Only lowercase ascii, add software terms

CHANGELOG.md

@@ -1,9 +1,13 @@
-[0.1.16]
+[0.1.17]
 
 - Added a download manager for adding many different dictionaries later
 - Using a larger en_us dictionary as default
 - Now checks on every change, instead of on save. May add an option later to toggle this off
 - Add a command to the LSP binary to clear cache
+- Don't give a code action when a word is not misspelled
+- Vendor OpenSSL
+- Add 'software_terms'
+- Only lowercase ascii letters when checking
 
 [0.1.15]
 

crates/codebook-config/src/lib.rs

@@ -46,7 +46,7 @@ impl<'de> Deserialize<'de> for ConfigSettings {
         D: serde::Deserializer<'de>,
     {
         fn to_lowercase_vec(v: Vec<String>) -> Vec<String> {
-            v.into_iter().map(|s| s.to_lowercase()).collect()
+            v.into_iter().map(|s| s.to_ascii_lowercase()).collect()
         }
         #[derive(Deserialize)]
         struct Helper {
@@ -163,7 +163,7 @@ impl CodebookConfig {
     /// Add a word to the allowlist and save the configuration
     pub fn add_word(&self, word: &str) -> Result<bool> {
         {
-            let word = word.to_lowercase();
+            let word = word.to_ascii_lowercase();
             let settings = &mut self.settings.write().unwrap();
             // Check if word already exists
             if settings.words.contains(&word.to_string()) {
@@ -289,7 +289,7 @@ impl CodebookConfig {
 
     /// Check if a word is in the custom allowlist
     pub fn is_allowed_word(&self, word: &str) -> bool {
-        let word = word.to_lowercase();
+        let word = word.to_ascii_lowercase();
         self.settings
             .read()
             .unwrap()
@@ -300,7 +300,7 @@ impl CodebookConfig {
 
     /// Check if a word should be flagged
     pub fn should_flag_word(&self, word: &str) -> bool {
-        let word = word.to_lowercase();
+        let word = word.to_ascii_lowercase();
         self.settings
             .read()
             .unwrap()

crates/codebook/src/dictionaries/dictionary.rs

@@ -3,6 +3,7 @@ use lru::LruCache;
 
 use std::{
     num::NonZeroUsize,
+    path::PathBuf,
     sync::{Arc, RwLock},
 };
 
@@ -122,7 +123,7 @@ pub struct TextDictionary {
 
 impl Dictionary for TextDictionary {
     fn check(&self, word: &str) -> bool {
-        let lower = word.to_lowercase();
+        let lower = word.to_ascii_lowercase();
         let words = self
             .word_list
             .lines()
@@ -142,9 +143,13 @@ impl Dictionary for TextDictionary {
 impl TextDictionary {
     pub fn new(word_list: &str) -> Self {
         Self {
-            word_list: word_list.to_owned(),
+            word_list: word_list.to_ascii_lowercase(),
         }
     }
+    pub fn new_from_path(path: &PathBuf) -> Self {
+        let word_list = std::fs::read_to_string(path).unwrap().to_ascii_lowercase();
+        Self { word_list }
+    }
 }
 
 #[cfg(test)]

crates/codebook/src/dictionaries/manager.rs

@@ -89,7 +89,7 @@ impl DictionaryManager {
                 return None;
             }
         };
-        let dict = TextDictionary::new(text_path.to_str().unwrap());
+        let dict = TextDictionary::new_from_path(&text_path);
         Some(Arc::new(dict))
     }
 }

