Add Java support

Author: lukejans

Cargo.lock

@@ -46,6 +46,7 @@ tree-sitter-css = "<0.25.0"
 tree-sitter-go = "<0.25.0"
 tree-sitter-haskell = "<0.25.0"
 tree-sitter-html = "<0.25.0"
+tree-sitter-java = "0.23.5"
 tree-sitter-javascript = "<0.25.0"
 tree-sitter-php = "<0.24.0"
 tree-sitter-python = "<0.25.0"

Cargo.toml

@@ -28,6 +28,7 @@ tree-sitter-css.workspace = true
 tree-sitter-go.workspace = true
 tree-sitter-haskell.workspace = true
 tree-sitter-html.workspace = true
+tree-sitter-java.workspace = true
 tree-sitter-javascript.workspace = true
 tree-sitter-php.workspace = true
 tree-sitter-python.workspace = true

crates/codebook/Cargo.toml

@@ -10,6 +10,7 @@ pub enum LanguageType {
     Go,
     Haskell,
     HTML,
+    Java,
     Javascript,
     Php,
     Python,
@@ -73,6 +74,13 @@ pub static LANGUAGE_SETTINGS: &[LanguageSetting] = &[
         query: include_str!("queries/python.scm"),
         extensions: &["py"],
     },
+    LanguageSetting {
+        type_: LanguageType::Java,
+        ids: &["java"],
+        dictionary_ids: &["java"],
+        query: include_str!("queries/java.scm"),
+        extensions: &["java"],
+    },
     LanguageSetting {
         type_: LanguageType::Javascript,
         ids: &["javascript", "javascriptreact"],
@@ -172,6 +180,7 @@ impl LanguageSetting {
             LanguageType::Go => Some(tree_sitter_go::LANGUAGE.into()),
             LanguageType::Haskell => Some(tree_sitter_haskell::LANGUAGE.into()),
             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::Php => Some(tree_sitter_php::LANGUAGE_PHP.into()),
             LanguageType::Python => Some(tree_sitter_python::LANGUAGE.into()),

crates/codebook/src/queries.rs

@@ -0,0 +1,24 @@
+[
+    (line_comment)
+    (block_comment)
+] @comment
+[
+    (character_literal)
+    (string_literal)
+] @string
+(variable_declarator
+    name: (identifier) @identifier)
+(interface_declaration
+    name: (identifier) @identifier)
+(class_declaration
+    name: (identifier) @identifier)
+(method_declaration
+    name: (identifier) @identifier)
+(enum_declaration
+    name: (identifier) @identifier)
+(enum_constant
+    name: (identifier) @identifier)
+(formal_parameter
+    name: (identifier) @identifier)
+(catch_formal_parameter
+    name: (identifier) @identifier)

