fix: preserve newlines in block comments to match JavaScript behavior

src/lib.rs

@@ -208,7 +208,9 @@ fn consume_line_comments(buf: &mut [u8], i: &mut usize) -> State {
     match memchr::memchr(b'\n', remaining) {
         Some(offset) => {
             *i += offset;
-            buf[cur..*i].fill(b' ');
+            // Preserve \r if it comes right before \n (Windows line endings)
+            let end = if *i > 0 && buf[*i - 1] == b'\r' { *i - 1 } else { *i };
+            buf[cur..end].fill(b' ');
             Top
         }
         None => {
@@ -227,18 +229,42 @@ fn consume_block_comments(buf: &mut [u8], i: &mut usize) -> State {
     match memchr::memchr(b'*', remaining) {
         Some(offset) => {
             *i += offset;
-            buf[cur..=*i].fill(b' ');
+            // Preserve newlines in block comments efficiently
+            fill_non_newlines(&mut buf[cur..=*i]);
             MaybeCommentEnd
         }
         None => {
             let len = buf.len();
             *i = len - 1;
-            buf[cur..len].fill(b' ');
+            // Preserve newlines in block comments efficiently
+            fill_non_newlines(&mut buf[cur..len]);
             InBlockComment
         }
     }
 }
 
+/// Fill a buffer with spaces, preserving newlines for performance
+#[inline]
+fn fill_non_newlines(buf: &mut [u8]) {
+    let mut pos = 0;
+    while pos < buf.len() {
+        // Find the next newline (\n or \r)
+        match memchr::memchr2(b'\n', b'\r', &buf[pos..]) {
+            Some(offset) => {
+                // Fill everything before the newline with spaces
+                buf[pos..pos + offset].fill(b' ');
+                // Skip the newline character
+                pos += offset + 1;
+            }
+            None => {
+                // No more newlines, fill the rest
+                buf[pos..].fill(b' ');
+                break;
+            }
+        }
+    }
+}
+
 #[inline(always)]
 fn top(c: &mut u8) -> State {
     match *c {
@@ -278,7 +304,10 @@ fn in_comment(c: &mut u8) -> Result<State> {
 #[inline]
 fn maybe_comment_end(c: &mut u8) -> State {
     let old = *c;
-    *c = b' ';
+    // Preserve newlines in block comments
+    if old != b'\n' && old != b'\r' {
+        *c = b' ';
+    }
     match old {
         b'/' => Top,
         b'*' => MaybeCommentEnd,

tests/main.rs

@@ -428,16 +428,13 @@ fn sindresorhus_replace_comments_with_whitespace() {
     assert_eq!(strip_string("/*//comment*/{\"a\":\"b\"}"), "             {\"a\":\"b\"}");
     assert_eq!(strip_string("{\"a\":\"b\"//comment\n}"), "{\"a\":\"b\"         \n}");
     assert_eq!(strip_string("{\"a\":\"b\"/*comment*/}"), "{\"a\":\"b\"           }");
-    // Note: The Rust implementation replaces newlines in block comments with spaces,
-    // unlike the JavaScript version which preserves them
     assert_eq!(
         strip_string("{\"a\"/*\n\n\ncomment\r\n*/:\"b\"}"),
-        "{\"a\"                :\"b\"}"
+        "{\"a\"  \n\n\n       \r\n  :\"b\"}"
     );
-    // Note: Same for multi-line comments
     assert_eq!(
         strip_string("/*!\n * comment\n */\n{\"a\":\"b\"}"),
-        "                  \n{\"a\":\"b\"}"
+        "   \n          \n   \n{\"a\":\"b\"}"
     );
     assert_eq!(strip_string("{/*comment*/\"a\":\"b\"}"), "{           \"a\":\"b\"}");
 }
@@ -471,8 +468,7 @@ fn sindresorhus_line_endings_no_comments() {
 #[test]
 fn sindresorhus_line_endings_single_line_comment() {
     assert_eq!(strip_string("{\"a\":\"b\"//c\n}"), "{\"a\":\"b\"   \n}");
-    // Note: The Rust implementation treats \r\n differently - it replaces the \r as part of comment
-    assert_eq!(strip_string("{\"a\":\"b\"//c\r\n}"), "{\"a\":\"b\"    \n}");
+    assert_eq!(strip_string("{\"a\":\"b\"//c\r\n}"), "{\"a\":\"b\"   \r\n}");
 }
 
 #[test]
@@ -483,14 +479,13 @@ fn sindresorhus_line_endings_single_line_block_comment() {
 
 #[test]
 fn sindresorhus_line_endings_multi_line_block_comment() {
-    // Note: The Rust implementation replaces newlines inside block comments with spaces
     assert_eq!(
         strip_string("{\"a\":\"b\",/*c\nc2*/\"x\":\"y\"\n}"),
-        "{\"a\":\"b\",        \"x\":\"y\"\n}"
+        "{\"a\":\"b\",   \n    \"x\":\"y\"\n}"
     );
     assert_eq!(
         strip_string("{\"a\":\"b\",/*c\r\nc2*/\"x\":\"y\"\r\n}"),
-        "{\"a\":\"b\",         \"x\":\"y\"\r\n}"
+        "{\"a\":\"b\",   \r\n    \"x\":\"y\"\r\n}"
     );
 }