pathlib mkdir hardening + remaining str concat fixes

deruyter92 · Jul 21, 2026 · Jul 20, 2026 · Jul 20, 2026 · Jul 20, 2026 · Jul 20, 2026
commit 6859773b8559987aae87df49a2ab85c2b66145b0
diff --git a/deeplabcut/generate_training_dataset/metadata.py b/deeplabcut/generate_training_dataset/metadata.py
@@ -298,7 +298,7 @@ def create(config: ProjectConfig | dict | Path | str) -> TrainingDatasetMetadata
                 f for f in trainset_path.iterdir() if re.match(r"Documentation_data-.+shuffle[0-9]+\.pickle", f.name)
             ]
         else:
-            trainset_path.mkdir(parents=True)
+            trainset_path.mkdir(parents=True, exist_ok=True)
             shuffle_docs = []
 
         prefix = cfg["Task"] + cfg["date"]

diff --git a/deeplabcut/modelzoo/generalized_data_converter/datasets/base_dlc.py b/deeplabcut/modelzoo/generalized_data_converter/datasets/base_dlc.py
@@ -32,7 +32,7 @@ def __init__(self, proj_root, dataset_name, shuffle=1, modelprefix=""):
         self.proj_root = proj_root
 
         if modelprefix:
-            config_file = Path(self.proj_root) / (modelprefix + "_config.yaml")
+            config_file = Path(self.proj_root) / f"{modelprefix}_config.yaml"
         else:
             config_file = Path(self.proj_root) / "config.yaml"
 

diff --git a/deeplabcut/modelzoo/video_inference.py b/deeplabcut/modelzoo/video_inference.py
@@ -436,15 +436,15 @@ def video_inference_superanimal(
 
             # we prepare the pseudo dataset in the same folder of the target video
             pseudo_dataset_folder = video_path.with_name(f"pseudo_{video_path.stem}")
-            pseudo_dataset_folder.mkdir(exist_ok=True)
+            pseudo_dataset_folder.mkdir(exist_ok=True, parents=True)
             model_folder = pseudo_dataset_folder / "checkpoints"
-            model_folder.mkdir(exist_ok=True)
+            model_folder.mkdir(exist_ok=True, parents=True)
 
             image_folder = pseudo_dataset_folder / "images"
             if image_folder.exists():
                 print(f"{image_folder} exists, skipping the frame extraction")
             else:
-                image_folder.mkdir()
+                image_folder.mkdir(exist_ok=True, parents=True)
                 print(f"Video frames being extracted to {image_folder} for video adaptation.")
                 video_to_frames(video_path, pseudo_dataset_folder, cropping=cropping)
 
@@ -455,7 +455,7 @@ def video_inference_superanimal(
                     f"Delete the folder if you want to re-construct pseudo annotations"
                 )
             else:
-                anno_folder.mkdir()
+                anno_folder.mkdir(exist_ok=True, parents=True)
 
                 if dest_folder is None:
                     pseudo_anno_dir = video_path.parent

diff --git a/deeplabcut/pose_estimation_pytorch/apis/videos.py b/deeplabcut/pose_estimation_pytorch/apis/videos.py
@@ -834,7 +834,7 @@ def _validate_destfolder(destfolder: str | None) -> None:
         output_folder = Path(destfolder)
         if not output_folder.exists():
             print(f"Creating the output folder {output_folder}")
-            output_folder.mkdir(parents=True)
+            output_folder.mkdir(parents=True, exist_ok=True)
 
         assert Path(output_folder).is_dir(), f"Output folder must be a directory: you passed '{output_folder}'"
 

diff --git a/deeplabcut/pose_estimation_tensorflow/core/openvino/session.py b/deeplabcut/pose_estimation_tensorflow/core/openvino/session.py
@@ -26,7 +26,7 @@
 class OpenVINOSession:
     def __init__(self, cfg, device):
         self.core = Core()
-        self.xml_path = cfg["init_weights"] + ".xml"
+        self.xml_path = str(Path(cfg["init_weights"]).with_suffix(".xml"))
         self.device = device
 
         # Convert a frozen graph to OpenVINO IR format
@@ -37,7 +37,7 @@ def __init__(self, cfg, device):
                     "--output_dir",
                     str(Path(cfg["init_weights"]).parent),
                     "--input_model",
-                    cfg["init_weights"] + ".pb",
+                    str(Path(cfg["init_weights"]).with_suffix(".pb")),
                     "--input_shape",
                     "[1, 747, 832, 3]",
                     "--extensions",

diff --git a/deeplabcut/utils/auxfun_multianimal.py b/deeplabcut/utils/auxfun_multianimal.py
@@ -282,7 +282,7 @@ def convert2_maDLC(config: str | Path, userfeedback=True, forceindividual=None):
 
         if askuser == "y" or askuser == "yes" or askuser == "Ja" or askuser == "ha":  # multilanguage support :)
             fn = folder / ("CollectedData_" + cfg["scorer"])
-            Data = pd.read_hdf(str(fn) + ".h5")
+            Data = pd.read_hdf(fn.with_suffix(".h5"))
             conversioncode.guarantee_multiindex_rows(Data)
             imindex = Data.index
 
@@ -341,13 +341,13 @@ def convert2_maDLC(config: str | Path, userfeedback=True, forceindividual=None):
                     dataFrame = pd.concat([dataFrame, frame], axis=1)
 
             Data.to_hdf(
-                fn + "singleanimal.h5",
+                fn.with_name(fn.name + "singleanimal.h5"),
                 key="df_with_missing",
             )
-            Data.to_csv(fn + "singleanimal.csv")
+            Data.to_csv(fn.with_name(fn.name + "singleanimal.csv"))
 
