refactor: Split out typos-dict

Cargo.toml

@@ -1,5 +1,5 @@
 [workspace]
-members = ["typos"]
+members = ["typos", "typos-dict"]
 
 [package]
 name = "typos-cli"
@@ -23,6 +23,7 @@ iterate_unstable = []
 
 [dependencies]
 typos = { version = "0.1", path = "typos" }
+typos-dict = { version = "0.1", path = "typos-dict" }
 failure = "0.1"
 structopt = "0.2"
 clap = "2"

benches/corrections.rs

@@ -4,12 +4,12 @@ extern crate test;
 
 #[bench]
 fn load_corrections(b: &mut test::Bencher) {
-    b.iter(|| typos::BuiltIn::new());
+    b.iter(|| typos_dict::BuiltIn::new());
 }
 
 #[bench]
 fn correct_word_hit(b: &mut test::Bencher) {
-    let corrections = typos::BuiltIn::new();
+    let corrections = typos_dict::BuiltIn::new();
     let input = typos::tokens::Word::new("successs", 0).unwrap();
     assert_eq!(
         corrections.correct_word(input),
@@ -20,7 +20,7 @@ fn correct_word_hit(b: &mut test::Bencher) {
 
 #[bench]
 fn correct_word_miss(b: &mut test::Bencher) {
-    let corrections = typos::BuiltIn::new();
+    let corrections = typos_dict::BuiltIn::new();
     let input = typos::tokens::Word::new("success", 0).unwrap();
     assert_eq!(corrections.correct_word(input), None);
     b.iter(|| corrections.correct_word(input));

benches/data.rs

@@ -28,4 +28,4 @@ fn main() {
 }
 ";
 
-pub const CORPUS: &str = include_str!("../typos/assets/words.csv");
+pub const CORPUS: &str = include_str!("../typos-dict/assets/words.csv");

benches/file.rs

@@ -12,7 +12,7 @@ fn process_empty(b: &mut test::Bencher) {
     let sample_path = temp.child("sample");
     sample_path.write_str(data::EMPTY).unwrap();
 
-    let corrections = typos::BuiltIn::new();
+    let corrections = typos_dict::BuiltIn::new();
     let parser = typos::tokens::Parser::new();
     let checks = typos::checks::CheckSettings::new().build(&corrections, &parser);
     b.iter(|| checks.check_file(sample_path.path(), true, typos::report::print_silent));
@@ -26,7 +26,7 @@ fn process_no_tokens(b: &mut test::Bencher) {
     let sample_path = temp.child("sample");
     sample_path.write_str(data::NO_TOKENS).unwrap();
 
-    let corrections = typos::BuiltIn::new();
+    let corrections = typos_dict::BuiltIn::new();
     let parser = typos::tokens::Parser::new();
     let checks = typos::checks::CheckSettings::new().build(&corrections, &parser);
     b.iter(|| checks.check_file(sample_path.path(), true, typos::report::print_silent));
@@ -40,7 +40,7 @@ fn process_single_token(b: &mut test::Bencher) {
     let sample_path = temp.child("sample");
     sample_path.write_str(data::SINGLE_TOKEN).unwrap();
 
-    let corrections = typos::BuiltIn::new();
+    let corrections = typos_dict::BuiltIn::new();
     let parser = typos::tokens::Parser::new();
     let checks = typos::checks::CheckSettings::new().build(&corrections, &parser);
     b.iter(|| checks.check_file(sample_path.path(), true, typos::report::print_silent));
@@ -54,7 +54,7 @@ fn process_sherlock(b: &mut test::Bencher) {
     let sample_path = temp.child("sample");
     sample_path.write_str(data::SHERLOCK).unwrap();
 
-    let corrections = typos::BuiltIn::new();
+    let corrections = typos_dict::BuiltIn::new();
     let parser = typos::tokens::Parser::new();
     let checks = typos::checks::CheckSettings::new().build(&corrections, &parser);
     b.iter(|| checks.check_file(sample_path.path(), true, typos::report::print_silent));
@@ -68,7 +68,7 @@ fn process_code(b: &mut test::Bencher) {
     let sample_path = temp.child("sample");
     sample_path.write_str(data::CODE).unwrap();
 
-    let corrections = typos::BuiltIn::new();
+    let corrections = typos_dict::BuiltIn::new();
     let parser = typos::tokens::Parser::new();
     let checks = typos::checks::CheckSettings::new().build(&corrections, &parser);
     b.iter(|| checks.check_file(sample_path.path(), true, typos::report::print_silent));
@@ -82,7 +82,7 @@ fn process_corpus(b: &mut test::Bencher) {
     let sample_path = temp.child("sample");
     sample_path.write_str(data::CORPUS).unwrap();
 
-    let corrections = typos::BuiltIn::new();
+    let corrections = typos_dict::BuiltIn::new();
     let parser = typos::tokens::Parser::new();
     let checks = typos::checks::CheckSettings::new().build(&corrections, &parser);
     b.iter(|| checks.check_file(sample_path.path(), true, typos::report::print_silent));

src/main.rs

@@ -310,7 +310,7 @@ fn run() -> Result<i32, failure::Error> {
         config.default.update(&args.overrides);
         let config = config;
 
