From f0f3d8004e92e653c484c18ac5efe05eb34d280e Mon Sep 17 00:00:00 2001
From: Michael Panchenko <michael.panchenko@oraios-ai.de>
Date: Sun, 21 Dec 2025 10:17:50 +0700
Subject: [PATCH] Improved output structure in GetSymbolsOverviewTool, major
 token reduction!

As before for jetbrains, grouped and without name_path. Using recursive structure for depth>0
---
 src/serena/tools/jetbrains_tools.py |  2 ++
 src/serena/tools/symbol_tools.py    | 46 +++++++++++++++++++++++++++--
 2 files changed, 46 insertions(+), 2 deletions(-)

diff --git a/src/serena/tools/jetbrains_tools.py b/src/serena/tools/jetbrains_tools.py
index 0c9f644e..846ccb6f 100644
--- a/src/serena/tools/jetbrains_tools.py
+++ b/src/serena/tools/jetbrains_tools.py
@@ -98,6 +98,8 @@ class JetBrainsGetSymbolsOverviewTool(Tool, ToolMarkerSymbolicRead, ToolMarkerOp
     Retrieves an overview of the top-level symbols within a specified file using the JetBrains backend
     """
 
+    # TODO: support depth param, structure output in the same way as in GetSymbolsOverviewTool (by extracting
+    #   the post-processing logic used there into a shared utility function)
     def apply(
         self,
         relative_path: str,
diff --git a/src/serena/tools/symbol_tools.py b/src/serena/tools/symbol_tools.py
index 75d0e5c9..64261874 100644
--- a/src/serena/tools/symbol_tools.py
+++ b/src/serena/tools/symbol_tools.py
@@ -7,6 +7,8 @@ from collections.abc import Sequence
 from copy import copy
 from typing import Any
 
+from black.trans import defaultdict
+
 from serena.tools import (
     SUCCESS_RESULT,
     Tool,
@@ -61,7 +63,15 @@ class GetSymbolsOverviewTool(Tool, ToolMarkerSymbolicRead):
         :param max_answer_chars: if the overview is longer than this number of characters,
             no content will be returned. -1 means the default value from the config will be used.
             Don't adjust unless there is really no other way to get the content required for the task.
-        :return: a JSON object containing info about top-level symbols in the file
+        :return: a JSON object containing symbols grouped by kind in a compact format.
+
+            For depth=0, the structure is: {"Kind1": ["name1", "name2"], "Kind2": ["name3"]}
+            For depth>0, symbols with children use nested dicts: {"Class": [{"ClassName": {"Method": ["method1", "method2"]}}]}
+
+            The name_path for any symbol can be inferred from the structure:
+            - Top-level symbols: name_path = name
+            - Nested symbols: name_path = parent_name + "/" + name
+            Example: "method1" under class "ClassName" has name_path "ClassName/method1"
         """
         symbol_retriever = self.create_language_server_symbol_retriever()
         file_path = os.path.join(self.project.project_root, relative_path)
@@ -73,9 +83,41 @@ class GetSymbolsOverviewTool(Tool, ToolMarkerSymbolicRead):
         if os.path.isdir(file_path):
             raise ValueError(f"Expected a file path, but got a directory path: {relative_path}. ")
         result = symbol_retriever.get_symbol_overview(relative_path, depth=depth)[relative_path]
-        result_json_str = self._to_json(result)
+        # Transform to compact format
+        compact_result = self._transform_symbols_to_compact_format(result)
+        result_json_str = self._to_json(compact_result)
         return self._limit_length(result_json_str, max_answer_chars)
 
+    @staticmethod
+    def _transform_symbols_to_compact_format(symbols: list[dict[str, Any]]) -> dict[str, list]:
+        """
+        Transform symbol overview from verbose format to compact grouped format.
+
+        Groups symbols by kind and uses names instead of full symbol objects.
+        For symbols with children, creates nested dictionaries.
+
+        The name_path can be inferred from the hierarchical structure:
+        - Top-level symbols: name_path = name
+        - Nested symbols: name_path = parent_name + "/" + name
+        For example, "convert" under class "ProjectType" has name_path "ProjectType/convert".
+        """
+        result = defaultdict(list)
+
+        for symbol in symbols:
+            kind = symbol.get("kind", "Unknown")
+            name = symbol.get("name", "unknown")
+            children = symbol.get("children", [])
+
+            if children:
+                # Symbol has children: create nested dict {name: children_dict}
+                children_dict = GetSymbolsOverviewTool._transform_symbols_to_compact_format(children)
+                result[kind].append({name: children_dict})
+            else:
+                # Symbol has no children: just add the name
+                result[kind].append(name)
+
+        return result
+
 
 class FindSymbolTool(Tool, ToolMarkerSymbolicRead):
     """