Merge pull request #9 from chris-delg/testing

Testing

.gitignore

@@ -1,5 +1,6 @@
 .DS_Store
 __pycache__/
+.pytest_cache
 .vscode/
 example_data/
 build/

benchmark.py

@@ -63,7 +63,7 @@ def run_write_tests(self, num_of_gigabytes: int, show_results: bool,
                         avg_graph: Optional[matplotlib.axes._axes.Axes] = None) -> None:
 
         # error checking to see if chosen lib exists in test
-        if choose_lib and choose_lib not in set(self.__zarr_writers.keys()):
+        if choose_lib != None and choose_lib not in set(self.__zarr_writers.keys()):
             raise ValueError(f"There is no library of name \"{choose_lib}\".") 
 
         gb_in_bytes = 1073741824 # represents number of bytes in a GB
@@ -94,7 +94,7 @@ def run_write_tests(self, num_of_gigabytes: int, show_results: bool,
 
                 # prints info to the terminal 
                 print(f"Multiplier on first dimension : {multiplier}x\n{lib_name} -> creating zarr : {total_time} seconds")
-                print(f"The zarr folder is of size {folder_size(writer.data_path)}\n\n")
+                print(f"The zarr folder is of size {formatted_folder_size(writer.data_path)}\n\n")
 
                 curr_data_size = np.prod(new_shape) # 3d array filled with 1 byte ints so multiplication gives accurate size in bytes
                 file_sizes.append(curr_data_size * 10**-9) # converts bytes to GB
@@ -115,13 +115,13 @@ def run_write_tests(self, num_of_gigabytes: int, show_results: bool,
             self.__print_results(additional_info=(f"Write Test GB Soft Cap: {num_of_gigabytes}GB"))
 
 
-    def run_append_test(self, num_of_gigabytes: int, show_results: bool,
+    def run_append_tests(self, num_of_gigabytes: int, show_results: bool,
                         choose_lib: Optional[str] = None,
                         graph: Optional[matplotlib.axes._axes.Axes] = None, 
                         avg_graph: Optional[matplotlib.axes._axes.Axes] = None) -> None:
 
         # error checking to see if chosen lib exists in test
-        if choose_lib and choose_lib not in set(self.__zarr_writers.keys()):
+        if choose_lib != None and choose_lib not in set(self.__zarr_writers.keys()):
             raise ValueError(f"There is no library of name \"{choose_lib}\".") 
 
         gb_in_bytes = 1073741824 # represents number of bytes in a GB
@@ -155,7 +155,7 @@ def run_append_test(self, num_of_gigabytes: int, show_results: bool,
 
                 # prints info to the terminal 
                 print(f"Multiplier on first dimension : {multiplier}x\n{lib_name} -> appending zarr : {total_time} seconds")
-                print(f"The zarr folder is of size {folder_size(writer.data_path)}\n\n")
+                print(f"The zarr folder is of size {formatted_folder_size(writer.data_path)}\n\n")
 
                 curr_data_size = np.prod(new_shape) # 3d array filled with 1 byte ints so multiplication gives accurate size in bytes
                 write_numbers.append(multiplier) # converts bytes to GB
@@ -180,7 +180,7 @@ def run_all_tests(self, append_test_gigabytes: int, write_test_gigabytes: int,
                       append_graph: Optional[matplotlib.axes._axes.Axes] = None, append_avg_graph: Optional[matplotlib.axes._axes.Axes] = None,
                       write_graph: Optional[matplotlib.axes._axes.Axes] = None, write_avg_graph: Optional[matplotlib.axes._axes.Axes] = None) -> None:
 
