From 5a2b53f9a612d1759e5b09c207a06890fe7c752a Mon Sep 17 00:00:00 2001
From: cookerpapa <jikimasu97@gmail.com>
Date: Mon, 9 Nov 2026 19:52:46 +0530
Subject: [PATCH] feat(graph): index Bash scripts with the bundled grammar
 (#354)

Re-landed from #354 with the conflicts against main resolved; the contributor's fork couldn't take the merge commit.

Co-Authored-By: Shrish Dwivedi <shrish@trailhq.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---
 README.md               |  2 ++
 src/graph/generic.ts    |  1 +
 test/graph-bash.test.ts | 60 +++++++++++++++++++++++++++++++++++++++++
 3 files changed, 63 insertions(+)
 create mode 100644 test/graph-bash.test.ts

diff --git a/README.md b/README.md
index 7f314cf35..6a08f6384 100644
--- a/README.md
+++ b/README.md
@@ -490,6 +490,8 @@ compiler-grade layer, all `$0` and deterministic (no model, no key):
   call edges via a generic tree-sitter extractor, one grammar per language:
   **Rust, C, C++, C#, Ruby, Scala, Elixir, Solidity,
   Dart, Clojure, Nix, Lua, Gleam, Julia**, and **OCaml, Zig, GLSL** (symbols only for now, no call edges).
+  **Bash** (`.sh`, `.bash`) also uses this tier for function symbols;
+  Bash call edges and extensionless shell scripts are not yet supported.
 
 - **Single-file components**: **Vue** (`.vue`), **Svelte** (`.svelte`) and
   **Astro** (`.astro`). The wrapper grammar only locates the `<script>` block
diff --git a/src/graph/generic.ts b/src/graph/generic.ts
index 2b3368c79..ffb1b3e69 100644
--- a/src/graph/generic.ts
+++ b/src/graph/generic.ts
@@ -66,6 +66,7 @@ export const GENERIC_LANGS: readonly GenericLang[] = [
   { name: "powershell", exts: [".ps1", ".psm1"], wasm: "powershell" },
   { name: "julia", exts: [".jl"], wasm: "julia" },
   { name: "glsl", exts: [".glsl"], wasm: "glsl" }, // #293; gdscript is #299
+  { name: "bash", exts: [".sh", ".bash"], wasm: "bash" },
 ];
 
 /**
diff --git a/test/graph-bash.test.ts b/test/graph-bash.test.ts
new file mode 100644
index 000000000..16693a119
--- /dev/null
+++ b/test/graph-bash.test.ts
@@ -0,0 +1,60 @@
+import { test } from "node:test";
+import assert from "node:assert/strict";
+import { mkdtempSync, rmSync, writeFileSync } from "node:fs";
+import { tmpdir } from "node:os";
+import { join } from "node:path";
+import { buildGraph } from "../src/graph/build.js";
+import { checkGraph } from "../src/graph/check.js";
+import { extractGeneric, genericLangOf, isWarm, warmGenericGrammars } from "../src/graph/generic.js";
+import { unsupportedExtensions } from "../src/graph/source-files.js";
+import { readGraph, wiringPath } from "../src/graph/write.js";
+
+const BASH = `#!/usr/bin/env bash
+greet() {
+  printf '%s\\n' "hello"
+}
+
+function main {
+  greet
+}
+`;
+
+test("Bash extensions load the bundled grammar and extract both function syntaxes", async () => {
+  for (const file of ["setup.sh", "library.bash", "SETUP.SH"]) {
+    assert.equal(genericLangOf(file)?.name, "bash");
+  }
+  assert.deepEqual(unsupportedExtensions([".sh", ".bash", "SH"]), []);
+  await warmGenericGrammars(["bash"]);
+  assert.ok(isWarm("bash"), "the bundled Bash grammar loads");
+  const { nodes } = extractGeneric("setup.sh", BASH, "bash");
+  assert.deepEqual(
+    nodes.filter((node) => node.kind !== "file").map((node) => [node.name, node.kind, node.span]),
+    [["greet", "function", "L2-L4"], ["main", "function", "L6-L8"]],
+  );
+});
+
+test("buildGraph indexes Bash files and checkGraph reports them in sync", async () => {
+  const dir = mkdtempSync(join(tmpdir(), "graft-bash-"));
+  try {
+    writeFileSync(join(dir, "setup.sh"), BASH);
+    writeFileSync(join(dir, "library.bash"), "helper() { echo ok; }\n");
+    writeFileSync(join(dir, "extensionless"), BASH);
+
+    const result = await buildGraph(dir, { reuse: false });
+    assert.deepEqual(result.languages, ["bash"]);
+    assert.deepEqual(result.errors, []);
+    const graph = readGraph(wiringPath(result.contextDir));
+    assert.ok(graph);
+    assert.deepEqual(
+      graph.nodes.filter((node) => node.kind === "file").map((node) => node.path).sort(),
+      ["library.bash", "setup.sh"],
+    );
+    assert.deepEqual(
+      graph.nodes.filter((node) => node.kind === "function").map((node) => node.name).sort(),
+      ["greet", "helper", "main"],
+    );
+    assert.equal((await checkGraph(dir)).ok, true);
+  } finally {
+    rmSync(dir, { recursive: true, force: true });
+  }
+});