Add tree-sitter Markdown support to skip fenced code blocks

Cargo.toml

@@ -57,6 +57,7 @@ tree-sitter-html = "<0.25.0"
 tree-sitter-java = "<0.25.0"
 tree-sitter-javascript = "<0.26.0"
 tree-sitter-lua = "<0.25.0"
+tree-sitter-md = "<0.6.0"
 tree-sitter-odin-codebook = "1.4.0"
 tree-sitter-php = "<0.25.0"
 tree-sitter-python = "<0.26.0"

crates/codebook/Cargo.toml

@@ -42,6 +42,7 @@ tree-sitter-java.workspace = true
 tree-sitter-javascript.workspace = true
 codebook-tree-sitter-latex.workspace = true
 tree-sitter-lua.workspace = true
+tree-sitter-md.workspace = true
 tree-sitter-odin-codebook.workspace = true
 tree-sitter-php.workspace = true
 tree-sitter-python.workspace = true

crates/codebook/src/queries.rs

@@ -18,6 +18,7 @@ pub enum LanguageType {
     Javascript,
     Latex,
     Lua,
+    Markdown,
     Odin,
     Php,
     Python,
@@ -199,6 +200,13 @@ pub static LANGUAGE_SETTINGS: &[LanguageSetting] = &[
         query: include_str!("queries/lua.scm"),
         extensions: &["lua"],
     },
+    LanguageSetting {
+        type_: LanguageType::Markdown,
+        ids: &["markdown"],
+        dictionary_ids: &[],
+        query: include_str!("queries/markdown.scm"),
+        extensions: &["md", "markdown"],
+    },
     LanguageSetting {
         type_: LanguageType::Bash,
         ids: &["bash", "shellscript", "sh", "shell script"],
@@ -291,6 +299,7 @@ impl LanguageSetting {
             LanguageType::Javascript => Some(tree_sitter_javascript::LANGUAGE.into()),
             LanguageType::Latex => Some(codebook_tree_sitter_latex::LANGUAGE.into()),
             LanguageType::Lua => Some(tree_sitter_lua::LANGUAGE.into()),
+            LanguageType::Markdown => Some(tree_sitter_md::LANGUAGE.into()),
             LanguageType::Odin => Some(tree_sitter_odin_codebook::LANGUAGE.into()),
             LanguageType::Php => Some(tree_sitter_php::LANGUAGE_PHP.into()),
             LanguageType::Python => Some(tree_sitter_python::LANGUAGE.into()),

crates/codebook/src/queries/markdown.scm

@@ -0,0 +1,2 @@
+(paragraph (inline) @string)
+(atx_heading (inline) @string)

crates/codebook/tests/test_markdown.rs

@@ -0,0 +1,136 @@
+use codebook::{
+    parser::{TextRange, WordLocation},
+    queries::LanguageType,
+};
+
+mod utils;
+
+#[test]
+fn test_markdown_paragraph() {
+    utils::init_logging();
+    let processor = utils::get_processor();
+    let sample_text = "Some paragraph text with a misspeled word.\n";
+    let expected = vec![WordLocation::new(
+        "misspeled".to_string(),
+        vec![TextRange {
+            start_byte: 27,
+            end_byte: 36,
+        }],
+    )];
+    let misspelled = processor
+        .spell_check(sample_text, Some(LanguageType::Markdown), None)
+        .to_vec();
+    println!("Misspelled words: {misspelled:?}");
+    assert_eq!(misspelled.len(), 1);
+    assert_eq!(misspelled[0].word, expected[0].word);
+    assert_eq!(misspelled[0].locations, expected[0].locations);
+}
+
+#[test]
+fn test_markdown_heading() {
+    utils::init_logging();
+    let processor = utils::get_processor();
+    let sample_text = "# A headng with a tyypo\n";
+    let misspelled = processor
+        .spell_check(sample_text, Some(LanguageType::Markdown), None)
+        .to_vec();
+    let words: Vec<&str> = misspelled.iter().map(|r| r.word.as_str()).collect();
+    println!("Misspelled words: {words:?}");
+    assert!(words.contains(&"headng"));
+    assert!(words.contains(&"tyypo"));
+}
+
+#[test]
+fn test_markdown_fenced_code_block_skipped() {
+    utils::init_logging();
+    let processor = utils::get_processor();
+    let sample_text = r#"# Hello World
+
+Some correct text here.
+
+```bash
+mkdir some_dir
+badwwword_in_code
+```
+
+More correct text here.
+"#;
+    let misspelled = processor
+        .spell_check(sample_text, Some(LanguageType::Markdown), None)
+        .to_vec();
+    let words: Vec<&str> = misspelled.iter().map(|r| r.word.as_str()).collect();
+    println!("Misspelled words: {words:?}");
+    // Words inside fenced code blocks should NOT be flagged
+    assert!(!words.contains(&"mkdir"));
+    assert!(!words.contains(&"badwwword"));
+    assert!(!words.contains(&"dir"));
+}
+
+#[test]
+fn test_markdown_fenced_code_block_with_typo_outside() {
+    utils::init_logging();
+    let processor = utils::get_processor();
+    let sample_text = r#"A paragrap with a tyypo.
+
+```python
+def some_functin():
+    pass
+```
+
+Another paragrap with a tyypo.
+"#;
+    let misspelled = processor
+        .spell_check(sample_text, Some(LanguageType::Markdown), None)
+        .to_vec();
+    let words: Vec<&str> = misspelled.iter().map(|r| r.word.as_str()).collect();
+    println!("Misspelled words: {words:?}");
+    // Typos in prose should be flagged
+    assert!(words.contains(&"paragrap"));
+    assert!(words.contains(&"tyypo"));
+    // Typos inside code blocks should NOT be flagged
+    assert!(!words.contains(&"functin"));
+}
+
+#[test]
+fn test_markdown_multiple_code_blocks() {
+    utils::init_logging();
+    let processor = utils::get_processor();
+    let sample_text = r#"Some text with a tyypo.
+
+```bash
+mkdir somedir
+```
+
+Middle text is corect.
+
+```python
+badspel = True
+```
+
+End text is also corect.
+"#;
+    let misspelled = processor
+        .spell_check(sample_text, Some(LanguageType::Markdown), None)
+        .to_vec();
+    let words: Vec<&str> = misspelled.iter().map(|r| r.word.as_str()).collect();
+    println!("Misspelled words: {words:?}");
+    assert!(words.contains(&"tyypo"));
+    assert!(words.contains(&"corect"));
+    assert!(!words.contains(&"mkdir"));
+    assert!(!words.contains(&"somedir"));
+    assert!(!words.contains(&"badspel"));
+}
+
+#[test]
+fn test_markdown_block_quote() {
+    utils::init_logging();
+    let processor = utils::get_processor();
+    let sample_text = "> A block quoet with a tyypo.\n";
+    let misspelled = processor
+        .spell_check(sample_text, Some(LanguageType::Markdown), None)
+        .to_vec();
+    let words: Vec<&str> = misspelled.iter().map(|r| r.word.as_str()).collect();
+    println!("Misspelled words: {words:?}");
+    assert!(words.contains(&"quoet"));
+    assert!(words.contains(&"tyypo"));
+}