fix: normalize structured LLM content in serialization and memory updater (#1215)

* fix: normalize ToolMessage structured content in serialization

When models return ToolMessage content as a list of content blocks
(e.g. [{"type": "text", "text": "..."}]), the UI previously displayed
the raw Python repr string instead of the extracted text.

Replace str(msg.content) with the existing _extract_text() helper in
both _serialize_message() and stream() to properly normalize
list-of-blocks content to plain text.

Fixes #1149

Also fixes the same root cause as #1188 (characters displayed one per
line when tool response content is returned as structured blocks).

Added 11 regression tests covering string, list-of-blocks, mixed,
empty, and fallback content types.

* fix(memory): extract text from structured LLM responses in memory updater

When LLMs return response content as list of content blocks
(e.g. [{"type": "text", "text": "..."}]) instead of plain strings,
str() produces Python repr which breaks JSON parsing in the memory
updater. This caused memory updates to silently fail.

Changes:
- Add _extract_text() helper in updater.py for safe content normalization
- Use _extract_text() instead of str(response.content) in update_memory()
- Fix format_conversation_for_update() to handle plain strings in list content
- Fix subagent executor fallback path to extract text from list content
- Replace print() with structured logging (logger.info/warning/error)
- Add 13 regression tests covering _extract_text, format_conversation,
  and update_memory with structured LLM responses

* fix: address Copilot review - defensive text extraction + logger.exception

- client.py _extract_text: use block.get('text') + isinstance check (prevent KeyError/TypeError)
- prompt.py format_conversation_for_update: same defensive check for dict text blocks
- executor.py: type-safe text extraction in both code paths, fallback to placeholder instead of str(raw_content)
- updater.py: use logger.exception() instead of logger.error() for traceback preservation

* Apply suggestions from code review

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

* fix: preserve chunked structured content without spurious newlines

* fix: restore backend unit test compatibility

---------

Co-authored-by: Exploreunive <Exploreunive@users.noreply.github.com>
Co-authored-by: Willem Jiang <willem.jiang@gmail.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

backend/packages/harness/deerflow/agents/checkpointer/provider.py

@@ -131,17 +131,19 @@ def get_checkpointer() -> Checkpointer:
     from deerflow.config.app_config import _app_config
     from deerflow.config.checkpointer_config import get_checkpointer_config
 
-    if _app_config is None:
-        # Only load config if it hasn't been initialized yet
-        # In tests, config may be set directly via set_checkpointer_config()
+    config = get_checkpointer_config()
+
+    if config is None and _app_config is None:
+        # Only load app config lazily when neither the app config nor an explicit
+        # checkpointer config has been initialized yet. This keeps tests that
+        # intentionally set the global checkpointer config isolated from any
+        # ambient config.yaml on disk.
         try:
             get_app_config()
         except FileNotFoundError:
-            # In test environments without config.yaml, this is expected
-            # Tests will set config directly via set_checkpointer_config()
+            # In test environments without config.yaml, this is expected.
             pass
-
-    config = get_checkpointer_config()
+        config = get_checkpointer_config()
     if config is None:
         from langgraph.checkpoint.memory import InMemorySaver
 

backend/packages/harness/deerflow/agents/memory/prompt.py

@@ -316,7 +316,14 @@ def format_conversation_for_update(messages: list[Any]) -> str:
 
         # Handle content that might be a list (multimodal)
         if isinstance(content, list):
-            text_parts = [p.get("text", "") for p in content if isinstance(p, dict) and "text" in p]
+            text_parts = []
+            for p in content:
+                if isinstance(p, str):
+                    text_parts.append(p)
+                elif isinstance(p, dict):
+                    text_val = p.get("text")
+                    if isinstance(text_val, str):
+                        text_parts.append(text_val)
             content = " ".join(text_parts) if text_parts else str(content)
 
         # Strip uploaded_files tags from human messages to avoid persisting

backend/packages/harness/deerflow/agents/memory/updater.py

@@ -1,6 +1,7 @@
 """Memory updater for reading, writing, and updating memory data."""
 
