From c963f680830bcbd4dd2ca063809073fba9f533d9 Mon Sep 17 00:00:00 2001
From: Ed Page <eopage@gmail.com>
Date: Tue, 6 Dec 2022 10:47:08 -0600
Subject: [PATCH] fix(dict): Remove nilable

See conversation in #613
---
 crates/typos-dict/assets/allowed.csv  |  3 +++
 crates/typos-dict/assets/words.csv    |  1 -
 crates/typos-dict/src/dict_codegen.rs |  2 --
 crates/typos-dict/tests/verify.rs     | 34 ++++++++++++++++++++++++---
 4 files changed, 34 insertions(+), 6 deletions(-)
 create mode 100644 crates/typos-dict/assets/allowed.csv

diff --git a/crates/typos-dict/assets/allowed.csv b/crates/typos-dict/assets/allowed.csv
new file mode 100644
index 0000000..68663ba
--- /dev/null
+++ b/crates/typos-dict/assets/allowed.csv
@@ -0,0 +1,3 @@
+nilable,used in ruby community
+thead,html tag
+hardlinked,filesystem term
diff --git a/crates/typos-dict/assets/words.csv b/crates/typos-dict/assets/words.csv
index 9dfbf37..5d8219f 100644
--- a/crates/typos-dict/assets/words.csv
+++ b/crates/typos-dict/assets/words.csv
@@ -33022,7 +33022,6 @@ nigthmares,nightmares
 nihilim,nihilism
 nihilisim,nihilism
 nihilsim,nihilism
-nilable,nillable
 nilihism,nihilism
 nimutes,minutes
 nin,inn,min,bin,nine
diff --git a/crates/typos-dict/src/dict_codegen.rs b/crates/typos-dict/src/dict_codegen.rs
index b9cfa86..380e8d6 100644
--- a/crates/typos-dict/src/dict_codegen.rs
+++ b/crates/typos-dict/src/dict_codegen.rs
@@ -68764,7 +68764,6 @@ pub static WORD_NI_CHILDREN: dictgen::DictTable<&'static [&'static str]> = dictg
         dictgen::InsensitiveStr::Ascii("hilim"),
         dictgen::InsensitiveStr::Ascii("hilisim"),
         dictgen::InsensitiveStr::Ascii("hilsim"),
-        dictgen::InsensitiveStr::Ascii("lable"),
         dictgen::InsensitiveStr::Ascii("lihism"),
         dictgen::InsensitiveStr::Ascii("mutes"),
         dictgen::InsensitiveStr::Ascii("n"),
@@ -68817,7 +68816,6 @@ pub static WORD_NI_CHILDREN: dictgen::DictTable<&'static [&'static str]> = dictg
         &["nihilism"],
         &["nihilism"],
         &["nihilism"],
-        &["nillable"],
         &["nihilism"],
         &["minutes"],
         &["inn", "min", "bin", "nine"],
diff --git a/crates/typos-dict/tests/verify.rs b/crates/typos-dict/tests/verify.rs
index dc31d7d..2855582 100644
--- a/crates/typos-dict/tests/verify.rs
+++ b/crates/typos-dict/tests/verify.rs
@@ -50,17 +50,26 @@ fn generate<W: std::io::Write>(file: &mut W, dict: &[u8]) {
         })
         .collect();
 
-    let disallowed_typos = varcon_words();
+    let varcon_words = varcon_words();
+    let allowed_words = allowed_words();
     let word_variants = proper_word_variants();
     let rows: Dict = rows
         .into_iter()
         .filter(|(typo, _)| {
-            let is_disallowed = disallowed_typos.contains(&unicase::UniCase::new(typo));
+            let is_disallowed = varcon_words.contains(&unicase::UniCase::new(typo));
             if is_disallowed {
-                eprintln!("{:?} is disallowed", typo);
+                eprintln!("{:?} is disallowed; in varcon", typo);
             }
             !is_disallowed
         })
+        .filter(|(typo, _)| {
+            if let Some(reason) = allowed_words.get(typo.as_ref()) {
+                eprintln!("{:?} is disallowed; {}", typo, reason);
+                false
+            } else {
+                true
+            }
+        })
         .map(|(typo, corrections)| {
             let mut new_corrections = vec![];
             for correction in corrections {
@@ -137,3 +146,22 @@ fn find_best_match<'c>(
     matches.sort_unstable();
     matches.into_iter().next().map(|(_, r)| r)
 }
+
+fn allowed_words() -> std::collections::HashMap<String, String> {
+    let allowed_path = "assets/allowed.csv";
+    let data = std::fs::read(allowed_path).unwrap();
+    csv::ReaderBuilder::new()
+        .has_headers(false)
+        .flexible(true)
+        .from_reader(data.as_slice())
+        .records()
+        .map(Result::unwrap)
+        .map(|r| {
+            let mut i = r.iter();
+            let mut typo = i.next().expect("typo").to_owned();
+            typo.make_ascii_lowercase();
+            let reason = i.next().expect("reason").to_owned();
+            (typo, reason)
+        })
+        .collect()
+}