Always expand tabs to four spaces in diagnostics

crates/ruff_db/Cargo.toml

@@ -42,7 +42,6 @@ serde_json = { workspace = true, optional = true }
 thiserror = { workspace = true }
 tracing = { workspace = true }
 tracing-subscriber = { workspace = true, optional = true }
-unicode-width = { workspace = true }
 zip = { workspace = true }
 
 [target.'cfg(target_arch="wasm32")'.dependencies]

crates/ruff_db/src/diagnostic/render.rs

@@ -585,8 +585,7 @@ impl<'r> RenderableSnippet<'r> {
         let EscapedSourceCode {
             text: snippet,
             annotations,
-        } = replace_whitespace_and_unprintable(snippet, annotations)
-            .fix_up_empty_spans_after_line_terminator();
+        } = replace_unprintable(snippet, annotations).fix_up_empty_spans_after_line_terminator();
 
         RenderableSnippet {
             snippet,
@@ -828,13 +827,18 @@ fn relativize_path<'p>(cwd: &SystemPath, path: &'p str) -> &'p str {
     path
 }
 
-/// Given some source code and annotation ranges, this routine replaces tabs
-/// with ASCII whitespace, and unprintable characters with printable
-/// representations of them.
+/// Given some source code and annotation ranges, this routine replaces
+/// unprintable characters with printable representations of them.
 ///
 /// The source code and annotations returned are updated to reflect changes made
 /// to the source code (if any).
-fn replace_whitespace_and_unprintable<'r>(
+///
+/// We don't need to normalize whitespace, such as converting tabs to spaces,
+/// because `annotate-snippets` handles that internally. Similarly, it's safe to
+/// modify the annotation ranges by inserting 3-byte Unicode replacements
+/// because `annotate-snippets` will account for their actual width when
+/// rendering and displaying the column to the user.
+fn replace_unprintable<'r>(
     source: &'r str,
     mut annotations: Vec<RenderableAnnotation<'r>>,
 ) -> EscapedSourceCode<'r> {
@@ -866,48 +870,17 @@ fn replace_whitespace_and_unprintable<'r>(
         }
     };
 
-    const TAB_SIZE: usize = 4;
-    let mut width = 0;
-    let mut column = 0;
     let mut last_end = 0;
     let mut result = String::new();
     for (index, c) in source.char_indices() {
-        let old_width = width;
-        match c {
-            '\n' | '\r' => {
-                width = 0;
-                column = 0;
-            }
-            '\t' => {
-                let tab_offset = TAB_SIZE - (column % TAB_SIZE);
-                width += tab_offset;
-                column += tab_offset;
+        if let Some(printable) = unprintable_replacement(c) {
+            result.push_str(&source[last_end..index]);
 
-                let tab_width =
-                    u32::try_from(width - old_width).expect("small width because of tab size");
-                result.push_str(&source[last_end..index]);
+            let len = printable.text_len().to_u32();
+            update_ranges(result.text_len().to_usize(), len);
 
-                update_ranges(result.text_len().to_usize(), tab_width);
-
-                for _ in 0..tab_width {
-                    result.push(' ');
-                }
-                last_end = index + 1;
-            }
-            _ => {
-                width += unicode_width::UnicodeWidthChar::width(c).unwrap_or(0);
-                column += 1;
-
-                if let Some(printable) = unprintable_replacement(c) {
-                    result.push_str(&source[last_end..index]);
-
-                    let len = printable.text_len().to_u32();
-                    update_ranges(result.text_len().to_usize(), len);
-
-                    result.push(printable);
-                    last_end = index + 1;
-                }
-            }
+            result.push(printable);
+            last_end = index + 1;
         }
     }
 

crates/ruff_db/src/diagnostic/render/full.rs

@@ -177,4 +177,25 @@ print()
 