-        self.run_append_test(num_of_gigabytes=append_test_gigabytes, show_results=False, choose_lib=choose_lib, graph=append_graph, avg_graph=append_avg_graph)
+        self.run_append_tests(num_of_gigabytes=append_test_gigabytes, show_results=False, choose_lib=choose_lib, graph=append_graph, avg_graph=append_avg_graph)
         self.run_write_tests(num_of_gigabytes=write_test_gigabytes, show_results=False, choose_lib=choose_lib, graph=write_graph, avg_graph=write_avg_graph)
         self.__print_results(additional_info=(f"Write Test GB Soft Cap: {write_test_gigabytes}GB | Append Test GB Soft Cap: {append_test_gigabytes}GB"))
 

conftest.py

@@ -0,0 +1 @@
+# needed so that pytest can identify this layer as the root layer of the project (will throw errors if deleted)

environment.yml

@@ -9,6 +9,7 @@ dependencies:
   - ome-zarr
   - zarr=2.15*
   - pip:
+      - pytest
       - cmake
       - ml-dtypes
       - tensorstore

tests/benchmark_test.py

@@ -0,0 +1,58 @@
+from benchmark import Benchmark
+import pytest
+
+
+@pytest.fixture
+def benchmark() -> Benchmark:
+    shape = [256, 256, 256]
+    chunks = [64, 64, 64]
+    benchmark = Benchmark(shape=shape, chunks=chunks)
+
+    return benchmark
+
+
+def test_benchmark_get_shape(benchmark: Benchmark):
+    assert benchmark.shape == [256, 256, 256]
+
+
+def test_benchmark_get_chunks(benchmark: Benchmark):
+    assert benchmark.chunks == [64, 64, 64]
+
+
+@pytest.mark.parametrize(
+    ("lib_name"),
+    [
+        (""),
+        ("."),
+        (" ")
+    ]
+)
+def test_if_incorrect_lib_name_in_write_tests_fails(benchmark: Benchmark, lib_name: str): 
+    with pytest.raises(ValueError):
+        benchmark.run_write_tests(num_of_gigabytes=1, show_results=False, choose_lib=lib_name) 
+
+
+@pytest.mark.parametrize(
+    ("lib_name"),
+    [
+        (""),
+        ("."),
+        (" ")
+    ]
+)
+def test_if_incorrect_lib_name_in_append_tests_fails(benchmark: Benchmark, lib_name: str):
+    with pytest.raises(ValueError):
+        benchmark.run_append_tests(num_of_gigabytes=1, show_results=False, choose_lib=lib_name)
+
+
+@pytest.mark.parametrize(
+    ("lib_name"),
+    [
+        (""),
+        ("."),
+        (" ")
+    ]
+)
+def test_if_incorrect_lib_name_in_run_all_tests_fails(benchmark: Benchmark, lib_name: str):
+    with pytest.raises(ValueError):
+        benchmark.run_all_tests(append_test_gigabytes=1, write_test_gigabytes=1, choose_lib=lib_name)  

tests/common_test.py

@@ -0,0 +1,57 @@
+from zarr_libraries.common import folder_size_in_bytes, formatted_folder_size
+import os 
+import pytest
+from tempfile import TemporaryDirectory
+
+
+@pytest.fixture
+def temp_dir():
+    with TemporaryDirectory() as temp_dir:
+        yield temp_dir
+
+
+@pytest.mark.parametrize(
+    ("num_mb"),
+    [
+        (1),
+        (10),
+        (50),
+        (100),
+        (500),
+        (750)
+    ]
+)
+def test_folder_size_in_bytes(temp_dir, num_mb: int):
+    mb_in_bytes = 1048576
+    MBs = num_mb * mb_in_bytes
+
+    # creating a test file of size variable size 
+    with open(temp_dir + "/test.bin", "wb") as f:
+        f.write(os.urandom(MBs))
+
+    # custom tolerance of 4096 bytes which is about 0.00390625 MB's
+    assert abs(folder_size_in_bytes(temp_dir) - MBs) <= 4096
+
+
+@pytest.mark.parametrize(
+    ("num_mb"),
+    [
+        (1),
+        (10),
+        (50),
+        (100),
+        (500),
+        (750)
+    ]
+)   
+def test_formatted_folder_size(temp_dir, num_mb: int):
+    mb_in_bytes = 1048576
+    MBs = num_mb * mb_in_bytes     
+
+    # creating a test file of size variable size 
+    with open(temp_dir + "/test.bin", "wb") as f:
+        f.write(os.urandom(MBs))
+
+    # checking if the formatted size w/o the decimal values is equal to expected
+    assert formatted_folder_size(temp_dir).split(".")[0] == str(num_mb)
+

