fix: don't ICE on LineOverflow when the line contains multibyte characters

Author: shulaoda

src/format_report_formatter.rs

@@ -106,7 +106,7 @@ fn annotation(error: &FormattingError) -> Option<Annotation<'_>> {
 
 fn error_kind_to_snippet_annotation_level(error_kind: &ErrorKind) -> Level {
     match error_kind {
-        ErrorKind::LineOverflow(..)
+        ErrorKind::LineOverflow { .. }
         | ErrorKind::TrailingWhitespace
         | ErrorKind::IoError(_)
         | ErrorKind::ModuleResolutionError(_)

src/formatting.rs

@@ -333,7 +333,7 @@ impl FormattingError {
 
     pub(crate) fn is_internal(&self) -> bool {
         match self.kind {
-            ErrorKind::LineOverflow(..)
+            ErrorKind::LineOverflow { .. }
             | ErrorKind::TrailingWhitespace
             | ErrorKind::IoError(_)
             | ErrorKind::ParseError
@@ -354,7 +354,13 @@ impl FormattingError {
     // (space, target)
     pub(crate) fn format_len(&self) -> (usize, usize) {
         match self.kind {
-            ErrorKind::LineOverflow(found, max) => (max, found - max),
+            ErrorKind::LineOverflow {
+                overflow_start_byte,
+                ..
+            } => (
+                overflow_start_byte,
+                self.line_buffer.len() - overflow_start_byte,
+            ),
             ErrorKind::TrailingWhitespace
             | ErrorKind::DeprecatedAttr
             | ErrorKind::BadAttr
@@ -564,8 +570,13 @@ impl<'a> FormatLines<'a> {
             }
 
             // Check for any line width errors we couldn't correct.
-            let error_kind = ErrorKind::LineOverflow(self.line_len, self.config.max_width());
-            if self.line_len > self.config.max_width()
+            let max_width = self.config.max_width();
+            let error_kind = ErrorKind::LineOverflow {
+                total_line_width: self.line_len,
+                max_width,
+                overflow_start_byte: self.byte_offset_at_col(max_width),
+            };
+            if self.line_len > max_width
                 && !self.is_skipped_line()
                 && self.should_report_error(kind, &error_kind)
             {
@@ -600,6 +611,22 @@ impl<'a> FormatLines<'a> {
         }
     }
 
+    /// Inverse of `Self::char`'s column accounting: walk `line_buffer` with
+    /// the same rule (tab = `tab_spaces` cols, every other char = 1 col) and
+    /// return the byte offset where the accumulated column count first reaches
+    /// `target_col`. Returns `line_buffer.len()` if the line is shorter.
+    fn byte_offset_at_col(&self, target_col: usize) -> usize {
+        let tab_spaces = self.config.tab_spaces();
+        let mut col = 0;
+        for (idx, ch) in self.line_buffer.char_indices() {
+            if col >= target_col {
+                return idx;
+            }
+            col += if ch == '\t' { tab_spaces } else { 1 };
+        }
+        self.line_buffer.len()
+    }
+
     fn push_err(&mut self, kind: ErrorKind, is_comment: bool, is_string: bool) {
         self.errors.push(FormattingError {
             line: self.cur_line,
@@ -621,7 +648,7 @@ impl<'a> FormatLines<'a> {
         };
 
         match error_kind {
-            ErrorKind::LineOverflow(..) => {
+            ErrorKind::LineOverflow { .. } => {
                 self.config.error_on_line_overflow() && allow_error_report
             }
             ErrorKind::TrailingWhitespace | ErrorKind::LostComment => allow_error_report,

src/lib.rs

@@ -106,12 +106,16 @@ pub(crate) mod visitor;
 /// these can currently be propagated to clients.
 #[derive(Error, Debug)]
 pub enum ErrorKind {
-    /// Line has exceeded character limit (found, maximum).
+    /// A line exceeded the configured maximum width.
     #[error(
         "line formatted, but exceeded maximum width \
-         (maximum: {1} (see `max_width` option), found: {0})"
+         (maximum: {max_width} (see `max_width` option), found: {total_line_width})"
     )]
-    LineOverflow(usize, usize),
+    LineOverflow {
+        total_line_width: usize,
+        max_width: usize,
+        overflow_start_byte: usize,
+    },
     /// Line ends in whitespace.
     #[error("left behind trailing whitespace")]
     TrailingWhitespace,
@@ -225,7 +229,7 @@ impl FormatReport {
         }
         for err in new_errors {
             match err.kind {
-                ErrorKind::LineOverflow(..) => {
+                ErrorKind::LineOverflow { .. } => {
                     errs.has_operational_errors = true;
                 }
                 ErrorKind::TrailingWhitespace => {

tests/cargo-fmt/source/issue_6850/Cargo.toml

@@ -0,0 +1,6 @@
+[package]
+name = "issue_6850"
+version = "0.1.0"
+edition = "2021"
+
+[dependencies]

tests/cargo-fmt/source/issue_6850/src/main.rs

@@ -0,0 +1,3 @@
+fn main() {
+    "☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃☃"
+}

tests/rustfmt/main.rs

@@ -203,6 +203,27 @@ fn dont_emit_ICE() {
     }
 }
 
+#[test]
+fn dont_panic_on_line_overflow_with_multibyte_chars() {
+    // See also https://github.com/rust-lang/rustfmt/issues/6850
+    let args = [
+        "--config",
+        "error_on_line_overflow=true,error_on_unformatted=true",
+        "tests/cargo-fmt/source/issue_6850/src/main.rs",
+    ];
+
+    let (_stdout, stderr) = rustfmt(&args);
+    assert!(stderr.contains(
+        "line formatted, but exceeded maximum width (maximum: 100 (see `max_width` option), found: 126)"
+    ));
+
+    let panic_re = regex::Regex::new("thread.*panicked").unwrap();
+    assert!(
+        !panic_re.is_match(&stderr),
+        "rustfmt panicked instead of reporting line overflow:\n{stderr}"
+    );
+}
+
 #[test]
 fn rustfmt_emits_error_when_control_brace_style_is_always_next_line() {
     // See also https://github.com/rust-lang/rustfmt/issues/5912