         Ok(())
     }
+
+    /// Ensure that the header column matches the column in the user's input, even if we've replaced
+    /// tabs with spaces for rendering purposes.
+    #[test]
+    fn tab_replacement() {
+        let mut env = TestEnvironment::new();
+        env.add("example.py", "def foo():\n\treturn 1");
+        env.format(DiagnosticFormat::Full);
+
+        let diagnostic = env.err().primary("example.py", "2:1", "2:9", "").build();
+
+        insta::assert_snapshot!(env.render(&diagnostic), @r"
+        error[test-diagnostic]: main diagnostic message
+         --> example.py:2:2
+          |
+        1 | def foo():
+        2 |     return 1
+          |     ^^^^^^^^
+          |
+        ");
+    }
 }

crates/ruff_linter/src/message/text.rs

@@ -13,7 +13,6 @@ use ruff_notebook::NotebookIndex;
 use ruff_source_file::OneIndexed;
 use ruff_text_size::{TextLen, TextRange, TextSize};
 
-use crate::line_width::{IndentWidth, LineWidthBuilder};
 use crate::message::diff::Diff;
 use crate::message::{Emitter, EmitterContext};
 use crate::settings::types::UnsafeFixes;
@@ -229,7 +228,7 @@ impl Display for MessageCodeFrame<'_> {
         let start_offset = source_code.line_start(start_index);
         let end_offset = source_code.line_end(end_index);
 
-        let source = replace_whitespace_and_unprintable(
+        let source = replace_unprintable(
             source_code.slice(TextRange::new(start_offset, end_offset)),
             self.message.expect_range() - start_offset,
         )
@@ -272,16 +271,20 @@ impl Display for MessageCodeFrame<'_> {
 }
 
 /// Given some source code and an annotation range, this routine replaces
-/// tabs with ASCII whitespace, and unprintable characters with printable
-/// representations of them.
+///  unprintable characters with printable representations of them.
 ///
 /// The source code returned has an annotation that is updated to reflect
 /// changes made to the source code (if any).
-fn replace_whitespace_and_unprintable(source: &str, annotation_range: TextRange) -> SourceCode {
+///
+/// We don't need to normalize whitespace, such as converting tabs to spaces,
+/// because `annotate-snippets` handles that internally. Similarly, it's safe to
+/// modify the annotation ranges by inserting 3-byte Unicode replacements
+/// because `annotate-snippets` will account for their actual width when
+/// rendering and displaying the column to the user.
+fn replace_unprintable(source: &str, annotation_range: TextRange) -> SourceCode {
     let mut result = String::new();
     let mut last_end = 0;
     let mut range = annotation_range;
-    let mut line_width = LineWidthBuilder::new(IndentWidth::default());
 
     // Updates the range given by the caller whenever a single byte (at
     // `index` in `source`) is replaced with `len` bytes.
@@ -310,19 +313,7 @@ fn replace_whitespace_and_unprintable(source: &str, annotation_range: TextRange)
     };
 
     for (index, c) in source.char_indices() {
-        let old_width = line_width.get();
-        line_width = line_width.add_char(c);
-
-        if matches!(c, '\t') {
-            let tab_width = u32::try_from(line_width.get() - old_width)
-                .expect("small width because of tab size");
-            result.push_str(&source[last_end..index]);
-            for _ in 0..tab_width {
-                result.push(' ');
-            }
-            last_end = index + 1;
-            update_range(index, tab_width);
-        } else if let Some(printable) = unprintable_replacement(c) {
+        if let Some(printable) = unprintable_replacement(c) {
             result.push_str(&source[last_end..index]);
             result.push(printable);
             last_end = index + 1;

crates/ruff_linter/src/rules/pycodestyle/snapshots/ruff_linter__rules__pycodestyle__tests__E101_E101.py.snap

@@ -15,8 +15,8 @@ E101.py:15:1: E101 Indentation contains mixed spaces and tabs
    |
 13 | def func_mixed_start_with_space():
 14 |     # E101
-15 |                print("mixed starts with space")
-   | ^^^^^^^^^^^^^^^ E101
+15 |                     print("mixed starts with space")
+   | ^^^^^^^^^^^^^^^^^^^^ E101
 16 |
 17 | def xyz():
    |
@@ -25,6 +25,6 @@ E101.py:19:1: E101 Indentation contains mixed spaces and tabs
    |
 17 | def xyz():
 18 |     # E101
-19 |     print("xyz");
-   | ^^^^ E101
+19 |        print("xyz");
+   | ^^^^^^^ E101
    |

crates/ruff_linter/src/rules/pycodestyle/snapshots/ruff_linter__rules__pycodestyle__tests__E201_E20.py.snap

