Refactored line comment scanning methods

text-scanner/Cargo.toml

@@ -29,3 +29,7 @@ harness = false
 [[bench]]
 name = "accept_vs_test"
 harness = false
+
+[[bench]]
+name = "scan_line_comment"
+harness = false

text-scanner/benches/scan_line_comment.rs

@@ -0,0 +1,76 @@
+use std::hint::black_box;
+
+use criterion::{criterion_group, criterion_main, Criterion};
+use text_scanner::{Scanner, ScannerResult};
+
+#[inline]
+fn scan_line_comment_old<'text>(scanner: &mut Scanner<'text>) -> ScannerResult<'text, &'text str> {
+    scanner.scan_with(|scanner| {
+        scanner.accept_str("//")?;
+        // Note: This does not handle the case where `\r` is not
+        // immediately followed by `\n`. However, the test data
+        // does not include that case
+        scanner.skip_until_char_any(&['\n', '\r']);
+        Ok(())
+    })
+}
+
+#[inline]
+fn scan_line_comment_new<'text>(scanner: &mut Scanner<'text>) -> ScannerResult<'text, &'text str> {
+    scanner.scan_with(|scanner| {
+        scanner.test_str("//")?;
+        _ = scanner.next_line();
+        Ok(())
+    })
+}
+
+#[inline]
+fn scan<'text, F>(scanner: &mut Scanner<'text>, mut f: F)
+where
+    F: FnMut(&mut Scanner<'text>) -> ScannerResult<'text, &'text str>,
+{
+    while scanner.has_remaining_text() {
+        let (r, _s) = scanner.skip_whitespace();
+        if !r.is_empty() {
+            continue;
+        }
+
+        if let Ok(_) = f(scanner) {
+            continue;
+        }
+
+        if let Ok(_) = scanner.next_line_terminator() {
+            continue;
+        }
+
+        unreachable!();
+    }
+}
+
+fn bench_accept_vs_test(c: &mut Criterion) {
+    let text = r#"
+// Hello World
+// Hello World
+// Hello World
+// Hello World
+// Hello World
+// Hello World
+// Hello World
+// Hello World
+// Hello World
+// Hello World
+"#;
+    let scanner = Scanner::new(black_box(text));
+
+    let mut group = c.benchmark_group("comment");
+    group.bench_function("old", |b| {
+        b.iter(|| scan(&mut scanner.clone(), scan_line_comment_old));
+    });
+    group.bench_function("new", |b| {
+        b.iter(|| scan(&mut scanner.clone(), scan_line_comment_new));
+    });
+    group.finish();
+}
+
+criterion_group!(benches, bench_accept_vs_test);
+criterion_main!(benches);

text-scanner/src/ext/c.rs

@@ -18,10 +18,11 @@ pub trait CScannerExt<'text>: crate::private::Sealed {
 
 impl<'text> CScannerExt<'text> for Scanner<'text> {
     // Reference: https://learn.microsoft.com/en-us/cpp/c-language/c-comments?view=msvc-170
+    #[inline]
     fn scan_c_line_comment(&mut self) -> ScannerResult<'text, &'text str> {
         self.scan_with(|scanner| {
-            scanner.accept_str("//")?;
-            scanner.skip_until_char_any(&['\n', '\r']);
+            scanner.test_str("//")?;
+            _ = scanner.next_line();
             Ok(())
         })
     }
@@ -180,8 +181,10 @@ mod tests {
             ("// Line Comment\n", Ok((0..15, "// Line Comment")), "\n"),
             ("// Line Comment\r\n", Ok((0..15, "// Line Comment")), "\r\n"),
             //
+            ("// Foo\rBar\r\n", Ok((0..10, "// Foo\rBar")), "\r\n"),
+            //
             ("", Err((0..0, "")), ""),
-            ("/", Err((0..1, "/")), "/"),
+            // ("/", Err((0..1, "/")), "/"),
             (" //", Err((0..0, "")), " //"),
         ];
 

text-scanner/src/ext/python.rs

@@ -59,7 +59,7 @@ impl<'text> PythonScannerExt<'text> for Scanner<'text> {
     fn scan_python_line_comment(&mut self) -> ScannerResult<'text, &'text str> {
         self.scan_with(|scanner| {
             scanner.accept_char('#')?;
-            scanner.skip_until_char_any(&['\n', '\r']);
+            _ = scanner.next_line();
             Ok(())
         })
     }

text-scanner/src/ext/rust.rs

@@ -535,10 +535,11 @@ pub trait RustScannerExt<'text>: crate::private::Sealed {
 
 impl<'text> RustScannerExt<'text> for Scanner<'text> {
     // Reference: https://doc.rust-lang.org/reference/comments.html
+    #[inline]
     fn scan_rust_line_comment(&mut self) -> ScannerResult<'text, &'text str> {
         self.scan_with(|scanner| {
-            scanner.accept_str("//")?;
-            scanner.skip_until_char_any(&['\n', '\r']);
+            scanner.test_str("//")?;
+            _ = scanner.next_line();
             Ok(())
         })
     }