hpcflow-new2 0.2.0a222__py3-none-any.whl → 0.2.0a223__py3-none-any.whl

hpcflow/_version.py

@@ -1 +1 @@
-__version__ = "0.2.0a222"
+__version__ = "0.2.0a223"

hpcflow/sdk/persistence/zarr.py

@@ -18,6 +18,7 @@ from numpy.ma.core import MaskedArray
 import zarr  # type: ignore
 from zarr.errors import BoundsCheckError  # type: ignore
 from zarr.storage import DirectoryStore, FSStore  # type: ignore
+from zarr.util import guess_chunks  # type: ignore
 from fsspec.implementations.zip import ZipFileSystem  # type: ignore
 from rich.console import Console
 from numcodecs import MsgPack, VLenArray, blosc, Blosc, Zstd  # type: ignore
@@ -59,6 +60,7 @@ from hpcflow.sdk.submission.submission import (
     SUBMISSION_SUBMIT_TIME_KEYS,
 )
 from hpcflow.sdk.utils.arrays import get_2D_idx, split_arr
+from hpcflow.sdk.utils.patches import override_module_attrs
 from hpcflow.sdk.utils.strings import shorten_list_str
 
 if TYPE_CHECKING:
@@ -89,6 +91,11 @@ if TYPE_CHECKING:
 ListAny: TypeAlias = "list[Any]"
 #: Zarr attribute mapping context.
 ZarrAttrs: TypeAlias = "dict[str, Any]"
+#: Soft lower limit for the number of bytes in an array chunk
+_ARRAY_CHUNK_MIN: int = 500 * 1024 * 1024  # 500 MiB
+#: Hard upper limit for the number of bytes in an array chunk. Should be lower than the
+#: maximum buffer size of the blosc encoder, if we're using it (2 GiB)
+_ARRAY_CHUNK_MAX: int = 1024 * 1024 * 1024  # 1 GiB
 _JS: TypeAlias = "dict[str, list[dict[str, dict]]]"
 
 
@@ -124,7 +131,13 @@ def _encode_numpy_array(
     new_idx = (
         max((int(i.removeprefix("arr_")) for i in param_arr_group.keys()), default=-1) + 1
     )
-    param_arr_group.create_dataset(name=f"arr_{new_idx}", data=obj, chunks=obj.shape)
+    with override_module_attrs(
+        "zarr.util", {"CHUNK_MIN": _ARRAY_CHUNK_MIN, "CHUNK_MAX": _ARRAY_CHUNK_MAX}
+    ):
+        # `guess_chunks` also ensures chunk shape is at least 1 in each dimension:
+        chunk_shape = guess_chunks(obj.shape, obj.dtype.itemsize)
+
+    param_arr_group.create_dataset(name=f"arr_{new_idx}", data=obj, chunks=chunk_shape)
     type_lookup["arrays"].append([path, new_idx])
 
     return len(type_lookup["arrays"]) - 1

hpcflow/sdk/utils/patches.py

@@ -1,4 +1,7 @@
+from contextlib import contextmanager
 from pathlib import Path
+import sys
+from typing import Any
 
 
 def resolve_path(path):
@@ -10,3 +13,19 @@ def resolve_path(path):
     """
     # TODO: this only seems to be used in a test; remove?
     return Path.cwd() / Path(path).resolve()  # cwd is ignored if already absolute
+
+
+@contextmanager
+def override_module_attrs(module_name: str, overrides: dict[str, Any]):
+    """Context manager to temporarily override module-level attributes. The module must be
+    imported (i.e. within `sys.modules`)."""
+
+    module = sys.modules[module_name]
+    original_values = {k: getattr(module, k) for k in overrides}
+    try:
+        for k, v in overrides.items():
+            setattr(module, k, v)
+        yield
+    finally:
+        for k, v in original_values.items():
+            setattr(module, k, v)

hpcflow/tests/unit/test_persistence.py

@@ -6,7 +6,11 @@ from typing import cast, TYPE_CHECKING
 import numpy as np
 import zarr  # type: ignore
 import pytest
-from hpcflow.sdk.core.test_utils import make_test_data_YAML_workflow, make_workflow
+from hpcflow.sdk.core.test_utils import (
+    make_schemas,
+    make_test_data_YAML_workflow,
+    make_workflow,
+)
 from hpcflow.sdk.persistence.json import (
     JSONPersistentStore,
     JsonStoreElement,
@@ -551,3 +555,38 @@ def test_zarr_encode_decode_jobscript_block_dependencies_large_one_to_one():
     arr = ZarrPersistentStore._encode_jobscript_block_dependencies(deps_t)
     deps_rt = ZarrPersistentStore._decode_jobscript_block_dependencies(arr)
     assert deps_rt == deps
+
+
+@pytest.mark.parametrize(
+    "array",
+    (
+        np.array([]),
+        np.empty(0),
+        np.empty((0, 1, 2)),
+        np.array([1, 2, 3]),
+        np.array([[1, 2, 3], [4, 5, 6]]),
+    ),
+)
+def test_zarr_save_persistent_array_shape(null_config, tmp_path, array):
+    s1 = make_schemas(({"p1": None}, ()))
+    t1 = hf.Task(schema=s1, inputs={"p1": array})
+    wk = hf.Workflow.from_template_data(
+        template_name="test_save_empty_array",
+        tasks=[t1],
+        path=tmp_path,
+    )
+    assert array.shape == wk.tasks[0].elements[0].get("inputs.p1")[:].shape
+
+
+def test_zarr_single_chunk_threshold(null_config, tmp_path):
+    # test very large arrays (> ~1 GB) are saved using multiple chunks
+    array = np.zeros(
+        268_435_456
+    )  # ~ 2.147483647 GB; greater than blosc's max buffer size
+    s1 = make_schemas(({"p1": None}, ()))
+    t1 = hf.Task(schema=s1, inputs={"p1": array})
+    wk = hf.Workflow.from_template_data(
+        template_name="test_large_array",
+        tasks=[t1],
+        path=tmp_path,
+    )