-        let dictionary = typos::BuiltIn::new();
+        let dictionary = typos_dict::BuiltIn::new();
 
         let parser = typos::tokens::ParserBuilder::new()
             .ignore_hex(config.default.ignore_hex())

typos-dict/Cargo.toml

@@ -0,0 +1,31 @@
+[package]
+name = "typos-dict"
+version = "0.1.0"
+authors = ["Ed Page <[email protected]>"]
+description = "Source Code Spelling Correction"
+repository = "https://github.com/epage/typos"
+documentation = "https://docs.rs/typos-dict"
+readme = "README.md"
+categories = ["development-tools", "text-processing"]
+keywords = ["development", "spelling"]
+license = "MIT"
+edition = "2018"
+
+[badges]
+travis-ci = { repository = "epage/typos" }
+appveyor  = { repository = "epage/typos" }
+
+[features]
+# Support quickly iterating
+iterate_unstable = []
+
+[dependencies]
+typos = { version = "0.1", path = "../typos" }
+phf = { version = "0.7", features = ["unicase"] }
+unicase = "1.1"
+log = "0.4"
+
+[build-dependencies]
+phf_codegen = "0.7"
+csv = "1.0"
+unicase = "1.1"

typos-dict/src/dict.rs

@@ -0,0 +1,97 @@
+use std::borrow::Cow;
+
+use unicase::UniCase;
+
+use typos::tokens::Case;
+
+#[derive(Default)]
+pub struct BuiltIn {}
+
+impl BuiltIn {
+    pub fn new() -> Self {
+        Self {}
+    }
+
+    pub fn correct_ident<'s, 'w>(
+        &'s self,
+        _ident: typos::tokens::Identifier<'w>,
+    ) -> Option<Cow<'s, str>> {
+        None
+    }
+
+    pub fn correct_word<'s, 'w>(&'s self, word: typos::tokens::Word<'w>) -> Option<Cow<'s, str>> {
+        map_lookup(&crate::dict_codegen::WORD_DICTIONARY, word.token())
+            .map(|s| case_correct(s, word.case()))
+    }
+}
+
+impl typos::Dictionary for BuiltIn {
+    fn correct_ident<'s, 'w>(
+        &'s self,
+        ident: typos::tokens::Identifier<'w>,
+    ) -> Option<Cow<'s, str>> {
+        BuiltIn::correct_ident(self, ident)
+    }
+
+    fn correct_word<'s, 'w>(&'s self, word: typos::tokens::Word<'w>) -> Option<Cow<'s, str>> {
+        BuiltIn::correct_word(self, word)
+    }
+}
+
+fn map_lookup(
+    map: &'static phf::Map<UniCase<&'static str>, &'static str>,
+    key: &str,
+) -> Option<&'static str> {
+    // This transmute should be safe as `get` will not store the reference with
+    // the expanded lifetime. This is due to `Borrow` being overly strict and
+    // can't have an impl for `&'static str` to `Borrow<&'a str>`.
+    //
+    //
+    // See https://github.com/rust-lang/rust/issues/28853#issuecomment-158735548
+    unsafe {
+        let key = ::std::mem::transmute::<_, &'static str>(key);
+        map.get(&UniCase(key)).cloned()
+    }
+}
+
+fn case_correct(correction: &str, case: Case) -> Cow<'_, str> {
+    match case {
+        Case::Lower | Case::None => correction.into(),
+        Case::Title => {
+            let mut title = String::with_capacity(correction.as_bytes().len());
+            let mut char_indices = correction.char_indices();
+            if let Some((_, c)) = char_indices.next() {
+                title.extend(c.to_uppercase());
+                if let Some((i, _)) = char_indices.next() {
+                    title.push_str(&correction[i..]);
+                }
+            }
+            title.into()
+        }
+        Case::Scream => correction
+            .chars()
+            .flat_map(|c| c.to_uppercase())
+            .collect::<String>()
+            .into(),
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+
+    #[test]
+    fn test_case_correct() {
+        let cases = [
+            ("foo", Case::Lower, "foo"),
+            ("foo", Case::None, "foo"),
+            ("foo", Case::Title, "Foo"),
+            ("foo", Case::Scream, "FOO"),
+            ("fOo", Case::None, "fOo"),
+        ];
+        for (correction, case, expected) in cases.iter() {
+            let actual = case_correct(correction, *case);
+            assert_eq!(*expected, actual);
+        }
+    }
+}

typos-dict/src/lib.rs

@@ -0,0 +1,4 @@
+mod dict;
+mod dict_codegen;
+
+pub use crate::dict::*;

typos/Cargo.toml

@@ -15,13 +15,8 @@ edition = "2018"
 travis-ci = { repository = "epage/typos" }
 appveyor  = { repository = "epage/typos" }
 
-[features]
-# Support quickly iterating
-iterate_unstable = []
-
 [dependencies]
 failure = "0.1"
-phf = { version = "0.7", features = ["unicase"] }
 regex = "1.0"
 lazy_static = "1.2.0"
 serde = { version = "1.0", features = ["derive"] }
@@ -32,8 +27,3 @@ bstr = "0.2"
 log = "0.4"
 unicode-segmentation = "1.3.0"
 derive_more = "0.15.0"
-
-[build-dependencies]
-phf_codegen = "0.7"
-csv = "1.0"
-unicase = "1.1"