tests/cpp_zarr_test.py

@@ -0,0 +1,32 @@
+from zarr_libraries.cpp_zarr.cpp_zarr import *
+from zarr_libraries.common import folder_size_in_bytes
+from pathlib import Path
+import shutil
+import numpy as np
+import pytest
+
+
+def test_getting_data_path_from_cpp_zarr():
+    cpp_zarr = Cpp_Zarr()
+
+    expected_path = str((Path(__file__).parent / "../zarr_libraries/example_data/cpp_zarr_data/test.zarr").resolve())
+    assert cpp_zarr.data_path == expected_path
+
+
+@pytest.mark.parametrize(
+    ("shape", "chunks"),
+    [
+        ([256, 256, 256], [64, 64, 64]),
+        ([64, 128, 256], [32, 64, 64]),
+    ]
+)
+def test_cpp_zarr_writes_correct_amount_of_data(shape: list, chunks: list):
+    cpp_zarr = Cpp_Zarr()
+
+    cpp_zarr.write_zarr(shape=shape, chunks=chunks)
+    expected_size = np.prod(shape)
+
+    # the actual size can be larger but can never be smaller that expected size
+    assert folder_size_in_bytes(cpp_zarr.data_path) >= expected_size
+    shutil.rmtree(cpp_zarr.data_path)    
+

tests/ome_zarr_test.py

@@ -0,0 +1,33 @@
+from zarr_libraries.ome_ngff.ome_zarr import Ome_Zarr
+from zarr_libraries.common import folder_size_in_bytes
+from pathlib import Path
+import numpy as np
+import shutil
+import pytest
+
+
+def test_getting_data_path_from_ome_zarr():
+    ome_zarr = Ome_Zarr()
+
+    expected_path = str((Path(__file__).parent / "../zarr_libraries/example_data/ome_zarr_data/test.zarr").resolve())
+    assert ome_zarr.data_path == expected_path
+
+
+@pytest.mark.parametrize(
+    ("shape", "chunks"),
+    [
+        ([256, 256, 256], [64, 64, 64]),
+        ([64, 128, 256], [32, 64, 64]),
+    ]
+)
+def test_ome_zarr_writes_correct_amount_of_data(shape: list, chunks: list):
+    ome_zarr = Ome_Zarr()
+
+    zarr_data = np.random.randint(low=0, high=256, size=shape, dtype=np.uint8)
+    ome_zarr.write_zarr(chunks=chunks, zarr_data=zarr_data)
+    expected_size = np.prod(shape)
+
+    # the actual size can be larger but can never be smaller that expected size
+    assert folder_size_in_bytes(ome_zarr.data_path) >= expected_size
+    shutil.rmtree(ome_zarr.data_path)
+