crates/codebook/src/queries/java.scm

@@ -0,0 +1,170 @@
+use codebook::{
+    parser::{TextRange, WordLocation},
+    queries::LanguageType,
+};
+
+mod utils;
+
+#[test]
+fn test_java_location() {
+    utils::init_logging();
+    let sample_text = r#"
+    // Singl-line comment
+    /* Blck comment */
+
+    interface ExamplInterface {
+        void doSomethng();
+    }
+
+    enum Statuss { ACTIV }
+
+    public class SoemJavaDemo implements ExamplInterface {
+
+        String messag = "Hello";
+
+        public void doSomethng(String smth) {
+            System.out.println("Doing " + smth + "...");
+        }
+
+        public static void main(String[] args) {
+            try {
+                int x = 1 / 0;
+            } catch (ArithmeticException errorr) {
+                System.out.println("Caught: " + errorr);
+                some.recoveryMthod();
+            }
+        }
+    }"#;
+
+    let expected = vec![
+        WordLocation::new(
+            "Singl".to_string(),
+            vec![TextRange {
+                start_char: 7,
+                end_char: 12,
+                line: 1,
+            }],
+        ),
+        WordLocation::new(
+            "Blck".to_string(),
+            vec![TextRange {
+                start_char: 7,
+                end_char: 11,
+                line: 2,
+            }],
+        ),
+        WordLocation::new(
+            "Exampl".to_string(),
+            vec![TextRange {
+                start_char: 14,
+                end_char: 20,
+                line: 4,
+            }],
+        ),
+        WordLocation::new(
+            "Somethng".to_string(),
+            vec![
+                TextRange {
+                    start_char: 15,
+                    end_char: 23,
+                    line: 5,
+                },
+                TextRange {
+                    start_char: 22,
+                    end_char: 30,
+                    line: 14,
+                },
+            ],
+        ),
+        WordLocation::new(
+            "Statuss".to_string(),
+            vec![TextRange {
+                start_char: 9,
+                end_char: 16,
+                line: 8,
+            }],
+        ),
+        WordLocation::new(
+            "ACTIV".to_string(),
+            vec![TextRange {
+                start_char: 19,
+                end_char: 24,
+                line: 8,
+            }],
+        ),
+        WordLocation::new(
+            "Soem".to_string(),
+            vec![TextRange {
+                start_char: 17,
+                end_char: 21,
+                line: 10,
+            }],
+        ),
+        WordLocation::new(
+            "messag".to_string(),
+            vec![TextRange {
+                start_char: 15,
+                end_char: 21,
+                line: 12,
+            }],
+        ),
+        WordLocation::new(
+            "smth".to_string(),
+            vec![TextRange {
+                start_char: 38,
+                end_char: 42,
+                line: 14,
+            }],
+        ),
+        WordLocation::new(
+            "errorr".to_string(),
+            vec![TextRange {
+                start_char: 41,
+                end_char: 47,
+                line: 21,
+            }],
+        ),
+    ];
+
+    let not_expected = [
+        "interface",
+        "void",
+        "enum",
+        "public",
+        "class",
+        "implements",
+        "String",
+        "System",
+        "out",
+        "println",
+        "static",
+        "main",
+        "try",
+        "catch",
+        "ArithmeticException",
+        "Hello",
+        "Doing",
+        "Caught",
+        "Mthod",
+    ];
+
+    let processor = utils::get_processor();
+    let misspelled = processor
+        .spell_check(sample_text, Some(LanguageType::Java), None)
+        .to_vec();
+
+    println!("Misspelled words: {misspelled:?}\n");
+
+    for e in &expected {
+        println!("Expecting: {e:?}");
+        let miss = misspelled
+            .iter()
+            .find(|r| r.word == e.word)
+            .expect("Word not found");
+        assert_eq!(miss.locations, e.locations);
+    }
+
+    for result in misspelled {
+        assert!(!not_expected.contains(&result.word.as_str()));
+    }
+}

crates/codebook/tests/test_java.rs

@@ -0,0 +1,48 @@
+public class example {
+    // single line comment with a spelling mistake: helllo
+
+    /**
+     * Block comment with a spelling mistake: byee
+     */
+
+    // variable declarators with spelling mistakes
+    short myFavoritNum = 404;
+    String myFavoiteString = "Hello, World!";
+
+    // a string with a spelling mistake
+    String myStr = "foooooooooooooooood";
+
+    // enum declaration with a spelling mistake
+    enum Levvel {
+        // enum constant with a spelling mistake
+        BEGINER,
+        INTERMEDIETE,
+        XPERT,
+    }
+
+    // interface declaration with a spelling mistake
+    public interface Innerexample {
+        // method declaration with a spelling mistake
+        void doSmething();
+    }
+
+    // class declaration with a spelling mistake
+    class PointlessClasss {
+    }
+
+    public static void main(String[] args) {
+        // catch formal parameter spelling mistake
+        try {
+            int result = 10 / 0;
+            System.out.println(result);
+        } catch (ArithmeticException uhoooh) {
+            System.out.println(uhoooh);
+        }
+    }
+
+    // method declaration and formal parameter with spelling mistakes
+    public String anthrMethod(String smth) {
+        // string literal with a spelling mistake
+        return "anthr method called with: " + smth;
+    }
+}