Enable prefix-based check code selection (#493)

Cargo.lock

@@ -466,6 +466,15 @@ dependencies = [
  "winapi 0.3.9",
 ]
 
+[[package]]
+name = "codegen"
+version = "0.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ff61280aed771c3070e7dcc9e050c66f1eb1e3b96431ba66f9f74641d02fc41d"
+dependencies = [
+ "indexmap",
+]
+
 [[package]]
 name = "codespan-reporting"
 version = "0.11.1"
@@ -2054,6 +2063,7 @@ dependencies = [
  "chrono",
  "clap",
  "clearscreen",
+ "codegen",
  "colored",
  "common-path",
  "dirs 4.0.0",

Cargo.toml

@@ -50,6 +50,7 @@ getrandom = { version = "0.2.7", features = ["js"] }
 
 [dev-dependencies]
 assert_cmd = { version = "2.0.4" }
+codegen = { version = "0.2.0" }
 insta = { version = "1.19.1", features = ["yaml"] }
 test-case = { version = "2.2.2" }
 

examples/generate_check_code_prefix.rs

@@ -0,0 +1,105 @@
+//! Generate the CheckCodePrefix enum.
+
+use std::collections::{BTreeMap, BTreeSet};
+
+use codegen::{Scope, Type, Variant};
+use itertools::Itertools;
+use strum::IntoEnumIterator;
+
+use ruff::checks::CheckCode;
+
+fn main() {
+    // Build up a map from prefix to matching CheckCodes.
+    let mut prefix_to_codes: BTreeMap<String, BTreeSet<CheckCode>> = Default::default();
+    for check_code in CheckCode::iter() {
+        let as_ref = check_code.as_ref().to_string();
+        for i in 1..=as_ref.len() {
+            let prefix = as_ref[..i].to_string();
+            let entry = prefix_to_codes
+                .entry(prefix)
+                .or_insert_with(|| Default::default());
+            entry.insert(check_code.clone());
+        }
+    }
+
+    let mut scope = Scope::new();
+
+    // Create the `CheckCodePrefix` definition.
+    let mut gen = scope
+        .new_enum("CheckCodePrefix")
+        .vis("pub")
+        .derive("EnumString")
+        .derive("Debug")
+        .derive("PartialEq")
+        .derive("Eq")
+        .derive("Clone")
+        .derive("Serialize")
+        .derive("Deserialize");
+    for (prefix, _) in &prefix_to_codes {
+        gen = gen.push_variant(Variant::new(prefix.to_string()));
+    }
+
+    // Create the `PrefixSpecificity` definition.
+    scope
+        .new_enum("PrefixSpecificity")
+        .vis("pub")
+        .derive("PartialEq")
+        .derive("Eq")
+        .derive("PartialOrd")
+        .derive("Ord")
+        .push_variant(Variant::new("Category"))
+        .push_variant(Variant::new("Hundreds"))
+        .push_variant(Variant::new("Tens"))
+        .push_variant(Variant::new("Explicit"));
+
+    // Create the `match` statement, to map from definition to relevant codes.
+    let mut gen = scope
+        .new_impl("CheckCodePrefix")
+        .new_fn("codes")
+        .arg_ref_self()
+        .ret(Type::new("Vec<CheckCode>"))
+        .vis("pub")
+        .line("match self {");
+    for (prefix, codes) in &prefix_to_codes {
+        gen = gen.line(format!(
+            "CheckCodePrefix::{prefix} => vec![{}],",
+            codes
+                .iter()
+                .map(|code| format!("CheckCode::{}", code.as_ref()))
+                .join(", ")
+        ));
+    }
+    gen.line("}");
+
+    // Create the `match` statement, to map from definition to specificity.
+    let mut gen = scope
+        .new_impl("CheckCodePrefix")
+        .new_fn("specificity")
+        .arg_ref_self()
+        .ret(Type::new("PrefixSpecificity"))
+        .vis("pub")
+        .line("match self {");
+    for (prefix, _) in &prefix_to_codes {
+        let specificity = match prefix.len() {
+            4 => "Explicit",
+            3 => "Tens",
+            2 => "Hundreds",
+            1 => "Category",
+            _ => panic!("Invalid prefix: {}", prefix),
+        };
+        gen = gen.line(format!(
+            "CheckCodePrefix::{prefix} => PrefixSpecificity::{},",
+            specificity
+        ));
+    }
+    gen.line("}");
+
+    println!("//! File automatically generated by examples/generate_check_code_prefix.rs.");
+    println!();
+    println!("use serde::{{Deserialize, Serialize}};");
+    println!("use strum_macros::EnumString;");
+    println!();
+    println!("use crate::checks::CheckCode;");
+    println!();
+    println!("{}", scope.to_string());
+}

