From 32ecf92afcf647f257b430c748cbe6bb5fa0fac4 Mon Sep 17 00:00:00 2001
From: Linlang <30293408+SunsetWolf@users.noreply.github.com>
Date: Mon, 3 Nov 2025 21:52:41 +0800
Subject: [PATCH] fix: handle mixed str and dict types in code_list (#1279)

* fix: handle mixed str and dict types in code_list

* fix: handle missing token_costs entry for loop 0 in summarize_win
---
 .../components/coder/factor_coder/evolving_strategy.py    | 7 ++++++-
 rdagent/log/ui/ds_trace.py                                | 8 ++++++--
 2 files changed, 12 insertions(+), 3 deletions(-)

diff --git a/rdagent/components/coder/factor_coder/evolving_strategy.py b/rdagent/components/coder/factor_coder/evolving_strategy.py
index 157f4148..338ed941 100644
--- a/rdagent/components/coder/factor_coder/evolving_strategy.py
+++ b/rdagent/components/coder/factor_coder/evolving_strategy.py
@@ -169,5 +169,10 @@ class FactorMultiProcessEvolvingStrategy(MultiProcessEvolvingStrategy):
                 continue
             if evo.sub_workspace_list[index] is None:
                 evo.sub_workspace_list[index] = FactorFBWorkspace(target_task=evo.sub_tasks[index])
-            evo.sub_workspace_list[index].inject_files(**{"factor.py": code_list[index]})
+            # Since the `implement_one_task` method is not standardized and the `code_list` has both `str` and `dict` data types,
+            # we ended up getting an `TypeError` here, so we chose to fix the problem temporarily with this dirty method.
+            if isinstance(code_list[index], dict):
+                evo.sub_workspace_list[index].inject_files(**code_list[index])
+            else:
+                evo.sub_workspace_list[index].inject_files(**{"factor.py": code_list[index]})
         return evo
diff --git a/rdagent/log/ui/ds_trace.py b/rdagent/log/ui/ds_trace.py
index f2648a5f..a489df3a 100644
--- a/rdagent/log/ui/ds_trace.py
+++ b/rdagent/log/ui/ds_trace.py
@@ -93,7 +93,7 @@ def load_data(log_path: Path):
             continue
         li, fn = extract_loopid_func_name(msg.tag)
         ei = extract_evoid(msg.tag)
-        if li:
+        if li is not None:
             li = int(li)
         if ei is not None:
             ei = int(ei)
@@ -784,7 +784,11 @@ def summarize_win():
                 for k, v in loop_data["direct_exp_gen"]["no_tag"].hypothesis.__dict__.items()
                 if k not in ["component", "hypothesis", "reason"] and v is not None
             }
-            df.loc[loop, "COST($)"] = sum(tc.content["cost"] for tc in state.token_costs[loop])
+            # In the test before 0.8.0 release, we found that when running `ui` of `data_science` (custom dataset),
+            # when `loop=0`, it doesn't exist in `state.token_costs.keys`, and we will get `KeyError` when running it,
+            # so we have fixed the problem with this dirty method for the time being.
+            if loop in state.token_costs:
+                df.loc[loop, "COST($)"] = sum(tc.content["cost"] for tc in state.token_costs[loop])
 
             # Time Stats
             exp_gen_time = timedelta()