From ee9f3f8e61ea8fc59f21a934c1ac37e489142a34 Mon Sep 17 00:00:00 2001
From: LJ <linghua@cocoindex.io>
Date: Thu, 17 Jul 2025 09:10:06 -0700
Subject: [PATCH] feat(tree-sitter-kotlin): support Kotlin in
 `SplitRecursively`

---
 Cargo.lock                             | 11 +++++++++++
 Cargo.toml                             |  3 +++
 src/ops/functions/split_recursively.rs |  7 +++++++
 3 files changed, 21 insertions(+)

diff --git a/Cargo.lock b/Cargo.lock
index 0dd5d8c7..f377bbd5 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1369,6 +1369,7 @@ dependencies = [
  "tree-sitter-java",
  "tree-sitter-javascript",
  "tree-sitter-json",
+ "tree-sitter-kotlin-ng",
  "tree-sitter-language",
  "tree-sitter-md",
  "tree-sitter-pascal",
@@ -5724,6 +5725,16 @@ dependencies = [
  "tree-sitter-language",
 ]
 
+[[package]]
+name = "tree-sitter-kotlin-ng"
+version = "1.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e800ebbda938acfbf224f4d2c34947a31994b1295ee6e819b65226c7b51b4450"
+dependencies = [
+ "cc",
+ "tree-sitter-language",
+]
+
 [[package]]
 name = "tree-sitter-language"
 version = "0.1.5"
diff --git a/Cargo.toml b/Cargo.toml
index 93196a58..262ad3c5 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -63,6 +63,7 @@ reqwest = { version = "0.12.15", default-features = false, features = [
     "rustls-tls",
 ] }
 async-openai = "0.28.0"
+
 tree-sitter = "0.25.3"
 tree-sitter-language = "0.1.5"
 # Per language tree-sitter parsers
@@ -76,6 +77,8 @@ tree-sitter-html = "0.23.2"
 tree-sitter-java = "0.23.5"
 tree-sitter-javascript = "0.23.1"
 tree-sitter-json = "0.24.8"
+# The other more popular crate tree-sitter-kotlin requires tree-sitter < 0.23 for now
+tree-sitter-kotlin-ng = "1.1.0"
 tree-sitter-md = "0.3.2"
 tree-sitter-pascal = "0.10.0"
 tree-sitter-php = "0.23.11"
diff --git a/src/ops/functions/split_recursively.rs b/src/ops/functions/split_recursively.rs
index 2d5842a9..30236ab9 100644
--- a/src/ops/functions/split_recursively.rs
+++ b/src/ops/functions/split_recursively.rs
@@ -149,6 +149,13 @@ static TREE_SITTER_LANGUAGE_BY_LANG: LazyLock<
         [],
     );
     add_treesitter_language(&mut map, "JSON", [".json"], tree_sitter_json::LANGUAGE, []);
+    add_treesitter_language(
+        &mut map,
+        "Kotlin",
+        [".kt", ".kts"],
+        tree_sitter_kotlin_ng::LANGUAGE,
+        [],
+    );
     add_treesitter_language(
         &mut map,
         "Markdown",