src/cli.rs

@@ -5,7 +5,7 @@ use clap::{command, Parser};
 use log::warn;
 use regex::Regex;
 
-use crate::checks::CheckCode;
+use crate::checks_gen::CheckCodePrefix;
 use crate::printer::SerializationFormat;
 use crate::pyproject::StrCheckCodePair;
 use crate::settings::PythonVersion;
@@ -43,16 +43,16 @@ pub struct Cli {
     pub no_cache: bool,
     /// List of error codes to enable.
     #[arg(long, value_delimiter = ',')]
-    pub select: Vec<CheckCode>,
+    pub select: Vec<CheckCodePrefix>,
     /// Like --select, but adds additional error codes on top of the selected ones.
     #[arg(long, value_delimiter = ',')]
-    pub extend_select: Vec<CheckCode>,
+    pub extend_select: Vec<CheckCodePrefix>,
     /// List of error codes to ignore.
     #[arg(long, value_delimiter = ',')]
-    pub ignore: Vec<CheckCode>,
+    pub ignore: Vec<CheckCodePrefix>,
     /// Like --ignore, but adds additional error codes on top of the ignored ones.
     #[arg(long, value_delimiter = ',')]
-    pub extend_ignore: Vec<CheckCode>,
+    pub extend_ignore: Vec<CheckCodePrefix>,
     /// List of paths, used to exclude files and/or directories from checks.
     #[arg(long, value_delimiter = ',')]
     pub exclude: Vec<String>,
@@ -106,9 +106,9 @@ impl fmt::Display for Warnable {
 /// Warn the user if they attempt to enable a code that won't be respected.
 pub fn warn_on(
     flag: Warnable,
-    codes: &[CheckCode],
-    cli_ignore: &[CheckCode],
-    cli_extend_ignore: &[CheckCode],
+    codes: &[CheckCodePrefix],
+    cli_ignore: &[CheckCodePrefix],
+    cli_extend_ignore: &[CheckCodePrefix],
     pyproject_settings: &RawSettings,
     pyproject_path: &Option<PathBuf>,
 ) {

src/fs.rs

@@ -133,7 +133,7 @@ pub fn ignores_from_path<'a>(
                 [&pattern_code_pair.pattern].into_iter(),
             )
         })
-        .map(|pattern_code_pair| &pattern_code_pair.code)
+        .flat_map(|pattern_code_pair| &pattern_code_pair.codes)
         .collect())
 }
 

src/lib.rs

@@ -18,6 +18,7 @@ pub mod check_ast;
 mod check_lines;
 mod check_tokens;
 pub mod checks;
+mod checks_gen;
 pub mod cli;
 pub mod code_gen;
 mod cst;

src/pyproject.rs

@@ -7,7 +7,7 @@ use path_absolutize::Absolutize;
 use serde::de;
 use serde::{Deserialize, Deserializer};
 
-use crate::checks::CheckCode;
+use crate::checks_gen::CheckCodePrefix;
 use crate::settings::PythonVersion;
 use crate::{flake8_quotes, fs};
 
@@ -34,13 +34,13 @@ pub struct Config {
     pub exclude: Option<Vec<String>>,
     #[serde(default)]
     pub extend_exclude: Vec<String>,
-    pub select: Option<Vec<CheckCode>>,
+    pub select: Option<Vec<CheckCodePrefix>>,
     #[serde(default)]
-    pub extend_select: Vec<CheckCode>,
+    pub extend_select: Vec<CheckCodePrefix>,
     #[serde(default)]
-    pub ignore: Vec<CheckCode>,
+    pub ignore: Vec<CheckCodePrefix>,
     #[serde(default)]
-    pub extend_ignore: Vec<CheckCode>,
+    pub extend_ignore: Vec<CheckCodePrefix>,
     #[serde(default)]
     pub per_file_ignores: Vec<StrCheckCodePair>,
     pub dummy_variable_rgx: Option<String>,
@@ -51,7 +51,7 @@ pub struct Config {
 #[derive(Clone, Debug, PartialEq, Eq)]
 pub struct StrCheckCodePair {
     pub pattern: String,
-    pub code: CheckCode,
+    pub code: CheckCodePrefix,
 }
 