crates/codebook/src/dictionaries/repo.rs

@@ -42,47 +42,54 @@ pub enum DictionaryRepo {
     Text(TextRepo),
 }
 
-static DICTIONARIES: LazyLock<Vec<DictionaryRepo>> = LazyLock::new(|| {
-    vec![DictionaryRepo::Hunspell(
+static HUNSPELL_DICTIONARIES: LazyLock<Vec<HunspellRepo>> = LazyLock::new(|| {
+    vec![
     HunspellRepo::new(
         "en_us",
         "https://raw.githubusercontent.com/streetsidesoftware/cspell-dicts/refs/heads/main/dictionaries/en_US/src/hunspell/en_US-large.aff",
         "https://raw.githubusercontent.com/streetsidesoftware/cspell-dicts/refs/heads/main/dictionaries/en_US/src/hunspell/en_US-large.dic",
-    )),
-    DictionaryRepo::Hunspell(
+    ),
     HunspellRepo::new(
         "en",
         "https://raw.githubusercontent.com/blopker/dictionaries/refs/heads/main/dictionaries/en/index.aff",
         "https://raw.githubusercontent.com/blopker/dictionaries/refs/heads/main/dictionaries/en/index.dic",
-    )),
-    DictionaryRepo::Hunspell(
+    ),
     HunspellRepo::new(
         "en_gb",
         "https://raw.githubusercontent.com/streetsidesoftware/cspell-dicts/refs/heads/main/dictionaries/en_GB/src/hunspell/en_GB-large.aff",
         "https://raw.githubusercontent.com/streetsidesoftware/cspell-dicts/refs/heads/main/dictionaries/en_GB/src/hunspell/en_GB-large.dic",
-    )),
-    DictionaryRepo::Text(
+    ),]
+});
+
+static TEXT_DICTIONARIES: LazyLock<Vec<TextRepo>> = LazyLock::new(|| {
+    vec![
     TextRepo::new(
         "rust",
         "https://raw.githubusercontent.com/streetsidesoftware/cspell-dicts/refs/heads/main/dictionaries/rust/dict/rust.txt",
-    )),
-    DictionaryRepo::Text(
+    ),
+    TextRepo::new(
+        "software_terms",
+        "https://raw.githubusercontent.com/streetsidesoftware/cspell-dicts/refs/heads/main/dictionaries/software-terms/dict/softwareTerms.txt",
+    ),
+    TextRepo::new(
+        "computing_acronyms",
+        "https://raw.githubusercontent.com/streetsidesoftware/cspell-dicts/refs/heads/main/dictionaries/software-terms/dict/computing-acronyms.txt",
+    ),
     TextRepo{
         name: "codebook".to_string(),
         text: Some(CODEBOOK_DICTIONARY),
         url: None
-    }),
-    ]
+    },]
 });
 
 pub fn get_repo(name: &str) -> Option<DictionaryRepo> {
-    let res = DICTIONARIES.iter().find(|d| match d {
-        DictionaryRepo::Hunspell(h) => h.name == name,
-        DictionaryRepo::Text(t) => t.name == name,
-    });
-
-    match res {
-        Some(d) => Some(d.clone()),
-        None => None,
+    let res = HUNSPELL_DICTIONARIES.iter().find(|d| d.name == name);
+    if res.is_some() {
+        return Some(DictionaryRepo::Hunspell(res.unwrap().clone()));
+    }
+    let res = TEXT_DICTIONARIES.iter().find(|d| d.name == name);
+    if res.is_some() {
+        return Some(DictionaryRepo::Text(res.unwrap().clone()));
     }
+    None
 }

crates/codebook/src/lib.rs

@@ -16,6 +16,9 @@ pub struct Codebook {
     manager: DictionaryManager,
 }
 
+// Custom 'codebook' dictionary could be removed later for a more general solution.
+static DEFAULT_DICTIONARIES: &[&str; 3] = &["codebook", "software_terms", "computing_acronyms"];
+
 impl Codebook {
     pub fn new(config: Arc<CodebookConfig>) -> Result<Self, Box<dyn std::error::Error>> {
         crate::log::init_logging();
@@ -76,8 +79,7 @@ impl Codebook {
             }
             None => {}
         };
-        // Push custom codebook dictionary. Could be removed later for a more general solution.
-        dictionary_ids.push("codebook".to_string());
+        dictionary_ids.extend(DEFAULT_DICTIONARIES.iter().map(|f| f.to_string()));
         let mut dictionaries = Vec::with_capacity(dictionary_ids.len());
         for dictionary_id in dictionary_ids {
             let dictionary = self.manager.get_dictionary(&dictionary_id);

word_lists/codebook.txt

@@ -17,17 +17,20 @@ htm
 jsx
 linter
 linux
+lru
 lsp
 macos
 msvc
 pbkdf
 pc
 prerelease
 pst
+redownload
 roadmap
 rpc
 semibold
 sha
+ssl
 tsx
 ttf
 typescript
@@ -36,5 +39,3 @@ uncomment
 validator
 validators
 webp
-redownload
-LRU

word_lists/combined.gen.txt

@@ -1997,6 +1997,7 @@ sqrt
 square
 src
 srv
+ssl
 st
 stable
 stack