tests/tensorstore_test.py

@@ -0,0 +1,60 @@
+from zarr_libraries.tensorstore.tensorstore_zarr import Tensorstore
+from zarr_libraries.common import folder_size_in_bytes
+from pathlib import Path
+import shutil
+import numpy as np
+import pytest
+
+
+def test_getting_data_path_from_tensorstore():
+    tensorstore = Tensorstore()
+
+    expected_path = str((Path(__file__).parent / "../zarr_libraries/example_data/tensorstore_data/test.zarr").resolve())
+    assert tensorstore.data_path == expected_path
+
+
+@pytest.mark.parametrize(
+    ("shape", "chunks"),
+    [
+        ([256, 256, 256], [64, 64, 64]),
+        ([64, 128, 256], [32, 64, 64]),
+    ]
+)   
+def test_tensorstore_writes_correct_amount_of_data(shape: list, chunks: list):
+    tensorstore = Tensorstore()
+
+    zarr_data = np.random.randint(low=0, high=256, size=shape, dtype=np.uint8)
+    tensorstore.write_zarr(shape=shape, chunks=chunks, zarr_data=zarr_data)
+    expected_size = np.prod(shape)
+
+    # the actual size can be larger but can never be smaller that expected size
+    assert folder_size_in_bytes(tensorstore.data_path) >= expected_size
+    shutil.rmtree(tensorstore.data_path)    
+
+
+@pytest.mark.parametrize(
+    ("shape", "chunks"),
+    [
+        ([256, 256, 256], [64, 64, 64]),
+        ([64, 128, 256], [32, 64, 64]),
+    ]
+)  
+def test_zarr_python_append_writes_correct_amount_of_data(shape: list, chunks: list):
+    tensorstore = Tensorstore()
+
+    zarr_data = np.random.randint(low=0, high=256, size=shape, dtype=np.uint8)
+    tensorstore.append_zarr(shape=shape, chunks=chunks, new_shape=shape, zarr_data=zarr_data, multiplier=1)
+    expected_size = np.prod(shape)
+
+    # checking the size of the first append
+    assert folder_size_in_bytes(tensorstore.data_path) >= expected_size
+
+    new_shape = [shape[0] * (2), *shape[1:]]
+    new_data = np.random.randint(low=0, high=256, size=shape, dtype=np.uint8)
+    tensorstore.append_zarr(shape=shape, chunks=chunks, new_shape=new_shape, zarr_data=new_data, multiplier=2)
+    expected_size = np.prod(new_shape)
+
+    # changing the shape and comparing the expected size after the second append
+    assert folder_size_in_bytes(tensorstore.data_path) >= expected_size
+    shutil.rmtree(tensorstore.data_path)
+

tests/zarr_python_test.py

@@ -0,0 +1,59 @@
+from zarr_libraries.zarr_python.zarr_python import Zarr_Python
+from zarr_libraries.common import folder_size_in_bytes
+from pathlib import Path
+import shutil
+import numpy as np
+import pytest
+
+
+def test_getting_data_path_from_zarr_python():
+    zarr_python = Zarr_Python()
+
+    expected_path = str((Path(__file__).parent / "../zarr_libraries/example_data/zarr_python_data/test.zarr").resolve())
+    assert zarr_python.data_path == expected_path
+
+
+@pytest.mark.parametrize(
+    ("shape", "chunks"),
+    [
+        ([256, 256, 256], [64, 64, 64]),
+        ([64, 128, 256], [32, 64, 64]),
+    ]
+)   
+def test_zarr_python_writes_correct_amount_of_data(shape: list, chunks: list):
+    zarr_python = Zarr_Python()
+
+    zarr_data = np.random.randint(low=0, high=256, size=shape, dtype=np.uint8)
+    zarr_python.write_zarr(shape=shape, chunks=chunks, zarr_data=zarr_data)
+    expected_size = np.prod(shape)
+
+    # the actual size can be larger but can never be smaller that expected size
+    assert folder_size_in_bytes(zarr_python.data_path) >= expected_size
+    shutil.rmtree(zarr_python.data_path)    
+
+
+@pytest.mark.parametrize(
+    ("shape", "chunks"),
+    [
+        ([256, 256, 256], [64, 64, 64]),
+        ([64, 128, 256], [32, 64, 64]),
+    ]
+)
+def test_zarr_python_append_writes_correct_amount_of_data(shape: list, chunks: list):
+    zarr_python = Zarr_Python()
+
+    zarr_data = np.random.randint(low=0, high=256, size=shape, dtype=np.uint8)
+    zarr_python.append_zarr(shape=shape, chunks=chunks, zarr_data=zarr_data)
+    expected_size = np.prod(shape)
+
+    # checking the size of the first append
+    assert folder_size_in_bytes(zarr_python.data_path) >= expected_size
+
+    new_shape = [shape[0] * (2), *shape[1:]]
+    new_data = np.random.randint(low=0, high=256, size=shape, dtype=np.uint8)
+    zarr_python.append_zarr(shape=shape, chunks=chunks, zarr_data=new_data)
+    expected_size = np.prod(new_shape)
+
+    # changing the shape and comparing the expected size after the second append
+    assert folder_size_in_bytes(zarr_python.data_path) >= expected_size
+    shutil.rmtree(zarr_python.data_path)