fix(coverage): better handling of multi-byte characters (#15159)

8 files changed, 129 insertions, 113 deletions

diff --git a/Cargo.lock b/Cargo.lock
index 46d6a3633..302fb63d3 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -838,7 +838,7 @@ dependencies = [
  "tempfile",
  "test_util",
  "text-size",
- "text_lines",
+ "text_lines 0.6.0",
  "tokio",
  "tokio-util",
  "tower-lsp",
@@ -881,7 +881,7 @@ dependencies = [
  "swc_ecma_transforms_typescript",
  "swc_ecma_utils",
  "swc_ecma_visit",
- "text_lines",
+ "text_lines 0.4.1",
  "url 2.2.2",
 ]
 
@@ -1370,7 +1370,7 @@ dependencies = [
  "dprint-core",
  "jsonc-parser",
  "serde",
- "text_lines",
+ "text_lines 0.4.1",
 ]
 
 [[package]]
@@ -1412,7 +1412,7 @@ dependencies = [
  "swc_common",
  "swc_ecma_ast",
  "swc_ecma_parser",
- "text_lines",
+ "text_lines 0.4.1",
 ]
 
 [[package]]
@@ -4581,6 +4581,12 @@ dependencies = [
 ]
 
 [[package]]
+name = "text_lines"
+version = "0.6.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7fd5828de7deaa782e1dd713006ae96b3bee32d3279b79eb67ecf8072c059bcf"
+
+[[package]]
 name = "textwrap"
 version = "0.15.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
diff --git a/cli/Cargo.toml b/cli/Cargo.toml
index f86ec4ea2..83eb7666d 100644
--- a/cli/Cargo.toml
+++ b/cli/Cargo.toml
@@ -94,7 +94,7 @@ serde = { version = "=1.0.139", features = ["derive"] }
 serde_repr = "=0.1.8"
 shell-escape = "=0.1.5"
 text-size = "=1.1.0"
-text_lines = "=0.4.1"
+text_lines = "=0.6.0"
 tokio = { version = "=1.19", features = ["full"] }
 tokio-util = "=0.7.2"
 tower-lsp = "=0.17.0"
diff --git a/cli/tests/testdata/coverage/complex_expected.lcov b/cli/tests/testdata/coverage/complex_expected.lcov
index 1bb79d277..7a3cd8d92 100644
--- a/cli/tests/testdata/coverage/complex_expected.lcov
+++ b/cli/tests/testdata/coverage/complex_expected.lcov
@@ -46,9 +46,9 @@ DA:64,0
 DA:65,0
 DA:66,0
 DA:67,0
-DA:68,1
+DA:68,0
 DA:71,0
 DA:74,1
-LH:23
+LH:22
 LF:38
 end_of_record
diff --git a/cli/tests/testdata/coverage/complex_expected.out b/cli/tests/testdata/coverage/complex_expected.out
index e318e2067..3f7c89e9b 100644
--- a/cli/tests/testdata/coverage/complex_expected.out
+++ b/cli/tests/testdata/coverage/complex_expected.out
@@ -1,4 +1,4 @@
-cover [WILDCARD]/coverage/complex.ts ... 60.526% (23/38)
+cover [WILDCARD]/coverage/complex.ts ... 57.895% (22/38)
   46 | export function unused(
   47 |   foo: string,
   48 |   bar: string,
@@ -15,5 +15,6 @@ cover [WILDCARD]/coverage/complex.ts ... 60.526% (23/38)
   65 |   return (
   66 |     0
   67 |   );
+  68 | }
 -----|-----
   71 | console.log("%s", () => 1);
diff --git a/cli/tools/coverage/json_types.rs b/cli/tools/coverage/json_types.rs
index 1e46cc7fd..6da17d58f 100644
--- a/cli/tools/coverage/json_types.rs
+++ b/cli/tools/coverage/json_types.rs
@@ -6,10 +6,12 @@ use serde::Serialize;
 #[derive(Debug, Eq, PartialEq, Serialize, Deserialize, Clone)]
 #[serde(rename_all = "camelCase")]
 pub struct CoverageRange {
-  /// Start byte index.
-  pub start_offset: usize,
-  /// End byte index.
-  pub end_offset: usize,
+  /// Start character index.
+  #[serde(rename = "startOffset")]
+  pub start_char_offset: usize,
+  /// End character index.
+  #[serde(rename = "endOffset")]
+  pub end_char_offset: usize,
   pub count: i64,
 }
 
