Merge pull request #107 from CellKai/fix-h5-resave

Fix function to not split time points by default

Author: ehrenfeu

src/imcflibs/imagej/bdv.py

@@ -807,7 +807,7 @@ def define_dataset_auto(
     file_path,
     bf_series_type,
     dataset_save_path=None,
-    timepoints_per_partition=1,
+    timepoints_per_partition=0,
     resave="Re-save as multiresolution HDF5",
     subsampling_factors=None,
     hdf5_chunk_sizes=None,
@@ -830,8 +830,9 @@ def define_dataset_auto(
         Defines how Bio-Formats interprets the series.
     timepoints_per_partition : int, optional
         Split the output dataset by timepoints. Use `0` for no split, resulting
-        in a single HDF5 file containing all timepoints. By default `1`,
-        resulting in a HDF5 per timepoints.
+        in a single HDF5 file containing all timepoints. Otherwise, choose the
+        number of timepoints per file. By default `0`.
+    
     resave : str, optional
         Allow the function to either re-save the images or simply create a
         merged xml. Use `Load raw data` to avoid re-saving, by default `Re-save
@@ -864,6 +865,12 @@ def define_dataset_auto(
         hdf5_chunk_sizes = "hdf5_chunk_sizes=" + hdf5_chunk_sizes + " "
     else:
         hdf5_chunk_sizes = ""
+    if timepoints_per_partition > 0:
+        split_timepoints = "split_hdf5 timepoints_per_partition=" + str(
+            timepoints_per_partition
+        ) + " "
+    else:
+        split_timepoints = ""
 
     if bf_series_type == "Angles":
         angle_rotation = "apply_angle_rotation "
@@ -901,10 +908,7 @@ def define_dataset_auto(
         + angle_rotation
         + subsampling_factors
         + hdf5_chunk_sizes
-        + "split_hdf5 "
-        + "timepoints_per_partition="
-        + str(timepoints_per_partition)
-        + " "
+        + split_timepoints
         + "setups_per_partition=0 "
         + "use_deflate_compression "
     )

tests/bdv/test_define_dataset_auto.py

@@ -81,6 +81,68 @@ def test_define_dataset_auto_tile(tmp_path, caplog):
     # Set the default values for dataset definitions
     options = set_default_values(project_filename, file_info["path"])
 
+    # Construct the options for dataset definitions
+    options = (
+        options
+        + "how_to_store_input_images=["
+        + "Re-save as multiresolution HDF5"
+        + "] "
+        + "load_raw_data_virtually "
+        + "metadata_save_path=["
+        + result_folder
+        + "] "
+        + "image_data_save_path=["
+        + result_folder
+        + "] "
+        + "check_stack_sizes "
+        + "setups_per_partition=0 "
+        + "use_deflate_compression "
+    )
+
+    # Construct the final call to ImageJ
+    final_call = "IJ.run(cmd=[%s], params=[%s])" % (cmd, options)
+
+    # Define the dataset using the "Auto-Loader" option
+    bdv.define_dataset_auto(project_filename, file_info["path"], bf_series_type)
+    # Check if the final call is in the log
+    assert final_call == caplog.messages[0]
+
+
+def test_define_dataset_auto_tile_split_timepoints(tmp_path, caplog):
+    """Test automatic dataset definition method for tile series.
+
+    Parameters
+    ----------
+    tmp_path : pytest.fixture
+        Temporary path for the test.
+    caplog : pytest.fixture
+        Log capturing fixture.
+    """
+
+    # Set the logging level to capture warnings
+    caplog.set_level(logging.WARNING)
+    # Clear the log
+    caplog.clear()
+
+    # Define the project and file names
+    project_filename = "proj_name"
+    file_path = tmp_path
+    file_info = pathtools.parse_path(file_path)
+
+    # Define the result and dataset save paths
+    result_folder = pathtools.join2(file_info["path"], project_filename)
+
+    # Default settings
+
+    # Define the type of Bio-Formats series
+    bf_series_type = "Tiles"
+
+    # Define the ImageJ command
+    cmd = "Define Multi-View Dataset"
+
+    # Set the default values for dataset definitions
+    options = set_default_values(project_filename, file_info["path"])
+
     # Construct the options for dataset definitions
     options = (
         options
@@ -105,7 +167,7 @@ def test_define_dataset_auto_tile(tmp_path, caplog):
     final_call = "IJ.run(cmd=[%s], params=[%s])" % (cmd, options)
 
     # Define the dataset using the "Auto-Loader" option
-    bdv.define_dataset_auto(project_filename, file_info["path"], bf_series_type)
+    bdv.define_dataset_auto(project_filename, file_info["path"], bf_series_type, timepoints_per_partition=1)
     # Check if the final call is in the log
     assert final_call == caplog.messages[0]
 
@@ -160,8 +222,6 @@ def test_define_dataset_auto_angle(tmp_path, caplog):
         + "] "
         + "check_stack_sizes "
         + "apply_angle_rotation "
-        + "split_hdf5 "
-        + "timepoints_per_partition=1 "
         + "setups_per_partition=0 "
         + "use_deflate_compression "
     )