From 44a0f32aede8f1d0ad15cce8d203ac67872adb50 Mon Sep 17 00:00:00 2001
From: Mike Bannister <notimpossiblemike@gmail.com>
Date: Mon, 16 Mar 2026 18:25:01 -0400
Subject: [PATCH] feat: add Bash tree-sitter language support

---
 .../docs/core-concepts/plugin-system.mdx      |  1 +
 src/kit/queries/bash/tags.scm                 |  5 ++
 src/kit/tree_sitter_symbol_extractor.py       |  4 ++
 tests/test_bash_symbols.py                    | 57 +++++++++++++++++++
 tests/test_symbol_extraction_multilang.py     |  1 +
 tests/test_tree_sitter_languages.py           |  1 +
 6 files changed, 69 insertions(+)
 create mode 100644 src/kit/queries/bash/tags.scm
 create mode 100644 tests/test_bash_symbols.py

diff --git a/docs/src/content/docs/core-concepts/plugin-system.mdx b/docs/src/content/docs/core-concepts/plugin-system.mdx
index 1d497da9..6b90ffa5 100644
--- a/docs/src/content/docs/core-concepts/plugin-system.mdx
+++ b/docs/src/content/docs/core-concepts/plugin-system.mdx
@@ -24,6 +24,7 @@ kit comes with built-in support for 12+ programming languages:
 - **Dart** (`.dart`) - Classes, functions, mixins, enums, extensions
 - **HCL/Terraform** (`.hcl`, `.tf`) - Resources, variables, modules
 - **Haskell** (`.hs`) - Module header, functions (including lambda-binds), common type-level declarations
+- **Bash** (`.sh`, `.bash`) - Function definitions
 
 Each language supports comprehensive symbol extraction including:
 - **Classes and interfaces** with inheritance relationships
diff --git a/src/kit/queries/bash/tags.scm b/src/kit/queries/bash/tags.scm
new file mode 100644
index 00000000..702962aa
--- /dev/null
+++ b/src/kit/queries/bash/tags.scm
@@ -0,0 +1,5 @@
+;; tags.scm for Bash symbol extraction (tree-sitter-bash)
+
+; Function definitions (covers both "function name()" and "name()" syntax)
+(function_definition
+  name: (word) @name) @definition.function
diff --git a/src/kit/tree_sitter_symbol_extractor.py b/src/kit/tree_sitter_symbol_extractor.py
index 273346f5..6e790d2f 100644
--- a/src/kit/tree_sitter_symbol_extractor.py
+++ b/src/kit/tree_sitter_symbol_extractor.py
@@ -34,6 +34,8 @@
     ".hxx": "cpp",
     ".zig": "zig",
     ".cs": "csharp",
+    ".sh": "bash",
+    ".bash": "bash",
 }
 
 
@@ -350,6 +352,8 @@ def reset_plugins(cls) -> None:
             ".hxx": "cpp",
             ".zig": "zig",
             ".cs": "csharp",
+            ".sh": "bash",
+            ".bash": "bash",
         }
         LANGUAGES.clear()
         LANGUAGES.update(original_languages)
diff --git a/tests/test_bash_symbols.py b/tests/test_bash_symbols.py
new file mode 100644
index 00000000..5cdf4804
--- /dev/null
+++ b/tests/test_bash_symbols.py
@@ -0,0 +1,57 @@
+import pytest
+
+from kit.tree_sitter_symbol_extractor import TreeSitterSymbolExtractor
+
+BASH_SAMPLE = """\
+function greet() {
+    echo "Hello, $1!"
+}
+
+say_hi() {
+    echo "Hi there"
+}
+"""
+
+
+def test_bash_parser_and_query_available():
+    parser = TreeSitterSymbolExtractor.get_parser(".sh")
+    query = TreeSitterSymbolExtractor.get_query(".sh")
+    if not parser or not query:
+        pytest.skip("Bash parser or query not available in this environment")
+
+    tree = parser.parse(BASH_SAMPLE.encode("utf-8"))
+    assert tree.root_node is not None
+
+
+def test_bash_symbols():
+    parser = TreeSitterSymbolExtractor.get_parser(".sh")
+    query = TreeSitterSymbolExtractor.get_query(".sh")
+    if not parser or not query:
+        pytest.skip("Bash parser or query not available in this environment")
+
+    symbols = TreeSitterSymbolExtractor.extract_symbols(".sh", BASH_SAMPLE)
+    names = {s["name"] for s in symbols}
+
+    assert "greet" in names
+    assert "say_hi" in names
+    assert all(s["type"] == "function" for s in symbols)
+
+
+def test_bash_extensions():
+    supported = TreeSitterSymbolExtractor.list_supported_languages()
+    assert "bash" in supported
+    assert ".sh" in supported["bash"]
+    assert ".bash" in supported["bash"]
+
+
+def test_bash_extension_extracts_symbols():
+    parser = TreeSitterSymbolExtractor.get_parser(".bash")
+    query = TreeSitterSymbolExtractor.get_query(".bash")
+    if not parser or not query:
+        pytest.skip("Bash parser or query not available in this environment")
+
+    symbols = TreeSitterSymbolExtractor.extract_symbols(".bash", BASH_SAMPLE)
+    names = {s["name"] for s in symbols}
+
+    assert "greet" in names
+    assert "say_hi" in names
diff --git a/tests/test_symbol_extraction_multilang.py b/tests/test_symbol_extraction_multilang.py
index 77767289..fd42129c 100644
--- a/tests/test_symbol_extraction_multilang.py
+++ b/tests/test_symbol_extraction_multilang.py
@@ -9,6 +9,7 @@
     ".java": "class Bar { void foo() {} }\n",
     ".rs": "fn foo() {}\nstruct Bar;\n",
     ".zig": "pub fn foo() void {}\npub const Bar = struct {};\n",
+    ".sh": "function foo() { echo hello; }\n",
 }
 
 
diff --git a/tests/test_tree_sitter_languages.py b/tests/test_tree_sitter_languages.py
index f08160a9..92eb8530 100644
--- a/tests/test_tree_sitter_languages.py
+++ b/tests/test_tree_sitter_languages.py
@@ -12,6 +12,7 @@
     "c": b"int foo() { return 42; }\n",
     "dart": b"int foo() { return 42; }\n",
     "zig": b"pub fn foo() void { }\n",
+    "bash": b"function foo() { echo hello; }\n",
 }