diff --git a/cli/tools/coverage/merge.rs b/cli/tools/coverage/merge.rs
index b0606e215..08a695e1d 100644
--- a/cli/tools/coverage/merge.rs
+++ b/cli/tools/coverage/merge.rs
@@ -54,15 +54,15 @@ pub fn merge_scripts(
     let first: &ScriptCoverage = &scripts[0];
     (first.script_id.clone(), first.url.clone())
   };
-  let mut range_to_funcs: BTreeMap<Range, Vec<FunctionCoverage>> =
+  let mut range_to_funcs: BTreeMap<CharRange, Vec<FunctionCoverage>> =
     BTreeMap::new();
   for script_cov in scripts {
     for func_cov in script_cov.functions {
       let root_range = {
         let root_range_cov: &CoverageRange = &func_cov.ranges[0];
-        Range {
-          start: root_range_cov.start_offset,
-          end: root_range_cov.end_offset,
+        CharRange {
+          start: root_range_cov.start_char_offset,
+          end: root_range_cov.end_char_offset,
         }
       };
       range_to_funcs
@@ -85,12 +85,12 @@ pub fn merge_scripts(
 }
 
 #[derive(Eq, PartialEq, Hash, Copy, Clone, Debug)]
-struct Range {
+struct CharRange {
   start: usize,
   end: usize,
 }
 
-impl Ord for Range {
+impl Ord for CharRange {
   fn cmp(&self, other: &Self) -> ::std::cmp::Ordering {
     if self.start != other.start {
       self.start.cmp(&other.start)
@@ -100,7 +100,7 @@ impl Ord for Range {
   }
 }
 
-impl PartialOrd for Range {
+impl PartialOrd for CharRange {
   fn partial_cmp(&self, other: &Self) -> Option<::std::cmp::Ordering> {
     if self.start != other.start {
       self.start.partial_cmp(&other.start)
@@ -249,7 +249,7 @@ fn merge_range_tree_children<'a>(
     Vec::with_capacity(parent_trees.len());
   let mut wrapped_children: Vec<Vec<&'a mut RangeTree<'a>>> =
     Vec::with_capacity(parent_trees.len());
-  let mut open_range: Option<Range> = None;
+  let mut open_range: Option<CharRange> = None;
 
