feat: add glob llm tool (#1495)

lua/avante/llm_tools.lua

@@ -38,6 +38,16 @@ local function has_permission_to_access(abs_path)
   return not Utils.is_ignored(rel_path, gitignore_patterns, gitignore_negate_patterns)
 end
 
+---@type AvanteLLMToolFunc<{ rel_path: string, pattern: string }>
+function M.glob(opts, on_log)
+  local abs_path = get_abs_path(opts.rel_path)
+  if not has_permission_to_access(abs_path) then return "", "No permission to access path: " .. abs_path end
+  if on_log then on_log("path: " .. abs_path) end
+  if on_log then on_log("pattern: " .. opts.pattern) end
+  local files = vim.fn.glob(abs_path .. "/" .. opts.pattern, true, true)
+  return vim.json.encode(files), nil
+end
+
 ---@type AvanteLLMToolFunc<{ rel_path: string, max_depth?: integer }>
 function M.list_files(opts, on_log)
   local abs_path = get_abs_path(opts.rel_path)
@@ -613,6 +623,38 @@ end
 
 ---@type AvanteLLMTool[]
 M._tools = {
+  {
+    name = "glob",
+    description = 'Fast file pattern matching using glob patterns like "**/*.js"',
+    param = {
+      type = "table",
+      fields = {
+        {
+          name = "pattern",
+          description = "Glob pattern",
+          type = "string",
+        },
+        {
+          name = "rel_path",
+          description = "Relative path to the directory, as cwd",
+          type = "string",
+        },
+      },
+    },
+    returns = {
+      {
+        name = "matches",
+        description = "List of matched files",
+        type = "string",
+      },
+      {
+        name = "err",
+        description = "Error message",
+        type = "string",
+        optional = true,
+      },
+    },
+  },
   {
     name = "rag_search",
     enabled = function() return Config.rag_service.enabled and RagService.is_ready() end,
@@ -643,7 +685,7 @@ M._tools = {
   },
   {
     name = "python",
-    description = "Run python code",
+    description = "Run python code. Can't use it to read files or modify files.",
     param = {
       type = "table",
       fields = {
@@ -858,7 +900,7 @@ M._tools = {
   },
   {
     name = "read_file",
-    description = "Read the contents of a file",
+    description = "Read the contents of a file. If the file content is already in the context, do not use this tool.",
     param = {
       type = "table",
       fields = {
@@ -1057,7 +1099,7 @@ M._tools = {
   },
   {
     name = "bash",
-    description = "Run a bash command in a directory",
+    description = "Run a bash command in a directory. Can't use search commands like find/grep or read tools like cat/ls.",
     param = {
       type = "table",
       fields = {

tests/llm_tools_spec.lua

@@ -272,4 +272,76 @@ describe("llm_tools", function()
       assert.equals("Hello from custom container\n", result)
     end)
   end)
+
+  describe("glob", function()
+    it("should find files matching the pattern", function()
+      -- Create some additional test files with different extensions for glob testing
+      os.execute("touch " .. test_dir .. "/file1.lua")
+      os.execute("touch " .. test_dir .. "/file2.lua")
+      os.execute("touch " .. test_dir .. "/file3.js")
+      os.execute("mkdir -p " .. test_dir .. "/nested")
+      os.execute("touch " .. test_dir .. "/nested/file4.lua")
+
+      -- Test for lua files in the root
+      local result, err = LlmTools.glob({ rel_path = ".", pattern = "*.lua" })
+      assert.is_nil(err)
+      local files = vim.json.decode(result)
+      assert.equals(2, #files)
+      assert.truthy(vim.tbl_contains(files, test_dir .. "/file1.lua"))
+      assert.truthy(vim.tbl_contains(files, test_dir .. "/file2.lua"))
+      assert.falsy(vim.tbl_contains(files, test_dir .. "/file3.js"))
+      assert.falsy(vim.tbl_contains(files, test_dir .. "/nested/file4.lua"))
+
+      -- Test with recursive pattern
+      local result2, err2 = LlmTools.glob({ rel_path = ".", pattern = "**/*.lua" })
+      assert.is_nil(err2)
+      local files2 = vim.json.decode(result2)
+      assert.equals(3, #files2)
+      assert.truthy(vim.tbl_contains(files2, test_dir .. "/file1.lua"))
+      assert.truthy(vim.tbl_contains(files2, test_dir .. "/file2.lua"))
+      assert.truthy(vim.tbl_contains(files2, test_dir .. "/nested/file4.lua"))
+    end)
+
+    it("should respect path permissions", function()
+      local result, err = LlmTools.glob({ rel_path = "../outside_project", pattern = "*.txt" })
+      assert.equals("", result)
+      assert.truthy(err:find("No permission to access path"))
+    end)
+
+    it("should handle patterns without matches", function()
+      local result, err = LlmTools.glob({ rel_path = ".", pattern = "*.nonexistent" })
+      assert.is_nil(err)
+      local files = vim.json.decode(result)
+      assert.equals(0, #files)
+    end)
+
+    it("should handle files in gitignored directories", function()
+      -- Create test files in ignored directory
+      os.execute("touch " .. test_dir .. "/test_dir2/ignored1.lua")
+      os.execute("touch " .. test_dir .. "/test_dir2/ignored2.lua")
+
+      -- Create test files in non-ignored directory
+      os.execute("touch " .. test_dir .. "/test_dir1/notignored1.lua")
+      os.execute("touch " .. test_dir .. "/test_dir1/notignored2.lua")
+
+      local result, err = LlmTools.glob({ rel_path = ".", pattern = "**/*.lua" })
+      assert.is_nil(err)
+      local files = vim.json.decode(result)
+
+      -- Check that files from non-ignored directory are found
+      local found_notignored = false
+      for _, file in ipairs(files) do
+        if file:find("test_dir1/notignored") then
+          found_notignored = true
+          break
+        end
+      end
+      assert.is_true(found_notignored)
+
+      -- Note: By default, vim.fn.glob does not respect gitignore files
+      -- This test simply verifies the glob function works as expected
+      -- If in the future, the function is modified to respect gitignore,
+      -- this test can be updated
+    end)
+  end)
 end)