[syntax-errors] Start detecting compile-time syntax errors

crates/ruff/tests/integration_test.rs

@@ -1054,6 +1054,52 @@ def mvce(keys, values):
     "#);
 }
 
+#[test]
+fn show_statistics_syntax_errors() {
+    let mut cmd = RuffCheck::default()
+        .args(["--statistics", "--target-version=py39", "--preview"])
+        .build();
+
+    // ParseError
+    assert_cmd_snapshot!(
+        cmd.pass_stdin("x ="),
+        @r"
+    success: false
+    exit_code: 1
+    ----- stdout -----
+    1		syntax-error
+    Found 1 error.
+
+    ----- stderr -----
+    ");
+
+    // match before 3.10, UnsupportedSyntaxError
+    assert_cmd_snapshot!(
+        cmd.pass_stdin("match 2:\n  case 1: ..."),
+        @r"
+    success: false
+    exit_code: 1
+    ----- stdout -----
+    1		syntax-error
+    Found 1 error.
+
+    ----- stderr -----
+    ");
+
+    // rebound comprehension variable, SemanticSyntaxError
+    assert_cmd_snapshot!(
+        cmd.pass_stdin("[x := 1 for x in range(0)]"),
+        @r"
+    success: false
+    exit_code: 1
+    ----- stdout -----
+    1		syntax-error
+    Found 1 error.
+
+    ----- stderr -----
+    ");
+}
+
 #[test]
 fn preview_enabled_prefix() {
     // All the RUF9XX test rules should be triggered

crates/ruff/tests/lint.rs

@@ -5491,3 +5491,68 @@ fn cookiecutter_globbing_no_project_root() -> Result<()> {
 
     Ok(())
 }
+
+/// Test that semantic syntax errors (1) are emitted, (2) are not cached, (3) don't affect the
+/// reporting of normal diagnostics, and (4) are not suppressed by `select = []` (or otherwise
+/// disabling all AST-based rules).
+#[test]
+fn semantic_syntax_errors() -> Result<()> {
+    let tempdir = TempDir::new()?;
+    let contents = "[(x := 1) for x in foo]";
+    fs::write(tempdir.path().join("main.py"), contents)?;
+
+    let mut cmd = Command::new(get_cargo_bin(BIN_NAME));
+    // inline STDIN_BASE_OPTIONS to remove --no-cache
+    cmd.args(["check", "--output-format", "concise"])
+        .arg("--preview")
+        .arg("--quiet") // suppress `debug build without --no-cache` warnings
+        .current_dir(&tempdir);
+
+    assert_cmd_snapshot!(
+        cmd,
+        @r"
+    success: false
+    exit_code: 1
+    ----- stdout -----
+    main.py:1:3: SyntaxError: assignment expression cannot rebind comprehension variable
+    main.py:1:20: F821 Undefined name `foo`
+
+    ----- stderr -----
+    "
+    );
+
+    // this should *not* be cached, like normal parse errors
+    assert_cmd_snapshot!(
+        cmd,
+        @r"
+    success: false
+    exit_code: 1
+    ----- stdout -----
+    main.py:1:3: SyntaxError: assignment expression cannot rebind comprehension variable
+    main.py:1:20: F821 Undefined name `foo`
+
+    ----- stderr -----
+    "
+    );
+
+    // ensure semantic errors are caught even without AST-based rules selected
+    assert_cmd_snapshot!(
+        Command::new(get_cargo_bin(BIN_NAME))
+            .args(STDIN_BASE_OPTIONS)
+            .args(["--config", "lint.select = []"])
+            .arg("--preview")
+            .arg("-")
+            .pass_stdin(contents),
+        @r"
+    success: false
+    exit_code: 1
+    ----- stdout -----
+    -:1:3: SyntaxError: assignment expression cannot rebind comprehension variable
+    Found 1 error.
+
+    ----- stderr -----
+    "
+    );
+
+    Ok(())
+}

crates/ruff_linter/src/checkers/ast/analyze/statement.rs

@@ -853,14 +853,6 @@ pub(crate) fn statement(stmt: &Stmt, checker: &mut Checker) {
                     if checker.enabled(Rule::FutureFeatureNotDefined) {
                         pyflakes::rules::future_feature_not_defined(checker, alias);
                     }
-                    if checker.enabled(Rule::LateFutureImport) {
-                        if checker.semantic.seen_futures_boundary() {
-                            checker.report_diagnostic(Diagnostic::new(
-                                pyflakes::rules::LateFutureImport,
-                                stmt.range(),
-                            ));
-                        }
-                    }
                 } else if &alias.name == "*" {
                     if checker.enabled(Rule::UndefinedLocalWithNestedImportStarUsage) {
                         if !matches!(checker.semantic.current_scope().kind, ScopeKind::Module) {

crates/ruff_linter/src/checkers/ast/mod.rs

@@ -26,6 +26,9 @@ use std::path::Path;
 
 use itertools::Itertools;
 use log::debug;
+use ruff_python_parser::semantic_errors::{
+    SemanticSyntaxChecker, SemanticSyntaxContext, SemanticSyntaxError, SemanticSyntaxErrorKind,
+};
 use rustc_hash::{FxHashMap, FxHashSet};
 
