Fix traceback

Author: youknowone

Lib/test/test_exceptions.py

@@ -1745,7 +1745,6 @@ def __del__(self):
                              f"deallocator {obj_repr}")
             self.assertIsNotNone(cm.unraisable.exc_traceback)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_unhandled(self):
         # Check for sensible reporting of unhandled exceptions
         for exc_type in (ValueError, BrokenStrException):
@@ -2283,7 +2282,6 @@ def test_multiline_not_highlighted(self):
 class SyntaxErrorTests(unittest.TestCase):
     maxDiff = None
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     @force_not_colorized
     def test_range_of_offsets(self):
         cases = [

Lib/test/test_syntax.py

@@ -2293,7 +2293,6 @@ def test_expression_with_assignment(self):
             offset=7
         )
 
-    @unittest.expectedFailure # TODO: RUSTPYTHON
     def test_curly_brace_after_primary_raises_immediately(self):
         self._check_error("f{}", "invalid syntax", mode="single")
 

Lib/test/test_traceback.py

@@ -88,7 +88,6 @@ def syntax_error_bad_indentation2(self):
     def tokenizer_error_with_caret_range(self):
         compile("blech  (  ", "?", "exec")
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: 11 != 14
     def test_caret(self):
         err = self.get_exception_format(self.syntax_error_with_caret,
                                         SyntaxError)
@@ -201,7 +200,6 @@ def f():
         finally:
             unlink(TESTFN)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: 3 != 4
     def test_bad_indentation(self):
         err = self.get_exception_format(self.syntax_error_bad_indentation,
                                         IndentationError)
@@ -1797,7 +1795,6 @@ class TestKeywordTypoSuggestions(unittest.TestCase):
         ("for x im n:\n  pass", "in"),
     ]
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_keyword_suggestions_from_file(self):
         with tempfile.TemporaryDirectory() as script_dir:
             for i, (code, expected_kw) in enumerate(self.TYPO_CASES):
@@ -1808,7 +1805,6 @@ def test_keyword_suggestions_from_file(self):
                     stderr_text = stderr.decode('utf-8')
                     self.assertIn(f"Did you mean '{expected_kw}'", stderr_text)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON
     def test_keyword_suggestions_from_command_string(self):
         for code, expected_kw in self.TYPO_CASES:
             with self.subTest(typo=expected_kw):
@@ -3352,7 +3348,6 @@ class MiscTracebackCases(unittest.TestCase):
     # Check non-printing functions in traceback module
     #
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: 1 != 0
     def test_clear(self):
         def outer():
             middle()
@@ -3574,7 +3569,6 @@ def format_frame_summary(self, frame_summary, colorize=False):
             f'  File "{__file__}", line {lno}, in f\n    1/0\n'
         )
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; Actual: _should_show_carets(13, 14, ['# this line will be used during rendering'], None)
     def test_summary_should_show_carets(self):
         # See: https://github.com/python/cpython/issues/122353
 
@@ -3731,7 +3725,6 @@ def test_context(self):
         self.assertEqual(type(exc_obj).__name__, exc.exc_type_str)
         self.assertEqual(str(exc_obj), str(exc))
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: 11 not greater than 1000
     def test_long_context_chain(self):
         def f():
             try:
@@ -4059,7 +4052,6 @@ def test_exception_group_format_exception_onlyi_recursive(self):
 
         self.assertEqual(formatted, expected)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; Diff is 2265 characters long. Set self.maxDiff to None to see it.
     def test_exception_group_format(self):
         teg = traceback.TracebackException.from_exception(self.eg)
 
@@ -4841,19 +4833,15 @@ class PurePythonSuggestionFormattingTests(
     traceback printing in traceback.py.
     """
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: "'bluch'" not found in "ImportError: cannot import name 'blach'"
     def test_import_from_suggestions_underscored(self):
         return super().test_import_from_suggestions_underscored()
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: "'bluch'" not found in "ImportError: cannot import name 'blech'"
     def test_import_from_suggestions_non_string(self):
         return super().test_import_from_suggestions_non_string()
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: "'bluchin'?" not found in "ImportError: cannot import name 'bluch'"
     def test_import_from_suggestions(self):
         return super().test_import_from_suggestions()
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: 'Did you mean' not found in "AttributeError: 'A' object has no attribute 'blich'"
     def test_attribute_error_inside_nested_getattr(self):
         return super().test_attribute_error_inside_nested_getattr()
 
