maximilianherde
commited on
Commit
•
b757b9d
1
Parent(s):
6149745
Upload assemble_data.py with huggingface_hub
Browse files- assemble_data.py +60 -0
assemble_data.py
ADDED
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import argparse
|
2 |
+
import os
|
3 |
+
from netCDF4 import Dataset
|
4 |
+
|
5 |
+
|
6 |
+
def assemble_data(input_dir, output_file):
|
7 |
+
nc_files = [f for f in os.listdir(input_dir) if f.endswith(".nc")]
|
8 |
+
nc_files.sort()
|
9 |
+
|
10 |
+
samples = [0]
|
11 |
+
with Dataset(os.path.join(input_dir, nc_files[0]), "r") as first_nc:
|
12 |
+
samples.append(first_nc.dimensions["sample"].size)
|
13 |
+
num_times = first_nc.dimensions["time"].size
|
14 |
+
try:
|
15 |
+
num_channels = first_nc.dimensions["channel"].size
|
16 |
+
except:
|
17 |
+
num_channels = None
|
18 |
+
x_size = first_nc.dimensions["x"].size
|
19 |
+
y_size = first_nc.dimensions["y"].size
|
20 |
+
dtype = first_nc.variables[nc_files[0].split("_")[0]].dtype
|
21 |
+
|
22 |
+
for nc_file in nc_files[1:]:
|
23 |
+
with Dataset(os.path.join(input_dir, nc_file), "r") as nc:
|
24 |
+
samples.append(nc.dimensions["sample"].size)
|
25 |
+
|
26 |
+
num_samples = sum(samples)
|
27 |
+
for i in range(1, len(samples)):
|
28 |
+
samples[i] += samples[i - 1]
|
29 |
+
with Dataset(output_file, "w") as out_nc:
|
30 |
+
out_nc.createDimension("sample", num_samples)
|
31 |
+
out_nc.createDimension("time", num_times)
|
32 |
+
if num_channels is not None:
|
33 |
+
out_nc.createDimension("channel", num_channels)
|
34 |
+
out_nc.createDimension("x", x_size)
|
35 |
+
out_nc.createDimension("y", y_size)
|
36 |
+
if num_channels is not None:
|
37 |
+
out_nc.createVariable(
|
38 |
+
nc_files[0].split("_")[0], dtype, ("sample", "time", "channel", "x", "y")
|
39 |
+
)
|
40 |
+
else:
|
41 |
+
out_nc.createVariable(
|
42 |
+
nc_files[0].split("_")[0], dtype, ("sample", "time", "x", "y")
|
43 |
+
)
|
44 |
+
|
45 |
+
for i, nc_file in enumerate(nc_files):
|
46 |
+
with Dataset(os.path.join(input_dir, nc_file), "r") as nc:
|
47 |
+
print(f"Processing {os.path.join(input_dir, nc_file)}")
|
48 |
+
variable = nc.variables[nc_file.split("_")[0]]
|
49 |
+
out_nc[nc_file.split("_")[0]][samples[i] : samples[i + 1]] = variable[:]
|
50 |
+
|
51 |
+
print(f"Saved data to {output_file}")
|
52 |
+
|
53 |
+
|
54 |
+
if __name__ == "__main__":
|
55 |
+
parser = argparse.ArgumentParser()
|
56 |
+
parser.add_argument("--input_dir", type=str, required=True)
|
57 |
+
parser.add_argument("--output_file", type=str, required=True)
|
58 |
+
args = parser.parse_args()
|
59 |
+
|
60 |
+
assemble_data(args.input_dir, args.output_file)
|