   for _parent_tree in parent_trees.iter() {
     flat_children.push(Vec::new());
@@ -318,7 +318,7 @@ fn merge_range_tree_children<'a>(
             .push(tree);
         }
         start_event_queue.set_pending_offset(open_range_end);
-        open_range = Some(Range {
+        open_range = Some(CharRange {
           start: event.offset,
           end: open_range_end,
         });
@@ -452,8 +452,8 @@ mod tests {
             function_name: String::from("lib"),
             is_block_coverage: true,
             ranges: vec![CoverageRange {
-              start_offset: 0,
-              end_offset: 9,
+              start_char_offset: 0,
+              end_char_offset: 9,
               count: 1,
             }],
           }],
@@ -467,8 +467,8 @@ mod tests {
             function_name: String::from("lib"),
             is_block_coverage: true,
             ranges: vec![CoverageRange {
-              start_offset: 0,
-              end_offset: 9,
+              start_char_offset: 0,
+              end_char_offset: 9,
               count: 2,
             }],
           }],
@@ -483,8 +483,8 @@ mod tests {
           function_name: String::from("lib"),
           is_block_coverage: true,
           ranges: vec![CoverageRange {
-            start_offset: 0,
-            end_offset: 9,
+            start_char_offset: 0,
+            end_char_offset: 9,
             count: 3,
           }],
         }],
@@ -506,13 +506,13 @@ mod tests {
             is_block_coverage: true,
             ranges: vec![
               CoverageRange {
-                start_offset: 0,
-                end_offset: 9,
+                start_char_offset: 0,
+                end_char_offset: 9,
                 count: 10,
               },
               CoverageRange {
-                start_offset: 3,
-                end_offset: 6,
+                start_char_offset: 3,
+                end_char_offset: 6,
                 count: 1,
               },
             ],
@@ -528,13 +528,13 @@ mod tests {
             is_block_coverage: true,
             ranges: vec![
               CoverageRange {
-                start_offset: 0,
-                end_offset: 9,
+                start_char_offset: 0,
+                end_char_offset: 9,
                 count: 20,
               },
               CoverageRange {
-                start_offset: 3,
-                end_offset: 6,
+                start_char_offset: 3,
+                end_char_offset: 6,
                 count: 2,
               },
             ],
@@ -551,13 +551,13 @@ mod tests {
           is_block_coverage: true,
           ranges: vec![
             CoverageRange {
-              start_offset: 0,
-              end_offset: 9,
+              start_char_offset: 0,
+              end_char_offset: 9,
               count: 30,
             },
             CoverageRange {
-              start_offset: 3,
-              end_offset: 6,
+              start_char_offset: 3,
+              end_char_offset: 6,
               count: 3,
             },
           ],
@@ -580,13 +580,13 @@ mod tests {
             is_block_coverage: true,
             ranges: vec![
               CoverageRange {
-                start_offset: 0,
-                end_offset: 9,
+                start_char_offset: 0,
+                end_char_offset: 9,
                 count: 10,
               },
               CoverageRange {
-                start_offset: 2,
-                end_offset: 5,
+                start_char_offset: 2,
+                end_char_offset: 5,
                 count: 1,
               },
             ],
@@ -602,13 +602,13 @@ mod tests {
             is_block_coverage: true,
             ranges: vec![
               CoverageRange {
-                start_offset: 0,
-                end_offset: 9,
+                start_char_offset: 0,
+                end_char_offset: 9,
                 count: 20,
               },
               CoverageRange {
-                start_offset: 4,
-                end_offset: 7,
+                start_char_offset: 4,
+                end_char_offset: 7,
                 count: 2,
               },
             ],
@@ -625,23 +625,23 @@ mod tests {
           is_block_coverage: true,
           ranges: vec![
             CoverageRange {
-              start_offset: 0,
-              end_offset: 9,
+              start_char_offset: 0,
+              end_char_offset: 9,
               count: 30,
             },
             CoverageRange {
-              start_offset: 2,
-              end_offset: 5,
+              start_char_offset: 2,
+              end_char_offset: 5,
               count: 21,
             },
             CoverageRange {
-              start_offset: 4,
-              end_offset: 5,
+              start_char_offset: 4,
+              end_char_offset: 5,
               count: 3,
             },
             CoverageRange {
-              start_offset: 5,
-              end_offset: 7,
+              start_char_offset: 5,
+              end_char_offset: 7,
