Add hidden --extension to override inference of source type from 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,
+    ExtensionPair, FilePattern, PatternPrefixPair, PerFileIgnore, PreviewMode, PythonVersion,
+    SerializationFormat, UnsafeFixes,
 };
 use ruff_linter::{RuleParser, RuleSelector, RuleSelectorParser};
 use ruff_workspace::configuration::{Configuration, RuleSelection};
@@ -351,6 +351,9 @@ pub struct CheckCommand {
         conflicts_with = "watch",
     )]
     pub show_settings: bool,
+    /// List of mappings from file extension to language (one of ["python", "ipynb", "pyi"]).
+    #[arg(long, value_delimiter = ',', hide = true)]
+    pub extension_override: Option<Vec<ExtensionPair>>,
     /// Dev-only argument to show fixes
     #[arg(long, hide = true)]
     pub ecosystem_ci: bool,
@@ -500,6 +503,7 @@ impl CheckCommand {
                 statistics: self.statistics,
                 stdin_filename: self.stdin_filename,
                 watch: self.watch,
+                extension_override: self.extension_override,
             },
             CliOverrides {
                 dummy_variable_rgx: self.dummy_variable_rgx,
@@ -598,6 +602,7 @@ pub struct CheckArguments {
     pub statistics: bool,
     pub stdin_filename: Option<PathBuf>,
     pub watch: bool,
+    pub extension_override: Option<Vec<ExtensionPair>>,
 }
 
 /// CLI settings that are distinct from configuration (commands, lists of files,

crates/ruff_cli/src/cache.rs

@@ -564,6 +564,7 @@ mod tests {
     use anyhow::Result;
     use filetime::{set_file_mtime, FileTime};
     use itertools::Itertools;
+    use rustc_hash::FxHashMap;
     use test_case::test_case;
 
     use ruff_cache::CACHE_DIR_NAME;
@@ -627,6 +628,7 @@ mod tests {
                     flags::Noqa::Enabled,
                     flags::FixMode::Generate,
                     UnsafeFixes::Enabled,
+                    &FxHashMap::default(),
                 )
                 .unwrap();
                 if diagnostics
@@ -673,6 +675,7 @@ mod tests {
                 flags::Noqa::Enabled,
                 flags::FixMode::Generate,
                 UnsafeFixes::Enabled,
+                &FxHashMap::default(),
             )
             .unwrap();
         }
@@ -1046,6 +1049,7 @@ mod tests {
                 flags::Noqa::Enabled,
                 flags::FixMode::Generate,
                 UnsafeFixes::Enabled,
+                &FxHashMap::default(),
             )
         }
 

crates/ruff_cli/src/commands/check.rs

@@ -14,7 +14,7 @@ use rustc_hash::FxHashMap;
 use ruff_diagnostics::Diagnostic;
 use ruff_linter::message::Message;
 use ruff_linter::registry::Rule;
-use ruff_linter::settings::types::UnsafeFixes;
+use ruff_linter::settings::types::{Language, UnsafeFixes};
 use ruff_linter::settings::{flags, LinterSettings};
 use ruff_linter::{fs, warn_user_once, IOError};
 use ruff_python_ast::imports::ImportMap;
@@ -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,
+    extension_override: &FxHashMap<String, Language>,
 ) -> 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,
+                    extension_override,
                 )
                 .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,
+    extension_override: &FxHashMap<String, Language>,
 ) -> 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,
+            extension_override,
+        )
     });
 
     match result {
@@ -280,6 +294,7 @@ mod test {
             flags::Noqa::Disabled,
             flags::FixMode::Generate,
             UnsafeFixes::Enabled,
+            &FxHashMap::default(),
         )
         .unwrap();
         let mut output = Vec::new();

crates/ruff_cli/src/commands/check_stdin.rs

@@ -4,7 +4,9 @@ use anyhow::Result;
 
