For Loop in Python 按顺序遍历每个文件并将其追加到数组中
For Loop in Python to move through each file sequentially and append it into an array
我有一个包含 72 个文件夹的目录，每个文件夹中有 245 个 numpy 文件。该目录名为 MP_Data。我想以任何顺序读取每个文件夹中的所有 245 个 numpy 文件。但是，文件夹中的循环应按数字顺序排列。
这72个文件夹的命名如下：（我想保持这样）
r0
r5
r10
r15
r20
r25
r30
r35
r40
r45
r50
r55
r60
r65
r70
r75
r80
r85
r90
r95
r100
r105
r110
r115
r120
r125
r130
r135
r140
r145
r150
r155
r160
r165
r170
r175
r180
r185
r190
r195
r200
r205
r210
r215
r220
r225
r230
r235
r240
r245
r250
r255
r260
r265
r270
r275
r280
r285
r290
r295
r300
r305
r310
r315
r320
r325
r330
r335
r340
r345
r350
r355
这是我到目前为止尝试过的方法，但似乎没有用
sequences, labels = [], []
directory_1= r'C:\Users\.....\angle-model\MP_DATA'
direc=[]
for root, subdirectories, filenames in (sorted(os.walk(directory_1))):
subdirectories=natsorted(subdirectories)
for dirs in subdirectories:
direc.append(dirs)
direc=np.array(direc)
for directory in direc:
directory_2=(os.path.join(directory_1, directory))
for action in actions:
window = []
for root, subdirectories, filenames in (sorted(os.walk(directory_2))):
filenames = natsorted(filenames)
for filename in filenames:
res=np.load(os.path.join(directory_2, filename))
window.append(res)
sequences.append(window)
labels.append(label_map[action])
以上代码卡在同一个文件夹中，不会移动到下一个文件夹。如果我打印文件名，它会重复。
['72.npy', '144.npy', '216.npy', '288.npy', '360.npy', '432.npy', '504.npy', '576.npy', '648.npy', '720.npy', '792.npy', '864.npy', '936.npy', '1008.npy', '1080.npy', '1152.npy', '1224.npy', '1296.npy', '1368.npy', '1440.npy', '1512.npy', '1584.npy', '1656.npy', '1728.npy', '1800.npy', '1872.npy', '1944.npy', '2016.npy', '2088.npy', '2160.npy', '2232.npy', '2304.npy', '2376.npy', '2448.npy', '2520.npy', '2592.npy', '2664.npy', '2736.npy', '2808.npy', '2880.npy', '2952.npy', '3024.npy', '3096.npy', '3168.npy', '3240.npy', '3312.npy', '3384.npy', '3456.npy', '3528.npy', '3600.npy', '3672.npy', '3744.npy', '3816.npy', '3888.npy', '3960.npy', '4032.npy', '4104.npy', '4176.npy', '4248.npy', '4320.npy', '4392.npy', '4464.npy', '4536.npy', '4608.npy', '4680.npy', '4752.npy', '4824.npy', '4896.npy', '4968.npy', '5040.npy', '5112.npy', '5184.npy', '5256.npy', '5328.npy', '5400.npy', '5472.npy', '5544.npy', '5616.npy', '5688.npy', '5760.npy', '5832.npy', '5904.npy', '5976.npy', '6048.npy', '6120.npy', '6192.npy', '6264.npy', '6336.npy', '6408.npy', '6480.npy', '6552.npy', '6624.npy', '6696.npy', '6768.npy', '6840.npy', '6912.npy', '6984.npy', '7056.npy', '7128.npy', '7200.npy', '7272.npy', '7344.npy', '7416.npy', '7488.npy', '7560.npy', '7632.npy', '7704.npy', '7776.npy', '7848.npy', '7920.npy', '7992.npy', '8064.npy', '8136.npy', '8208.npy', '8280.npy', '8352.npy', '8424.npy', '8496.npy', '8568.npy', '8640.npy', '8712.npy', '8784.npy', '8856.npy', '8928.npy', '9000.npy', '9072.npy', '9144.npy', '9216.npy', '9288.npy', '9360.npy', '9432.npy', '9504.npy', '9576.npy', '9648.npy', '9720.npy', '9792.npy', '9864.npy', '9936.npy', '10008.npy', '10080.npy', '10152.npy', '10224.npy', '10296.npy', '10368.npy', '10440.npy', '10512.npy', '10584.npy', '10656.npy', '10728.npy', '10800.npy', '10872.npy', '10944.npy', '11016.npy', '11088.npy', '11160.npy', '11232.npy', '11304.npy', '11376.npy', '11448.npy', '11520.npy', '11