{hpcflow_new2-0.2.0a222.dist-info → hpcflow_new2-0.2.0a223.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: hpcflow-new2
-Version: 0.2.0a222
+Version: 0.2.0a223
 Summary: Computational workflow management
 License: MPL-2.0
 Author: aplowman

{hpcflow_new2-0.2.0a222.dist-info → hpcflow_new2-0.2.0a223.dist-info}/RECORD

@@ -1,7 +1,7 @@
 hpcflow/__init__.py,sha256=WIETuRHeOp2SqUqHUzpjQ-lk9acbYv-6aWOhZPRdlhs,64
 hpcflow/__pyinstaller/__init__.py,sha256=YOzBlPSck6slucv6lJM9K80JtsJWxXRL00cv6tRj3oc,98
 hpcflow/__pyinstaller/hook-hpcflow.py,sha256=P2b-8QdQqkSS7cJB6CB3CudUuJ9iZzTh2fQF4hNdCa4,1118
-hpcflow/_version.py,sha256=phew7JkOXGWTg0xqEm-B1aQnoPYpQVwtKpvMwtWxwkU,26
+hpcflow/_version.py,sha256=ExqG_C9ZO0pdvjMEloB9_jw5T23HBOJ1y0ajtaOl5UY,26
 hpcflow/app.py,sha256=gl2viVS65PbpDhUp2DARaYHFDqDWQjuoyB3ikrCNRW4,1367
 hpcflow/cli.py,sha256=G2J3D9v6MnMWOWMMWK6UEKLn_6wnV9lT_qygEBBxg-I,66
 hpcflow/data/demo_data_manifest/__init__.py,sha256=Hsq0jT8EXM13wu1MpGy5FQgyuz56ygep4VWOnulFn50,41
@@ -125,7 +125,7 @@ hpcflow/sdk/persistence/pending.py,sha256=JB42Emk2cN639bPlNtQtoFhe5WXGWDJeFm7aFU
 hpcflow/sdk/persistence/store_resource.py,sha256=P-VZF7gMIsVvgOsHJEplK1Se4jHsaAqbdoKoQo-7_LU,5903
 hpcflow/sdk/persistence/types.py,sha256=c4vd8RW8bg8Meoz1HNXjGT8Za9Hpfl1Sxp4pLAgdV0g,9125
 hpcflow/sdk/persistence/utils.py,sha256=TuQp0BKwiy1-5WTgiMaYXn00bZb6HWHAjs4BBFE9khw,1804
-hpcflow/sdk/persistence/zarr.py,sha256=yCsiNteaok-eM9x5jogeDoCPdnIpx1Yv7LgYm8qoVeA,88267
+hpcflow/sdk/persistence/zarr.py,sha256=ozJSx0W8L5yVRVvF7gM7M8psF5vd_-dNDnXNa-Rh3sI,88979
 hpcflow/sdk/runtime.py,sha256=vPNu4_DXYMnRyJenkVLq8yt6_XG4tCuwX69Oew4OWmo,11269
 hpcflow/sdk/submission/__init__.py,sha256=79xJXVkddsuj3uJz3BV9iOnAV7vCeJvLrhKccOA_dnU,67
 hpcflow/sdk/submission/enums.py,sha256=DykNHISQDmfa0oDggj6iIKugeFXJd8iF9NwqV3MN2oA,2049
@@ -146,7 +146,7 @@ hpcflow/sdk/typing.py,sha256=gu2h7a0RJs_r5UtV8AFVh9TNjo8kGe1n6UOxeH3zdgg,5060
 hpcflow/sdk/utils/arrays.py,sha256=7znIaqfl_bGElZ6U_da0vumWPez6ITyNrSDu_AlACDI,2084
 hpcflow/sdk/utils/deferred_file.py,sha256=LYIjTPjAmbOflwUaq_-o-O6DZJCQzRHX8dl_rIWTO5c,1381
 hpcflow/sdk/utils/hashing.py,sha256=40H7yO-oYgrt6RCdlTDDzJV9YIGiQeDx6zGpMJrRKZo,484
-hpcflow/sdk/utils/patches.py,sha256=jpbodcOjhE5ccrHPz5-L-r9yC7s2ymQccIdJi6Zd2Jo,398
+hpcflow/sdk/utils/patches.py,sha256=sVK8TjeeIEBgngm5PobxkKcIiuDORLaOzI_8ua1H3bw,988