 use ruff_linter::packaging;
 use ruff_linter::settings::flags;
+use ruff_linter::settings::types::Language;
 use ruff_workspace::resolver::{match_exclusion, python_file_at_path, PyprojectConfig};
+use rustc_hash::FxHashMap;
 
 use crate::args::CliOverrides;
 use crate::diagnostics::{lint_stdin, Diagnostics};
@@ -17,6 +19,7 @@ pub(crate) fn check_stdin(
     overrides: &CliOverrides,
     noqa: flags::Noqa,
     fix_mode: flags::FixMode,
+    extension_override: &FxHashMap<String, Language>,
 ) -> Result<Diagnostics> {
     if let Some(filename) = filename {
         if !python_file_at_path(filename, pyproject_config, overrides)? {
@@ -42,6 +45,7 @@ pub(crate) fn check_stdin(
         &pyproject_config.settings,
         noqa,
         fix_mode,
+        extension_override,
     )?;
     diagnostics.messages.sort_unstable();
     Ok(diagnostics)

crates/ruff_cli/src/diagnostics.rs

@@ -17,13 +17,13 @@ use ruff_linter::logging::DisplayParseError;
 use ruff_linter::message::Message;
 use ruff_linter::pyproject_toml::lint_pyproject_toml;
 use ruff_linter::registry::AsRule;
-use ruff_linter::settings::types::UnsafeFixes;
+use ruff_linter::settings::types::{Language, UnsafeFixes};
 use ruff_linter::settings::{flags, LinterSettings};
 use ruff_linter::source_kind::{SourceError, SourceKind};
 use ruff_linter::{fs, IOError, SyntaxError};
 use ruff_notebook::{Notebook, NotebookError, NotebookIndex};
 use ruff_python_ast::imports::ImportMap;
-use ruff_python_ast::{SourceType, TomlSourceType};
+use ruff_python_ast::{PySourceType, SourceType, TomlSourceType};
 use ruff_source_file::{LineIndex, SourceCode, SourceFileBuilder};
 use ruff_text_size::{TextRange, TextSize};
 use ruff_workspace::Settings;
@@ -177,7 +177,23 @@ impl AddAssign for FixMap {
     }
 }
 
+fn get_override_source_type(
+    path: Option<&Path>,
+    extension: &FxHashMap<String, Language>,
+) -> Option<PySourceType> {
+    let Some(ext) = path.and_then(|p| p.extension()).and_then(|p| p.to_str()) else {
+        return None;
+    };
+    match extension.get(ext) {
+        Some(Language::Python) => Some(PySourceType::Python),
+        Some(Language::Ipynb) => Some(PySourceType::Ipynb),
+        Some(Language::Pyi) => Some(PySourceType::Stub),
+        None => None,
+    }
+}
+
 /// 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 +202,7 @@ pub(crate) fn lint_path(
     noqa: flags::Noqa,
     fix_mode: flags::FixMode,
     unsafe_fixes: UnsafeFixes,
+    extension_override: &FxHashMap<String, Language>,
 ) -> Result<Diagnostics> {
     // Check the cache.
     let caching = match cache {
@@ -220,32 +237,35 @@ pub(crate) fn lint_path(
     };
 
