Debug import segfault

pyproject.toml

@@ -33,8 +33,8 @@ training = [
 
 # CASA tools for Measurement Set I/O
 casa = [
-    "python-casacore; sys_platform != 'darwin'",
-    "casatools; sys_platform == 'darwin'",
+    "casatools>=6.5.0",
+    "casatasks>=6.5.0",
 ]
 
 # Development tools

rfi_toolbox/__init__.py

@@ -26,12 +26,14 @@
 __email__ = "pjaganna@nrao.edu"
 
 # Eager imports (required for multiprocessing pickle compatibility)
-from . import datasets, preprocessing
+import datasets as datasets
+import preprocessing as preprocessing
 
 # Lazy imports for other modules to avoid circular dependencies
 def __getattr__(name):
     """Lazy import for optional modules."""
     import importlib
+    print(f"[DEBUG __getattr__] Lazy loading: {name}")
 
     # List of valid lazy-loaded modules
     valid_modules = {
@@ -40,10 +42,12 @@ def __getattr__(name):
     }
 
     if name in valid_modules:
+        print(f"[DEBUG __getattr__] Calling importlib.import_module for: {name}")
         # Use importlib to avoid triggering __getattr__ recursion
         mod = importlib.import_module(f".{name}", __name__)
+        print(f"[DEBUG __getattr__] Successfully imported: {name}")
         # Cache in globals to avoid repeated imports
         globals()[name] = mod
         return mod
 
-    raise AttributeError(f"module '{__name__}' has no attribute '{name}'")
+    raise AttributeError(f"module '{__name__}' has no attribute '{name}'")

rfi_toolbox/datasets/rfi_mask_dataset.py

@@ -15,22 +15,12 @@
 from tqdm import tqdm
 
 try:
-    import casacore.tables as ct
-
-    use_casacore = True
+    from casatools import table
     CASA_AVAILABLE = True
 except ImportError:
-    try:
-        from casatools import table
-
-        use_casacore = False
-        CASA_AVAILABLE = True
-    except ImportError:
-        # CASA not available - class will fail at instantiation if use_ms=True
-        ct = None
-        table = None
-        use_casacore = False
-        CASA_AVAILABLE = False
+    # CASA not available - class will fail at instantiation if use_ms=True
+    table = None
+    CASA_AVAILABLE = False
 
 
 class RFIMaskDataset(Dataset):
@@ -73,13 +63,11 @@ def __init__(
             if not CASA_AVAILABLE:
                 raise ImportError(
                     "CASA is required for use_ms=True but is not installed.\n"
-                    "Install with: pip install python-casacore  OR  pip install casatools"
+                    "Install with: pip install rfi-toolbox[casa]"
                 )
-            if use_casacore:
-                self.tb = ct.table(ms_name, readonly=True)
-            else:
-                tb = table()
-                self.tb = tb.open(ms_name, readonly=True)
+            tb = table()
+            tb.open(ms_name)
+            self.tb = tb
             self.num_antennas = self.tb.getcol("ANTENNA1").max() + 1
             self.spw_array = np.unique(self.tb.getcol("DATA_DESC_ID"))
 

rfi_toolbox/io/__init__.py

@@ -4,16 +4,28 @@
 Handles CASA measurement set I/O and synthetic data injection.
 """
 
+print("[DEBUG io/__init__] Starting io module import")
+
+print("[DEBUG io/__init__] Attempting to import MSLoader")
 try:
     from .ms_loader import MSLoader
+
+    print("[DEBUG io/__init__] MSLoader imported successfully")
     _HAS_CASA = True
-except ImportError:
+except ImportError as e:
+    print(f"[DEBUG io/__init__] MSLoader import failed: {e}")
     _HAS_CASA = False
     MSLoader = None
 
+print("[DEBUG io/__init__] Attempting to import inject_synthetic_data")
 try:
     from .ms_injection import inject_synthetic_data
-except ImportError:
+
+    print("[DEBUG io/__init__] inject_synthetic_data imported successfully")
+except ImportError as e:
+    print(f"[DEBUG io/__init__] inject_synthetic_data import failed: {e}")
     inject_synthetic_data = None
 
+print("[DEBUG io/__init__] io module import complete")
+
 __all__ = ["MSLoader", "inject_synthetic_data"]

rfi_toolbox/io/ms_injection.py

@@ -11,13 +11,20 @@
 import numpy as np
 from tqdm import tqdm
 
+print("[DEBUG ms_injection] Basic imports complete")
+
+print("[DEBUG ms_injection] Attempting casatools import")
 try:
     from casatools import table
 
+    print("[DEBUG ms_injection] casatools.table imported successfully")
     CASA_AVAILABLE = True
-except ImportError:
+except ImportError as e:
+    print(f"[DEBUG ms_injection] casatools import failed: {e}")
     CASA_AVAILABLE = False
 
+print("[DEBUG ms_injection] ms_injection module import complete")
+
 
 def inject_synthetic_data(
     template_ms_path,
@@ -44,14 +51,17 @@ def inject_synthetic_data(
     """
     if not CASA_AVAILABLE:
         raise ImportError(
-            "casatools is required for MS injection. " "Install with: pip install rfi-toolbox[casa]"
+            "casatools is required for MS injection. "
+            "Install with: pip install rfi-toolbox[casa]"
         )
 
     template_ms_path = Path(template_ms_path)
 
     # Default output path
     if output_ms_path is None:
-        output_ms_path = template_ms_path.parent / f"{template_ms_path.stem}.synthetic.ms"
+        output_ms_path = (
+            template_ms_path.parent / f"{template_ms_path.stem}.synthetic.ms"
+        )
     else:
         output_ms_path = Path(output_ms_path)
 
@@ -104,7 +114,10 @@ def inject_synthetic_data(
     # For simplicity, assume all SPWs have same channel count
     # and we're filling all SPWs with the same data
     if len(set(channels_per_spw)) > 1:
-        print("  WARNING: MS has SPWs with different channel counts. " "Using first SPW only.")
+        print(
+            "  WARNING: MS has SPWs with different channel counts. "
+            "Using first SPW only."
+        )
 
     channels_in_spw = channels_per_spw[0]
 
@@ -130,7 +143,9 @@ def inject_synthetic_data(
 
         for spw_idx in range(num_spw):
             # Query this baseline + SPW
-            subtable = tb.query(f"DATA_DESC_ID=={spw_idx} && ANTENNA1=={ant1} && ANTENNA2=={ant2}")
+            subtable = tb.query(
+                f"DATA_DESC_ID=={spw_idx} && ANTENNA1=={ant1} && ANTENNA2=={ant2}"
+            )
 
             if subtable.nrows() == 0:
                 print(f"  WARNING: No rows for baseline ({ant1},{ant2}), SPW {spw_idx}")
@@ -251,7 +266,9 @@ def inject_synthetic_data(
                         subtable.putcell("DATA", row_idx, cell_val)
                     except Exception as e:
                         subtable.close()
-                        raise RuntimeError(f"Failed to write DATA row {row_idx}: {e}") from e
+                        raise RuntimeError(
+                            f"Failed to write DATA row {row_idx}: {e}"
+                        ) from e
 
             subtable.close()