592.npy', '11664.npy', '11736.npy', '11808.npy', '11880.npy', '11952.npy', '12024.npy', '12096.npy', '12168.npy', '12240.npy', '12312.npy', '12384.npy', '12456.npy', '12528.npy', '12600.npy', '12672.npy', '12744.npy', '12816.npy', '12888.npy', '12960.npy', '13032.npy', '13104.npy', '13176.npy', '13248.npy', '13320.npy', '13392.npy', '13464.npy', '13536.npy', '13608.npy', '13680.npy', '13752.npy', '13824.npy', '13896.npy', '13968.npy', '14040.npy', '14112.npy', '14184.npy', '14256.npy', '14328.npy', '14400.npy', '14472.npy', '14544.npy', '14616.npy', '14688.npy', '14760.npy', '14832.npy', '14904.npy', '14976.npy', '15048.npy', '15120.npy', '15192.npy', '15264.npy', '15336.npy', '15408.npy', '15480.npy', '15552.npy', '15624.npy', '15696.npy', '15768.npy', '15840.npy', '15912.npy', '15984.npy', '16056.npy', '16128.npy', '16200.npy', '16272.npy', '16344.npy', '16416.npy', '16488.npy', '16560.npy', '16632.npy', '16704.npy', '16776.npy', '16848.npy', '16920.npy', '16992.npy', '17064.npy', '17136.npy', '17208.npy', '17280.npy', '17352.npy', '17424.npy', '17496.npy', '17568.npy']
.
.
.
.
['72.npy', '144.npy', '216.npy', '288.npy', '360.npy', '432.npy', '504.npy', '576.npy', '648.npy', '720.npy', '792.npy', '864.npy', '936.npy', '1008.npy', '1080.npy', '1152.npy', '1224.npy', '1296.npy', '1368.npy', '1440.npy', '1512.npy', '1584.npy', '1656.npy', '1728.npy', '1800.npy', '1872.npy', '1944.npy', '2016.npy', '2088.npy', '2160.npy', '2232.npy', '2304.npy', '2376.npy', '2448.npy', '2520.npy', '2592.npy', '2664.npy', '2736.npy', '2808.npy', '2880.npy', '2952.npy', '3024.npy', '3096.npy', '3168.npy', '3240.npy', '3312.npy', '3384.npy', '3456.npy', '3528.npy', '3600.npy', '3672.npy', '3744.npy', '3816.npy', '3888.npy', '3960.npy', '4032.npy', '4104.npy', '4176.npy', '4248.npy', '4320.npy', '4392.npy', '4464.npy', '4536.npy', '4608.npy', '4680.npy', '4752.npy', '4824.npy', '4896.npy', '4968.npy', '5040.npy', '5112.npy', '5184.npy', '5256.npy', '5328.npy', '5400.npy', '5472.npy', '5544.npy', '5616.npy', '5688.npy', '5760.npy', '5832.npy', '5904.npy', '5976.npy', '6048.npy', '6120.npy', '6192.npy', '6264.npy', '6336.npy', '6408.npy', '6480.npy', '6552.npy', '6624.npy', '6696.npy', '6768.npy', '6840.npy', '6912.npy', '6984.npy', '7056.npy', '7128.npy', '7200.npy', '7272.npy', '7344.npy', '7416.npy', '7488.npy', '7560.npy', '7632.npy', '7704.npy', '7776.npy', '7848.npy', '7920.npy', '7992.npy', '8064.npy', '8136.npy', '8208.npy', '8280.npy', '8352.npy', '8424.npy', '8496.npy', '8568.npy', '8640.npy', '8712.npy', '8784.npy', '8856.npy', '8928.npy', '9000.npy', '9072.npy', '9144.npy', '9216.npy', '9288.npy', '9360.npy', '9432.npy', '9504.npy', '9576.npy', '9648.npy', '9720.npy', '9792.npy', '9864.npy', '9936.npy', '10008.npy', '10080.npy', '10152.npy', '10224.npy', '10296.npy', '10368.npy', '10440.npy', '10512.npy', '10584.npy', '10656.npy', '10728.npy', '10800.npy', '10872.npy', '10944.npy', '11016.npy', '11088.npy', '11160.npy', '11232.npy', '11304.npy', '11376.npy', '11448.npy', '11520.npy', '11592.npy', '11664.npy', '11736.npy', '11808.npy', '11880.npy', '11952.npy', '12024.npy', '12096.npy', '12168.npy', '12240.npy', '12312.npy', '12384.npy', '12456.npy', '12528.npy', '12600.npy', '