MCPcopy
hub / github.com/dmlc/dgl / _chunk_numpy_array

Function _chunk_numpy_array

tests/tools/pytest_utils.py:13–35  ·  view source on GitHub ↗
(arr, fmt_meta, chunk_sizes, path_fmt, vector_rows=False)

Source from the content-addressed store, hash-verified

11
12
13def _chunk_numpy_array(arr, fmt_meta, chunk_sizes, path_fmt, vector_rows=False):
14 paths = []
15 offset = 0
16
17 for j, n in enumerate(chunk_sizes):
18 path = os.path.abspath(path_fmt % j)
19 arr_chunk = arr[offset : offset + n]
20 shape = arr_chunk.shape
21 logging.info("Chunking %d-%d" % (offset, offset + n))
22 # If requested we write multi-column arrays as single-column vector Parquet files
23 array_parser = array_readwriter.get_array_parser(**fmt_meta)
24 if (
25 isinstance(array_parser, ParquetArrayParser)
26 and len(shape) > 1
27 and shape[1] > 1
28 ):
29 array_parser.write(path, arr_chunk, vector_rows=vector_rows)
30 else:
31 array_parser.write(path, arr_chunk)
32 offset += n
33 paths.append(path)
34
35 return paths
36
37
38def _initialize_num_chunks(g, num_chunks, kwargs=None):

Callers 1

_chunk_graphFunction · 0.85

Calls 3

infoMethod · 0.80
appendMethod · 0.80
writeMethod · 0.45

Tested by

no test coverage detected