(result) => {
+                assert!(result.map.is_some());
+                let map = result.map.unwrap();
+
+                let source_map = sourcemap::SourceMap::from_slice(map.as_bytes())
+                    .expect("failed to deserialize sourcemap");
+
+                // "export"
+                let export_line_col =
+                    line_col("export", &result.code).expect("failed to find `export`");
+                let token = source_map
+                    .lookup_token(export_line_col.0, export_line_col.1)
+                    .expect("failed to find token");
+                assert_eq!(token.get_src(), (5, 0));
+
+                // "if"
+                let if_line_col = line_col("if", &result.code).expect("failed to find `export`");
+                let token = source_map
+                    .lookup_token(if_line_col.0, export_line_col.1)
+                    .expect("failed to find token");
+                assert_eq!(token.get_src(), (6, 2));
+            }
+            Err(err) => {
+                panic!("Error: {:#?}", err);
+            }
+        }
+
+        Ok(())
+    });
+}

crates/swc_common/src/syntax_pos.rs

@@ -1220,7 +1220,10 @@ pub struct LineInfo {
 /// Used to create a `.map` file.
 #[derive(Copy, Clone, Debug, PartialEq, Eq, Hash)]
 pub struct LineCol {
+    /// Index of line, starting from 0.
     pub line: u32,
+
+    /// UTF-16 column in line, starting from 0.
     pub col: u32,
 }
 

crates/swc_common/src/syntax_pos/analyze_source_file.rs

@@ -283,10 +283,28 @@ mod tests {
     );
 
     test!(
-        case: windows_crlf,
-        text: "012345678\r\nabcdef012345678\r\na",
+        case: unix_lf,
+        text: "/**\n * foo\n */\n012345678\nabcdef012345678\na",
         source_file_start_pos: 0,
-        lines: vec![0, 11, 28],
+        lines: vec![0, 4, 11, 15, 25, 41],
+        multi_byte_chars: vec![],
+        non_narrow_chars: vec![],
+    );
+
+    test!(
+        case: windows_cr,
+        text: "/**\r * foo\r */\r012345678\rabcdef012345678\ra",
+        source_file_start_pos: 0,
+        lines: vec![0, 4, 11, 15, 25, 41],
+        multi_byte_chars: vec![],
+        non_narrow_chars: vec![],
+    );
+
+    test!(
+        case: windows_crlf,
+        text: "/**\r\n * foo\r\n */\r\n012345678\r\nabcdef012345678\r\na",
+        source_file_start_pos: 0,
+        lines: vec![0, 5, 13, 18, 29, 46],
         multi_byte_chars: vec![],
         non_narrow_chars: vec![],
     );

crates/swc_ecma_codegen/src/text_writer/basic_impl.rs

@@ -51,6 +51,9 @@ impl<'a, W: Write> JsWriter<'a, W> {
         for _ in 0..self.indent {
             self.raw_write(INDENT)?;
         }
+        if self.srcmap.is_some() {
+            self.line_pos += INDENT.len() * self.indent;
+        }
 
         Ok(())
     }
@@ -59,11 +62,7 @@ impl<'a, W: Write> JsWriter<'a, W> {
     fn raw_write(&mut self, data: &str) -> Result {
         // #[cfg(debug_assertions)]
         // tracing::trace!("Write: `{}`", data);
-
         self.wr.write_all(data.as_bytes())?;
-        if self.srcmap.is_some() {
-            self.line_pos += data.chars().count();
-        }
 
         Ok(())
     }
@@ -86,6 +85,7 @@ impl<'a, W: Write> JsWriter<'a, W> {
             }
 
             self.raw_write(data)?;
+            self.update_pos(data);
 
             if let Some(span) = span {
                 self.srcmap(span.hi());
@@ -95,6 +95,21 @@ impl<'a, W: Write> JsWriter<'a, W> {
         Ok(())
     }
 