12672.npy', '12744.npy', '12816.npy', '12888.npy', '12960.npy', '13032.npy', '13104.npy', '13176.npy', '13248.npy', '13320.npy', '13392.npy', '13464.npy', '13536.npy', '13608.npy', '13680.npy', '13752.npy', '13824.npy', '13896.npy', '13968.npy', '14040.npy', '14112.npy', '14184.npy', '14256.npy', '14328.npy', '14400.npy', '14472.npy', '14544.npy', '14616.npy', '14688.npy', '14760.npy', '14832.npy', '14904.npy', '14976.npy', '15048.npy', '15120.npy', '15192.npy', '15264.npy', '15336.npy', '15408.npy', '15480.npy', '15552.npy', '15624.npy', '15696.npy', '15768.npy', '15840.npy', '15912.npy', '15984.npy', '16056.npy', '16128.npy', '16200.npy', '16272.npy', '16344.npy', '16416.npy', '16488.npy', '16560.npy', '16632.npy', '16704.npy', '16776.npy', '16848.npy', '16920.npy', '16992.npy', '17064.npy', '17136.npy', '17208.npy', '17280.npy', '17352.npy', '17424.npy', '17496.npy', '17568.npy']
动作中的动作只是标签名称，用于在MP_Data 目录中创建文件夹。所以这里是 actions
的代码
DATA_PATH = os.path.join('MP_DATA')
actions=np.array(['r0', 'r5', 'r10', 'r15', 'r20', 'r25', 'r30', 'r35', 'r40', 'r45', 'r50', 'r55', 'r60', 'r65', 'r70', 'r75', 'r80', 'r85', 'r90', 'r95', 'r100', 'r105', 'r110', 'r115', 'r120', 'r125', 'r130', 'r135', 'r140', 'r145', 'r150', 'r155', 'r160', 'r165', 'r170', 'r175', 'r180', 'r185', 'r190', 'r195', 'r200', 'r205', 'r210', 'r215', 'r220', 'r225', 'r230', 'r235', 'r240', 'r245', 'r250', 'r255', 'r260', 'r265', 'r270', 'r275', 'r280', 'r285', 'r290', 'r295', 'r300', 'r305', 'r310', 'r315', 'r320', 'r325', 'r330', 'r335', 'r340', 'r345', 'r350', 'r355'])
print(len(actions))
for action in actions:
try:
os.makedirs(os.path.join(DATA_PATH, action))
except:
pass
您应该将 action 循环移到 directory 循环中。如果不这样做，action 循环仅使用 directory 循环的最后一个目录。
您的代码可以简化（除了 actions，您没有向我们展示）类似
import os
from collections import defaultdict
root = "/tmp/mp_data"
# Map labels (subdirectories of root) to data
data_per_label = defaultdict(list)
# Get all top-level directories within `root`
label_dirs = [
name for name in os.listdir(root) if os.path.isdir(os.path.join(root, name))
]
print(f"{label_dirs=}")
# Loop over each label directory
for label in label_dirs:
label_dir = os.path.join(root, label)
# Loop over each filename in the label directory
for filename in os.listdir(label_dir):
# Take care to only look at .npy files
if filename.endswith(".npy"):
filepath = os.path.join(label_dir, filename)
print(f"{label=} {filename=} {filepath=}")
data = filename # replace with np.load(filename)
data_per_label[label].append(data)
print(data_per_label)
给定一棵树
/tmp/mp_data
├── r1
│ └── a.npy
├── r2
│ └── b.npy
└── r3
└── c.npy
这会打印出来
label_dirs=['r1', 'r3', 'r2']
label='r1' filename='a.npy' filepath='/tmp/mp_data/r1/a.npy'
label='r3' filename='c.npy' filepath='/tmp/mp_data/r3/c.npy'
label='r2' filename='b.npy' filepath='/tmp/mp_data/r2/b.npy'
defaultdict(<class 'list'>, {'r1': ['/tmp/mp_data/r1/a.npy'], 'r3': ['/tmp/mp_data/r3/c.npy'], 'r2': ['/tmp/mp_data/r2/b.npy']})
这比你做的要简单得多。
have a directory that has 72 folders, and within each folder are 245 numpy files. The directory is called MP_Data. I would like to read all the 245 numpy files within each folder in any order. However, the loop through the folders should be in numerical order.