     debug!("Checking: {}", path.display());
-
-    let source_type = match SourceType::from(path) {
-        SourceType::Toml(TomlSourceType::Pyproject) => {
-            let messages = if settings
-                .rules
-                .iter_enabled()
-                .any(|rule_code| rule_code.lint_source().is_pyproject_toml())
-            {
-                let contents = match std::fs::read_to_string(path).map_err(SourceError::from) {
-                    Ok(contents) => contents,
-                    Err(err) => {
-                        return Ok(Diagnostics::from_source_error(&err, Some(path), settings));
-                    }
+    let source_type = match get_override_source_type(Some(path), extension_override) {
+        Some(source_type) => source_type,
+        None => match SourceType::from(path) {
+            SourceType::Toml(TomlSourceType::Pyproject) => {
+                let messages = if settings
+                    .rules
+                    .iter_enabled()
+                    .any(|rule_code| rule_code.lint_source().is_pyproject_toml())
+                {
+                    let contents = match std::fs::read_to_string(path).map_err(SourceError::from) {
+                        Ok(contents) => contents,
+                        Err(err) => {
+                            return Ok(Diagnostics::from_source_error(&err, Some(path), settings));
+                        }
+                    };
+                    let source_file =
+                        SourceFileBuilder::new(path.to_string_lossy(), contents).finish();
+                    lint_pyproject_toml(source_file, settings)
+                } else {
+                    vec![]
                 };
-                let source_file = SourceFileBuilder::new(path.to_string_lossy(), contents).finish();
-                lint_pyproject_toml(source_file, settings)
-            } else {
-                vec![]
-            };
-            return Ok(Diagnostics {
-                messages,
-                ..Diagnostics::default()
-            });
-        }
-        SourceType::Toml(_) => return Ok(Diagnostics::default()),
-        SourceType::Python(source_type) => source_type,
+                return Ok(Diagnostics {
+                    messages,
+                    ..Diagnostics::default()
+                });
+            }
+            SourceType::Toml(_) => return Ok(Diagnostics::default()),
+            SourceType::Python(source_type) => source_type,
+        },
     };
 
     // Extract the sources from the file.
@@ -368,10 +388,17 @@ pub(crate) fn lint_stdin(
     settings: &Settings,
     noqa: flags::Noqa,
     fix_mode: flags::FixMode,
+    extension_override: &FxHashMap<String, Language>,
 ) -> 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 = if let Some(source_type) = get_override_source_type(path, extension_override)
+    {
+        source_type
+    } else {
+        let SourceType::Python(source_type) = path.map(SourceType::from).unwrap_or_default() else {
+            return Ok(Diagnostics::default());
+        };
+        source_type
     };
 
     // Extract the sources from the file.

crates/ruff_cli/src/lib.rs

@@ -17,6 +17,7 @@ use ruff_linter::settings::flags::FixMode;
 use ruff_linter::settings::types::SerializationFormat;
 use ruff_linter::{fs, warn_user, warn_user_once};
 use ruff_workspace::Settings;
+use rustc_hash::FxHashMap;
 
 use crate::args::{Args, CheckCommand, Command, FormatCommand, HelpFormat};
 use crate::printer::{Flags as PrinterFlags, Printer};
@@ -314,6 +315,11 @@ pub fn check(args: CheckCommand, log_level: LogLevel) -> Result<ExitStatus> {
         unsafe_fixes,
         printer_flags,
     );
+    let extension_override = cli
+        .extension_override
+        .map(|eo| eo.into_iter().map(|y| (y.extension, y.language)))
+        .map(FxHashMap::from_iter)
+        .unwrap_or_default();
 
     if cli.watch {
         if output_format != SerializationFormat::Text {
@@ -342,6 +348,7 @@ pub fn check(args: CheckCommand, log_level: LogLevel) -> Result<ExitStatus> {
             noqa.into(),
             fix_mode,
             unsafe_fixes,
+            &extension_override,
         )?;
         printer.write_continuously(&mut writer, &messages)?;
 
@@ -375,6 +382,7 @@ pub fn check(args: CheckCommand, log_level: LogLevel) -> Result<ExitStatus> {
                         noqa.into(),
                         fix_mode,
                         unsafe_fixes,
+                        &extension_override,
                     )?;
                     printer.write_continuously(&mut writer, &messages)?;
                 }
@@ -392,6 +400,7 @@ pub fn check(args: CheckCommand, log_level: LogLevel) -> Result<ExitStatus> {
                 &overrides,
                 noqa.into(),
                 fix_mode,
+                &extension_override,
             )?
         } else {
             commands::check::check(
@@ -402,6 +411,7 @@ pub fn check(args: CheckCommand, log_level: LogLevel) -> Result<ExitStatus> {
                 noqa.into(),
                 fix_mode,
                 unsafe_fixes,
+                &extension_override,
             )?
         };