Merge pull request #1189 from epage/bench

perf(dict): Add benches

Cargo.lock

@@ -1496,6 +1496,7 @@ dependencies = [
  "codegenrs",
  "csv",
  "dictgen",
+ "divan",
  "edit-distance",
  "indexmap",
  "itertools 0.13.0",

crates/typos-dict/Cargo.toml

@@ -24,10 +24,15 @@ 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"
+divan = "0.1.16"
 
 [lints]
 workspace = true
+
+[[bench]]
+name = "benches"
+harness = false

crates/typos-dict/benches/benches/main.rs

@@ -0,0 +1,51 @@
+#![allow(clippy::wildcard_imports)]
+
+mod map_codegen;
+mod table_codegen;
+mod trie_codegen;
+
+mod miss {
+    use super::*;
+
+    const MISS: &str = "finalizes";
+
+    #[divan::bench(args = [unicase::UniCase::new(MISS)])]
+    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)
+    }
+
+    #[divan::bench(args = [unicase::UniCase::new(MISS)])]
+    fn table(word: unicase::UniCase<&str>) -> Option<&'static &[&str]> {
+        table_codegen::WORD.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 trie(word: unicase::UniCase<&str>) -> Option<&'static &[&str]> {
+        trie_codegen::WORD_TRIE.find(&word)
+    }
+
+    #[divan::bench(args = [unicase::UniCase::new(HIT)])]
+    fn table(word: unicase::UniCase<&str>) -> Option<&'static &[&str]> {
+        table_codegen::WORD.find(&word)
+    }
+}
+
+fn main() {
+    divan::main();
+}

crates/typos-dict/tests/codegen.rs

@@ -1,15 +1,41 @@
 #[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()
+    );
+
+    let mut table_content = vec![];
+    generate_table(&mut table_content, "WORD", DICT);
+    let table_content = String::from_utf8(table_content).unwrap();
+    let table_content = codegenrs::rustfmt(&table_content, None).unwrap();
+    snapbox::assert_data_eq!(
+        &table_content,
+        snapbox::file!["../benches/benches/table_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 +70,73 @@ 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();
+}
+
+fn generate_table<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_table(
+        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();
+}