Render tabs as 4 spaces in diagnostics (#4132)

crates/ruff/src/message/text.rs

@@ -8,7 +8,8 @@ use bitflags::bitflags;
 use colored::Colorize;
 use ruff_diagnostics::DiagnosticKind;
 use ruff_python_ast::source_code::{OneIndexed, SourceLocation};
-use ruff_text_size::TextRange;
+use ruff_text_size::{TextRange, TextSize};
+use std::borrow::Cow;
 use std::fmt::{Display, Formatter};
 use std::io::Write;
 
@@ -172,6 +173,7 @@ impl Display for MessageCodeFrame<'_> {
         };
 
         let source_code = file.to_source_code();
+
         let content_start_index = source_code.line_index(range.start());
         let mut start_index = content_start_index.saturating_sub(2);
 
@@ -200,26 +202,23 @@ impl Display for MessageCodeFrame<'_> {
         let start_offset = source_code.line_start(start_index);
         let end_offset = source_code.line_end(end_index);
 
-        let source_text = source_code.slice(TextRange::new(start_offset, end_offset));
+        let source = replace_whitespace(
+            source_code.slice(TextRange::new(start_offset, end_offset)),
+            range - start_offset,
+        );
 
-        let annotation_start_offset = range.start() - start_offset;
+        let start_char = source.text[TextRange::up_to(source.annotation_range.start())]
-        let annotation_end_offset = range.end() - start_offset;
-
-        let start_char = source_text[TextRange::up_to(annotation_start_offset)]
             .chars()
             .count();
 
-        let char_length = source_text
+        let char_length = source.text[source.annotation_range].chars().count();
-            [TextRange::new(annotation_start_offset, annotation_end_offset)]
-        .chars()
-        .count();
 
         let label = kind.rule().noqa_code().to_string();
 
         let snippet = Snippet {
             title: None,
             slices: vec![Slice {
-                source: source_text,
+                source: &source.text,
                 line_start: content_start_index.get(),
                 annotations: vec![SourceAnnotation {
                     label: &label,
@@ -245,6 +244,60 @@ impl Display for MessageCodeFrame<'_> {
     }
 }
 
+fn replace_whitespace(source: &str, annotation_range: TextRange) -> SourceCode {
+    static TAB_SIZE: TextSize = TextSize::new(4);
+
+    let mut result = String::new();
+    let mut last_end = 0;
+    let mut range = annotation_range;
+    let mut column = 0;
+
+    for (index, m) in source.match_indices(['\t', '\n', '\r']) {
+        match m {
+            "\t" => {
+                let tab_width = TAB_SIZE - TextSize::new(column % 4);
+
+                if index < usize::from(annotation_range.start()) {
+                    range += tab_width - TextSize::new(1);
+                } else if index < usize::from(annotation_range.end()) {
+                    range = range.add_end(tab_width - TextSize::new(1));
+                }
+
+                result.push_str(&source[last_end..index]);
+
+                for _ in 0..u32::from(tab_width) {
+                    result.push(' ');
+                }
+
+                last_end = index + 1;
+            }
+            "\n" | "\r" => {
+                column = 0;
+            }
+            _ => unreachable!(),
+        }
+    }
+
+    // No tabs
+    if result.is_empty() {
+        SourceCode {
+            annotation_range,
+            text: Cow::Borrowed(source),
+        }
+    } else {
+        result.push_str(&source[last_end..]);
+        SourceCode {
+            annotation_range: range,
+            text: Cow::Owned(result),
+        }
+    }
+}
+
+struct SourceCode<'a> {
+    text: Cow<'a, str>,
+    annotation_range: TextRange,
+}
+
 #[cfg(test)]
 mod tests {
     use crate::message::tests::{capture_emitter_output, create_messages};

crates/ruff/src/rules/pycodestyle/snapshots/ruff__rules__pycodestyle__tests__E101_E101.py.snap

