Use FNV hasher in more places (#732)

Cargo.lock

@@ -935,6 +935,7 @@ dependencies = [
  "anyhow",
  "clap 4.0.22",
  "configparser",
+ "fnv",
  "once_cell",
  "regex",
  "ruff",

flake8_to_ruff/Cargo.toml

@@ -10,6 +10,7 @@ name = "flake8_to_ruff"
 anyhow = { version = "1.0.66" }
 clap = { version = "4.0.1", features = ["derive"] }
 configparser = { version = "3.0.2" }
+fnv = { version = "1.0.7" }
 once_cell = { version = "1.16.0" }
 regex = { version = "1.6.0" }
 ruff = { path = "..", default-features = false }

flake8_to_ruff/src/parser.rs

@@ -1,7 +1,7 @@
-use std::collections::BTreeMap;
 use std::str::FromStr;
 
 use anyhow::Result;
+use fnv::FnvHashMap;
 use once_cell::sync::Lazy;
 use regex::Regex;
 use ruff::checks_gen::CheckCodePrefix;
@@ -179,8 +179,8 @@ pub fn parse_files_to_codes_mapping(value: &str) -> Result<Vec<PatternPrefixPair
 /// Collect a list of `PatternPrefixPair` structs as a `BTreeMap`.
 pub fn collect_per_file_ignores(
     pairs: Vec<PatternPrefixPair>,
-) -> BTreeMap<String, Vec<CheckCodePrefix>> {
+) -> FnvHashMap<String, Vec<CheckCodePrefix>> {
-    let mut per_file_ignores: BTreeMap<String, Vec<CheckCodePrefix>> = BTreeMap::new();
+    let mut per_file_ignores: FnvHashMap<String, Vec<CheckCodePrefix>> = FnvHashMap::default();
     for pair in pairs {
         per_file_ignores
             .entry(pair.pattern)

src/autofix/fixer.rs

@@ -61,7 +61,7 @@ fn apply_fixes<'a>(
 ) -> Cow<'a, str> {
     let mut output = RopeBuilder::new();
     let mut last_pos: Location = Location::new(1, 0);
-    let mut applied: BTreeSet<&Patch> = BTreeSet::new();
+    let mut applied: BTreeSet<&Patch> = BTreeSet::default();
 
     for fix in fixes.sorted_by_key(|fix| fix.patch.location) {
         // If we already applied an identical fix as part of another correction, skip

src/cli.rs

@@ -1,8 +1,8 @@
-use std::collections::BTreeMap;
 use std::fmt;
 use std::path::PathBuf;
 
 use clap::{command, Parser};
+use fnv::FnvHashMap;
 use log::warn;
 use regex::Regex;
 
@@ -188,7 +188,7 @@ pub fn collect_per_file_ignores(
     pairs: Vec<PatternPrefixPair>,
     project_root: &Option<PathBuf>,
 ) -> Vec<PerFileIgnore> {
-    let mut per_file_ignores: BTreeMap<String, Vec<CheckCodePrefix>> = BTreeMap::new();
+    let mut per_file_ignores: FnvHashMap<String, Vec<CheckCodePrefix>> = FnvHashMap::default();
     for pair in pairs {
         per_file_ignores
             .entry(pair.pattern)

src/docstrings/google.rs

@@ -1,11 +1,10 @@
 //! Abstractions for Google-style docstrings.
 
-use std::collections::BTreeSet;
+use fnv::FnvHashSet;
-
 use once_cell::sync::Lazy;
 
-pub(crate) static GOOGLE_SECTION_NAMES: Lazy<BTreeSet<&'static str>> = Lazy::new(|| {
+pub(crate) static GOOGLE_SECTION_NAMES: Lazy<FnvHashSet<&'static str>> = Lazy::new(|| {
-    BTreeSet::from([
+    FnvHashSet::from_iter([
         "Args",
         "Arguments",
         "Attention",
@@ -37,8 +36,9 @@ pub(crate) static GOOGLE_SECTION_NAMES: Lazy<BTreeSet<&'static str>> = Lazy::new
     ])
 });
 
-pub(crate) static LOWERCASE_GOOGLE_SECTION_NAMES: Lazy<BTreeSet<&'static str>> = Lazy::new(|| {
+pub(crate) static LOWERCASE_GOOGLE_SECTION_NAMES: Lazy<FnvHashSet<&'static str>> =
-    BTreeSet::from([
+    Lazy::new(|| {
+        FnvHashSet::from_iter([
             "args",
             "arguments",
             "attention",
@@ -68,4 +68,4 @@ pub(crate) static LOWERCASE_GOOGLE_SECTION_NAMES: Lazy<BTreeSet<&'static str>> =
             "yield",
             "yields",
         ])
-});
+    });

src/docstrings/numpy.rs

@@ -1,11 +1,10 @@
 //! Abstractions for NumPy-style docstrings.
 