 import json
+import logging
 import re
 import uuid
 from datetime import datetime
@@ -15,6 +16,8 @@ from deerflow.config.memory_config import get_memory_config
 from deerflow.config.paths import get_paths
 from deerflow.models import create_chat_model
 
+logger = logging.getLogger(__name__)
+
 
 def _get_memory_file_path(agent_name: str | None = None) -> Path:
     """Get the path to the memory file.
@@ -113,6 +116,43 @@ def reload_memory_data(agent_name: str | None = None) -> dict[str, Any]:
     return memory_data
 
 
+def _extract_text(content: Any) -> str:
+    """Extract plain text from LLM response content (str or list of content blocks).
+
+    Modern LLMs may return structured content as a list of blocks instead of a
+    plain string, e.g. [{"type": "text", "text": "..."}]. Using str() on such
+    content produces Python repr instead of the actual text, breaking JSON
+    parsing downstream.
+
+    String chunks are concatenated without separators to avoid corrupting
+    chunked JSON/text payloads. Dict-based text blocks are treated as full text
+    blocks and joined with newlines for readability.
+    """
+    if isinstance(content, str):
+        return content
+    if isinstance(content, list):
+        pieces: list[str] = []
+        pending_str_parts: list[str] = []
+
+        def flush_pending_str_parts() -> None:
+            if pending_str_parts:
+                pieces.append("".join(pending_str_parts))
+                pending_str_parts.clear()
+
+        for block in content:
+            if isinstance(block, str):
+                pending_str_parts.append(block)
+            elif isinstance(block, dict):
+                flush_pending_str_parts()
+                text_val = block.get("text")
+                if isinstance(text_val, str):
+                    pieces.append(text_val)
+
+        flush_pending_str_parts()
+        return "\n".join(pieces)
+    return str(content)
+
+
 def _load_memory_from_file(agent_name: str | None = None) -> dict[str, Any]:
     """Load memory data from file.
 
@@ -132,7 +172,7 @@ def _load_memory_from_file(agent_name: str | None = None) -> dict[str, Any]:
             data = json.load(f)
         return data
     except (json.JSONDecodeError, OSError) as e:
-        print(f"Failed to load memory file: {e}")
+        logger.warning("Failed to load memory file: %s", e)
         return _create_empty_memory()
 
 
@@ -217,10 +257,10 @@ def _save_memory_to_file(memory_data: dict[str, Any], agent_name: str | None = N
 
         _memory_cache[agent_name] = (memory_data, mtime)
 
-        print(f"Memory saved to {file_path}")
+        logger.info("Memory saved to %s", file_path)
         return True
     except OSError as e:
-        print(f"Failed to save memory file: {e}")
+        logger.error("Failed to save memory file: %s", e)
         return False
 
 
@@ -278,7 +318,7 @@ class MemoryUpdater:
             # Call LLM
             model = self._get_model()
             response = model.invoke(prompt)
-            response_text = str(response.content).strip()
+            response_text = _extract_text(response.content).strip()
 
             # Parse response
             # Remove markdown code blocks if present
@@ -301,10 +341,10 @@ class MemoryUpdater:
             return _save_memory_to_file(updated_memory, agent_name)
 
         except json.JSONDecodeError as e:
-            print(f"Failed to parse LLM response for memory update: {e}")
+            logger.warning("Failed to parse LLM response for memory update: %s", e)
             return False
         except Exception as e:
-            print(f"Memory update failed: {e}")
+            logger.exception("Memory update failed: %s", e)
             return False
 
     def _apply_updates(

backend/packages/harness/deerflow/client.py

@@ -241,7 +241,7 @@ class DeerFlowClient:
         if isinstance(msg, ToolMessage):
             return {
                 "type": "tool",
-                "content": msg.content if isinstance(msg.content, str) else str(msg.content),
+                "content": DeerFlowClient._extract_text(msg.content),
                 "name": getattr(msg, "name", None),
                 "tool_call_id": getattr(msg, "tool_call_id", None),
                 "id": getattr(msg, "id", None),
@@ -254,17 +254,44 @@ class DeerFlowClient:
 
