Add Lua support

Author: blopker

CHANGELOG.md

@@ -1,3 +1,7 @@
+[0.3.8]
+
+- Add support for Lua language
+
 [0.3.7]
 
 - Better error message for user-supplied regex

Cargo.lock

@@ -50,6 +50,7 @@ tree-sitter-haskell = "<0.25.0"
 tree-sitter-html = "<0.25.0"
 tree-sitter-java = "<0.25.0"
 tree-sitter-javascript = "<0.25.0"
+tree-sitter-lua = "<0.25.0"
 tree-sitter-php = "<0.24.0"
 tree-sitter-python = "<0.25.0"
 tree-sitter-r = "1.1.0"

Cargo.toml

@@ -129,6 +129,7 @@ Codebook is in active development. As better dictionaries are added, words that
 | Haskell | ⚠️ |
 | Java | ✅ |
 | JavaScript | ✅ |
+| Lua | ✅ |
 | Markdown | ✅ |
 | PHP | ⚠️ |
 | Plain Text | ✅ |

README.md

@@ -39,6 +39,7 @@ tree-sitter-haskell.workspace = true
 tree-sitter-html.workspace = true
 tree-sitter-java.workspace = true
 tree-sitter-javascript.workspace = true
+tree-sitter-lua.workspace = true
 tree-sitter-php.workspace = true
 tree-sitter-python.workspace = true
 tree-sitter-r.workspace = true

crates/codebook/Cargo.toml

@@ -2988,4 +2988,4 @@ zip
 zone
 zoom
 zsh
-zshrc
+zshrc

crates/codebook/src/dictionaries/combined.gen.txt

@@ -12,6 +12,7 @@ pub enum LanguageType {
     HTML,
     Java,
     Javascript,
+    Lua,
     Php,
     Python,
     R,
@@ -137,6 +138,13 @@ pub static LANGUAGE_SETTINGS: &[LanguageSetting] = &[
         query: include_str!("queries/ruby.scm"),
         extensions: &["rb"],
     },
+    LanguageSetting {
+        type_: LanguageType::Lua,
+        ids: &["lua"],
+        dictionary_ids: &["lua"],
+        query: include_str!("queries/lua.scm"),
+        extensions: &["lua"],
+    },
     LanguageSetting {
         type_: LanguageType::Bash,
         ids: &["bash", "shellscript", "sh", "shell script"],
@@ -182,6 +190,7 @@ impl LanguageSetting {
             LanguageType::HTML => Some(tree_sitter_html::LANGUAGE.into()),
             LanguageType::Java => Some(tree_sitter_java::LANGUAGE.into()),
             LanguageType::Javascript => Some(tree_sitter_javascript::LANGUAGE.into()),
+            LanguageType::Lua => Some(tree_sitter_lua::LANGUAGE.into()),
             LanguageType::Php => Some(tree_sitter_php::LANGUAGE_PHP.into()),
             LanguageType::Python => Some(tree_sitter_python::LANGUAGE.into()),
             LanguageType::R => Some(tree_sitter_r::LANGUAGE.into()),

crates/codebook/src/queries.rs

@@ -0,0 +1,31 @@
+; Comments
+(comment) @comment
+
+; Strings
+(string) @string
+
+; Function declarations
+(function_declaration
+  name: (identifier) @identifier)
+
+(function_declaration
+    (method_index_expression
+        method: (identifier) @identifier))
+
+; Variable assignments
+(assignment_statement
+  (variable_list
+    (identifier) @identifier))
+
+(assignment_statement
+    (variable_list
+        (dot_index_expression
+            field: (identifier) @identifier)))
+
+; Function parameters
+(parameters
+  (identifier) @identifier)
+
+; Table fields
+(field
+  name: (identifier) @identifier)

crates/codebook/src/queries/lua.scm

@@ -14,7 +14,7 @@ pub struct SplitRef<'a> {
     pub start_char: u32,
 }
 
-pub fn split(s: &str) -> Vec<SplitRef> {
+pub fn split(s: &str) -> Vec<SplitRef<'_>> {
     if s.is_empty() {
         return Vec::new();
     }

crates/codebook/src/splitter.rs

@@ -0,0 +1,39 @@
+-- Lua exampl file with intentional misspellings
+
+-- Simple variable assignments
+mesage = "Helo Wrold!"
+countr = 0
+
+-- Function declaration with misspelling
+function calculatr(numbr1, numbr2, operashun)
+    local result = 0
+
+    if operashun == "+" then
+        result = numbr1 + numbr2
+    elseif operashun == "-" then
+        result = numbr1 - numbr2
+    end
+
+    return result
+end
+
+-- Table definition
+local UserAccont = {}
+
+-- Function with parameters
+function UserAccont:calculat_intrest(rate)
+    return 1000 * rate
+end
+
+-- Table with fields
+local config = {
+    mesage = "test",
+    countr = 42,
+    intrest = 0.05
+}
+
+-- Assignment statement
+local numbr = 100
+
+-- More assignments
+local operashun = "add"