Use Cursor for shebang parsing (#5716)

## Summary

Better to leverage the shared functionality we get from `Cursor`. It's
also a little bit faster, which is very cool.

crates/ruff/src/checkers/physical_lines.rs

@@ -7,9 +7,9 @@ use ruff_diagnostics::Diagnostic;
 use ruff_python_ast::source_code::{Indexer, Locator, Stylist};
 use ruff_python_whitespace::UniversalNewlines;
 
+use crate::comments::shebang::ShebangDirective;
 use crate::registry::Rule;
 use crate::rules::flake8_copyright::rules::missing_copyright_notice;
-use crate::rules::flake8_executable::helpers::ShebangDirective;
 use crate::rules::flake8_executable::rules::{
     shebang_missing, shebang_newline, shebang_not_executable, shebang_python, shebang_whitespace,
 };

crates/ruff/src/comments/mod.rs

@@ -0,0 +1 @@
+pub(crate) mod shebang;

crates/ruff/src/comments/shebang.rs

@@ -0,0 +1,67 @@
+use ruff_python_whitespace::{is_python_whitespace, Cursor};
+use ruff_text_size::{TextLen, TextSize};
+
+/// A shebang directive (e.g., `#!/usr/bin/env python3`).
+#[derive(Debug, PartialEq, Eq)]
+pub(crate) struct ShebangDirective<'a> {
+    /// The offset of the directive contents (e.g., `/usr/bin/env python3`) from the start of the
+    /// line.
+    pub(crate) offset: TextSize,
+    /// The contents of the directive (e.g., `"/usr/bin/env python3"`).
+    pub(crate) contents: &'a str,
+}
+
+impl<'a> ShebangDirective<'a> {
+    /// Parse a shebang directive from a line, or return `None` if the line does not contain a
+    /// shebang directive.
+    pub(crate) fn try_extract(line: &'a str) -> Option<Self> {
+        let mut cursor = Cursor::new(line);
+
+        // Trim whitespace.
+        cursor.eat_while(is_python_whitespace);
+
+        // Trim the `#!` prefix.
+        if !cursor.eat_char('#') {
+            return None;
+        }
+        if !cursor.eat_char('!') {
+            return None;
+        }
+
+        Some(Self {
+            offset: line.text_len() - cursor.text_len(),
+            contents: cursor.chars().as_str(),
+        })
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use insta::assert_debug_snapshot;
+
+    use super::ShebangDirective;
+
+    #[test]
+    fn shebang_non_match() {
+        let source = "not a match";
+        assert_debug_snapshot!(ShebangDirective::try_extract(source));
+    }
+
+    #[test]
+    fn shebang_end_of_line() {
+        let source = "print('test')  #!/usr/bin/python";
+        assert_debug_snapshot!(ShebangDirective::try_extract(source));
+    }
+
+    #[test]
+    fn shebang_match() {
+        let source = "#!/usr/bin/env python";
+        assert_debug_snapshot!(ShebangDirective::try_extract(source));
+    }
+
+    #[test]
+    fn shebang_leading_space() {
+        let source = "  #!/usr/bin/env python";
+        assert_debug_snapshot!(ShebangDirective::try_extract(source));
+    }
+}

crates/ruff/src/comments/snapshots/ruff__comments__shebang__tests__shebang_end_of_line.snap

@@ -1,5 +1,5 @@
 ---
-source: crates/ruff/src/rules/flake8_executable/helpers.rs
+source: crates/ruff/src/comments/shebang.rs
 expression: "ShebangDirective::try_extract(source)"
 ---
 None

crates/ruff/src/comments/snapshots/ruff__comments__shebang__tests__shebang_leading_space.snap

@@ -1,5 +1,5 @@
 ---
-source: crates/ruff/src/rules/flake8_executable/helpers.rs
+source: crates/ruff/src/comments/shebang.rs
 expression: "ShebangDirective::try_extract(source)"
 ---
 Some(

crates/ruff/src/comments/snapshots/ruff__comments__shebang__tests__shebang_match.snap

@@ -1,5 +1,5 @@
 ---
-source: crates/ruff/src/rules/flake8_executable/helpers.rs
+source: crates/ruff/src/comments/shebang.rs
 expression: "ShebangDirective::try_extract(source)"
 ---
 Some(

crates/ruff/src/comments/snapshots/ruff__comments__shebang__tests__shebang_non_match.snap

@@ -1,5 +1,5 @@
 ---
-source: crates/ruff/src/rules/flake8_executable/helpers.rs
+source: crates/ruff/src/comments/shebang.rs
 expression: "ShebangDirective::try_extract(source)"
 ---
 None

crates/ruff/src/lib.rs

@@ -14,6 +14,7 @@ pub const VERSION: &str = env!("CARGO_PKG_VERSION");
 mod autofix;
 mod checkers;
 mod codes;
+mod comments;
 mod cst;
 pub mod directives;
 mod doc_lines;

crates/ruff/src/rules/flake8_executable/helpers.rs

@@ -1,92 +1,12 @@
-#[cfg(target_family = "unix")]
+#![cfg(target_family = "unix")]
+
 use std::os::unix::fs::PermissionsExt;
-#[cfg(target_family = "unix")]
 use std::path::Path;
 
