Rust: fix panic when the last character in a range is multi-byte

Author: aibaars

rust/extractor/src/translate/base.rs

@@ -3,7 +3,7 @@ use crate::trap::{Label, TrapClass};
 use codeql_extractor::trap::{self};
 use ra_ap_ide_db::line_index::{LineCol, LineIndex};
 use ra_ap_syntax::ast::RangeItem;
-use ra_ap_syntax::{ast, SyntaxError, TextRange, TextSize};
+use ra_ap_syntax::{ast, SyntaxError, TextRange};
 pub trait TextValue {
     fn try_get_text(&self) -> Option<String>;
 }
@@ -70,12 +70,21 @@ impl Translator {
     }
     pub fn location(&self, range: TextRange) -> (LineCol, LineCol) {
         let start = self.line_index.line_col(range.start());
-        let end = self.line_index.line_col(
-            range
-                .end()
-                .checked_sub(TextSize::new(1))
-                .unwrap_or(range.end()),
-        );
+        let range_end = range.end();
+        // QL end positions are inclusive, while TextRange offsets are exclusive and point at the position
+        // right after the last character of the range. We need to shift the end offset one character to the left to
+        // get the right inclusive QL position. Unfortunately, simply subtracting `1` from the end-offset may cause
+        // the offset to point in the middle of a mult-byte character, resulting in a `panic`. Therefore we use `try_line_col`
+        // with decreasing offsets to find the start of the last character included in the range.
+        for i in 1..4 {
+            if let Some(end) = range_end
+                .checked_sub(i.into())
+                .and_then(|x| self.line_index.try_line_col(x))
+            {
+                return (start, end);
+            }
+        }
+        let end = self.line_index.line_col(range_end);
         (start, end)
     }
     pub fn emit_location<T: TrapClass>(&mut self, label: Label<T>, node: impl ast::AstNode) {

rust/ql/test/extractor-tests/utf8/ast.expected

@@ -0,0 +1,39 @@
+| lib.rs:1:1:3:22 | SourceFile |
+| lib.rs:2:1:2:8 | Module |
+| lib.rs:2:5:2:7 | Name |
+| lib.rs:3:1:3:8 | Module |
+| lib.rs:3:5:3:8 | Name |
+| lib.rs:3:10:3:20 | NameRef |
+| lib.rs:3:10:3:20 | Path |
+| lib.rs:3:10:3:20 | PathSegment |
+| lib.rs:3:10:3:21 | MacroCall |
+| utf8-identifiers.rs:1:1:4:6 | foo |
+| utf8-identifiers.rs:1:1:12:2 | SourceFile |
+| utf8-identifiers.rs:1:4:1:6 | Name |
+| utf8-identifiers.rs:1:7:4:1 | GenericParamList |
+| utf8-identifiers.rs:2:5:2:6 | Lifetime |
+| utf8-identifiers.rs:2:5:2:6 | LifetimeParam |
+| utf8-identifiers.rs:3:5:3:5 | Name |
+| utf8-identifiers.rs:3:5:3:5 | TypeParam |
+| utf8-identifiers.rs:4:2:4:3 | ParamList |
+| utf8-identifiers.rs:4:5:4:6 | BlockExpr |
+| utf8-identifiers.rs:4:5:4:6 | StmtList |
+| utf8-identifiers.rs:6:1:8:1 | Struct |
+| utf8-identifiers.rs:6:8:6:8 | Name |
+| utf8-identifiers.rs:6:10:8:1 | RecordFieldList |
+| utf8-identifiers.rs:7:5:7:5 | Name |
+| utf8-identifiers.rs:7:5:7:13 | RecordField |
+| utf8-identifiers.rs:7:9:7:13 | NameRef |
+| utf8-identifiers.rs:7:9:7:13 | Path |
+| utf8-identifiers.rs:7:9:7:13 | PathSegment |
+| utf8-identifiers.rs:7:9:7:13 | PathType |
+| utf8-identifiers.rs:10:1:10:3 | Visibility |
+| utf8-identifiers.rs:10:1:12:1 | main |
+| utf8-identifiers.rs:10:8:10:11 | Name |
+| utf8-identifiers.rs:10:12:10:13 | ParamList |
+| utf8-identifiers.rs:10:15:12:1 | BlockExpr |
+| utf8-identifiers.rs:10:15:12:1 | StmtList |
+| utf8-identifiers.rs:11:5:11:24 | LetStmt |
+| utf8-identifiers.rs:11:9:11:9 | IdentPat |
+| utf8-identifiers.rs:11:9:11:9 | Name |
+| utf8-identifiers.rs:11:14:11:23 | LiteralExpr |

rust/ql/test/extractor-tests/utf8/ast.ql

@@ -0,0 +1,3 @@
+import codeql.rust.elements
+
+select any(AstNode n)

rust/ql/test/extractor-tests/utf8/utf8-identifiers.rs

@@ -0,0 +1,12 @@
+fn foo<
+    'β,
+    γ
+>() {}
+
+struct X {
+    δ: usize
+}
+
+pub fn main() {
+    let α = 0.00001f64;
+}