来自 os.walk 的文档：
When topdown is True, the caller can modify the dirnames list in-place (perhaps using del or slice assignment), and walk() will only recurse into the subdirectories whose names remain in dirnames; this can be used to prune the search, impose a specific order of visiting, or even to inform walk() about directories the caller creates or renames before it resumes walk() again.
因此，我们使用该功能使 os.walk 以正确的顺序访问所有子目录和文件，并在遇到文件时读取它们。我们应该只需要os.walk一次。因此：
window = []
for current, subdirectories, filenames in sorted(os.walk(directory_1)):
subdirectories[:] = natsorted(subdirectories)
for filename in filenames:
window.append(np.load(os.path.join(current, filename)))
读取每个文件一次并构建数据列表后，我们就可以为每个文件克隆该数据 action in actions:
from copy import deepcopy
sequences = [deepcopy(window) for action in actions]
labels = [label_map[action] for action in actions]
假设您实际上需要为每个操作单独的数据，也就是说。这将取决于您程序的其余部分。
我有一个包含 72 个文件夹的目录，每个文件夹中有 245 个 numpy 文件。该目录名为 MP_Data。我想以任何顺序读取每个文件夹中的所有 245 个 numpy 文件。但是，文件夹中的循环应按数字顺序排列。
这72个文件夹的命名如下：（我想保持这样）
r0
r5
r10
r15
r20
r25
r30
r35
r40
r45
r50
r55
r60
r65
r70
r75
r80
r85
r90
r95
r100
r105
r110
r115
r120
r125
r130
r135
r140
r145
r150
r155
r160
r165
r170
r175
r180
r185
r190
r195
r200
r205
r210
r215
r220
r225
r230
r235
r240
r245
r250
r255
r260
r265
r270
r275
r280
r285
r290
r295
r300
r305
r310
r315
r320
r325
r330
r335
r340
r345
r350
r355
这是我到目前为止尝试过的方法，但似乎没有用
sequences, labels = [], []
directory_1= r'C:\Users\.....\angle-model\MP_DATA'
direc=[]
for root, subdirectories, filenames in (sorted(os.walk(directory_1))):
subdirectories=natsorted(subdirectories)
for dirs in subdirectories:
direc.append(dirs)
direc=np.array(direc)
for directory in direc:
directory_2=(os.path.join(directory_1, directory))
for action in actions:
window = []
for root, subdirectories, filenames in (sorted(os.walk(directory_2))):
filenames = natsorted(filenames)
for filename in filenames:
res=np.load(os.path.join(directory_2, filename))
window.append(res)
sequences.append(window)
labels.append(label_map[action])
以上代码卡在同一个文件夹中，不会移动到下一个文件夹。如果我打印文件名，它会重复。
['72.npy', '144.npy', '216.npy', '288.npy', '360.npy', '432.npy', '504.npy', '576.npy', '648.npy', '720.npy', '792.npy', '864.npy', '936.npy', '1008.npy', '1080.npy', '1152.npy', '1224.npy', '1296.npy', '1368.npy', '1440.npy', '1512.npy', '1584.npy', '1656.npy', '1728.npy', '1800.npy', '1872.npy', '1944.npy', '2016.npy', '2088.npy', '2160.npy', '2232.npy', '2304.npy', '2376.npy', '2448.npy', '2520.npy', '2592.npy', '2664.npy', '2736.npy', '2808.npy', '2880.npy', '2952.npy', '3024