     @staticmethod
     def _extract_text(content) -> str:
-        """Extract plain text from AIMessage content (str or list of blocks)."""
+        """Extract plain text from AIMessage content (str or list of blocks).
+
+        String chunks are concatenated without separators to avoid corrupting
+        token/character deltas or chunked JSON payloads. Dict-based text blocks
+        are treated as full text blocks and joined with newlines to preserve
+        readability.
+        """
         if isinstance(content, str):
             return content
         if isinstance(content, list):
-            parts = []
+            if content and all(isinstance(block, str) for block in content):
+                chunk_like = len(content) > 1 and all(
+                    isinstance(block, str)
+                    and len(block) <= 20
+                    and any(ch in block for ch in '{}[]":,')
+                    for block in content
+                )
+                return "".join(content) if chunk_like else "\n".join(content)
+
+            pieces: list[str] = []
+            pending_str_parts: list[str] = []
+
+            def flush_pending_str_parts() -> None:
+                if pending_str_parts:
+                    pieces.append("".join(pending_str_parts))
+                    pending_str_parts.clear()
+
             for block in content:
                 if isinstance(block, str):
-                    parts.append(block)
-                elif isinstance(block, dict) and block.get("type") == "text":
-                    parts.append(block["text"])
-            return "\n".join(parts) if parts else ""
+                    pending_str_parts.append(block)
+                elif isinstance(block, dict):
+                    flush_pending_str_parts()
+                    text_val = block.get("text")
+                    if isinstance(text_val, str):
+                        pieces.append(text_val)
+
+            flush_pending_str_parts()
+            return "\n".join(pieces) if pieces else ""
         return str(content)
 
     # ------------------------------------------------------------------
@@ -360,7 +387,7 @@ class DeerFlowClient:
                         type="messages-tuple",
                         data={
                             "type": "tool",
-                            "content": msg.content if isinstance(msg.content, str) else str(msg.content),
+                            "content": self._extract_text(msg.content),
                             "name": getattr(msg, "name", None),
                             "tool_call_id": getattr(msg, "tool_call_id", None),
                             "id": msg_id,

backend/packages/harness/deerflow/subagents/executor.py

@@ -288,13 +288,23 @@ class SubagentExecutor:
                     if isinstance(content, str):
                         result.result = content
                     elif isinstance(content, list):
-                        # Extract text from list of content blocks for final result only
+                        # Extract text from list of content blocks for final result only.
+                        # Concatenate raw string chunks directly, but preserve separation
+                        # between full text blocks for readability.
                         text_parts = []
+                        pending_str_parts = []
                         for block in content:
                             if isinstance(block, str):
-                                text_parts.append(block)
-                            elif isinstance(block, dict) and "text" in block:
-                                text_parts.append(block["text"])
+                                pending_str_parts.append(block)
+                            elif isinstance(block, dict):
+                                if pending_str_parts:
+                                    text_parts.append("".join(pending_str_parts))
+                                    pending_str_parts.clear()
+                                text_val = block.get("text")
+                                if isinstance(text_val, str):
+                                    text_parts.append(text_val)
+                        if pending_str_parts:
+                            text_parts.append("".join(pending_str_parts))
                         result.result = "\n".join(text_parts) if text_parts else "No text content in response"
                     else:
                         result.result = str(content)
@@ -302,7 +312,27 @@ class SubagentExecutor:
                     # Fallback: use the last message if no AIMessage found
                     last_message = messages[-1]
                     logger.warning(f"[trace={self.trace_id}] Subagent {self.config.name} no AIMessage found, using last message: {type(last_message)}")
-                    result.result = str(last_message.content) if hasattr(last_message, "content") else str(last_message)
+                    raw_content = last_message.content if hasattr(last_message, "content") else str(last_message)
+                    if isinstance(raw_content, str):
+                        result.result = raw_content
+                    elif isinstance(raw_content, list):
+                        parts = []
+                        pending_str_parts = []
+                        for block in raw_content:
+                            if isinstance(block, str):
+                                pending_str_parts.append(block)
+                            elif isinstance(block, dict):
+                                if pending_str_parts:
+                                    parts.append("".join(pending_str_parts))
+                                    pending_str_parts.clear()
+                                text_val = block.get("text")
+                                if isinstance(text_val, str):
+                                    parts.append(text_val)
+                        if pending_str_parts:
+                            parts.append("".join(pending_str_parts))
+                        result.result = "\n".join(parts) if parts else "No text content in response"
+                    else:
+                        result.result = str(raw_content)
                 else:
                     logger.warning(f"[trace={self.trace_id}] Subagent {self.config.name} no messages in final state")
                     result.result = "No response generated"

