refactor!: Move dict implementation into CLI

Cargo.lock

@@ -140,7 +140,6 @@ name = "codespell-codegen"
 version = "0.1.1"
 dependencies = [
  "codegenrs 0.1.5 (registry+https://github.com/rust-lang/crates.io-index)",
- "codespell-dict 0.1.1",
  "itertools 0.8.0 (registry+https://github.com/rust-lang/crates.io-index)",
  "phf 0.8.0 (registry+https://github.com/rust-lang/crates.io-index)",
  "phf_codegen 0.8.0 (registry+https://github.com/rust-lang/crates.io-index)",
@@ -154,7 +153,6 @@ version = "0.1.1"
 dependencies = [
  "log 0.4.8 (registry+https://github.com/rust-lang/crates.io-index)",
  "phf 0.8.0 (registry+https://github.com/rust-lang/crates.io-index)",
- "typos 0.1.1",
  "unicase 2.5.1 (registry+https://github.com/rust-lang/crates.io-index)",
 ]
 
@@ -788,11 +786,13 @@ dependencies = [
  "failure 0.1.6 (registry+https://github.com/rust-lang/crates.io-index)",
  "ignore 0.4.10 (registry+https://github.com/rust-lang/crates.io-index)",
  "log 0.4.8 (registry+https://github.com/rust-lang/crates.io-index)",
+ "phf 0.8.0 (registry+https://github.com/rust-lang/crates.io-index)",
  "serde 1.0.101 (registry+https://github.com/rust-lang/crates.io-index)",
  "structopt 0.3.3 (registry+https://github.com/rust-lang/crates.io-index)",
  "toml 0.4.10 (registry+https://github.com/rust-lang/crates.io-index)",
  "typos 0.1.1",
  "typos-dict 0.1.1",
+ "unicase 2.5.1 (registry+https://github.com/rust-lang/crates.io-index)",
 ]
 
 [[package]]
@@ -804,7 +804,6 @@ dependencies = [
  "phf 0.8.0 (registry+https://github.com/rust-lang/crates.io-index)",
  "phf_codegen 0.8.0 (registry+https://github.com/rust-lang/crates.io-index)",
  "structopt 0.3.3 (registry+https://github.com/rust-lang/crates.io-index)",
- "typos-dict 0.1.1",
  "unicase 2.5.1 (registry+https://github.com/rust-lang/crates.io-index)",
 ]
 
@@ -814,7 +813,6 @@ version = "0.1.1"
 dependencies = [
  "log 0.4.8 (registry+https://github.com/rust-lang/crates.io-index)",
  "phf 0.8.0 (registry+https://github.com/rust-lang/crates.io-index)",
- "typos 0.1.1",
  "unicase 2.5.1 (registry+https://github.com/rust-lang/crates.io-index)",
 ]
 

Cargo.toml

@@ -29,6 +29,8 @@ codecov = { repository = "crate-ci/typos" }
 [dependencies]
 typos = { version = "0.1", path = "typos" }
 typos-dict = { version = "0.1", path = "dict/typos" }
+phf = { version = "0.8", features = ["unicase"] }
+unicase = "2.5"
 failure = "0.1"
 structopt = "0.3"
 clap = "2"

dict/codespell/Cargo.toml

@@ -15,7 +15,6 @@ azure-devops = { project = "crate-ci", pipeline = "typos" }
 codecov = { repository = "crate-ci/typos" }
 
 [dependencies]
-typos = { version = "0.1", path = "../../typos" }
 phf = { version = "0.8", features = ["unicase"] }
 unicase = "2.5"
 log = "0.4"

dict/codespell/codegen/src/main.rs

@@ -37,7 +37,7 @@ fn generate<W: std::io::Write>(file: &mut W) {
 
     writeln!(
         file,
-        "pub(crate) static WORD_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &[&'static str]> = ",
+        "pub static WORD_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &[&'static str]> = ",
     )
     .unwrap();
     let mut builder = phf_codegen::Map::new();

dict/codespell/src/dict.rs

@@ -1,98 +0,0 @@
-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>> {
-        // HACK: Just assuming the first element is "good enough"
-        map_lookup(&crate::dict_codegen::WORD_DICTIONARY, word.token())
-            .map(|s| case_correct(s[0], 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 [&'static str]>,
-    key: &str,
-) -> Option<&'static [&'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::new(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);
-        }
-    }
-}

dict/codespell/src/dict_codegen.rs

@@ -1,7 +1,7 @@
 // This file is code-genned by codespell-codegen
 
 use unicase::UniCase;
-pub(crate) static WORD_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &[&'static str]> =
+pub static WORD_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &[&'static str]> =
     ::phf::Map {
         key: 3213172566270843353,
         disps: ::phf::Slice::Static(&[

dict/codespell/src/lib.rs

@@ -1,4 +1,3 @@
-mod dict;
 mod dict_codegen;
 
-pub use crate::dict::*;
+pub use crate::dict_codegen::*;

dict/typos/Cargo.toml

@@ -15,7 +15,6 @@ azure-devops = { project = "crate-ci", pipeline = "typos" }
 codecov = { repository = "crate-ci/typos" }
 
 [dependencies]
-typos = { version = "0.1", path = "../../typos" }
 phf = { version = "0.8", features = ["unicase"] }
 unicase = "2.5"
 log = "0.4"

dict/typos/codegen/src/main.rs

@@ -14,7 +14,7 @@ fn generate<W: std::io::Write>(file: &mut W) {
 
     writeln!(
         file,
-        "pub(crate) static WORD_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &'static str> = "
+        "pub static WORD_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &'static str> = "
     )
     .unwrap();
     let mut builder = phf_codegen::Map::new();

dict/typos/src/dict_codegen.rs

@@ -1,8 +1,7 @@
 // This file is code-genned by typos-codegen
 
 use unicase::UniCase;
-pub(crate) static WORD_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &'static str> =
+pub static WORD_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &'static str> = ::phf::Map {
-    ::phf::Map {
     key: 3213172566270843353,
     disps: ::phf::Slice::Static(&[
         (0, 0),
@@ -35612,4 +35611,4 @@ pub(crate) static WORD_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &'st
         (UniCase::ascii("narcissisist"), "narcissist"),
         (UniCase::ascii("decotations"), "decorations"),
     ]),
-    };
+};

dict/typos/src/lib.rs

@@ -1,4 +1,3 @@
-mod dict;
 mod dict_codegen;
 
-pub use crate::dict::*;
+pub use crate::dict_codegen::*;

src/dict.rs

@@ -20,8 +20,7 @@ impl BuiltIn {
     }
 
     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_lookup(&typos_dict::WORD_DICTIONARY, word.token()).map(|s| case_correct(s, word.case()))
-            .map(|s| case_correct(s, word.case()))
     }
 }
 

src/main.rs

@@ -7,6 +7,7 @@ use std::io::Write;
 use structopt::StructOpt;
 
 mod config;
+mod dict;
 
 arg_enum! {
     #[derive(Debug, Copy, Clone, PartialEq, Eq)]
@@ -318,7 +319,7 @@ fn run() -> Result<i32, failure::Error> {
         config.default.update(&args.overrides);
         let config = config;
 
-        let dictionary = typos_dict::BuiltIn::new();
+        let dictionary = crate::dict::BuiltIn::new();
 
         let parser = typos::tokens::ParserBuilder::new()
             .ignore_hex(config.default.ignore_hex())