@@ -4969,7 +4957,6 @@ class MyList(list):
 class TestColorizedTraceback(unittest.TestCase):
     maxDiff = None
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: "y = \x1b[31mx['a']['b']\x1b[0m\x1b[1;31m['c']\x1b[0m" not found in 'Traceback (most recent call last):\n  File \x1b[35m"/Users/al03219714/Projects/RustPython/crates/pylib/Lib/test/test_traceback.py"\x1b[0m, line \x1b[35m4764\x1b[0m, in \x1b[35mtest_colorized_traceback\x1b[0m\n    \x1b[31mbar\x1b[0m\x1b[1;31m()\x1b[0m\n    \x1b[31m~~~\x1b[0m\x1b[1;31m^^\x1b[0m\n    bar = <function TestColorizedTraceback.test_colorized_traceback.<locals>.bar at 0xb57b09180>\n    baz1 = <function TestColorizedTraceback.test_colorized_traceback.<locals>.baz1 at 0xb57b09e00>\n    baz2 = <function TestColorizedTraceback.test_colorized_traceback.<locals>.baz2 at 0xb57b09cc0>\n    e = TypeError("\'NoneType\' object is not subscriptable")\n    foo = <function TestColorizedTraceback.test_colorized_traceback.<locals>.foo at 0xb57b08140>\n    self = <test.test_traceback.TestColorizedTraceback testMethod=test_colorized_traceback>\n  File \x1b[35m"/Users/al03219714/Projects/RustPython/crates/pylib/Lib/test/test_traceback.py"\x1b[0m, line \x1b[35m4760\x1b[0m, in \x1b[35mbar\x1b[0m\n    return baz1(1,\n            2,3\n            ,4)\n    baz1 = <function TestColorizedTraceback.test_colorized_traceback.<locals>.baz1 at 0xb57b09e00>\n  File \x1b[35m"/Users/al03219714/Projects/RustPython/crates/pylib/Lib/test/test_traceback.py"\x1b[0m, line \x1b[35m4757\x1b[0m, in \x1b[35mbaz1\x1b[0m\n    return baz2(1,2,3,4)\n    args = (1, 2, 3, 4)\n    baz2 = <function TestColorizedTraceback.test_colorized_traceback.<locals>.baz2 at 0xb57b09cc0>\n  File \x1b[35m"/Users/al03219714/Projects/RustPython/crates/pylib/Lib/test/test_traceback.py"\x1b[0m, line \x1b[35m4754\x1b[0m, in \x1b[35mbaz2\x1b[0m\n    return \x1b[31m(lambda *args: foo(*args))\x1b[0m\x1b[1;31m(1,2,3,4)\x1b[0m\n           \x1b[31m~~~~~~~~~~~~~~~~~~~~~~~~~~\x1b[0m\x1b[1;31m^^^^^^^^^\x1b[0m\n    args = (1, 2, 3, 4)\n    foo = <function TestColorizedTraceback.test_colorized_traceback.<locals>.foo at 0xb57b08140>\n  File \x1b[35m"/Users/al03219714/Projects/RustPython/crates/pylib/Lib/test/test_traceback.py"\x1b[0m, line \x1b[35m4754\x1b[0m, in \x1b[35m<lambda>\x1b[0m\n    return (lambda *args: \x1b[31mfoo\x1b[0m\x1b[1;31m(*args)\x1b[0m)(1,2,3,4)\n                          \x1b[31m~~~\x1b[0m\x1b[1;31m^^^^^^^\x1b[0m\n    args = (1, 2, 3, 4)\n    foo = <function TestColorizedTraceback.test_colorized_traceback.<locals>.foo at 0xb57b08140>\n  File \x1b[35m"/Users/al03219714/Projects/RustPython/crates/pylib/Lib/test/test_traceback.py"\x1b[0m, line \x1b[35m4751\x1b[0m, in \x1b[35mfoo\x1b[0m\n    y = x[\'a\'][\'b\'][\x1b[1;31m\'c\'\x1b[0m]\n                    \x1b[1;31m^^^\x1b[0m\n    args = (1, 2, 3, 4)\n    x = {\'a\': {\'b\': None}}\n\x1b[1;35mTypeError\x1b[0m: \x1b[35m\'NoneType\' object is not subscriptable\x1b[0m\n'
     def test_colorized_traceback(self):
         def foo(*args):
             x = {'a':{'b': None}}
@@ -5002,7 +4989,6 @@ def bar():
         self.assertIn("return baz1(1,\n            2,3\n            ,4)", lines)
         self.assertIn(red + "bar" + reset + boldr + "()" + reset, lines)
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; AssertionError: '  File \x1b[35m"<string>"\x1b[0m, line \x1b[35m1\x1b[0m\n    a \x1b[1;31m$\x1b[0m b\n      \x1b[1;31m^\x1b[0m\n\x1b[1;35mSyntaxError\x1b[0m: \x1b[35minvalid syntax\x1b[0m\n' not found in 'Traceback (most recent call last):\n  File \x1b[35m"/Users/al03219714/Projects/RustPython/crates/pylib/Lib/test/test_traceback.py"\x1b[0m, line \x1b[35m4782\x1b[0m, in \x1b[35mtest_colorized_syntax_error\x1b[0m\n    \x1b[31mcompile\x1b[0m\x1b[1;31m("a $ b", "<string>", "exec")\x1b[0m\n    \x1b[31m~~~~~~~\x1b[0m\x1b[1;31m^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\x1b[0m\n    e = SyntaxError(\'got unexpected token $\')\n    self = <test.test_traceback.TestColorizedTraceback testMethod=test_colorized_syntax_error>\n  File \x1b[35m"<string>"\x1b[0m, line \x1b[35m1\x1b[0m\n    a \x1b[1;31m$\x1b[0m b\n      \x1b[1;31m^\x1b[0m\n\x1b[1;35mSyntaxError\x1b[0m: \x1b[35mgot unexpected token $\x1b[0m\n'
     def test_colorized_syntax_error(self):
         try:
             compile("a $ b", "<string>", "exec")
@@ -5053,7 +5039,6 @@ def expected(t, m, fn, l, f, E, e, z):
             ]
         self.assertEqual(actual, expected(**colors))
 
