Implement advanced performance optimizations with significant speedups

Co-authored-by: Boshen <[email protected]>

Author: Copilot

src/lib.rs

@@ -31,15 +31,14 @@ use std::io::{ErrorKind, Read, Result};
 enum State {
     Top,
     InString,
-    StringEscape,
     InComment,
     InBlockComment,
     MaybeCommentEnd,
     InLineComment,
 }
 
 use State::{
-    InBlockComment, InComment, InLineComment, InString, MaybeCommentEnd, StringEscape, Top,
+    InBlockComment, InComment, InLineComment, InString, MaybeCommentEnd, Top,
 };
 
 /// A [`Read`] that transforms another [`Read`] so that it changes all comments to spaces so that a downstream json parser
@@ -159,24 +158,15 @@ pub fn strip(s: &mut str) -> Result<()> {
 /// Fast check if the input contains any comment-starting characters
 #[inline]
 fn has_potential_comments(s: &[u8], settings: CommentSettings) -> bool {
-    // Combine all potential comment starting characters into one search
-    let mut chars_to_find = Vec::with_capacity(2);
-    
-    if settings.block_comments || settings.slash_line_comments {
-        chars_to_find.push(b'/');
-    }
-    if settings.hash_line_comments {
-        chars_to_find.push(b'#');
-    }
-    
-    if chars_to_find.is_empty() {
-        return false;
-    }
-    
-    if chars_to_find.len() == 1 {
-        memchr::memchr(chars_to_find[0], s).is_some()
-    } else {
-        memchr::memchr2(chars_to_find[0], chars_to_find[1], s).is_some()
+    // Use the most efficient memchr variant based on enabled comment types
+    match (
+        settings.block_comments || settings.slash_line_comments,
+        settings.hash_line_comments,
+    ) {
+        (true, true) => memchr::memchr2(b'/', b'#', s).is_some(),
+        (true, false) => memchr::memchr(b'/', s).is_some(),
+        (false, true) => memchr::memchr(b'#', s).is_some(),
+        (false, false) => false,
     }
 }
 
@@ -336,8 +326,7 @@ fn consume_comment_whitespace_until_maybe_bracket(
                 }
                 return Ok(*c == b'}' || *c == b']');
             }
-            InString => in_string(*c),
-            StringEscape => InString,
+            InString => skip_string_content(buf, i),
             InComment => in_comment(c, settings)?,
             InBlockComment => consume_block_comments(buf, i),
             MaybeCommentEnd => maybe_comment_end(c),
@@ -348,12 +337,13 @@ fn consume_comment_whitespace_until_maybe_bracket(
     Ok(false)
 }
 
+
 fn strip_buf(state: &mut State, buf: &mut [u8], settings: CommentSettings) -> Result<()> {
     let mut i = 0;
     let len = buf.len();
     while i < len {
-        let c = &mut buf[i];
         if matches!(state, Top) {
+            let c = &mut buf[i];
             let cur = i;
             *state = top(c, settings);
             if settings.trailing_commas
@@ -363,10 +353,13 @@ fn strip_buf(state: &mut State, buf: &mut [u8], settings: CommentSettings) -> Re
                 buf[cur] = b' ';
             }
         } else {
+            let c = &mut buf[i];
             *state = match state {
                 Top => unreachable!(),
-                InString => in_string(*c),
-                StringEscape => InString,
+                InString => {
+                    // Use optimized string skipping for long strings
+                    skip_string_content(buf, &mut i)
+                }
                 InComment => in_comment(c, settings)?,
                 InBlockComment => consume_block_comments(buf, &mut i),
                 MaybeCommentEnd => maybe_comment_end(c),
@@ -430,12 +423,34 @@ fn top(c: &mut u8, settings: CommentSettings) -> State {
     }
 }
 
+
+/// Optimized string content skipping using memchr
 #[inline]
-fn in_string(c: u8) -> State {
-    match c {
-        b'"' => Top,
-        b'\\' => StringEscape,
-        _ => InString,
+fn skip_string_content(buf: &[u8], i: &mut usize) -> State {
+    // Find the next quote or backslash
+    match memchr::memchr2(b'"', b'\\', &buf[*i..]) {
+        Some(offset) => {
+            *i += offset;
+            match buf[*i] {
+                b'"' => {
+                    // Found closing quote
+                    Top
+                }
+                b'\\' => {
+                    // Found escape, skip next character if it exists
+                    if *i + 1 < buf.len() {
+                        *i += 1;
+                    }
+                    InString
+                }
+                _ => unreachable!(),
+            }
+        }
+        None => {
+            // No more quotes or escapes in this buffer
+            *i = buf.len() - 1;
+            InString
+        }
     }
 }