.npy', '3096.npy', '3168.npy', '3240.npy', '3312.npy', '3384.npy', '3456.npy', '3528.npy', '3600.npy', '3672.npy', '3744.npy', '3816.npy', '3888.npy', '3960.npy', '4032.npy', '4104.npy', '4176.npy', '4248.npy', '4320.npy', '4392.npy', '4464.npy', '4536.npy', '4608.npy', '4680.npy', '4752.npy', '4824.npy', '4896.npy', '4968.npy', '5040.npy', '5112.npy', '5184.npy', '5256.npy', '5328.npy', '5400.npy', '5472.npy', '5544.npy', '5616.npy', '5688.npy', '5760.npy', '5832.npy', '5904.npy', '5976.npy', '6048.npy', '6120.npy', '6192.npy', '6264.npy', '6336.npy', '6408.npy', '6480.npy', '6552.npy', '6624.npy', '6696.npy', '6768.npy', '6840.npy', '6912.npy', '6984.npy', '7056.npy', '7128.npy', '7200.npy', '7272.npy', '7344.npy', '7416.npy', '7488.npy', '7560.npy', '7632.npy', '7704.npy', '7776.npy', '7848.npy', '7920.npy', '7992.npy', '8064.npy', '8136.npy', '8208.npy', '8280.npy', '8352.npy', '8424.npy', '8496.npy', '8568.npy', '8640.npy', '8712.npy', '8784.npy', '8856.npy', '8928.npy', '9000.npy', '9072.npy', '9144.npy', '9216.npy', '9288.npy', '9360.npy', '9432.npy', '9504.npy', '9576.npy', '9648.npy', '9720.npy', '9792.npy', '9864.npy', '9936.npy', '10008.npy', '10080.npy', '10152.npy', '10224.npy', '10296.npy', '10368.npy', '10440.npy', '10512.npy', '10584.npy', '10656.npy', '10728.npy', '10800.npy', '10872.npy', '10944.npy', '11016.npy', '11088.npy', '11160.npy', '11232.npy', '11304.npy', '11376.npy', '11448.npy', '11520.npy', '11592.npy', '11664.npy', '11736.npy', '11808.npy', '11880.npy', '11952.npy', '12024.npy', '12096.npy', '12168.npy', '12240.npy', '12312.npy', '12384.npy', '12456.npy', '12528.npy', '12600.npy', '12672.npy', '12744.npy', '12816.npy', '12888.npy', '12960.npy', '13032.npy', '13104.npy', '13176.npy', '13248.npy', '13320.npy', '13392.npy', '13464.npy', '13536.npy', '13608.npy', '13680.npy', '13752.npy', '13824.npy', '13896.npy', '13968.npy', '14040.npy', '14112.npy', '14184.npy', '14256.npy', '14328.npy', '14400.npy', '14472.npy', '14544.npy', '14616.npy', '14688.npy', '14760.npy', '14832.npy', '14904.npy', '14976.npy', '15048.npy', '15120.npy', '15192.npy', '15264.npy', '15336.npy', '15408.npy', '15480.npy', '15552.npy', '15624.npy', '15696.npy', '15768.npy', '15840.npy', '15912.npy', '15984.npy', '16056.npy', '16128.npy', '16200.npy', '16272.npy', '16344.npy', '16416.npy', '16488.npy', '16560.npy', '16632.npy', '16704.npy', '16776.npy', '16848.npy', '16920.npy', '16992.npy', '17064.npy', '17136.npy', '17208.npy', '17280.npy', '17352.npy', '17424.npy', '17496.npy', '17568.npy']
.
.
.
.