-    @unittest.expectedFailure  # TODO: RUSTPYTHON; Diff is 1795 characters long. Set self.maxDiff to None to see it.
     def test_colorized_traceback_from_exception_group(self):
         def foo():
             exceptions = []

crates/codegen/src/compile.rs

@@ -493,6 +493,9 @@ impl Compiler {
         slice: &ast::Expr,
         ctx: ast::ExprContext,
     ) -> CompileResult<()> {
+        // Save the subscript expression's source range
+        let subscript_range = self.current_source_range;
+
         // 1. Check subscripter and index for Load context
         // 2. VISIT value
         // 3. Handle two-element slice specially
@@ -515,6 +518,8 @@ impl Compiler {
                     "should_use_slice_optimization should only return true for ast::Expr::Slice"
                 ),
             };
+            // Restore full subscript expression range before emitting
+            self.set_source_range(subscript_range);
             match ctx {
                 ast::ExprContext::Load => {
                     emit!(self, Instruction::BinarySlice);
@@ -527,6 +532,8 @@ impl Compiler {
         } else {
             // VISIT(c, expr, e->v.Subscript.slice)
             self.compile_expression(slice)?;
+            // Restore full subscript expression range before emitting
+            self.set_source_range(subscript_range);
 
             // Emit appropriate instruction based on context
             match ctx {
@@ -6603,7 +6610,8 @@ impl Compiler {
                 self.compile_expression(left)?;
                 self.compile_expression(right)?;
 
-                // Perform operation:
+                // Restore full expression range before emitting the operation
+                self.set_source_range(range);
                 self.compile_op(op, false);
             }
             ast::Expr::Subscript(ast::ExprSubscript {
@@ -6614,7 +6622,8 @@ impl Compiler {
             ast::Expr::UnaryOp(ast::ExprUnaryOp { op, operand, .. }) => {
                 self.compile_expression(operand)?;
 
-                // Perform operation:
+                // Restore full expression range before emitting the operation
+                self.set_source_range(range);
                 match op {
                     ast::UnaryOp::UAdd => emit!(
                         self,