 hpcflow/sdk/utils/strings.py,sha256=MGdVeQpEZjo_9_VSMNI2ArAesQpgAx4-RkHL7IAQInQ,1966
 hpcflow/tests/api/test_api.py,sha256=h0HT9W0Jd1pChrXYaBOVwGThaI3slGkloS0sbq2YX88,962
 hpcflow/tests/conftest.py,sha256=xtSqhOxjZYioiAPvrKwf7NFicZoA4BR9Si4J1A8mWHw,4083
@@ -188,7 +188,7 @@ hpcflow/tests/unit/test_meta_task.py,sha256=FEWLuUrP-6e6BO0CRJe2fQDDOZUjz1jfljL_
 hpcflow/tests/unit/test_multi_path_sequences.py,sha256=6cVYD2e7jMt-m4sr3Z6ixnKewvs-s--slP344trKIWM,9079
 hpcflow/tests/unit/test_object_list.py,sha256=uWBou5QCyScaM3DLqL_7RCguMdYZKMkeZR0V8Lpterk,3342
 hpcflow/tests/unit/test_parameter.py,sha256=_ImfDhmroXzPKwIuKiehsAw9qQPPHC9FQa8jN7OJqoA,7440
-hpcflow/tests/unit/test_persistence.py,sha256=cwQoa2czgZ0ZDtCRoRSdkfoFMgMQ5qjnef_mhsMWzrc,17472
+hpcflow/tests/unit/test_persistence.py,sha256=bpBiL4YPMosbZk18Ug1JRqYJ69HHJzDYUff_2wDi-5w,18544
 hpcflow/tests/unit/test_resources.py,sha256=48zC3A5YaSu0MlLlGiohowDpVZNgORsikVX8mP5agvI,8011
 hpcflow/tests/unit/test_run.py,sha256=jFRC1-wV5G2ZCEU-KBKYeHVnJdDBUXX6MbxTrB-RJSE,6205
 hpcflow/tests/unit/test_run_directories.py,sha256=VA2xhDRe7agKO7q1Y8kP0101BstkjLiC1RReKrMI6jE,1088
@@ -218,8 +218,8 @@ hpcflow/tests/workflows/test_submission.py,sha256=SUbBUbD8C8LSulrI7aETkzP9RqED48
 hpcflow/tests/workflows/test_workflows.py,sha256=9z3rtXjA5iMOp4C0q4TkD_9kLzwourCY-obpeOtnNt0,18927
 hpcflow/tests/workflows/test_zip.py,sha256=MzEwsIAYV_1A3bD0XRo23zUwUKVzkkmNd8_cil6YdWQ,578
 hpcflow/viz_demo.ipynb,sha256=6D9uBbWK3oMfbaf93Tnv5riFPtW-2miUTWNr9kGcnd4,228913
-hpcflow_new2-0.2.0a222.dist-info/LICENSE,sha256=Xhxf_KsrJNJFGMogumZhXSTPhUOVHCWf7nU-TDzqg0E,16763
-hpcflow_new2-0.2.0a222.dist-info/METADATA,sha256=93L_wYR_GPmtZ6f2vhlM4QwFdppn3S1p6EjiupYAbx4,2663
-hpcflow_new2-0.2.0a222.dist-info/WHEEL,sha256=kLuE8m1WYU0Ig0_YEGrXyTtiJvKPpLpDEiChiNyei5Y,88
-hpcflow_new2-0.2.0a222.dist-info/entry_points.txt,sha256=aoGtCnFdfPcXfBdu2zZyMOJoz6fPgdR0elqsgrE-USU,106
-hpcflow_new2-0.2.0a222.dist-info/RECORD,,
+hpcflow_new2-0.2.0a223.dist-info/LICENSE,sha256=Xhxf_KsrJNJFGMogumZhXSTPhUOVHCWf7nU-TDzqg0E,16763
+hpcflow_new2-0.2.0a223.dist-info/METADATA,sha256=yMUy1XOWyJ9GWx93QPFA-RlqpLVVl7YeXikl2Vx7zIc,2663
+hpcflow_new2-0.2.0a223.dist-info/WHEEL,sha256=kLuE8m1WYU0Ig0_YEGrXyTtiJvKPpLpDEiChiNyei5Y,88
+hpcflow_new2-0.2.0a223.dist-info/entry_points.txt,sha256=aoGtCnFdfPcXfBdu2zZyMOJoz6fPgdR0elqsgrE-USU,106
+hpcflow_new2-0.2.0a223.dist-info/RECORD,,