Merge pull request #252 from epage/dict

fix(dict): Handle cases from Linux
crate-ci · May 18, 2021 · e6c595c · e6c595c
2 parents 444d2cc + 639e65b
commit e6c595c
Show file tree

Hide file tree

Showing 7 changed files with 33,748 additions and 33,671 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/crates/typos-dict/assets/words.csv b/crates/typos-dict/assets/words.csv
@@ -13021,7 +13021,6 @@ handelbars,handlebars
 handicaped,handicapped
 handwritng,handwriting
 harasments,harassments
-hardlinked,hardline
 harmoniacs,harmonic
 harmonisch,harmonic
 harrasment,harassment
@@ -23557,7 +23556,7 @@ referens,references
 referere,referee
 referign,refering
 refering,referring
-refernce,references
+refernce,reference
 reffered,referred
 refilles,refills
 refillls,refills
@@ -27976,7 +27975,7 @@ tast,taste
 tath,that
 tehy,they
 tghe,the
-ther,there
+ther,there,their,the
 thge,the
 thna,than
 thne,then

diff --git a/crates/typos-dict/codegen/Cargo.toml b/crates/typos-dict/codegen/Cargo.toml
@@ -18,6 +18,7 @@ codecov = { repository = "crate-ci/typos" }
 phf = { version = "0.8", features = ["unicase"] }
 phf_codegen = "0.8"
 csv = "1.1"
+itertools = "0.10"
 unicase = "2.5"
 codegenrs = "1.0"
 structopt = "0.3"
diff --git a/crates/typos-dict/codegen/src/main.rs b/crates/typos-dict/codegen/src/main.rs
@@ -18,20 +18,26 @@ fn generate<W: std::io::Write>(file: &mut W) {
 
     writeln!(
         file,
-        "pub static WORD_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &'static str> = "
+        "pub static WORD_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &'static [&'static str]> = "
     )
     .unwrap();
     let mut builder = phf_codegen::Map::new();
     let records: Vec<_> = csv::ReaderBuilder::new()
         .has_headers(false)
+        .flexible(true)
         .from_reader(DICT)
         .records()
         .map(|r| r.unwrap())
         .collect();
     for record in &records {
-        smallest = std::cmp::min(smallest, record[0].len());
-        largest = std::cmp::max(largest, record[0].len());
-        let value = format!(r#""{}""#, &record[1]);
+        let mut record_fields = record.iter();
+        let key = record_fields.next().unwrap();
+        smallest = std::cmp::min(smallest, key.len());
+        largest = std::cmp::max(largest, key.len());
+        let value = format!(
+            "&[{}]",
+            itertools::join(record_fields.map(|field| format!(r#""{}""#, field)), ", ")
+        );
         builder.entry(unicase::UniCase::new(&record[0]), &value);
     }
     let codegenned = builder.build();