crates/compiler/src/lib.rs

@@ -46,22 +46,75 @@ pub enum CompileError {
 impl CompileError {
     pub fn from_ruff_parse_error(error: parser::ParseError, source_file: &SourceFile) -> Self {
         let source_code = source_file.to_source_code();
-        let location = source_code.source_location(error.location.start(), PositionEncoding::Utf8);
-        let mut end_location =
-            source_code.source_location(error.location.end(), PositionEncoding::Utf8);
-
-        // If the error range ends at the start of a new line (column 1),
-        // adjust it to the end of the previous line
-        if end_location.character_offset.get() == 1 && end_location.line > location.line {
-            // Get the end of the previous line
-            let prev_line_end = error.location.end() - ruff_text_size::TextSize::from(1);
-            end_location = source_code.source_location(prev_line_end, PositionEncoding::Utf8);
-            // Adjust column to be after the last character
-            end_location.character_offset = end_location.character_offset.saturating_add(1);
-        }
+        let source_text = source_file.source_text();
+
+        // For EOF errors (unclosed brackets), find the unclosed bracket position
+        // and adjust both the error location and message
+        let (error_type, location, end_location) = if matches!(
+            &error.error,
+            parser::ParseErrorType::Lexical(parser::LexicalErrorType::Eof)
+        ) {
+            if let Some((bracket_char, bracket_offset)) =
+                find_unclosed_bracket(source_text)
+            {
+                let bracket_text_size = ruff_text_size::TextSize::new(bracket_offset as u32);
+                let loc =
+                    source_code.source_location(bracket_text_size, PositionEncoding::Utf8);
+                let end_loc = SourceLocation {
+                    line: loc.line,
+                    character_offset: loc.character_offset.saturating_add(1),
+                };
+                let msg = format!("'{}' was never closed", bracket_char);
+                (
+                    parser::ParseErrorType::OtherError(msg),
+                    loc,
+                    end_loc,
+                )
+            } else {
+                let loc = source_code
+                    .source_location(error.location.start(), PositionEncoding::Utf8);
+                let end_loc = source_code
+                    .source_location(error.location.end(), PositionEncoding::Utf8);
+                (error.error, loc, end_loc)
+            }
+        } else if matches!(
+            &error.error,
+            parser::ParseErrorType::Lexical(parser::LexicalErrorType::IndentationError)
+        ) {
+            // For IndentationError, point the offset to the end of the line content
+            // (matching CPython behavior) instead of the beginning
+            let loc =
+                source_code.source_location(error.location.start(), PositionEncoding::Utf8);
+            let line_idx = loc.line.to_zero_indexed();
+            let line = source_text.split('\n').nth(line_idx).unwrap_or("");
+            let line_end_col = line.len() + 1; // 1-indexed, past last char (the newline)
+            let end_loc = SourceLocation {
+                line: loc.line,
+                character_offset: ruff_source_file::OneIndexed::new(line_end_col)
+                    .unwrap_or(loc.character_offset),
+            };
+            (error.error, end_loc, end_loc)
+        } else {
+            let loc =
+                source_code.source_location(error.location.start(), PositionEncoding::Utf8);
+            let mut end_loc =
+                source_code.source_location(error.location.end(), PositionEncoding::Utf8);
+
+            // If the error range ends at the start of a new line (column 1),
+            // adjust it to the end of the previous line
+            if end_loc.character_offset.get() == 1 && end_loc.line > loc.line {
+                let prev_line_end =
+                    error.location.end() - ruff_text_size::TextSize::from(1);
+                end_loc =
+                    source_code.source_location(prev_line_end, PositionEncoding::Utf8);
+                end_loc.character_offset = end_loc.character_offset.saturating_add(1);
+            }
+
+            (error.error, loc, end_loc)
+        };
 
         Self::Parse(ParseError {
-            error: error.error,
+            error: error_type,
             raw_location: error.location,
             location,
             end_location,
@@ -102,6 +155,94 @@ impl CompileError {
     }
 }
 
+/// Find the last unclosed opening bracket in source code.
+/// Returns the bracket character and its byte offset, or None if all brackets are balanced.
+fn find_unclosed_bracket(source: &str) -> Option<(char, usize)> {
+    let mut stack: Vec<(char, usize)> = Vec::new();
+    let mut in_string = false;
+    let mut string_quote = '\0';
+    let mut triple_quote = false;
+    let mut escape_next = false;
+
+    let chars: Vec<char> = source.chars().collect();
+    let mut i = 0;
+    let mut byte_offset = 0;
+
+    while i < chars.len() {
+        let ch = chars[i];
+        let ch_len = ch.len_utf8();
+
+        if escape_next {
+            escape_next = false;
+            byte_offset += ch_len;
+            i += 1;
+            continue;
+        }
+
+        if in_string {
+            if ch == '\\' {
+                escape_next = true;
+            } else if triple_quote {
+                if ch == string_quote
+                    && i + 2 < chars.len()
+                    && chars[i + 1] == string_quote
+                    && chars[i + 2] == string_quote
+                {
+                    in_string = false;
+                    byte_offset += ch_len * 3;
+                    i += 3;
+                    continue;
+                }
+            } else if ch == string_quote {
+                in_string = false;
+            }
+            byte_offset += ch_len;
+            i += 1;
+            continue;
+        }
+
+        // Check for comments
+        if ch == '#' {
+            // Skip to end of line
+            while i < chars.len() && chars[i] != '\n' {
+                byte_offset += chars[i].len_utf8();
+                i += 1;
+            }
+            continue;
+        }
+
+        // Check for string start
+        if ch == '\'' || ch == '"' {
+            string_quote = ch;
+            if i + 2 < chars.len() && chars[i + 1] == ch && chars[i + 2] == ch {
+                triple_quote = true;
+                in_string = true;
+                byte_offset += ch_len * 3;
+                i += 3;
+                continue;
+            }
+            triple_quote = false;
+            in_string = true;
+            byte_offset += ch_len;
+            i += 1;
+            continue;
+        }
+
+        match ch {
+            '(' | '[' | '{' => stack.push((ch, byte_offset)),
+            ')' | ']' | '}' => {
+                stack.pop();
+            }
+            _ => {}
+        }
+
+        byte_offset += ch_len;
+        i += 1;
+    }
+
+    stack.last().copied()
+}
+
 /// Compile a given source code into a bytecode object.
 pub fn compile(
     source: &str,