-            dataFrame.to_hdf(fn + ".h5", key="df_with_missing")
-            dataFrame.to_csv(fn + ".csv")
+            dataFrame.to_hdf(fn.with_suffix(".h5"), key="df_with_missing")
+            dataFrame.to_csv(fn.with_suffix(".csv"))
 
 
 def convert_single2multiplelegacyAM(config, userfeedback=True, target=None):
@@ -370,7 +370,7 @@ def convert_single2multiplelegacyAM(config, userfeedback=True, target=None):
 
         if askuser == "y" or askuser == "yes" or askuser == "Ja" or askuser == "ha":  # multilanguage support :)
             fn = folder / ("CollectedData_" + cfg["scorer"])
-            Data = pd.read_hdf(str(fn) + ".h5")
+            Data = pd.read_hdf(fn.with_suffix(".h5"))
             conversioncode.guarantee_multiindex_rows(Data)
             imindex = Data.index
 
@@ -413,16 +413,16 @@ def convert_single2multiplelegacyAM(config, userfeedback=True, target=None):
                         DataFrame = pd.concat([DataFrame, dataFrame], axis=1)
 
                 Data.to_hdf(
-                    fn + "multianimal.h5",
+                    fn.with_name(fn.name + "multianimal.h5"),
                     key="df_with_missing",
                 )
-                Data.to_csv(fn + "multianimal.csv")
+                Data.to_csv(fn.with_name(fn.name + "multianimal.csv"))
 
                 DataFrame.to_hdf(
-                    fn + ".h5",
+                    fn.with_suffix(".h5"),
                     key="df_with_missing",
                 )
-                DataFrame.to_csv(fn + ".csv")
+                DataFrame.to_csv(fn.with_suffix(".csv"))
             elif target is None or target == "multi":
                 print("This is a single animal data set, converting to multi...", folder)
                 for prfxindex, prefix in enumerate(prefixes):
@@ -500,16 +500,16 @@ def convert_single2multiplelegacyAM(config, userfeedback=True, target=None):
                         DataFrame = pd.concat([DataFrame, dataFrame], axis=1)
 
                 Data.to_hdf(
-                    fn + "singleanimal.h5",
+                    fn.with_name(fn.name + "singleanimal.h5"),
                     key="df_with_missing",
                 )
-                Data.to_csv(fn + "singleanimal.csv")
+                Data.to_csv(fn.with_name(fn.name + "singleanimal.csv"))
 
                 DataFrame.to_hdf(
-                    fn + ".h5",
+                    fn.with_suffix(".h5"),
                     key="df_with_missing",
                 )
-                DataFrame.to_csv(fn + ".csv")
+                DataFrame.to_csv(fn.with_suffix(".csv"))
 
 
 def form_default_inferencecfg(cfg):

diff --git a/deeplabcut/utils/auxfun_videos.py b/deeplabcut/utils/auxfun_videos.py
@@ -357,7 +357,7 @@ def imread(image_path, mode="skimage"):
         return img_as_ubyte(image)
 
     elif mode == "cv2":
-        return cv2.imread(image_path, cv2.IMREAD_UNCHANGED)[..., ::-1]  # ~10% faster than using cv2.cvtColor
+        return cv2.imread(str(image_path), cv2.IMREAD_UNCHANGED)[..., ::-1]  # ~10% faster than using cv2.cvtColor
 
 
 # https://docs.opencv.org/3.4.0/da/d54/group__imgproc__transform.html#ga5bb5a1fea74ea38e1a5445ca803ff121

diff --git a/deeplabcut/utils/auxiliaryfunctions.py b/deeplabcut/utils/auxiliaryfunctions.py
@@ -134,9 +134,9 @@ def attempt_to_make_folder(foldername, recursive=False):
         return
 
     if recursive:
-        foldername.mkdir(parents=True)
+        foldername.mkdir(parents=True, exist_ok=True)
     else:
-        foldername.mkdir()
+        foldername.mkdir(exist_ok=True, parents=True)
 
 
 def read_pickle(filename):
@@ -642,7 +642,7 @@ def find_analyzed_data(folder, videoname: str, scorer: str, filtered=False, trac
     candidates = []
     for file in grab_files_in_folder(folder, "h5"):
         stem = Path(file).stem.replace("_filtered", "")
-        starts_by_scorer = file.startswith((videoname + scorer, videoname + scorer_legacy))
+        starts_by_scorer = Path(file).name.startswith((videoname + scorer, videoname + scorer_legacy))
         if tracker:
             matches_tracker = stem.endswith(tracker)
         else:

diff --git a/deeplabcut/utils/make_labeled_video.py b/deeplabcut/utils/make_labeled_video.py
@@ -1261,7 +1261,7 @@ def _create_video_from_tracks(video, tracks, destfolder, output_name, pcutoff, s
     from tqdm import tqdm
 
     if not Path(destfolder).is_dir():
-        Path(destfolder).mkdir()
+        Path(destfolder).mkdir(exist_ok=True, parents=True)
 
     vid = VideoWriter(video)
     nframes = len(vid)

diff --git a/deeplabcut/utils/pseudo_label.py b/deeplabcut/utils/pseudo_label.py
@@ -415,7 +415,7 @@ def dlc3predictions_2_annotation_from_video(
     assert len(predictions) == len(image_paths)
     imageid2annotations = defaultdict(list)
     for image_id, (prediction, image_path) in enumerate(zip(predictions, image_paths, strict=False)):
-        image_obj = cv2.imread(image_path)
+        image_obj = cv2.imread(str(image_path))
         height, width, channels = image_obj.shape
         imagename = Path(image_path).name
         image = {