 use ruff_diagnostics::{Diagnostic, IsolationLevel};
@@ -63,6 +66,7 @@ use crate::importer::Importer;
 use crate::noqa::NoqaMapping;
 use crate::package::PackageRoot;
 use crate::registry::Rule;
+use crate::rules::pyflakes::rules::LateFutureImport;
 use crate::rules::{flake8_pyi, flake8_type_checking, pyflakes, pyupgrade};
 use crate::settings::{flags, LinterSettings};
 use crate::{docstrings, noqa, Locator};
@@ -225,6 +229,9 @@ pub(crate) struct Checker<'a> {
     docstring_state: DocstringState,
     /// The target [`PythonVersion`] for version-dependent checks
     target_version: PythonVersion,
+
+    #[allow(clippy::struct_field_names)]
+    syntax_checker: SemanticSyntaxChecker,
 }
 
 impl<'a> Checker<'a> {
@@ -272,6 +279,7 @@ impl<'a> Checker<'a> {
             last_stmt_end: TextSize::default(),
             docstring_state: DocstringState::default(),
             target_version,
+            syntax_checker: SemanticSyntaxChecker::new(),
         }
     }
 }
@@ -514,8 +522,20 @@ impl<'a> Checker<'a> {
     }
 }
 
+impl SemanticSyntaxContext for Checker<'_> {
+    fn seen_docstring_boundary(&self) -> bool {
+        self.semantic.seen_module_docstring_boundary()
+    }
+
+    fn python_version(&self) -> PythonVersion {
+        self.target_version
+    }
+}
+
 impl<'a> Visitor<'a> for Checker<'a> {
     fn visit_stmt(&mut self, stmt: &'a Stmt) {
+        self.syntax_checker.visit_stmt(stmt, self);
+
         // Step 0: Pre-processing
         self.semantic.push_node(stmt);
 
@@ -550,17 +570,13 @@ impl<'a> Visitor<'a> for Checker<'a> {
                     {
                         self.semantic.flags |= SemanticModelFlags::FUTURE_ANNOTATIONS;
                     }
-                } else {
-                    self.semantic.flags |= SemanticModelFlags::FUTURES_BOUNDARY;
                 }
             }
             Stmt::Import(_) => {
                 self.semantic.flags |= SemanticModelFlags::MODULE_DOCSTRING_BOUNDARY;
-                self.semantic.flags |= SemanticModelFlags::FUTURES_BOUNDARY;
             }
             _ => {
                 self.semantic.flags |= SemanticModelFlags::MODULE_DOCSTRING_BOUNDARY;
-                self.semantic.flags |= SemanticModelFlags::FUTURES_BOUNDARY;
                 if !(self.semantic.seen_import_boundary()
                     || stmt.is_ipy_escape_command_stmt()
                     || helpers::is_assignment_to_a_dunder(stmt)
@@ -1131,6 +1147,8 @@ impl<'a> Visitor<'a> for Checker<'a> {
     }
 
     fn visit_expr(&mut self, expr: &'a Expr) {
+        self.syntax_checker.visit_expr(expr, self);
+
         // Step 0: Pre-processing
         if self.source_type.is_stub()
             && self.semantic.in_class_base()
@@ -2674,7 +2692,7 @@ pub(crate) fn check_ast(
     cell_offsets: Option<&CellOffsets>,
     notebook_index: Option<&NotebookIndex>,
     target_version: PythonVersion,
-) -> Vec<Diagnostic> {
+) -> (Vec<Diagnostic>, Vec<SemanticSyntaxError>) {
     let module_path = package
         .map(PackageRoot::path)
         .and_then(|package| to_module_path(package, path));
@@ -2739,5 +2757,34 @@ pub(crate) fn check_ast(
     checker.analyze.scopes.push(ScopeId::global());
     analyze::deferred_scopes(&checker);
 
-    checker.diagnostics.take()
+    let Checker {
+        diagnostics,
+        syntax_checker,
+        settings,
+        ..
+    } = checker;
+
+    // partition the semantic syntax errors into Diagnostics and regular errors
+    let mut diagnostics = diagnostics.take();
+    let mut semantic_syntax_errors = Vec::new();
+    for semantic_syntax_error in syntax_checker.finish() {
+        match semantic_syntax_error.kind {
+            SemanticSyntaxErrorKind::LateFutureImport => {
+                if settings.rules.enabled(Rule::LateFutureImport) {
+                    diagnostics.push(Diagnostic::new(
+                        LateFutureImport,
+                        semantic_syntax_error.range,
+                    ));
+                }
+            }
+            SemanticSyntaxErrorKind::ReboundComprehensionVariable
+                if settings.preview.is_enabled() =>
+            {
+                semantic_syntax_errors.push(semantic_syntax_error);
+            }
+            SemanticSyntaxErrorKind::ReboundComprehensionVariable => {}
+        }
+    }
+
+    (diagnostics, semantic_syntax_errors)
 }