-use std::collections::BTreeSet;
+use fnv::FnvHashSet;
-
 use once_cell::sync::Lazy;
 
-pub(crate) static LOWERCASE_NUMPY_SECTION_NAMES: Lazy<BTreeSet<&'static str>> = Lazy::new(|| {
+pub(crate) static LOWERCASE_NUMPY_SECTION_NAMES: Lazy<FnvHashSet<&'static str>> = Lazy::new(|| {
-    BTreeSet::from([
+    FnvHashSet::from_iter([
         "short summary",
         "extended summary",
         "parameters",
@@ -22,8 +21,8 @@ pub(crate) static LOWERCASE_NUMPY_SECTION_NAMES: Lazy<BTreeSet<&'static str>> =
     ])
 });
 
-pub(crate) static NUMPY_SECTION_NAMES: Lazy<BTreeSet<&'static str>> = Lazy::new(|| {
+pub(crate) static NUMPY_SECTION_NAMES: Lazy<FnvHashSet<&'static str>> = Lazy::new(|| {
-    BTreeSet::from([
+    FnvHashSet::from_iter([
         "Short Summary",
         "Extended Summary",
         "Parameters",

src/docstrings/styles.rs

@@ -1,5 +1,4 @@
-use std::collections::BTreeSet;
+use fnv::FnvHashSet;
-
 use once_cell::sync::Lazy;
 
 use crate::docstrings::google::{GOOGLE_SECTION_NAMES, LOWERCASE_GOOGLE_SECTION_NAMES};
@@ -11,14 +10,14 @@ pub(crate) enum SectionStyle {
 }
 
 impl SectionStyle {
-    pub(crate) fn section_names(&self) -> &Lazy<BTreeSet<&'static str>> {
+    pub(crate) fn section_names(&self) -> &Lazy<FnvHashSet<&'static str>> {
         match self {
             SectionStyle::NumPy => &NUMPY_SECTION_NAMES,
             SectionStyle::Google => &GOOGLE_SECTION_NAMES,
         }
     }
 
-    pub(crate) fn lowercase_section_names(&self) -> &Lazy<BTreeSet<&'static str>> {
+    pub(crate) fn lowercase_section_names(&self) -> &Lazy<FnvHashSet<&'static str>> {
         match self {
             SectionStyle::NumPy => &LOWERCASE_NUMPY_SECTION_NAMES,
             SectionStyle::Google => &LOWERCASE_GOOGLE_SECTION_NAMES,

src/fs.rs

@@ -1,11 +1,11 @@
 use std::borrow::Cow;
-use std::collections::BTreeSet;
 use std::fs::File;
 use std::io::{BufReader, Read};
 use std::ops::Deref;
 use std::path::{Path, PathBuf};
 
 use anyhow::{anyhow, Result};
+use fnv::FnvHashSet;
 use log::debug;
 use path_absolutize::{path_dedot, Absolutize};
 use walkdir::{DirEntry, WalkDir};
@@ -121,7 +121,7 @@ pub fn iter_python_files<'a>(
 pub(crate) fn ignores_from_path<'a>(
     path: &Path,
     pattern_code_pairs: &'a [PerFileIgnore],
-) -> Result<BTreeSet<&'a CheckCode>> {
+) -> Result<FnvHashSet<&'a CheckCode>> {
     let (file_path, file_basename) = extract_path_names(path)?;
     Ok(pattern_code_pairs
         .iter()

src/pydocstyle/plugins.rs

@@ -1,5 +1,6 @@
 use std::collections::BTreeSet;
 
+use fnv::FnvHashSet;
 use itertools::Itertools;
 use once_cell::sync::Lazy;
 use regex::Regex;
@@ -1287,7 +1288,11 @@ fn common_section(
     blanks_and_section_underline(checker, definition, context);
 }
 
