fix: add missing stacklevel=2 to 14 warnings.warn() calls (Fixes #1164)

pyhealth/data/__init__.py

@@ -5,4 +5,4 @@ class Visit:
     """This class is deprecated and should not be used."""
     def __init__(self, *args, **kwargs):
         import warnings
-        warnings.warn("The Visit class is deprecated and will be removed in a future version.", DeprecationWarning)
+        warnings.warn("The Visit class is deprecated and will be removed in a future version.", DeprecationWarning, stacklevel=2)

pyhealth/datasets/__init__.py

@@ -7,6 +7,7 @@ def __init__(self, *args, **kwargs):
         warnings.warn(
             "The BaseEHRDataset class is deprecated and will be removed in a future version.",
             DeprecationWarning,
+            stacklevel=2,
         )
 
 
@@ -19,6 +20,7 @@ def __init__(self, *args, **kwargs):
         warnings.warn(
             "The BaseSignalDataset class is deprecated and will be removed in a future version.",
             DeprecationWarning,
+            stacklevel=2,
         )
 
 
@@ -31,6 +33,7 @@ def __init__(self, *args, **kwargs):
         warnings.warn(
             "The SampleEHRDataset class is deprecated and will be removed in a future version.",
             DeprecationWarning,
+            stacklevel=2,
         )
 
 
@@ -43,6 +46,7 @@ def __init__(self, *args, **kwargs):
         warnings.warn(
             "The SampleSignalDataset class is deprecated and will be removed in a future version.",
             DeprecationWarning,
+            stacklevel=2,
         )
 
 

pyhealth/datasets/mimic3.py

@@ -60,6 +60,7 @@ def __init__(
                 "Events from prescriptions table only have date timestamp (no specific time). "
                 "This may affect temporal ordering of events.",
                 UserWarning,
+                stacklevel=2,
             )
         super().__init__(
             root=root,

pyhealth/datasets/mimic4.py

@@ -122,12 +122,14 @@ def __init__(
                 "Events from discharge table only have date timestamp (no specific time). "
                 "This may affect temporal ordering of events.",
                 UserWarning,
+                stacklevel=2,
             )
         if "discharge_detail" in tables:
             warnings.warn(
                 "Events from discharge_detail table only have date timestamp (no specific time). "
                 "This may affect temporal ordering of events.",
                 UserWarning,
+                stacklevel=2,
             )
         log_memory_usage(f"Before initializing {dataset_name}")
         super().__init__(

pyhealth/models/concare.py

@@ -954,6 +954,7 @@ def __init__(
                     "and sum-pooled per visit for memory efficiency. This reduces "
                     "per-visit code representations to a single scalar.",
                     UserWarning,
+                    stacklevel=2,
                 )
             else:
                 input_dim = proc.size()

pyhealth/models/stagenet.py

@@ -440,6 +440,7 @@ def forward_from_embedding(
                     f"StageNet format with time intervals for "
                     f"better performance.",
                     UserWarning,
+                    stacklevel=2,
                 )
             else:
                 time = time.to(self.device)
@@ -453,6 +454,7 @@ def forward_from_embedding(
                     f"Feature '{feature_key}' does not have mask "
                     f"information. Default mask will be created from "
                     f"embedded values. But it may not be accurate.",
+                    stacklevel=2,
                 )
                 mask = (value.abs().sum(dim=-1) != 0).int()
             elif not processor.is_token() and value.dim() == mask.dim():

pyhealth/models/stagenet_mha.py

@@ -516,6 +516,7 @@ def forward_from_embedding(
                     f"StageNet format with time intervals for "
                     f"better performance.",
                     UserWarning,
+                    stacklevel=2,
                 )
             else:
                 time = time.to(self.device)
@@ -528,6 +529,7 @@ def forward_from_embedding(
                     f"Feature '{feature_key}' does not have mask "
                     f"information. Default mask will be created from "
                     f"embedded values. But it may not be accurate.",
+                    stacklevel=2,
                 )
                 mask = (value.abs().sum(dim=-1) != 0).int()
             elif not processor.is_token() and value.dim() == mask.dim():

pyhealth/models/text_embedding.py

@@ -244,6 +244,7 @@ def _chunk_and_encode(
                 f"Text produced {original_chunks} chunks, truncated to {self.max_chunks}. "
                 f"Consider increasing max_chunks or summarizing input.",
                 UserWarning,
+                stacklevel=2,
             )
 
         # Step 4: Pad chunks to uniform length for batched encoding