Add hidden --parser option to override the input format, rather than inferring it from the file extension

crates/ruff_cli/src/args.rs

@@ -8,8 +8,8 @@ use ruff_linter::line_width::LineLength;
 use ruff_linter::logging::LogLevel;
 use ruff_linter::registry::Rule;
 use ruff_linter::settings::types::{
-    FilePattern, PatternPrefixPair, PerFileIgnore, PreviewMode, PythonVersion, SerializationFormat,
-    UnsafeFixes,
+    FilePattern, ParserType, PatternPrefixPair, PerFileIgnore, PreviewMode, PythonVersion,
+    SerializationFormat, UnsafeFixes,
 };
 use ruff_linter::{RuleParser, RuleSelector, RuleSelectorParser};
 use ruff_workspace::configuration::{Configuration, RuleSelection};
@@ -347,6 +347,10 @@ pub struct CheckCommand {
     /// Dev-only argument to show fixes
     #[arg(long, hide = true)]
     pub ecosystem_ci: bool,
+
+    /// Override file format
+    #[arg(long, hide = true)]
+    pub parser: Option<ParserType>,
 }
 
 #[derive(Clone, Debug, clap::Parser)]
@@ -496,6 +500,7 @@ impl CheckCommand {
                 statistics: self.statistics,
                 stdin_filename: self.stdin_filename,
                 watch: self.watch,
+                parser: self.parser,
             },
             CliOverrides {
                 dummy_variable_rgx: self.dummy_variable_rgx,
@@ -594,6 +599,7 @@ pub struct CheckArguments {
     pub statistics: bool,
     pub stdin_filename: Option<PathBuf>,
     pub watch: bool,
+    pub parser: Option<ParserType>,
 }
 
 /// CLI settings that are distinct from configuration (commands, lists of files,

crates/ruff_cli/src/cache.rs

@@ -630,6 +630,7 @@ mod tests {
                     flags::Noqa::Enabled,
                     flags::FixMode::Generate,
                     UnsafeFixes::Enabled,
+                    flags::Parser::Auto,
                 )
                 .unwrap();
                 if diagnostics
@@ -676,6 +677,7 @@ mod tests {
                 flags::Noqa::Enabled,
                 flags::FixMode::Generate,
                 UnsafeFixes::Enabled,
+                flags::Parser::Auto,
             )
             .unwrap();
         }
@@ -1049,6 +1051,7 @@ mod tests {
                 flags::Noqa::Enabled,
                 flags::FixMode::Generate,
                 UnsafeFixes::Enabled,
+                flags::Parser::Auto,
             )
         }
 

crates/ruff_cli/src/commands/check.rs

@@ -30,6 +30,7 @@ use crate::diagnostics::Diagnostics;
 use crate::panic::catch_unwind;
 
 /// Run the linter over a collection of files.
+#[allow(clippy::too_many_arguments)]
 pub(crate) fn check(
     files: &[PathBuf],
     pyproject_config: &PyprojectConfig,
@@ -38,6 +39,7 @@ pub(crate) fn check(
     noqa: flags::Noqa,
     fix_mode: flags::FixMode,
     unsafe_fixes: UnsafeFixes,
+    parser: flags::Parser,
 ) -> Result<Diagnostics> {
     // Collect all the Python files to check.
     let start = Instant::now();
@@ -103,6 +105,7 @@ pub(crate) fn check(
                     noqa,
                     fix_mode,
                     unsafe_fixes,
+                    parser,
                 )
                 .map_err(|e| {
                     (Some(path.to_path_buf()), {
@@ -184,6 +187,7 @@ pub(crate) fn check(
 
 /// Wraps [`lint_path`](crate::diagnostics::lint_path) in a [`catch_unwind`](std::panic::catch_unwind) and emits
 /// a diagnostic if the linting the file panics.
+#[allow(clippy::too_many_arguments)]
 fn lint_path(
     path: &Path,
     package: Option<&Path>,
@@ -192,9 +196,19 @@ fn lint_path(
     noqa: flags::Noqa,
     fix_mode: flags::FixMode,
     unsafe_fixes: UnsafeFixes,
+    parser: flags::Parser,
 ) -> Result<Diagnostics> {
     let result = catch_unwind(|| {
-        crate::diagnostics::lint_path(path, package, settings, cache, noqa, fix_mode, unsafe_fixes)
+        crate::diagnostics::lint_path(
+            path,
+            package,
+            settings,
+            cache,
+            noqa,
+            fix_mode,
+            unsafe_fixes,
+            parser,
+        )
     });
 