               count: 12,
             },
           ],
@@ -664,23 +664,23 @@ mod tests {
             is_block_coverage: true,
             ranges: vec![
               CoverageRange {
-                start_offset: 0,
-                end_offset: 9,
+                start_char_offset: 0,
+                end_char_offset: 9,
                 count: 1,
               },
               CoverageRange {
-                start_offset: 1,
-                end_offset: 8,
+                start_char_offset: 1,
+                end_char_offset: 8,
                 count: 6,
               },
               CoverageRange {
-                start_offset: 1,
-                end_offset: 5,
+                start_char_offset: 1,
+                end_char_offset: 5,
                 count: 5,
               },
               CoverageRange {
-                start_offset: 5,
-                end_offset: 8,
+                start_char_offset: 5,
+                end_char_offset: 8,
                 count: 7,
               },
             ],
@@ -696,23 +696,23 @@ mod tests {
             is_block_coverage: true,
             ranges: vec![
               CoverageRange {
-                start_offset: 0,
-                end_offset: 9,
+                start_char_offset: 0,
+                end_char_offset: 9,
                 count: 4,
               },
               CoverageRange {
-                start_offset: 1,
-                end_offset: 8,
+                start_char_offset: 1,
+                end_char_offset: 8,
                 count: 8,
               },
               CoverageRange {
-                start_offset: 1,
-                end_offset: 5,
+                start_char_offset: 1,
+                end_char_offset: 5,
                 count: 9,
               },
               CoverageRange {
-                start_offset: 5,
-                end_offset: 8,
+                start_char_offset: 5,
+                end_char_offset: 8,
                 count: 7,
               },
             ],
@@ -729,13 +729,13 @@ mod tests {
           is_block_coverage: true,
           ranges: vec![
             CoverageRange {
-              start_offset: 0,
-              end_offset: 9,
+              start_char_offset: 0,
+              end_char_offset: 9,
               count: 5,
             },
             CoverageRange {
-              start_offset: 1,
-              end_offset: 8,
+              start_char_offset: 1,
+              end_char_offset: 8,
               count: 14,
             },
           ],
@@ -758,13 +758,13 @@ mod tests {
             is_block_coverage: true,
             ranges: vec![
               CoverageRange {
-                start_offset: 0,
-                end_offset: 7,
+                start_char_offset: 0,
+                end_char_offset: 7,
                 count: 10,
               },
               CoverageRange {
-                start_offset: 0,
-                end_offset: 4,
+                start_char_offset: 0,
+                end_char_offset: 4,
                 count: 1,
               },
             ],
@@ -780,18 +780,18 @@ mod tests {
             is_block_coverage: true,
             ranges: vec![
               CoverageRange {
-                start_offset: 0,
-                end_offset: 7,
+                start_char_offset: 0,
+                end_char_offset: 7,
                 count: 20,
               },
               CoverageRange {
-                start_offset: 1,
-                end_offset: 6,
+                start_char_offset: 1,
+                end_char_offset: 6,
                 count: 11,
               },
               CoverageRange {
-                start_offset: 2,
-                end_offset: 5,
+                start_char_offset: 2,
+                end_char_offset: 5,
                 count: 2,
               },
             ],
@@ -808,23 +808,23 @@ mod tests {
           is_block_coverage: true,
           ranges: vec![
             CoverageRange {
-              start_offset: 0,
-              end_offset: 7,
+              start_char_offset: 0,
+              end_char_offset: 7,
               count: 30,
             },
             CoverageRange {
-              start_offset: 0,
-              end_offset: 6,
+              start_char_offset: 0,
+              end_char_offset: 6,
               count: 21,
             },
             CoverageRange {
-              start_offset: 1,
-              end_offset: 5,
+              start_char_offset: 1,
+              end_char_offset: 5,
               count: 12,
             },
             CoverageRange {
-              start_offset: 2,
-              end_offset: 4,
+              start_char_offset: 2,
+              end_char_offset: 4,
               count: 3,
             },
           ],
diff --git a/cli/tools/coverage/mod.rs b/cli/tools/coverage/mod.rs
index 0b1a9bc35..bba7271f5 100644
--- a/cli/tools/coverage/mod.rs
+++ b/cli/tools/coverage/mod.rs
@@ -202,16 +202,18 @@ fn generate_coverage_report(
       continue;
     }
 
