fix(service): propagate text_keys to ops via get_init_configs in _set…

data_juicer/config/config.py

@@ -1681,8 +1681,6 @@ def get_init_configs(cfg: Union[Namespace, Dict], load_configs_only: bool = True
     """
     set init configs of data-juicer for cfg
     """
-    temp_dir = tempfile.gettempdir()
-    temp_file = os.path.join(temp_dir, "job_dj_config.json")
     if isinstance(cfg, Namespace):
         cfg = namespace_to_dict(cfg)
 
@@ -1703,10 +1701,13 @@ def get_init_configs(cfg: Union[Namespace, Dict], load_configs_only: bool = True
         for attr in internal_attrs:
             cfg.pop(attr, None)
 
-    # create a temp config file
-    with open(temp_file, "w") as f:
-        json.dump(prepare_cfgs_for_export(cfg), f)
-    inited_dj_cfg = init_configs(["--config", temp_file], load_configs_only=load_configs_only)
+    # Use a unique temporary file per call to avoid race conditions when
+    # multiple requests are processed concurrently (e.g. in API service mode).
+    # The file is automatically deleted after the with-block exits.
+    with tempfile.NamedTemporaryFile(mode="w", suffix=".json", prefix="job_dj_config_", delete=True) as temp_f:
+        json.dump(prepare_cfgs_for_export(cfg), temp_f)
+        temp_f.flush()
+        inited_dj_cfg = init_configs(["--config", temp_f.name], load_configs_only=load_configs_only)
     return inited_dj_cfg
 
 

service.py

@@ -7,10 +7,9 @@
 from typing import Dict
 from urllib.parse import parse_qs
 
-from jsonargparse import Namespace
 from pydantic import validate_call
 
-from data_juicer.config.config import get_default_cfg
+from data_juicer.config.config import get_default_cfg, get_init_configs
 from data_juicer.core.data.dataset_builder import DatasetBuilder
 from data_juicer.core.exporter import Exporter
 from data_juicer.utils.lazy_loader import LazyLoader
@@ -143,13 +142,12 @@ def _parse_json_dumps(params: Dict, prefix="<json_dumps>"):
 
 
 def _setup_cfg(params: Dict):
-    """convert string `cfg` to Namespace"""
-    # TODO: Traverse method's signature and convert any arguments \
-    #  that should be Namespace but are passed as str
-    if cfg_str := params.get("cfg"):
-        if isinstance(cfg_str, str):
-            cfg = Namespace(**json.loads(cfg_str))
-            params["cfg"] = cfg
+    """convert string or dict `cfg` to a fully initialized Namespace"""
+    cfg_val = params.get("cfg")
+    if cfg_val is not None and isinstance(cfg_val, (str, dict)):
+        if isinstance(cfg_val, str):
+            cfg_val = json.loads(cfg_val)
+        params["cfg"] = get_init_configs(cfg_val, load_configs_only=True)
     return params