2025-08-02 09:35:02 +02:00
3 changed files with 44 additions and 5 deletions
--- a/src/ast.rs
+++ b/src/ast.rs
@ -3,7 +3,7 @@ use std::iter::zip;
 use std::{
    borrow::Cow,
    cell::UnsafeCell,
-    cmp::{max, min},
+    cmp::max,
    fmt::Display,
    hash::{Hash, Hasher},
    ops::Range,
@ -21,6 +21,7 @@ use tree_sitter::{
 use typed_arena::Arena;

 use crate::{
+    StrExt,
    lang_profile::{CommutativeParent, LangProfile, ParentType},
    signature::{Signature, SignatureDefinition},
 };
@ -311,13 +312,19 @@ impl<'a> AstNode<'a> {
        let local_source = &global_source[range.start..range.end];
        if node.is_error() {
            let full_range = node.range();
+
+            // it can be that byte 32 doesn't lie on char boundary,
+            // so increase the index until it does
+            #[expect(unstable_name_collisions)]
+            let idx = local_source.ceil_char_boundary(32);
+
            return Err(format!(
                "parse error at {}:{}..{}:{}, starting with: {}",
                full_range.start_point.row,
                full_range.start_point.column,
                full_range.end_point.row,
                full_range.end_point.column,
-                &local_source[..min(32, local_source.len())]
+                &local_source[..idx]
            ));
        }

@ -1153,12 +1160,25 @@ mod tests {
        let ctx = ctx();
        let lang_profile = LangProfile::detect_from_filename("test.json")
            .expect("could not load language profile");
+
        let parse = AstNode::parse("[\n {,\n]", lang_profile, &ctx.arena, &ctx.ref_arena);

        assert_eq!(
            parse,
            Err("parse error at 1:1..1:3, starting with: {,".to_string())
        );
+
+        let parse = AstNode::parse(
+            "属于个人的非赢利性开源项目",
+            lang_profile,
+            &ctx.arena,
+            &ctx.ref_arena,
+        );
+
+        assert_eq!(
+            parse,
+            Err("parse error at 0:0..0:39, starting with: 属于个人的非赢利性开源".to_string())
+        );
    }

    #[test]
--- a/src/lib.rs
+++ b/src/lib.rs
@ -32,11 +32,11 @@ pub(crate) mod merged_tree;
 pub(crate) mod multimap;
 pub mod newline;
 pub(crate) mod parsed_merge;
-mod path_buf_ext;
 pub(crate) mod pcs;
 pub mod settings;
 pub(crate) mod signature;
 mod solve;
+mod std_ext;
 pub(crate) mod structured;
 pub mod supported_langs;
 #[cfg(test)]
@ -59,7 +59,7 @@ use settings::DisplaySettings;
 use structured::structured_merge;
 use supported_langs::SUPPORTED_LANGUAGES;

-pub use path_buf_ext::PathBufExt;
+pub use std_ext::{PathBufExt, StrExt};

 /// Current way to disable Mergiraf
 /// ## Usage
--- a/src/path_buf_ext.rs
+++ b/src/path_buf_ext.rs
@ -1,9 +1,10 @@
+//! currently unstable things from stdlib, vendored in
+
 use std::{
    ffi::OsStr,
    path::{Path, PathBuf},
 };

-/// a temporary trait to implement currently unstable methods
 pub trait PathBufExt {
    /// <https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.leak>
    fn leak<'a>(self) -> &'a mut Path;
@ -21,3 +22,21 @@ impl PathBufExt for PathBuf {
        Self::from(new_path)
    }
 }
+
+pub trait StrExt {
+    /// <https://doc.rust-lang.org/std/primitive.str.html#method.ceil_char_boundary>
+    fn ceil_char_boundary(self, index: usize) -> usize;
+}
+
+impl StrExt for &'_ str {
+    fn ceil_char_boundary(self, index: usize) -> usize {
+        let len = self.len();
+        if index > len {
+            len
+        } else {
+            (index..len)
+                .find(|&i| self.is_char_boundary(i))
+                .expect("`i = len` must be a char boundary") // otherwise `self` wouldn't have been a valid `&str` to begin with
+        }
+    }
+}