+    #[inline]
+    fn update_pos(&mut self, s: &str) {
+        if self.srcmap.is_some() {
+            let line_start_of_s = compute_line_starts(s);
+            self.line_count += line_start_of_s.line_count;
+
+            let chars = s[line_start_of_s.byte_pos..].encode_utf16().count();
+            if line_start_of_s.line_count > 0 {
+                self.line_pos = chars;
+            } else {
+                self.line_pos += chars;
+            }
+        }
+    }
+
     #[inline]
     fn srcmap(&mut self, byte_pos: BytePos) {
         if byte_pos.is_dummy() && byte_pos != BytePos(u32::MAX) {
@@ -183,8 +198,10 @@ impl<'a, W: Write> WriteJs for JsWriter<'a, W> {
         let pending = self.pending_srcmap.take();
         if !self.line_start {
             self.raw_write(self.new_line)?;
-            self.line_count += 1;
-            self.line_pos = 0;
+            if self.srcmap.is_some() {
+                self.line_count += 1;
+                self.line_pos = 0;
+            }
             self.line_start = true;
 
             if let Some(pending) = pending {
@@ -200,18 +217,7 @@ impl<'a, W: Write> WriteJs for JsWriter<'a, W> {
     fn write_lit(&mut self, span: Span, s: &str) -> Result {
         if !s.is_empty() {
             self.srcmap(span.lo());
-
             self.write(None, s)?;
-
-            if self.srcmap.is_some() {
-                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;
-                    let last_line_byte_index = line_start_of_s.last().cloned().unwrap_or(0);
-                    self.line_pos = s[last_line_byte_index..].chars().count();
-                }
-            }
-
             self.srcmap(span.hi());
         }
 
@@ -222,14 +228,6 @@ impl<'a, W: Write> WriteJs for JsWriter<'a, W> {
     #[cfg_attr(debug_assertions, tracing::instrument(skip_all))]
     fn write_comment(&mut self, s: &str) -> Result {
         self.write(None, s)?;
-        if self.srcmap.is_some() {
-            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;
-                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(())
     }
 
@@ -239,16 +237,6 @@ impl<'a, W: Write> WriteJs for JsWriter<'a, W> {
         if !s.is_empty() {
             self.srcmap(span.lo());
             self.write(None, s)?;
-
-            if self.srcmap.is_some() {
-                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;
-                    let last_line_byte_index = line_start_of_s.last().cloned().unwrap_or(0);
-                    self.line_pos = s[last_line_byte_index..].chars().count();
-                }
-            }
-
             self.srcmap(span.hi());
         }
 
@@ -285,10 +273,12 @@ impl<'a, W: Write> WriteJs for JsWriter<'a, W> {
     #[inline]
     #[cfg_attr(debug_assertions, tracing::instrument(skip_all))]
     fn add_srcmap(&mut self, pos: BytePos) -> Result {
-        if self.line_start {
-            self.pending_srcmap = Some(pos);
-        } else {
-            self.srcmap(pos);
+        if self.srcmap.is_some() {
+            if self.line_start {
+                self.pending_srcmap = Some(pos);
+            } else {
+                self.srcmap(pos);
+            }
         }
         Ok(())
     }
@@ -300,23 +290,31 @@ impl<'a, W: Write> WriteJs for JsWriter<'a, W> {
     }
 }
 
-fn compute_line_starts(s: &str) -> Vec<usize> {
-    let mut res = vec![];
-
+#[derive(Debug)]
+struct LineStart {
+    line_count: usize,
+    byte_pos: usize,
+}
+fn compute_line_starts(s: &str) -> LineStart {
+    let mut count = 0;
     let mut line_start = 0;
 
-    let mut chars = s.char_indices().peekable();
+    let mut chars = s.as_bytes().iter().enumerate().peekable();
 
     while let Some((pos, c)) = chars.next() {
         match c {
-            '\r' => {
-                if let Some(&(_, '\n')) = chars.peek() {
+            b'\r' => {
+                count += 1;
+                if let Some(&(_, b'\n')) = chars.peek() {
                     let _ = chars.next();
+                    line_start = pos + 2
+                } else {
+                    line_start = pos + 1
                 }
             }
 
-            '\n' => {
-                res.push(line_start);
+            b'\n' => {
+                count += 1;
                 line_start = pos + 1;
             }
 
@@ -324,7 +322,8 @@ fn compute_line_starts(s: &str) -> Vec<usize> {
         }
     }
 
-    // Last line.
-    res.push(line_start);
-    res
+    LineStart {
+        line_count: count,
+        byte_pos: line_start,
+    }
 }