@@ -47,7 +47,7 @@ E20.py:6:15: E201 [*] Whitespace after '{'
 6 | spam(ham[1], { eggs: 2})
   |               ^ E201
 7 | #: E201:1:6
-8 | spam(   ham[1], {eggs: 2})
+8 | spam(    ham[1], {eggs: 2})
   |
   = help: Remove whitespace before '{'
 
@@ -65,10 +65,10 @@ E20.py:8:6: E201 [*] Whitespace after '('
    |
  6 | spam(ham[1], { eggs: 2})
  7 | #: E201:1:6
- 8 | spam(   ham[1], {eggs: 2})
-   |      ^^^ E201
+ 8 | spam(    ham[1], {eggs: 2})
+   |      ^^^^ E201
  9 | #: E201:1:10
-10 | spam(ham[   1], {eggs: 2})
+10 | spam(ham[    1], {eggs: 2})
    |
    = help: Remove whitespace before '('
 
@@ -84,12 +84,12 @@ E20.py:8:6: E201 [*] Whitespace after '('
 
 E20.py:10:10: E201 [*] Whitespace after '['
    |
- 8 | spam(   ham[1], {eggs: 2})
+ 8 | spam(    ham[1], {eggs: 2})
  9 | #: E201:1:10
-10 | spam(ham[   1], {eggs: 2})
-   |          ^^^ E201
+10 | spam(ham[    1], {eggs: 2})
+   |          ^^^^ E201
 11 | #: E201:1:15
-12 | spam(ham[1], {  eggs: 2})
+12 | spam(ham[1], {    eggs: 2})
    |
    = help: Remove whitespace before '['
 
@@ -105,10 +105,10 @@ E20.py:10:10: E201 [*] Whitespace after '['
 
 E20.py:12:15: E201 [*] Whitespace after '{'
    |
-10 | spam(ham[   1], {eggs: 2})
+10 | spam(ham[    1], {eggs: 2})
 11 | #: E201:1:15
-12 | spam(ham[1], {  eggs: 2})
-   |               ^^ E201
+12 | spam(ham[1], {    eggs: 2})
+   |               ^^^^ E201
 13 | #: Okay
 14 | spam(ham[1], {eggs: 2})
    |

crates/ruff_linter/src/rules/pycodestyle/snapshots/ruff_linter__rules__pycodestyle__tests__E202_E20.py.snap

@@ -49,7 +49,7 @@ E20.py:23:11: E202 [*] Whitespace before ']'
 23 | spam(ham[1 ], {eggs: 2})
    |           ^ E202
 24 | #: E202:1:23
-25 | spam(ham[1], {eggs: 2}  )
+25 | spam(ham[1], {eggs: 2}    )
    |
    = help: Remove whitespace before ']'
 
@@ -67,10 +67,10 @@ E20.py:25:23: E202 [*] Whitespace before ')'
    |
 23 | spam(ham[1 ], {eggs: 2})
 24 | #: E202:1:23
-25 | spam(ham[1], {eggs: 2}  )
-   |                       ^^ E202
+25 | spam(ham[1], {eggs: 2}    )
+   |                       ^^^^ E202
 26 | #: E202:1:22
-27 | spam(ham[1], {eggs: 2   })
+27 | spam(ham[1], {eggs: 2    })
    |
    = help: Remove whitespace before ')'
 
@@ -86,12 +86,12 @@ E20.py:25:23: E202 [*] Whitespace before ')'
 
 E20.py:27:22: E202 [*] Whitespace before '}'
    |
-25 | spam(ham[1], {eggs: 2}  )
+25 | spam(ham[1], {eggs: 2}    )
 26 | #: E202:1:22
-27 | spam(ham[1], {eggs: 2   })
-   |                      ^^^ E202
+27 | spam(ham[1], {eggs: 2    })
+   |                      ^^^^ E202
 28 | #: E202:1:11
-29 | spam(ham[1  ], {eggs: 2})
+29 | spam(ham[1    ], {eggs: 2})
    |
    = help: Remove whitespace before '}'
 