-    let source_line_index =
-      text_lines.line_index(function.ranges[0].start_offset);
+    let dest_line_index = text_lines.line_index(
+      text_lines
+        .byte_index_from_char_index(function.ranges[0].start_char_offset),
+    );
     let line_index = if let Some(source_map) = maybe_source_map.as_ref() {
       source_map
         .tokens()
-        .find(|token| token.get_dst_line() as usize == source_line_index)
+        .find(|token| token.get_dst_line() as usize == dest_line_index)
         .map(|token| token.get_src_line() as usize)
         .unwrap_or(0)
     } else {
-      source_line_index
+      dest_line_index
     };
 
     coverage_report.named_functions.push(FunctionCoverageItem {
@@ -224,7 +226,9 @@ fn generate_coverage_report(
   for (block_number, function) in script_coverage.functions.iter().enumerate() {
     let block_hits = function.ranges[0].count;
     for (branch_number, range) in function.ranges[1..].iter().enumerate() {
-      let source_line_index = text_lines.line_index(range.start_offset);
+      let source_line_index = text_lines.line_index(
+        text_lines.byte_index_from_char_index(range.start_char_offset),
+      );
       let line_index = if let Some(source_map) = maybe_source_map.as_ref() {
         source_map
           .tokens()
@@ -264,11 +268,14 @@ fn generate_coverage_report(
   // parts of a line in color (word diff style) instead of the entire line.
   let mut line_counts = Vec::with_capacity(text_lines.lines_count());
   for line_index in 0..text_lines.lines_count() {
-    let line_start_offset = text_lines.line_start(line_index);
-    let line_end_offset = text_lines.line_end(line_index);
+    let line_start_byte_offset = text_lines.line_start(line_index);
+    let line_start_char_offset = text_lines.char_index(line_start_byte_offset);
+    let line_end_byte_offset = text_lines.line_end(line_index);
+    let line_end_char_offset = text_lines.char_index(line_end_byte_offset);
     let ignore = comment_ranges.iter().any(|range| {
-      range.start <= line_start_offset && range.end >= line_end_offset
-    }) || script_source[line_start_offset..line_end_offset]
+      range.start <= line_start_byte_offset && range.end >= line_end_byte_offset
+    }) || script_source
+      [line_start_byte_offset..line_end_byte_offset]
       .trim()
       .is_empty();
     let mut count = 0;
@@ -280,8 +287,8 @@ fn generate_coverage_report(
       // as long as the code has been evaluated.
       for function in &script_coverage.functions {
         for range in &function.ranges {
-          if range.start_offset <= line_start_offset
-            && range.end_offset >= line_end_offset
+          if range.start_char_offset <= line_start_char_offset
+            && range.end_char_offset >= line_end_char_offset
           {
             count += range.count;
           }
@@ -295,8 +302,8 @@ fn generate_coverage_report(
             continue;
           }
 
-          let overlaps = range.start_offset < line_end_offset
-            && range.end_offset > line_start_offset;
+          let overlaps = range.start_char_offset < line_end_char_offset
+            && range.end_char_offset > line_start_char_offset;
           if overlaps {
             count = 0;
           }
diff --git a/cli/tools/coverage/range_tree.rs b/cli/tools/coverage/range_tree.rs
index 24d0a9ffc..aca8939ee 100644
--- a/cli/tools/coverage/range_tree.rs
+++ b/cli/tools/coverage/range_tree.rs
@@ -134,8 +134,8 @@ impl<'rt> RangeTree<'rt> {
     while let Some((cur, parent_count)) = stack.pop() {
       let count: i64 = parent_count + cur.delta;
       ranges.push(CoverageRange {
-        start_offset: cur.start,
-        end_offset: cur.end,
+        start_char_offset: cur.start,
+        end_char_offset: cur.end,
         count,
       });
       for child in cur.children.iter().rev() {
@@ -165,14 +165,14 @@ impl<'rt> RangeTree<'rt> {
   ) -> Option<&'a mut RangeTree<'a>> {
     let has_range: bool = match ranges.peek() {
       None => false,
-      Some(range) => range.start_offset < parent_end,
+      Some(range) => range.start_char_offset < parent_end,
     };
     if !has_range {
       return None;
     }
     let range = ranges.next().unwrap();
-    let start: usize = range.start_offset;
-    let end: usize = range.end_offset;
+    let start: usize = range.start_char_offset;
+    let end: usize = range.end_char_offset;
     let count: i64 = range.count;
     let delta: i64 = count - parent_count;
     let mut children: Vec<&mut RangeTree> = Vec::new();
@@ -193,8 +193,8 @@ mod tests {
   fn from_sorted_ranges_empty() {
     let rta = RangeTreeArena::new();
     let inputs: Vec<CoverageRange> = vec![CoverageRange {
-      start_offset: 0,
-      end_offset: 9,
+      start_char_offset: 0,
+      end_char_offset: 9,
       count: 1,
     }];
     let actual: Option<&mut RangeTree> =