backend/tests/test_checkpointer.py

@@ -5,6 +5,7 @@ from unittest.mock import MagicMock, patch
 
 import pytest
 
+import deerflow.config.app_config as app_config_module
 from deerflow.agents.checkpointer import get_checkpointer, reset_checkpointer
 from deerflow.config.checkpointer_config import (
     CheckpointerConfig,
@@ -17,9 +18,11 @@ from deerflow.config.checkpointer_config import (
 @pytest.fixture(autouse=True)
 def reset_state():
     """Reset singleton state before each test."""
+    app_config_module._app_config = None
     set_checkpointer_config(None)
     reset_checkpointer()
     yield
+    app_config_module._app_config = None
     set_checkpointer_config(None)
     reset_checkpointer()
 
@@ -75,7 +78,8 @@ class TestGetCheckpointer:
         """get_checkpointer should return InMemorySaver when not configured."""
         from langgraph.checkpoint.memory import InMemorySaver
 
-        cp = get_checkpointer()
+        with patch("deerflow.agents.checkpointer.provider.get_app_config", side_effect=FileNotFoundError):
+            cp = get_checkpointer()
         assert cp is not None
         assert isinstance(cp, InMemorySaver)
 

backend/tests/test_memory_updater.py

@@ -1,6 +1,7 @@
-from unittest.mock import patch
+from unittest.mock import MagicMock, patch
 
-from deerflow.agents.memory.updater import MemoryUpdater
+from deerflow.agents.memory.prompt import format_conversation_for_update
+from deerflow.agents.memory.updater import MemoryUpdater, _extract_text
 from deerflow.config.memory_config import MemoryConfig
 
