perf(dict): Compare map to trie

This commit is contained in:
Ed Page 2024-12-24 21:34:34 -06:00
parent 661825438c
commit fad1637b6c
5 changed files with 371740 additions and 14 deletions

View file

@ -24,7 +24,7 @@ itertools = "0.13"
edit-distance = "2.1"
unicase = "2.7"
codegenrs = "3.0"
dictgen = { version = "^0.2", path = "../dictgen", features = ["codegen"] }
dictgen = { version = "^0.2", path = "../dictgen", features = ["codegen", "map"] }
varcon = { version = "^1.0", path = "../varcon" }
snapbox = "0.6.5"
indexmap = "2.2.6"

View file

@ -1,19 +1,37 @@
#![allow(clippy::wildcard_imports)]
const MISS: &str = "finalizes";
const HIT: &str = "finallizes";
mod map_codegen;
mod trie_codegen;
mod trie {
mod miss {
use super::*;
const MISS: &str = "finalizes";
#[divan::bench(args = [unicase::UniCase::new(MISS)])]
fn miss(word: unicase::UniCase<&str>) -> Option<&'static &[&str]> {
typos_dict::WORD_TRIE.find(&word)
fn map(word: unicase::UniCase<&str>) -> Option<&'static &[&str]> {
map_codegen::WORD.find(&word)
}
#[divan::bench(args = [unicase::UniCase::new(MISS)])]
fn trie(word: unicase::UniCase<&str>) -> Option<&'static &[&str]> {
trie_codegen::WORD_TRIE.find(&word)
}
}
mod hit {
use super::*;
const HIT: &str = "finallizes";
#[divan::bench(args = [unicase::UniCase::new(HIT)])]
fn map(word: unicase::UniCase<&str>) -> Option<&'static &[&str]> {
map_codegen::WORD.find(&word)
}
#[divan::bench(args = [unicase::UniCase::new(HIT)])]
fn hit(word: unicase::UniCase<&str>) -> Option<&'static &[&str]> {
typos_dict::WORD_TRIE.find(&word)
fn trie(word: unicase::UniCase<&str>) -> Option<&'static &[&str]> {
trie_codegen::WORD_TRIE.find(&word)
}
}

File diff suppressed because it is too large Load diff

File diff suppressed because it is too large Load diff

View file

@ -1,15 +1,32 @@
#[test]
fn codegen() {
let mut content = vec![];
const DICT: &[u8] = include_bytes!("../assets/words.csv");
generate(&mut content, "WORD", DICT);
let content = String::from_utf8(content).unwrap();
let content = codegenrs::rustfmt(&content, None).unwrap();
snapbox::assert_data_eq!(content, snapbox::file!["../src/word_codegen.rs"].raw());
let mut trie_content = vec![];
generate_trie(&mut trie_content, "WORD", DICT);
let trie_content = String::from_utf8(trie_content).unwrap();
let trie_content = codegenrs::rustfmt(&trie_content, None).unwrap();
snapbox::assert_data_eq!(
&trie_content,
snapbox::file!["../benches/benches/trie_codegen.rs"].raw()
);
let mut map_content = vec![];
generate_map(&mut map_content, "WORD", DICT);
let map_content = String::from_utf8(map_content).unwrap();
let map_content = codegenrs::rustfmt(&map_content, None).unwrap();
snapbox::assert_data_eq!(
&map_content,
snapbox::file!["../benches/benches/map_codegen.rs"].raw()
);
snapbox::assert_data_eq!(
&trie_content,
snapbox::file!["../src/word_codegen.rs"].raw()
);
}
fn generate<W: std::io::Write>(file: &mut W, prefix: &str, dict: &[u8]) {
fn generate_trie<W: std::io::Write>(file: &mut W, prefix: &str, dict: &[u8]) {
writeln!(
file,
"// This file is @generated by {}",
@ -44,3 +61,38 @@ fn generate<W: std::io::Write>(file: &mut W, prefix: &str, dict: &[u8]) {
)
.unwrap();
}
fn generate_map<W: std::io::Write>(file: &mut W, prefix: &str, dict: &[u8]) {
writeln!(
file,
"// This file is @generated by {}",
file!().replace('\\', "/")
)
.unwrap();
writeln!(file, "#![allow(clippy::unreadable_literal)]",).unwrap();
writeln!(file, "#![allow(unreachable_pub)]",).unwrap();
writeln!(file).unwrap();
let records: Vec<_> = csv::ReaderBuilder::new()
.has_headers(false)
.flexible(true)
.from_reader(dict)
.records()
.map(|r| r.unwrap())
.collect();
dictgen::generate_map(
file,
prefix,
"&'static [&'static str]",
records.iter().map(|record| {
let mut record_fields = record.iter();
let key = record_fields.next().unwrap();
let value = format!(
"&[{}]",
itertools::join(record_fields.map(|field| format!(r#""{field}""#)), ", ")
);
(key, value)
}),
)
.unwrap();
}