Allow shebang comments at start-of-file (#4473)

crates/ruff/resources/test/fixtures/pycodestyle/E26.py

@@ -64,3 +64,11 @@ a = 42  #  (One space one NBSP)
 #: E262:2:9
 #  (Two spaces) Ok for block comment
 a = 42  #  (Two spaces)
+
+#: E265:5:1
+### Means test is not done yet
+# E   Means test is giving error (E)
+# F   Means test is failing (F)
+# EF  Means test is giving error and Failing
+#!   Means test is segfaulting
+# 8   Means test runs forever

crates/ruff/resources/test/fixtures/pycodestyle/shebang.py

@@ -0,0 +1,4 @@
+#!/usr/bin/python
+#
+#!
+#:

crates/ruff/src/checkers/logical_lines.rs

@@ -67,7 +67,7 @@ pub(crate) fn check_logical_lines(
         }
 
         if line.flags().contains(TokenFlags::COMMENT) {
-            whitespace_before_comment(&line, locator, prev_line.is_none(), &mut context);
+            whitespace_before_comment(&line, locator, &mut context);
         }
 
         if line.flags().contains(TokenFlags::BRACKET) {

crates/ruff/src/rules/pycodestyle/mod.rs

@@ -128,6 +128,21 @@ mod tests {
         Ok(())
     }
 
+    #[test]
+    fn shebang() -> Result<()> {
+        let diagnostics = test_path(
+            Path::new("pycodestyle/shebang.py"),
+            &settings::Settings::for_rules(vec![
+                Rule::TooFewSpacesBeforeInlineComment,
+                Rule::NoSpaceAfterInlineComment,
+                Rule::NoSpaceAfterBlockComment,
+                Rule::MultipleLeadingHashesForBlockComment,
+            ]),
+        )?;
+        assert_messages!(diagnostics);
+        Ok(())
+    }
+
     #[test_case(false)]
     #[test_case(true)]
     fn task_tags(ignore_overlong_task_comments: bool) -> Result<()> {

crates/ruff/src/rules/pycodestyle/rules/logical_lines/mod.rs

@@ -151,6 +151,11 @@ pub(crate) struct LogicalLine<'a> {
 }
 
 impl<'a> LogicalLine<'a> {
+    /// Returns `true` if this line is positioned at the start of the file.
+    pub(crate) const fn is_start_of_file(&self) -> bool {
+        self.line.tokens_start == 0
+    }
+
     /// Returns `true` if this is a comment only line
     pub(crate) fn is_comment_only(&self) -> bool {
         self.flags() == TokenFlags::COMMENT

crates/ruff/src/rules/pycodestyle/rules/logical_lines/whitespace_before_comment.rs

@@ -139,7 +139,6 @@ impl Violation for MultipleLeadingHashesForBlockComment {
 pub(crate) fn whitespace_before_comment(
     line: &LogicalLine,
     locator: &Locator,
-    is_first_row: bool,
     context: &mut LogicalLinesContext,
 ) {
     let mut prev_end = TextSize::default();
@@ -149,13 +148,13 @@ pub(crate) fn whitespace_before_comment(
         if let TokenKind::Comment = kind {
             let range = token.range();
 
-            let line = locator.slice(TextRange::new(
+            let line_text = locator.slice(TextRange::new(
                 locator.line_start(range.start()),
                 range.start(),
             ));
-            let text = locator.slice(range);
+            let token_text = locator.slice(range);
 
-            let is_inline_comment = !line.trim().is_empty();
+            let is_inline_comment = !line_text.trim().is_empty();
             if is_inline_comment {
                 if range.start() - prev_end < "  ".text_len() {
                     context.push(
@@ -166,7 +165,7 @@ pub(crate) fn whitespace_before_comment(
             }
 
             // Split into the portion before and after the first space.
-            let mut parts = text.splitn(2, ' ');
+            let mut parts = token_text.splitn(2, ' ');
             let symbol = parts.next().unwrap_or("");
             let comment = parts.next().unwrap_or("");
 
@@ -182,7 +181,7 @@ pub(crate) fn whitespace_before_comment(
                     context.push(NoSpaceAfterInlineComment, range);
                 }
             } else if let Some(bad_prefix) = bad_prefix {
-                if bad_prefix != '!' || !is_first_row {
+                if bad_prefix != '!' || !line.is_start_of_file() {
                     if bad_prefix != '#' {
                         context.push(NoSpaceAfterBlockComment, range);
                     } else if !comment.is_empty() {

crates/ruff/src/rules/pycodestyle/snapshots/ruff__rules__pycodestyle__tests__E262_E26.py.snap

@@ -47,6 +47,8 @@ E26.py:66:9: E262 Inline comment should start with `# `
 67 | #  (Two spaces) Ok for block comment
 68 | a = 42  #  (Two spaces)
    |         ^^^^^^^^^^^^^^^ E262
+69 | 
+70 | #: E265:5:1
    |
 
 

crates/ruff/src/rules/pycodestyle/snapshots/ruff__rules__pycodestyle__tests__E265_E26.py.snap

@@ -40,4 +40,13 @@ E26.py:32:1: E265 Block comment should start with `# `
 35 | pass  # an inline comment
    |
 
+E26.py:73:1: E265 Block comment should start with `# `
+   |
+73 | # F   Means test is failing (F)
+74 | # EF  Means test is giving error and Failing
+75 | #!   Means test is segfaulting
+   | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ E265
+76 | # 8   Means test runs forever
+   |
+
 

crates/ruff/src/rules/pycodestyle/snapshots/ruff__rules__pycodestyle__tests__E266_E26.py.snap

@@ -30,4 +30,13 @@ E26.py:26:1: E266 Too many leading `#` before block comment
 30 | #########################################
    |
 
+E26.py:69:1: E266 Too many leading `#` before block comment
+   |
+69 | #: E265:5:1
+70 | ### Means test is not done yet
+   | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ E266
+71 | # E   Means test is giving error (E)
+72 | # F   Means test is failing (F)
+   |
+
 

crates/ruff/src/rules/pycodestyle/snapshots/ruff__rules__pycodestyle__tests__shebang.snap

@@ -0,0 +1,13 @@
+---
+source: crates/ruff/src/rules/pycodestyle/mod.rs
+---
+shebang.py:3:1: E265 Block comment should start with `# `
+  |
+3 | #!/usr/bin/python
+4 | #
+5 | #!
+  | ^^ E265
+6 | #:
+  |
+
+