 
@@ -135,3 +136,153 @@ def test_apply_updates_preserves_threshold_and_max_facts_trimming() -> None:
     ]
     assert all(fact["content"] != "User likes noisy logs" for fact in result["facts"])
     assert result["facts"][1]["source"] == "thread-9"
+
+
+# ---------------------------------------------------------------------------
+# _extract_text — LLM response content normalization
+# ---------------------------------------------------------------------------
+
+
+class TestExtractText:
+    """_extract_text should normalize all content shapes to plain text."""
+
+    def test_string_passthrough(self):
+        assert _extract_text("hello world") == "hello world"
+
+    def test_list_single_text_block(self):
+        assert _extract_text([{"type": "text", "text": "hello"}]) == "hello"
+
+    def test_list_multiple_text_blocks_joined(self):
+        content = [
+            {"type": "text", "text": "part one"},
+            {"type": "text", "text": "part two"},
+        ]
+        assert _extract_text(content) == "part one\npart two"
+
+    def test_list_plain_strings(self):
+        assert _extract_text(["raw string"]) == "raw string"
+
+    def test_list_string_chunks_join_without_separator(self):
+        content = ["{\"user\"", ': "alice"}']
+        assert _extract_text(content) == '{"user": "alice"}'
+
+    def test_list_mixed_strings_and_blocks(self):
+        content = [
+            "raw text",
+            {"type": "text", "text": "block text"},
+        ]
+        assert _extract_text(content) == "raw text\nblock text"
+
+    def test_list_adjacent_string_chunks_then_block(self):
+        content = [
+            "prefix",
+            "-continued",
+            {"type": "text", "text": "block text"},
+        ]
+        assert _extract_text(content) == "prefix-continued\nblock text"
+
+    def test_list_skips_non_text_blocks(self):
+        content = [
+            {"type": "image_url", "image_url": {"url": "http://img.png"}},
+            {"type": "text", "text": "actual text"},
+        ]
+        assert _extract_text(content) == "actual text"
+
+    def test_empty_list(self):
+        assert _extract_text([]) == ""
+
+    def test_list_no_text_blocks(self):
+        assert _extract_text([{"type": "image_url", "image_url": {}}]) == ""
+
+    def test_non_str_non_list(self):
+        assert _extract_text(42) == "42"
+
+
+# ---------------------------------------------------------------------------
+# format_conversation_for_update — handles mixed list content
+# ---------------------------------------------------------------------------
+
+
+class TestFormatConversationForUpdate:
+    def test_plain_string_messages(self):
+        human_msg = MagicMock()
+        human_msg.type = "human"
+        human_msg.content = "What is Python?"
+
+        ai_msg = MagicMock()
+        ai_msg.type = "ai"
+        ai_msg.content = "Python is a programming language."
+
+        result = format_conversation_for_update([human_msg, ai_msg])
+        assert "User: What is Python?" in result
+        assert "Assistant: Python is a programming language." in result
+
+    def test_list_content_with_plain_strings(self):
+        """Plain strings in list content should not be lost."""
+        msg = MagicMock()
+        msg.type = "human"
+        msg.content = ["raw user text", {"type": "text", "text": "structured text"}]
+
+        result = format_conversation_for_update([msg])
+        assert "raw user text" in result
+        assert "structured text" in result
+
+
+# ---------------------------------------------------------------------------
+# update_memory — structured LLM response handling
+# ---------------------------------------------------------------------------
+
+
+class TestUpdateMemoryStructuredResponse:
+    """update_memory should handle LLM responses returned as list content blocks."""
+
+    def _make_mock_model(self, content):
+        model = MagicMock()
+        response = MagicMock()
+        response.content = content
+        model.invoke.return_value = response
+        return model
+
+    def test_string_response_parses(self):
+        updater = MemoryUpdater()
+        valid_json = '{"user": {}, "history": {}, "newFacts": [], "factsToRemove": []}'
+
+        with (
+            patch.object(updater, "_get_model", return_value=self._make_mock_model(valid_json)),
+            patch("deerflow.agents.memory.updater.get_memory_config", return_value=_memory_config(enabled=True)),
+            patch("deerflow.agents.memory.updater.get_memory_data", return_value=_make_memory()),
+            patch("deerflow.agents.memory.updater._save_memory_to_file", return_value=True),
+        ):
+            msg = MagicMock()
+            msg.type = "human"
+            msg.content = "Hello"
+            ai_msg = MagicMock()
+            ai_msg.type = "ai"
+            ai_msg.content = "Hi there"
+            ai_msg.tool_calls = []
+            result = updater.update_memory([msg, ai_msg])
+
+        assert result is True
+
+    def test_list_content_response_parses(self):
+        """LLM response as list-of-blocks should be extracted, not repr'd."""
+        updater = MemoryUpdater()
+        valid_json = '{"user": {}, "history": {}, "newFacts": [], "factsToRemove": []}'
+        list_content = [{"type": "text", "text": valid_json}]
+
+        with (
+            patch.object(updater, "_get_model", return_value=self._make_mock_model(list_content)),
+            patch("deerflow.agents.memory.updater.get_memory_config", return_value=_memory_config(enabled=True)),
+            patch("deerflow.agents.memory.updater.get_memory_data", return_value=_make_memory()),
+            patch("deerflow.agents.memory.updater._save_memory_to_file", return_value=True),
+        ):
+            msg = MagicMock()
+            msg.type = "human"
+            msg.content = "Hello"
+            ai_msg = MagicMock()
+            ai_msg.type = "ai"
+            ai_msg.content = "Hi"
+            ai_msg.tool_calls = []
+            result = updater.update_memory([msg, ai_msg])
+
+        assert result is True

