From de7e379cae0cf4910bd771fd495092472c6c5740 Mon Sep 17 00:00:00 2001
From: konstin <konstin@mailbox.org>
Date: Thu, 6 Apr 2023 12:58:36 +0200
Subject: [PATCH] Fix unicode handling in PLE2515

Previously, we used byte indices when we should have used char indices, causing crashes when the were non-ascii characters before our replaces

Fixes #3716
---
 .../fixtures/pylint/invalid_characters.py     | Bin 972 -> 1106 bytes
 .../pylint/rules/invalid_string_characters.rs |   7 +-
 ..._tests__PLE2515_invalid_characters.py.snap |  63 ++++++++++++++++++
 3 files changed, 68 insertions(+), 2 deletions(-)

diff --git a/crates/ruff/resources/test/fixtures/pylint/invalid_characters.py b/crates/ruff/resources/test/fixtures/pylint/invalid_characters.py
index ecc681b112288e93baa23e2f27c688d3d7fb907d..6233ed9353308d93ea948e16755fa23d395c9552 100644
GIT binary patch
delta 143
zcmX@Zeu-nl8RmMfs`BE3_{6l5)S~#@(wvgaq{@=i_~eYlqQqpNn1Zc>(t}N_A2oC<
zVW@;Ficc>}EXYXB1?tNwEk@{iu(|iarVS4^?R&6k1(?0#!KVEWHthhjr#{%+rSOQo
Ip&M*A0OWf{9RL6T

delta 7
Ocmcb_afW@v8D;<txC1Q!

diff --git a/crates/ruff/src/rules/pylint/rules/invalid_string_characters.rs b/crates/ruff/src/rules/pylint/rules/invalid_string_characters.rs
index 8eac9f2621631..17454d599421d 100644
--- a/crates/ruff/src/rules/pylint/rules/invalid_string_characters.rs
+++ b/crates/ruff/src/rules/pylint/rules/invalid_string_characters.rs
@@ -184,7 +184,9 @@ pub fn invalid_string_characters(
     let text = locator.slice(Range::new(start, end));
 
     for (row, line) in UniversalNewlineIterator::from(text).enumerate() {
-        for (column, match_) in line.match_indices(&['\x08', '\x1A', '\x1B', '\0', '\u{200b}']) {
+        for (column_bytes, match_) in
+            line.match_indices(&['\x08', '\x1A', '\x1B', '\0', '\u{200b}'])
+        {
             let (replacement, rule): (&str, DiagnosticKind) = match match_.chars().next().unwrap() {
                 '\x08' => ("\\b", InvalidCharacterBackspace.into()),
                 '\x1A' => ("\\x1A", InvalidCharacterSub.into()),
@@ -193,7 +195,8 @@ pub fn invalid_string_characters(
                 '\u{200b}' => ("\\u200b", InvalidCharacterZeroWidthSpace.into()),
                 _ => unreachable!(),
             };
-            let location = helpers::to_absolute(Location::new(row + 1, column), start);
+            let column_chars = line[..column_bytes].chars().count();
+            let location = helpers::to_absolute(Location::new(row + 1, column_chars), start);
             let end_location = Location::new(location.row(), location.column() + 1);
             let mut diagnostic = Diagnostic::new(rule, Range::new(location, end_location));
             if autofix {
diff --git a/crates/ruff/src/rules/pylint/snapshots/ruff__rules__pylint__tests__PLE2515_invalid_characters.py.snap b/crates/ruff/src/rules/pylint/snapshots/ruff__rules__pylint__tests__PLE2515_invalid_characters.py.snap
index 4bc299394025f..272ae53441add 100644
--- a/crates/ruff/src/rules/pylint/snapshots/ruff__rules__pylint__tests__PLE2515_invalid_characters.py.snap
+++ b/crates/ruff/src/rules/pylint/snapshots/ruff__rules__pylint__tests__PLE2515_invalid_characters.py.snap
@@ -23,4 +23,67 @@ expression: diagnostics
           row: 34
           column: 13
   parent: ~
+- kind:
+    name: InvalidCharacterZeroWidthSpace
+    body: "Invalid unescaped character zero-width-space, use \"\\u200B\" instead"
+    suggestion: Replace with escape sequence
+    fixable: true
+  location:
+    row: 38
+    column: 35
+  end_location:
+    row: 38
+    column: 36
+  fix:
+    edits:
+      - content: "\\u200b"
+        location:
+          row: 38
+          column: 35
+        end_location:
+          row: 38
+          column: 36
+  parent: ~
+- kind:
+    name: InvalidCharacterZeroWidthSpace
+    body: "Invalid unescaped character zero-width-space, use \"\\u200B\" instead"
+    suggestion: Replace with escape sequence
+    fixable: true
+  location:
+    row: 39
+    column: 59
+  end_location:
+    row: 39
+    column: 60
+  fix:
+    edits:
+      - content: "\\u200b"
+        location:
+          row: 39
+          column: 59
+        end_location:
+          row: 39
+          column: 60
+  parent: ~
+- kind:
+    name: InvalidCharacterZeroWidthSpace
+    body: "Invalid unescaped character zero-width-space, use \"\\u200B\" instead"
+    suggestion: Replace with escape sequence
+    fixable: true
+  location:
+    row: 39
+    column: 60
+  end_location:
+    row: 39
+    column: 61
+  fix:
+    edits:
+      - content: "\\u200b"
+        location:
+          row: 39
+          column: 60
+        end_location:
+          row: 39
+          column: 61
+  parent: ~