Autoformat unit characters

astral-sh · Jun 3, 2023 · 5a9f4aa · 5a9f4aa
1 parent e82160a
commit 5a9f4aa
Show file tree

Hide file tree

Showing 5 changed files with 43 additions and 5 deletions.
diff --git a/crates/ruff/resources/test/fixtures/ruff/confusables.py b/crates/ruff/resources/test/fixtures/ruff/confusables.py
@@ -29,3 +29,12 @@ def f():
 # consisting of a single ambiguous character, while the second character is a "word
 # boundary" (whitespace) that it itself ambiguous.
 x = "Р усский"
+
+
+# At runtime the attribute will be stored as Greek small letter mu instead of
+# micro sign because of PEP 3131's NFKC normalization
+class Labware:
+    µL = 1.5
+
+
+assert getattr(Labware(), "µL") == 1.5
diff --git a/crates/ruff/src/rules/ruff/rules/ambiguous_unicode_character.rs b/crates/ruff/src/rules/ruff/rules/ambiguous_unicode_character.rs
@@ -150,7 +150,7 @@ pub(crate) fn ambiguous_unicode_character(
                     let candidate = Candidate::new(
                         TextSize::try_from(relative_offset).unwrap() + range.start(),
                         current_char,
-                        representant as char,
+                        char::from_u32(representant).unwrap(),
                     );
                     if let Some(diagnostic) = candidate.into_diagnostic(context, settings) {
                         diagnostics.push(diagnostic);
@@ -165,7 +165,7 @@ pub(crate) fn ambiguous_unicode_character(
             word_candidates.push(Candidate::new(
                 TextSize::try_from(relative_offset).unwrap() + range.start(),
                 current_char,
-                representant as char,
+                char::from_u32(representant).unwrap(),
             ));
         } else {
             // The current word contains at least one unambiguous unicode character.

diff --git a/crates/ruff/src/rules/ruff/rules/confusables.rs b/crates/ruff/src/rules/ruff/rules/confusables.rs
@@ -4,7 +4,7 @@ use rustc_hash::FxHashMap;
 
 /// Via: <https://github.com/hediet/vscode-unicode-data/blob/main/out/ambiguous.json>
 /// See: <https://github.com/microsoft/vscode/blob/095ddabc52b82498ee7f718a34f9dd11d59099a8/src/vs/base/common/strings.ts#L1094>
-pub(crate) static CONFUSABLES: Lazy<FxHashMap<u32, u8>> = Lazy::new(|| {
+pub(crate) static CONFUSABLES: Lazy<FxHashMap<u32, u32>> = Lazy::new(|| {
     #[allow(clippy::unreadable_literal)]
     FxHashMap::from_iter([
         (8232, 32),
@@ -2116,5 +2116,8 @@ pub(crate) static CONFUSABLES: Lazy<FxHashMap<u32, u8>> = Lazy::new(|| {
         (1059, 89),
         (65283, 35),
         (65307, 59),
+        (0x212B, 0x00C5), // ANGSTROM SIGN → LATIN CAPITAL LETTER A WITH RING ABOVE
+        (0x2126, 0x03A9), // OHM SIGN → GREEK CAPITAL LETTER OMEGA
+        (0x00B5, 0x03BC), // MICRO SIGN → GREEK SMALL LETTER MU
     ])
 });
diff --git a/crates/ruff/src/rules/ruff/snapshots/ruff__rules__ruff__tests__confusables.snap b/crates/ruff/src/rules/ruff/snapshots/ruff__rules__ruff__tests__confusables.snap
@@ -110,6 +110,9 @@ confusables.py:31:6: RUF001 [*] String contains ambiguous `Р` (CYRILLIC CAPITAL
 30 30 | # boundary" (whitespace) that it itself ambiguous.
 31    |-x = "Р усский"
    31 |+x = "P усский"
+32 32 | 
+33 33 | 
+34 34 | # At runtime the attribute will be stored as Greek small letter mu instead of
 
 confusables.py:31:7: RUF001 [*] String contains ambiguous ` ` (EN QUAD). Did you mean ` ` (SPACE)?
    |
@@ -126,5 +129,22 @@ confusables.py:31:7: RUF001 [*] String contains ambiguous ` ` (EN QUAD). Did y
 30 30 | # boundary" (whitespace) that it itself ambiguous.
 31    |-x = "Р усский"
    31 |+x = "Р усский"
+32 32 | 
+33 33 | 
+34 34 | # At runtime the attribute will be stored as Greek small letter mu instead of
+
+confusables.py:40:28: RUF001 [*] String contains ambiguous `µ` (MICRO SIGN). Did you mean `μ` (GREEK SMALL LETTER MU)?
+   |
+40 | assert getattr(Labware(), "µL") == 1.5
+   |                            ^ RUF001
+   |
+   = help: Replace `µ` (MICRO SIGN) with `μ` (GREEK SMALL LETTER MU)
+
+ℹ Suggested fix
+37 37 |     µL = 1.5
+38 38 | 
+39 39 | 
+40    |-assert getattr(Labware(), "µL") == 1.5
+   40 |+assert getattr(Labware(), "μL") == 1.5
 
 
diff --git a/scripts/update_ambiguous_characters.py b/scripts/update_ambiguous_characters.py
@@ -32,15 +32,21 @@ def format_confusables_rs(raw_data: dict) -> str:
 
 /// Via: <https://github.com/hediet/vscode-unicode-data/blob/main/out/ambiguous.json>
 /// See: <https://github.com/microsoft/vscode/blob/095ddabc52b82498ee7f718a34f9dd11d59099a8/src/vs/base/common/strings.ts#L1094>
-pub(crate) static CONFUSABLES: Lazy<FxHashMap<u32, u8>> = Lazy::new(|| {
+pub(crate) static CONFUSABLES: Lazy<FxHashMap<u32, u32>> = Lazy::new(|| {
     #[allow(clippy::unreadable_literal)]
     FxHashMap::from_iter([
 """.lstrip()
     tuples = []
     for _category, items in raw_data.items():
         for i in range(0, len(items), 2):
             tuples.append(f"({items[i]}, {items[i + 1]}),")
-    postlude = """])});"""
+    postlude = """
+        (0x212B, 0x00C5), // ANGSTROM SIGN → LATIN CAPITAL LETTER A WITH RING ABOVE
+        (0x2126, 0x03A9), // OHM SIGN → GREEK CAPITAL LETTER OMEGA
+        (0x00B5, 0x03BC), // MICRO SIGN → GREEK SMALL LETTER MU
+    ])
+});
+"""
 
     print(f"{len(tuples)} confusable tuples.")