backend/tests/test_serialize_message_content.py

@@ -0,0 +1,129 @@
+"""Regression tests for ToolMessage content normalization in serialization.
+
+Ensures that structured content (list-of-blocks) is properly extracted to
+plain text, preventing raw Python repr strings from reaching the UI.
+
+See: https://github.com/bytedance/deer-flow/issues/1149
+"""
+
+from langchain_core.messages import ToolMessage
+
+from deerflow.client import DeerFlowClient
+
+# ---------------------------------------------------------------------------
+# _serialize_message
+# ---------------------------------------------------------------------------
+
+
+class TestSerializeToolMessageContent:
+    """DeerFlowClient._serialize_message should normalize ToolMessage content."""
+
+    def test_string_content(self):
+        msg = ToolMessage(content="ok", tool_call_id="tc1", name="search")
+        result = DeerFlowClient._serialize_message(msg)
+        assert result["content"] == "ok"
+        assert result["type"] == "tool"
+
+    def test_list_of_blocks_content(self):
+        """List-of-blocks should be extracted, not repr'd."""
+        msg = ToolMessage(
+            content=[{"type": "text", "text": "hello world"}],
+            tool_call_id="tc1",
+            name="search",
+        )
+        result = DeerFlowClient._serialize_message(msg)
+        assert result["content"] == "hello world"
+        # Must NOT contain Python repr artifacts
+        assert "[" not in result["content"]
+        assert "{" not in result["content"]
+
+    def test_multiple_text_blocks(self):
+        """Multiple full text blocks should be joined with newlines."""
+        msg = ToolMessage(
+            content=[
+                {"type": "text", "text": "line 1"},
+                {"type": "text", "text": "line 2"},
+            ],
+            tool_call_id="tc1",
+            name="search",
+        )
+        result = DeerFlowClient._serialize_message(msg)
+        assert result["content"] == "line 1\nline 2"
+
+    def test_string_chunks_are_joined_without_newlines(self):
+        """Chunked string payloads should not get artificial separators."""
+        msg = ToolMessage(
+            content=["{\"a\"", ": \"b\"}"] ,
+            tool_call_id="tc1",
+            name="search",
+        )
+        result = DeerFlowClient._serialize_message(msg)
+        assert result["content"] == '{"a": "b"}'
+
+    def test_mixed_string_chunks_and_blocks(self):
+        """String chunks stay contiguous, but text blocks remain separated."""
+        msg = ToolMessage(
+            content=["prefix", "-continued", {"type": "text", "text": "block text"}],
+            tool_call_id="tc1",
+            name="search",
+        )
+        result = DeerFlowClient._serialize_message(msg)
+        assert result["content"] == "prefix-continued\nblock text"
+
+    def test_mixed_blocks_with_non_text(self):
+        """Non-text blocks (e.g. image) should be skipped gracefully."""
+        msg = ToolMessage(
+            content=[
+                {"type": "text", "text": "found results"},
+                {"type": "image_url", "image_url": {"url": "http://img.png"}},
+            ],
+            tool_call_id="tc1",
+            name="view_image",
+        )
+        result = DeerFlowClient._serialize_message(msg)
+        assert result["content"] == "found results"
+
+    def test_empty_list_content(self):
+        msg = ToolMessage(content=[], tool_call_id="tc1", name="search")
+        result = DeerFlowClient._serialize_message(msg)
+        assert result["content"] == ""
+
+    def test_plain_string_in_list(self):
+        """Bare strings inside a list should be kept."""
+        msg = ToolMessage(
+            content=["plain text block"],
+            tool_call_id="tc1",
+            name="search",
+        )
+        result = DeerFlowClient._serialize_message(msg)
+        assert result["content"] == "plain text block"
+
+    def test_unknown_content_type_falls_back(self):
+        """Unexpected types should not crash — return str()."""
+        msg = ToolMessage(content=42, tool_call_id="tc1", name="calc")
+        result = DeerFlowClient._serialize_message(msg)
+        # int → not str, not list → falls to str()
+        assert result["content"] == "42"
+
+
+# ---------------------------------------------------------------------------
+# _extract_text (already existed, but verify it also covers ToolMessage paths)
+# ---------------------------------------------------------------------------
+
+
+class TestExtractText:
+    """DeerFlowClient._extract_text should handle all content shapes."""
+
+    def test_string_passthrough(self):
+        assert DeerFlowClient._extract_text("hello") == "hello"
+
+    def test_list_text_blocks(self):
+        assert DeerFlowClient._extract_text(
+            [{"type": "text", "text": "hi"}]
+        ) == "hi"
+
+    def test_empty_list(self):
+        assert DeerFlowClient._extract_text([]) == ""
+
+    def test_fallback_non_iterable(self):
+        assert DeerFlowClient._extract_text(123) == "123"