@@ -107,10 +107,10 @@ E20.py:27:22: E202 [*] Whitespace before '}'
 
 E20.py:29:11: E202 [*] Whitespace before ']'
    |
-27 | spam(ham[1], {eggs: 2   })
+27 | spam(ham[1], {eggs: 2    })
 28 | #: E202:1:11
-29 | spam(ham[1  ], {eggs: 2})
-   |           ^^ E202
+29 | spam(ham[1    ], {eggs: 2})
+   |           ^^^^ E202
 30 | #: Okay
 31 | spam(ham[1], {eggs: 2})
    |

crates/ruff_linter/src/rules/pycodestyle/snapshots/ruff_linter__rules__pycodestyle__tests__E203_E20.py.snap

@@ -25,8 +25,8 @@ E20.py:55:10: E203 [*] Whitespace before ':'
    |
 53 |     x, y = y, x
 54 | #: E203:1:10
-55 | if x == 4   :
-   |          ^^^ E203
+55 | if x == 4    :
+   |          ^^^^ E203
 56 |     print(x, y)
 57 |     x, y = y, x
    |
@@ -67,8 +67,8 @@ E20.py:63:16: E203 [*] Whitespace before ';'
    |
 61 | #: E203:2:15 E702:2:16
 62 | if x == 4:
-63 |     print(x, y) ; x, y = y, x
-   |                ^ E203
+63 |     print(x, y)    ; x, y = y, x
+   |                ^^^^ E203
 64 | #: E203:3:13
 65 | if x == 4:
    |

crates/ruff_linter/src/rules/pycodestyle/snapshots/ruff_linter__rules__pycodestyle__tests__E223_E22.py.snap

@@ -1,13 +1,12 @@
 ---
 source: crates/ruff_linter/src/rules/pycodestyle/mod.rs
-snapshot_kind: text
 ---
 E22.py:43:2: E223 [*] Tab before operator
    |
 41 | #: E223
 42 | foobart = 4
-43 | a   = 3  # aligned with tab
-   |  ^^^ E223
+43 | a    = 3  # aligned with tab
+   |  ^^^^ E223
 44 | #:
    |
    = help: Replace with single space

crates/ruff_linter/src/rules/pycodestyle/snapshots/ruff_linter__rules__pycodestyle__tests__E242_E24.py.snap

@@ -1,13 +1,12 @@
 ---
 source: crates/ruff_linter/src/rules/pycodestyle/mod.rs
-snapshot_kind: text
 ---
 E24.py:6:8: E242 [*] Tab after comma
   |
 4 | b = (1, 20)
 5 | #: E242
-6 | a = (1, 2)  # tab before 2
-  |        ^ E242
+6 | a = (1,    2)  # tab before 2
+  |        ^^^^ E242
 7 | #: Okay
 8 | b = (1, 20)  # space before 20
   |

crates/ruff_linter/src/rules/pycodestyle/snapshots/ruff_linter__rules__pycodestyle__tests__E271_E27.py.snap

@@ -66,7 +66,7 @@ E27.py:8:3: E271 [*] Multiple spaces after keyword
 
 E27.py:15:6: E271 [*] Multiple spaces after keyword
    |
-13 | True        and False
+13 | True        and    False
 14 | #: E271
 15 | a and  b
    |      ^^ E271

crates/ruff_linter/src/rules/pycodestyle/snapshots/ruff_linter__rules__pycodestyle__tests__E272_E27.py.snap

@@ -1,6 +1,5 @@
 ---
 source: crates/ruff_linter/src/rules/pycodestyle/mod.rs
-snapshot_kind: text
 ---
 E27.py:21:2: E272 [*] Multiple spaces before keyword
    |
@@ -51,7 +50,7 @@ E27.py:25:5: E272 [*] Multiple spaces before keyword
 25 | this  and False
    |     ^^ E272
 26 | #: E273
-27 | a and   b
+27 | a and    b
    |
    = help: Replace with single space