     match result {
@@ -280,6 +294,7 @@ mod test {
             flags::Noqa::Disabled,
             flags::FixMode::Generate,
             UnsafeFixes::Enabled,
+            flags::Parser::Auto,
         )
         .unwrap();
         let mut output = Vec::new();

crates/ruff_cli/src/commands/check_stdin.rs

@@ -17,6 +17,7 @@ pub(crate) fn check_stdin(
     overrides: &CliOverrides,
     noqa: flags::Noqa,
     fix_mode: flags::FixMode,
+    parser: flags::Parser,
 ) -> Result<Diagnostics> {
     if let Some(filename) = filename {
         if !python_file_at_path(filename, pyproject_config, overrides)? {
@@ -42,6 +43,7 @@ pub(crate) fn check_stdin(
         &pyproject_config.settings,
         noqa,
         fix_mode,
+        parser,
     )?;
     diagnostics.messages.sort_unstable();
     Ok(diagnostics)

crates/ruff_cli/src/diagnostics.rs

@@ -177,7 +177,21 @@ impl AddAssign for FixMap {
     }
 }
 
+fn parser_source_type(
+    parser: flags::Parser,
+    source_type: ruff_python_ast::PySourceType,
+) -> ruff_python_ast::PySourceType {
+    match (parser, source_type) {
+        (flags::Parser::Auto, _) => source_type,
+        (flags::Parser::Ipynb, _) => ruff_python_ast::PySourceType::Ipynb,
+        (flags::Parser::Python, ruff_python_ast::PySourceType::Stub) => {
+            ruff_python_ast::PySourceType::Stub
+        }
+        (flags::Parser::Python, _) => ruff_python_ast::PySourceType::Python,
+    }
+}
 /// Lint the source code at the given `Path`.
+#[allow(clippy::too_many_arguments)]
 pub(crate) fn lint_path(
     path: &Path,
     package: Option<&Path>,
@@ -186,6 +200,7 @@ pub(crate) fn lint_path(
     noqa: flags::Noqa,
     fix_mode: flags::FixMode,
     unsafe_fixes: UnsafeFixes,
+    parser: flags::Parser,
 ) -> Result<Diagnostics> {
     // Check the cache.
     // TODO(charlie): `fixer::Mode::Apply` and `fixer::Mode::Diff` both have
@@ -194,7 +209,7 @@ pub(crate) fn lint_path(
     // write the fixes to disk, thus invalidating the cache. But it's a bit hard
     // to reason about. We need to come up with a better solution here.)
     let caching = match cache {
-        Some(cache) if noqa.into() && fix_mode.is_generate() => {
+        Some(cache) if noqa.into() && fix_mode.is_generate() && parser.is_auto() => {
             let relative_path = cache
                 .relative_path(path)
                 .expect("wrong package cache for file");
@@ -240,7 +255,7 @@ pub(crate) fn lint_path(
             });
         }
         SourceType::Toml(_) => return Ok(Diagnostics::default()),
-        SourceType::Python(source_type) => source_type,
+        SourceType::Python(source_type) => parser_source_type(parser, source_type),
     };
 
     // Extract the sources from the file.
@@ -353,11 +368,13 @@ pub(crate) fn lint_stdin(
     settings: &Settings,
     noqa: flags::Noqa,
     fix_mode: flags::FixMode,
+    parser: flags::Parser,
 ) -> Result<Diagnostics> {
     // TODO(charlie): Support `pyproject.toml`.
     let SourceType::Python(source_type) = path.map(SourceType::from).unwrap_or_default() else {
         return Ok(Diagnostics::default());
     };
+    let source_type = parser_source_type(parser, source_type);
 
     // Extract the sources from the file.
     let source_kind = match SourceKind::from_source_code(contents, source_type) {

crates/ruff_cli/src/lib.rs

@@ -13,8 +13,8 @@ use log::warn;
 use notify::{recommended_watcher, RecursiveMode, Watcher};
 
 use ruff_linter::logging::{set_up_logging, LogLevel};
-use ruff_linter::settings::flags::FixMode;
-use ruff_linter::settings::types::SerializationFormat;
+use ruff_linter::settings::flags::{FixMode, Parser};
+use ruff_linter::settings::types::{ParserType, SerializationFormat};
 use ruff_linter::{fs, warn_user, warn_user_once};
 use ruff_workspace::Settings;
 