-#[cfg(target_family = "unix")]
 use anyhow::Result;
-use ruff_text_size::{TextLen, TextSize};
 
-/// A shebang directive (e.g., `#!/usr/bin/env python3`).
-#[derive(Debug, PartialEq, Eq)]
-pub(crate) struct ShebangDirective<'a> {
-    /// The offset of the directive contents (e.g., `/usr/bin/env python3`) from the start of the
-    /// line.
-    pub(crate) offset: TextSize,
-    /// The contents of the directive (e.g., `"/usr/bin/env python3"`).
-    pub(crate) contents: &'a str,
-}
-
-impl<'a> ShebangDirective<'a> {
-    /// Parse a shebang directive from a line, or return `None` if the line does not contain a
-    /// shebang directive.
-    pub(crate) fn try_extract(line: &'a str) -> Option<Self> {
-        // Trim whitespace.
-        let directive = Self::lex_whitespace(line);
-
-        // Trim the `#!` prefix.
-        let directive = Self::lex_char(directive, '#')?;
-        let directive = Self::lex_char(directive, '!')?;
-
-        Some(Self {
-            offset: line.text_len() - directive.text_len(),
-            contents: directive,
-        })
-    }
-
-    /// Lex optional leading whitespace.
-    #[inline]
-    fn lex_whitespace(line: &str) -> &str {
-        line.trim_start()
-    }
-
-    /// Lex a specific character, or return `None` if the character is not the first character in
-    /// the line.
-    #[inline]
-    fn lex_char(line: &str, c: char) -> Option<&str> {
-        let mut chars = line.chars();
-        if chars.next() == Some(c) {
-            Some(chars.as_str())
-        } else {
-            None
-        }
-    }
-}
-
-#[cfg(target_family = "unix")]
 pub(super) fn is_executable(filepath: &Path) -> Result<bool> {
     let metadata = filepath.metadata()?;
     let permissions = metadata.permissions();
     Ok(permissions.mode() & 0o111 != 0)
 }
-
-#[cfg(test)]
-mod tests {
-    use insta::assert_debug_snapshot;
-
-    use crate::rules::flake8_executable::helpers::ShebangDirective;
-
-    #[test]
-    fn shebang_non_match() {
-        let source = "not a match";
-        assert_debug_snapshot!(ShebangDirective::try_extract(source));
-    }
-
-    #[test]
-    fn shebang_end_of_line() {
-        let source = "print('test')  #!/usr/bin/python";
-        assert_debug_snapshot!(ShebangDirective::try_extract(source));
-    }
-
-    #[test]
-    fn shebang_match() {
-        let source = "#!/usr/bin/env python";
-        assert_debug_snapshot!(ShebangDirective::try_extract(source));
-    }
-
-    #[test]
-    fn shebang_leading_space() {
-        let source = "  #!/usr/bin/env python";
-        assert_debug_snapshot!(ShebangDirective::try_extract(source));
-    }
-}

crates/ruff/src/rules/flake8_executable/rules/shebang_newline.rs

@@ -3,7 +3,7 @@ use ruff_text_size::{TextLen, TextRange};
 use ruff_diagnostics::{Diagnostic, Violation};
 use ruff_macros::{derive_message_formats, violation};
 
-use crate::rules::flake8_executable::helpers::ShebangDirective;
+use crate::comments::shebang::ShebangDirective;
 
 /// ## What it does
 /// Checks for a shebang directive that is not at the beginning of the file.

crates/ruff/src/rules/flake8_executable/rules/shebang_not_executable.rs

@@ -7,10 +7,10 @@ use ruff_text_size::{TextLen, TextRange, TextSize};
 use ruff_diagnostics::{Diagnostic, Violation};
 use ruff_macros::{derive_message_formats, violation};
 
+use crate::comments::shebang::ShebangDirective;
 use crate::registry::AsRule;
 #[cfg(target_family = "unix")]
 use crate::rules::flake8_executable::helpers::is_executable;
-use crate::rules::flake8_executable::helpers::ShebangDirective;
 
 /// ## What it does
 /// Checks for a shebang directive in a file that is not executable.

crates/ruff/src/rules/flake8_executable/rules/shebang_python.rs

@@ -3,7 +3,7 @@ use ruff_text_size::{TextLen, TextRange, TextSize};
 use ruff_diagnostics::{Diagnostic, Violation};
 use ruff_macros::{derive_message_formats, violation};
 
-use crate::rules::flake8_executable::helpers::ShebangDirective;
+use crate::comments::shebang::ShebangDirective;
 
 /// ## What it does
 /// Checks for a shebang directive in `.py` files that does not contain `python`.

crates/ruff/src/rules/flake8_executable/rules/shebang_whitespace.rs

@@ -1,10 +1,11 @@
-use ruff_text_size::{TextRange, TextSize};
 use std::ops::Sub;
 
+use ruff_text_size::{TextRange, TextSize};
+
 use ruff_diagnostics::{AlwaysAutofixableViolation, Diagnostic, Edit, Fix};
 use ruff_macros::{derive_message_formats, violation};
 
-use crate::rules::flake8_executable::helpers::ShebangDirective;
+use crate::comments::shebang::ShebangDirective;
 
 /// ## What it does
 /// Checks for whitespace before a shebang directive.