 impl StrCheckCodePair {
@@ -84,7 +84,7 @@ impl FromStr for StrCheckCodePair {
             }
             (tokens[0].trim(), tokens[1].trim())
         };
-        let code = CheckCode::from_str(code_string)?;
+        let code = CheckCodePrefix::from_str(code_string)?;
         let pattern = pattern_str.into();
         Ok(Self { pattern, code })
     }
@@ -157,7 +157,7 @@ mod tests {
 
     use anyhow::Result;
 
-    use crate::checks::CheckCode;
+    use crate::checks_gen::CheckCodePrefix;
     use crate::flake8_quotes;
     use crate::flake8_quotes::settings::Quote;
     use crate::pyproject::{
@@ -269,7 +269,7 @@ select = ["E501"]
                     line_length: None,
                     exclude: None,
                     extend_exclude: vec![],
-                    select: Some(vec![CheckCode::E501]),
+                    select: Some(vec![CheckCodePrefix::E501]),
                     extend_select: vec![],
                     ignore: vec![],
                     extend_ignore: vec![],
@@ -297,8 +297,8 @@ ignore = ["E501"]
                     exclude: None,
                     extend_exclude: vec![],
                     select: None,
-                    extend_select: vec![CheckCode::M001],
-                    ignore: vec![CheckCode::E501],
+                    extend_select: vec![CheckCodePrefix::M001],
+                    ignore: vec![CheckCodePrefix::E501],
                     extend_ignore: vec![],
                     per_file_ignores: vec![],
                     dummy_variable_rgx: None,
@@ -372,7 +372,7 @@ other-attribute = 1
                 extend_ignore: vec![],
                 per_file_ignores: vec![StrCheckCodePair {
                     pattern: "__init__.py".to_string(),
-                    code: CheckCode::F401
+                    code: CheckCodePrefix::F401
                 }],
                 dummy_variable_rgx: None,
                 target_version: None,

src/settings.rs

@@ -8,9 +8,9 @@ use glob::Pattern;
 use once_cell::sync::Lazy;
 use regex::Regex;
 use serde::{Deserialize, Serialize};
-use strum::IntoEnumIterator;
 
-use crate::checks::{CheckCategory, CheckCode};
+use crate::checks::CheckCode;
+use crate::checks_gen::{CheckCodePrefix, PrefixSpecificity};
 use crate::pyproject::{load_config, StrCheckCodePair};
 use crate::{flake8_quotes, fs};
 
@@ -74,14 +74,14 @@ impl FilePattern {
 #[derive(Debug, Clone, Hash)]
 pub struct PerFileIgnore {
     pub pattern: FilePattern,
-    pub code: CheckCode,
+    pub codes: BTreeSet<CheckCode>,
 }
 
 impl PerFileIgnore {
     pub fn new(user_in: StrCheckCodePair, project_root: &Option<PathBuf>) -> Self {
         let pattern = FilePattern::from_user(user_in.pattern.as_str(), project_root);
-        let code = user_in.code;
-        Self { pattern, code }
+        let codes = BTreeSet::from_iter(user_in.code.codes());
+        Self { pattern, codes }
     }
 }
 