-fn missing_args(checker: &mut Checker, definition: &Definition, docstrings_args: &BTreeSet<&str>) {
+fn missing_args(
+    checker: &mut Checker,
+    definition: &Definition,
+    docstrings_args: &FnvHashSet<&str>,
+) {
     if let DefinitionKind::Function(parent)
     | DefinitionKind::NestedFunction(parent)
     | DefinitionKind::Method(parent) = definition.kind
@@ -1377,7 +1382,7 @@ fn args_section(checker: &mut Checker, definition: &Definition, context: &Sectio
         checker,
         definition,
         // Collect the list of arguments documented in the docstring.
-        &BTreeSet::from_iter(args_sections.iter().filter_map(|section| {
+        &FnvHashSet::from_iter(args_sections.iter().filter_map(|section| {
             match GOOGLE_ARGS_REGEX.captures(section.as_str()) {
                 Some(caps) => caps.get(1).map(|arg_name| arg_name.as_str()),
                 None => None,
@@ -1388,7 +1393,7 @@ fn args_section(checker: &mut Checker, definition: &Definition, context: &Sectio
 
 fn parameters_section(checker: &mut Checker, definition: &Definition, context: &SectionContext) {
     // Collect the list of arguments documented in the docstring.
-    let mut docstring_args: BTreeSet<&str> = Default::default();
+    let mut docstring_args: FnvHashSet<&str> = FnvHashSet::default();
     let section_level_indent = helpers::leading_space(context.line);
     for i in 1..context.following_lines.len() {
         let current_line = context.following_lines[i - 1];

src/pyflakes/checks.rs

@@ -1,5 +1,4 @@
-use std::collections::BTreeSet;
+use fnv::FnvHashSet;
-
 use regex::Regex;
 use rustpython_parser::ast::{
     Arg, Arguments, Constant, Excepthandler, ExcepthandlerKind, Expr, ExprKind, Stmt, StmtKind,
@@ -113,7 +112,7 @@ pub fn duplicate_arguments(arguments: &Arguments) -> Vec<Check> {
     }
 
     // Search for duplicates.
-    let mut idents: BTreeSet<&str> = BTreeSet::new();
+    let mut idents: FnvHashSet<&str> = FnvHashSet::default();
     for arg in all_arguments {
         let ident = &arg.node.arg;
         if idents.contains(ident.as_str()) {

src/python/sys.rs

@@ -1,10 +1,9 @@
-use std::collections::BTreeSet;
+use fnv::FnvHashSet;
-
 use once_cell::sync::Lazy;
 
 // See: https://pycqa.github.io/isort/docs/configuration/options.html#known-standard-library
-pub static KNOWN_STANDARD_LIBRARY: Lazy<BTreeSet<&'static str>> = Lazy::new(|| {
+pub static KNOWN_STANDARD_LIBRARY: Lazy<FnvHashSet<&'static str>> = Lazy::new(|| {
-    BTreeSet::from([
+    FnvHashSet::from_iter([
         "_ast",
         "_dummy_thread",
         "_thread",

src/settings/options.rs

@@ -1,7 +1,6 @@
 //! Options that the user can provide via pyproject.toml.
 
-use std::collections::BTreeMap;
+use fnv::FnvHashMap;
-
 use serde::{Deserialize, Serialize};
 
 use crate::checks_gen::CheckCodePrefix;
@@ -29,5 +28,5 @@ pub struct Options {
     pub isort: Option<isort::settings::Options>,
     pub pep8_naming: Option<pep8_naming::settings::Options>,
     // Tables are required to go last.
-    pub per_file_ignores: Option<BTreeMap<String, Vec<CheckCodePrefix>>>,
+    pub per_file_ignores: Option<FnvHashMap<String, Vec<CheckCodePrefix>>>,
 }

src/settings/pyproject.rs

@@ -96,12 +96,12 @@ pub fn load_options(pyproject: &Option<PathBuf>) -> Result<Options> {
 
 #[cfg(test)]
 mod tests {
-    use std::collections::BTreeMap;
     use std::env::current_dir;
     use std::path::PathBuf;
     use std::str::FromStr;
 
     use anyhow::Result;
+    use fnv::FnvHashMap;
 
     use crate::checks_gen::CheckCodePrefix;
     use crate::flake8_quotes::settings::Quote;
@@ -346,7 +346,7 @@ other-attribute = 1
                 extend_select: None,
                 ignore: None,
                 extend_ignore: None,
-                per_file_ignores: Some(BTreeMap::from([(
+                per_file_ignores: Some(FnvHashMap::from_iter([(
                     "__init__.py".to_string(),
                     vec![CheckCodePrefix::F401]
                 ),])),