@@ -6,7 +6,7 @@ E101.py:11:1: E101 Indentation contains mixed spaces and tabs
 11 | def func_mixed_start_with_tab():
 12 |     # E101
 13 |       print("mixed starts with tab")
-   | ^^ E101
+   | ^^^^^^ E101
 14 | 
 15 | def func_mixed_start_with_space():
    |
@@ -16,7 +16,7 @@ E101.py:15:1: E101 Indentation contains mixed spaces and tabs
 15 | def func_mixed_start_with_space():
 16 |     # E101
 17 |                     print("mixed starts with space")
-   | ^^^^^^^^ E101
+   | ^^^^^^^^^^^^^^^^^^^^ E101
 18 | 
 19 | def xyz():
    |
@@ -26,7 +26,7 @@ E101.py:19:1: E101 Indentation contains mixed spaces and tabs
 19 | def xyz():
 20 |     # E101
 21 |        print("xyz");
-   | ^^^ E101
+   | ^^^^^^^ E101
    |
 
 

crates/ruff/src/rules/pycodestyle/snapshots/ruff__rules__pycodestyle__tests__E117_E11.py.snap

@@ -26,7 +26,7 @@ E11.py:42:1: E117 Over-indented
 42 | #: E117 W191
 43 | def start():
 44 |         print()
-   |  E117
+   | ^^^^^^^^ E117
    |
 
 

crates/ruff/src/rules/pycodestyle/snapshots/ruff__rules__pycodestyle__tests__W191_W19.py.snap

@@ -6,7 +6,7 @@ W19.py:3:1: W191 Indentation contains tabs
 3 | #: W191
 4 | if False:
 5 |     print  # indented with 1 tab
-  |  W191
+  | ^^^^ W191
 6 | #:
   |
 
@@ -15,7 +15,7 @@ W19.py:9:1: W191 Indentation contains tabs
  9 | #: W191
 10 | y = x == 2 \
 11 |     or x == 3
-   |  W191
+   | ^^^^ W191
 12 | #: E101 W191 W504
 13 | if (
    |
@@ -25,7 +25,7 @@ W19.py:16:1: W191 Indentation contains tabs
 16 |         ) or
 17 |         y == 4):
 18 |     pass
-   |  W191
+   | ^^^^ W191
 19 | #: E101 W191
 20 | if x == 2 \
    |
@@ -35,7 +35,7 @@ W19.py:21:1: W191 Indentation contains tabs
 21 |     or y > 1 \
 22 |         or x == 3:
 23 |     pass
-   |  W191
+   | ^^^^ W191
 24 | #: E101 W191
 25 | if x == 2 \
    |
@@ -45,7 +45,7 @@ W19.py:26:1: W191 Indentation contains tabs
 26 |         or y > 1 \
 27 |         or x == 3:
 28 |     pass
-   |  W191
+   | ^^^^ W191
 29 | #:
    |
 
@@ -54,7 +54,7 @@ W19.py:32:1: W191 Indentation contains tabs
 32 | if (foo == bar and
 33 |         baz == bop):
 34 |     pass
-   |  W191
+   | ^^^^ W191
 35 | #: E101 W191 W504
 36 | if (
    |
@@ -64,7 +64,7 @@ W19.py:38:1: W191 Indentation contains tabs
 38 |     baz == bop
 39 | ):
 40 |     pass
-   |  W191
+   | ^^^^ W191
 41 | #:
    |
 
@@ -73,7 +73,7 @@ W19.py:44:1: W191 Indentation contains tabs
 44 | if start[1] > end_col and not (
 45 |         over_indent == 4 and indent_next):
 46 |     return (0, "E121 continuation line over-"
-   |  W191
+   | ^^^^ W191
 47 |             "indented for visual indent")
 48 | #:
    |
@@ -83,7 +83,7 @@ W19.py:45:1: W191 Indentation contains tabs
 45 |         over_indent == 4 and indent_next):
 46 |     return (0, "E121 continuation line over-"
 47 |             "indented for visual indent")
-   | ^^^^^^^^ W191
+   | ^^^^^^^^^^^^ W191
 48 | #:
    |
 