@@ -90,12 +90,12 @@ pub struct RawSettings {
     pub dummy_variable_rgx: Regex,
     pub exclude: Vec<FilePattern>,
     pub extend_exclude: Vec<FilePattern>,
-    pub extend_ignore: Vec<CheckCode>,
-    pub extend_select: Vec<CheckCode>,
-    pub ignore: Vec<CheckCode>,
+    pub extend_ignore: Vec<CheckCodePrefix>,
+    pub extend_select: Vec<CheckCodePrefix>,
+    pub ignore: Vec<CheckCodePrefix>,
     pub line_length: usize,
     pub per_file_ignores: Vec<PerFileIgnore>,
-    pub select: Vec<CheckCode>,
+    pub select: Vec<CheckCodePrefix>,
     pub target_version: PythonVersion,
     // Plugins
     pub flake8_quotes: flake8_quotes::settings::Settings,
@@ -157,16 +157,9 @@ impl RawSettings {
                 .map(|path| FilePattern::from_user(path, project_root))
                 .collect(),
             extend_ignore: config.extend_ignore,
-            select: config.select.unwrap_or_else(|| {
-                CheckCode::iter()
-                    .filter(|code| {
-                        matches!(
-                            code.category(),
-                            CheckCategory::PycodestyleError | CheckCategory::Pyflakes
-                        )
-                    })
-                    .collect()
-            }),
+            select: config
+                .select
+                .unwrap_or_else(|| vec![CheckCodePrefix::E, CheckCodePrefix::F]),
             extend_select: config.extend_select,
             ignore: config.ignore,
             line_length: config.line_length.unwrap_or(88),
@@ -197,21 +190,58 @@ pub struct Settings {
     pub flake8_quotes: flake8_quotes::settings::Settings,
 }
 
+/// Given a set of selected and ignored prefixes, resolve the set of enabled error codes.
+fn resolve_codes(
+    select: &[CheckCodePrefix],
+    extend_select: &[CheckCodePrefix],
+    ignore: &[CheckCodePrefix],
+    extend_ignore: &[CheckCodePrefix],
+) -> BTreeSet<CheckCode> {
+    let mut codes: BTreeSet<CheckCode> = BTreeSet::new();
+    for specificity in [
+        PrefixSpecificity::Category,
+        PrefixSpecificity::Hundreds,
+        PrefixSpecificity::Tens,
+        PrefixSpecificity::Explicit,
+    ] {
+        for prefix in select {
+            if prefix.specificity() == specificity {
+                codes.extend(prefix.codes());
+            }
+        }
+        for prefix in extend_select {
+            if prefix.specificity() == specificity {
+                codes.extend(prefix.codes());
+            }
+        }
+        for prefix in ignore {
+            if prefix.specificity() == specificity {
+                for code in prefix.codes() {
+                    codes.remove(&code);
+                }
+            }
+        }
+        for prefix in extend_ignore {
+            if prefix.specificity() == specificity {
+                for code in prefix.codes() {
+                    codes.remove(&code);
+                }
+            }
+        }
+    }
+    codes
+}
+
 impl Settings {
     pub fn from_raw(settings: RawSettings) -> Self {
-        // Materialize the set of enabled CheckCodes.
-        let mut enabled: BTreeSet<CheckCode> = BTreeSet::new();
-        enabled.extend(settings.select);
-        enabled.extend(settings.extend_select);
-        for code in &settings.ignore {
-            enabled.remove(code);
-        }
-        for code in &settings.extend_ignore {
-            enabled.remove(code);
-        }
         Self {
             dummy_variable_rgx: settings.dummy_variable_rgx,
-            enabled,
+            enabled: resolve_codes(
+                &settings.select,
+                &settings.extend_select,
+                &settings.ignore,
+                &settings.extend_ignore,
+            ),
             exclude: settings.exclude,
             extend_exclude: settings.extend_exclude,
             flake8_quotes: settings.flake8_quotes,
@@ -290,12 +320,12 @@ pub struct CurrentSettings {
     pub dummy_variable_rgx: Regex,
     pub exclude: Vec<Exclusion>,
     pub extend_exclude: Vec<Exclusion>,
-    pub extend_ignore: Vec<CheckCode>,
-    pub extend_select: Vec<CheckCode>,
-    pub ignore: Vec<CheckCode>,
+    pub extend_ignore: Vec<CheckCodePrefix>,
+    pub extend_select: Vec<CheckCodePrefix>,
+    pub ignore: Vec<CheckCodePrefix>,
     pub line_length: usize,
     pub per_file_ignores: Vec<PerFileIgnore>,
-    pub select: Vec<CheckCode>,
+    pub select: Vec<CheckCodePrefix>,
     pub target_version: PythonVersion,
     // Plugins
     pub flake8_quotes: flake8_quotes::settings::Settings,