@@ -259,6 +259,12 @@ pub fn check(args: CheckCommand, log_level: LogLevel) -> Result<ExitStatus> {
     if show_source {
         printer_flags |= PrinterFlags::SHOW_SOURCE;
     }
+    let parser = match cli.parser {
+        Some(ParserType::Ipynb) => Parser::Ipynb,
+        Some(ParserType::Py) => Parser::Python,
+        None => Parser::Auto,
+    };
+
     if cli.ecosystem_ci {
         warn_user!(
             "The formatting of fixes emitted by this option is a work-in-progress, subject to \
@@ -325,6 +331,7 @@ pub fn check(args: CheckCommand, log_level: LogLevel) -> Result<ExitStatus> {
             noqa.into(),
             fix_mode,
             unsafe_fixes,
+            parser,
         )?;
         printer.write_continuously(&mut writer, &messages)?;
 
@@ -358,6 +365,7 @@ pub fn check(args: CheckCommand, log_level: LogLevel) -> Result<ExitStatus> {
                         noqa.into(),
                         fix_mode,
                         unsafe_fixes,
+                        Parser::Auto,
                     )?;
                     printer.write_continuously(&mut writer, &messages)?;
                 }
@@ -375,6 +383,7 @@ pub fn check(args: CheckCommand, log_level: LogLevel) -> Result<ExitStatus> {
                 &overrides,
                 noqa.into(),
                 fix_mode,
+                parser,
             )?
         } else {
             commands::check::check(
@@ -385,6 +394,7 @@ pub fn check(args: CheckCommand, log_level: LogLevel) -> Result<ExitStatus> {
                 noqa.into(),
                 fix_mode,
                 unsafe_fixes,
+                parser,
             )?
         };
 

crates/ruff_cli/tests/integration_test.rs

@@ -321,6 +321,120 @@ fn stdin_fix_jupyter() {
     "###);
 }
 
+#[test]
+fn stdin_override_parser_py() {
+    assert_cmd_snapshot!(Command::new(get_cargo_bin(BIN_NAME))
+        .args(STDIN_BASE_OPTIONS)
+        .args(["--stdin-filename", "F401.ipynb", "--parser", "py"])
+        .pass_stdin("import os\n"), @r###"
+    success: false
+    exit_code: 1
+    ----- stdout -----
+    F401.ipynb:1:8: F401 [*] `os` imported but unused
+    Found 1 error.
+    [*] 1 fixable with the `--fix` option.
+
+    ----- stderr -----
+    "###);
+}
+
+#[test]
+fn stdin_override_parser_ipynb() {
+    let args = ["--parser", "ipynb", "--stdin-filename", "Jupyter.py"];
+    assert_cmd_snapshot!(Command::new(get_cargo_bin(BIN_NAME))
+        .args(STDIN_BASE_OPTIONS)
+        .args(args)
+        .pass_stdin(r#"{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "dccc687c-96e2-4604-b957-a8a89b5bec06",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import os"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "19e1b029-f516-4662-a9b9-623b93edac1a",
+   "metadata": {},
+   "source": [
+    "Foo"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "cdce7b92-b0fb-4c02-86f6-e233b26fa84f",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import sys"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "e40b33d2-7fe4-46c5-bdf0-8802f3052565",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "1\n"
+     ]
+    }
+   ],
+   "source": [
+    "print(1)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "a1899bc8-d46f-4ec0-b1d1-e1ca0f04bf60",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.2"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}"#), @r###"
+    success: false
+    exit_code: 1
+    ----- stdout -----
+    Jupyter.py:cell 1:1:8: F401 [*] `os` imported but unused
+    Jupyter.py:cell 3:1:8: F401 [*] `sys` imported but unused
+    Found 2 errors.
+    [*] 2 fixable with the `--fix` option.
+
+    ----- stderr -----
+    "###);
+}
+
 #[test]
 fn stdin_fix_when_not_fixable_should_still_print_contents() {
     let args = ["--fix"];

crates/ruff_linter/src/settings/flags.rs

@@ -16,3 +16,16 @@ pub enum Cache {
     Enabled,
     Disabled,
 }
+
+#[derive(Debug, Copy, Clone, Hash, is_macro::Is)]
+pub enum Parser {
+    Auto,
+    Python,
+    Ipynb,
+}
+
+impl Default for Parser {
+    fn default() -> Self {
+        Self::Auto
+    }
+}