@@ -92,7 +92,7 @@ W19.py:54:1: W191 Indentation contains tabs
 54 |         var_one, var_two, var_three,
 55 |         var_four):
 56 |     print(var_one)
-   |  W191
+   | ^^^^ W191
 57 | #: E101 W191 W504
 58 | if ((row < 0 or self.moduleCount <= row or
    |
@@ -102,7 +102,7 @@ W19.py:58:1: W191 Indentation contains tabs
 58 | if ((row < 0 or self.moduleCount <= row or
 59 |      col < 0 or self.moduleCount <= col)):
 60 |     raise Exception("%s,%s - %s" % (row, col, self.moduleCount))
-   |  W191
+   | ^^^^ W191
 61 | #: E101 E101 E101 E101 W191 W191 W191 W191 W191 W191
 62 | if bar:
    |
@@ -112,7 +112,7 @@ W19.py:61:1: W191 Indentation contains tabs
 61 | #: E101 E101 E101 E101 W191 W191 W191 W191 W191 W191
 62 | if bar:
 63 |     return (
-   |  W191
+   | ^^^^ W191
 64 |         start, 'E121 lines starting with a '
 65 |         'closing bracket should be indented '
    |
@@ -122,7 +122,7 @@ W19.py:62:1: W191 Indentation contains tabs
 62 | if bar:
 63 |     return (
 64 |         start, 'E121 lines starting with a '
-   | ^^^^ W191
+   | ^^^^^^^^ W191
 65 |         'closing bracket should be indented '
 66 |         "to match that of the opening "
    |
@@ -132,7 +132,7 @@ W19.py:63:1: W191 Indentation contains tabs
 63 |     return (
 64 |         start, 'E121 lines starting with a '
 65 |         'closing bracket should be indented '
-   | ^^^^ W191
+   | ^^^^^^^^ W191
 66 |         "to match that of the opening "
 67 |         "bracket's line"
    |
@@ -142,7 +142,7 @@ W19.py:64:1: W191 Indentation contains tabs
 64 |         start, 'E121 lines starting with a '
 65 |         'closing bracket should be indented '
 66 |         "to match that of the opening "
-   | ^^^^ W191
+   | ^^^^^^^^ W191
 67 |         "bracket's line"
 68 |     )
    |
@@ -152,7 +152,7 @@ W19.py:65:1: W191 Indentation contains tabs
 65 |         'closing bracket should be indented '
 66 |         "to match that of the opening "
 67 |         "bracket's line"
-   | ^^^^ W191
+   | ^^^^^^^^ W191
 68 |     )
 69 | #
    |
@@ -162,7 +162,7 @@ W19.py:66:1: W191 Indentation contains tabs
 66 |         "to match that of the opening "
 67 |         "bracket's line"
 68 |     )
-   |  W191
+   | ^^^^ W191
 69 | #
 70 | #: E101 W191 W504
    |
@@ -172,7 +172,7 @@ W19.py:73:1: W191 Indentation contains tabs
 73 |      foo.bar("bop")
 74 |      )):
 75 |     print "yes"
-   |  W191
+   | ^^^^ W191
 76 | #: E101 W191 W504
 77 | # also ok, but starting to look like LISP
    |
@@ -182,7 +182,7 @@ W19.py:78:1: W191 Indentation contains tabs
 78 | if ((foo.bar("baz") and
 79 |      foo.bar("bop"))):
 80 |     print "yes"
-   |  W191
+   | ^^^^ W191
 81 | #: E101 W191 W504
 82 | if (a == 2 or
    |
@@ -192,7 +192,7 @@ W19.py:83:1: W191 Indentation contains tabs
 83 |     b == "abc def ghi"
 84 |          "jkl mno"):
 85 |     return True
-   |  W191
+   | ^^^^ W191
 86 | #: E101 W191 W504
 87 | if (a == 2 or
    |
@@ -202,7 +202,7 @@ W19.py:88:1: W191 Indentation contains tabs
 88 |     b == """abc def ghi
 89 | jkl mno"""):
 90 |     return True
-   |  W191
+   | ^^^^ W191
 91 | #: W191:2:1 W191:3:1 E101:3:2
 92 | if length > options.max_line_length:
    |
@@ -212,7 +212,7 @@ W19.py:91:1: W191 Indentation contains tabs
 91 | #: W191:2:1 W191:3:1 E101:3:2
 92 | if length > options.max_line_length:
 93 |     return options.max_line_length, \
-   |  W191
+   | ^^^^ W191
 94 |         "E501 line too long (%d characters)" % length
    |
 
@@ -221,7 +221,7 @@ W19.py:92:1: W191 Indentation contains tabs
 92 | if length > options.max_line_length:
 93 |     return options.max_line_length, \
 94 |         "E501 line too long (%d characters)" % length
-   | ^^^^ W191
+   | ^^^^^^^^ W191
    |
 
 W19.py:98:1: W191 Indentation contains tabs
@@ -229,7 +229,7 @@ W19.py:98:1: W191 Indentation contains tabs
  98 | #: E101 W191 W191 W504
  99 | if os.path.exists(os.path.join(path, PEP8_BIN)):
 100 |     cmd = ([os.path.join(path, PEP8_BIN)] +
-    |  W191
+    | ^^^^ W191
 101 |            self._pep8_options(targetfile))
 102 | #: W191 - okay
     |
@@ -239,7 +239,7 @@ W19.py:99:1: W191 Indentation contains tabs
  99 | if os.path.exists(os.path.join(path, PEP8_BIN)):
 100 |     cmd = ([os.path.join(path, PEP8_BIN)] +
 101 |            self._pep8_options(targetfile))
-    | ^^^^^^^ W191
+    | ^^^^^^^^^^^ W191
 102 | #: W191 - okay
 103 | '''
     |
@@ -249,7 +249,7 @@ W19.py:125:1: W191 Indentation contains tabs
 125 | if foo is None and bar is "bop" and \
 126 |         blah == 'yeah':
 127 |     blah = 'yeahnah'
-    |  W191
+    | ^^^^ W191
     |
 
 W19.py:131:1: W191 Indentation contains tabs
@@ -257,7 +257,7 @@ W19.py:131:1: W191 Indentation contains tabs
 131 | #: W191 W191 W191
 132 | if True:
 133 |     foo(
-    |  W191
+    | ^^^^ W191
 134 |         1,
 135 |         2)
     |
@@ -267,7 +267,7 @@ W19.py:132:1: W191 Indentation contains tabs
 132 | if True:
 133 |     foo(
 134 |         1,
-    |  W191
+    | ^^^^^^^^ W191
 135 |         2)
 136 | #: W191 W191 W191 W191 W191
     |
@@ -277,7 +277,7 @@ W19.py:133:1: W191 Indentation contains tabs
 133 |     foo(
 134 |         1,
 135 |         2)
-    |  W191
+    | ^^^^^^^^ W191
 136 | #: W191 W191 W191 W191 W191
 137 | def test_keys(self):
     |
@@ -287,7 +287,7 @@ W19.py:136:1: W191 Indentation contains tabs
 136 | #: W191 W191 W191 W191 W191
 137 | def test_keys(self):
 138 |     """areas.json - All regions are accounted for."""
-    |  W191
+    | ^^^^ W191
 139 |     expected = set([
 140 |         u'Norrbotten',
     |
@@ -297,7 +297,7 @@ W19.py:137:1: W191 Indentation contains tabs
 137 | def test_keys(self):
 138 |     """areas.json - All regions are accounted for."""
 139 |     expected = set([
-    |  W191
+    | ^^^^ W191
 140 |         u'Norrbotten',
 141 |         u'V\xe4sterbotten',
     |
@@ -307,7 +307,7 @@ W19.py:138:1: W191 Indentation contains tabs
 138 |     """areas.json - All regions are accounted for."""
 139 |     expected = set([
 140 |         u'Norrbotten',
-    |  W191
+    | ^^^^^^^^ W191
 141 |         u'V\xe4sterbotten',
 142 |     ])
     |
@@ -317,7 +317,7 @@ W19.py:139:1: W191 Indentation contains tabs
 139 |     expected = set([
 140 |         u'Norrbotten',
 141 |         u'V\xe4sterbotten',
-    |  W191
+    | ^^^^^^^^ W191
 142 |     ])
 143 | #: W191
     |
@@ -327,7 +327,7 @@ W19.py:140:1: W191 Indentation contains tabs
 140 |         u'Norrbotten',
 141 |         u'V\xe4sterbotten',
 142 |     ])
-    |  W191
+    | ^^^^ W191
 143 | #: W191
 144 | x = [
     |
@@ -337,7 +337,7 @@ W19.py:143:1: W191 Indentation contains tabs
 143 | #: W191
 144 | x = [
 145 |     'abc'
-    |  W191
+    | ^^^^ W191
 146 | ]
 147 | #: W191 - okay
     |