['72.npy', '144.npy', '216.npy', '288.npy', '360.npy', '432.npy', '504.npy', '576.npy', '648.npy', '720.npy', '792.npy', '864.npy', '936.npy', '1008.npy', '1080.npy', '1152.npy', '1224.npy', '1296.npy', '1368.npy', '1440.npy', '1512.npy', '1584.npy', '1656.npy', '1728.npy', '1800.npy', '1872.npy', '1944.npy', '2016.npy', '2088.npy', '2160.npy', '2232.npy', '2304.npy', '2376.npy', '2448.npy', '2520.npy', '2592.npy', '2664.npy', '2736.npy', '2808.npy', '2880.npy', '2952.npy', '3024.npy', '3096.npy', '3168.npy', '3240.npy', '3312.npy', '3384.npy', '3456.npy', '3528.npy', '3600.npy', '3672.npy', '3744.npy', '3816.npy', '3888.npy', '3960.npy', '4032.npy', '4104.npy', '4176.npy', '4248.npy', '4320.npy', '4392.npy', '4464.npy', '4536.npy', '4608.npy', '4680.npy', '4752.npy', '4824.npy', '4896.npy', '4968.npy', '5040.npy', '5112.npy', '5184.npy', '5256.npy', '5328.npy', '5400.npy', '5472.npy', '5544.npy', '5616.npy', '5688.npy', '5760.npy', '5832.npy', '5904.npy', '5976.npy', '6048.npy', '6120.npy', '6192.npy', '6264.npy', '6336.npy', '6408.npy', '6480.npy', '6552.npy', '6624.npy', '6696.npy', '6768.npy', '6840.npy', '6912.npy', '6984.npy', '7056.npy', '7128.npy', '7200.npy', '7272.npy', '7344.npy', '7416.npy', '7488.npy', '7560.npy', '7632.npy', '7704.npy', '7776.npy', '7848.npy', '7920.npy', '7992.npy', '8064.npy', '8136.npy', '8208.npy', '8280.npy', '8352.npy', '8424.npy', '8496.npy', '8568.npy', '8640.npy', '8712.npy', '8784.npy', '8856.npy', '8928.npy', '9000.npy', '9072.npy', '9144.npy', '9216.npy', '9288.npy', '9360.npy', '9432.npy', '9504.npy', '9576.npy', '9648.npy', '9720.npy', '9792.npy', '9864.npy', '9936.npy', '10008.npy', '10080.npy', '10152.npy', '10224.npy', '10296.npy', '10368.npy', '10440.npy', '10512.npy', '10584.npy', '10656.npy', '10728.npy', '10800.npy', '10872.npy', '10944.npy', '11016.npy', '11088.npy', '11160.npy', '11232.npy', '11304.npy', '11376.npy', '11448.npy', '11520.npy', '11592.npy', '11664.npy', '11736.npy', '11808.npy', '11880.npy', '11952.npy', '12024.npy', '12096.npy', '12168.npy', '12240.npy', '12312.npy', '12384.npy', '12456.npy', '12528.npy', '12600.npy', '12672.npy', '12744.npy', '12816.npy', '12888.npy', '12960.npy', '13032.npy', '13104.npy', '13176.npy', '13248.npy', '13320.npy', '13392.npy', '13464.npy', '13536.npy', '13608.npy', '13680.npy', '13752.npy', '13824.npy', '13896.npy', '13968.npy', '14040.npy', '14112.npy', '14184.npy', '14256.npy', '14328.npy', '14400.npy', '14472.npy', '14544.npy', '14616.npy', '14688.npy', '14760.npy', '14832.npy', '14904.npy', '14976.npy', '15048.npy', '15120.npy', '15192.npy', '15264.npy', '15336.npy', '15408.npy', '15480.npy', '15552.npy', '15624.npy', '15696.npy', '15768.npy', '15840.npy', '15912.npy', '15984.npy', '16056.npy', '16128.npy', '16200.npy', '16272.npy', '16344.npy', '16416.npy', '16488.npy', '16560.npy', '16632.npy', '16704.npy', '16776.npy', '16848.npy', '16920.npy', '16992.npy', '17064.npy', '17136.npy', '17208.npy', '17280.npy', '17352.npy', '17424.npy', '17496.npy', '17568.npy']
动作中的动作只是标签名称，用于在MP_Data 目录中创建文件夹。所以这里是 actions
的代码DATA_PATH = os.path.join('MP_DATA')
actions=np.array(['r0', 'r5', 'r10', 'r15', 'r20', 'r25', 'r30', 'r35', 'r40', 'r45', 'r50', 'r55', 'r60', 'r65', 'r70', 'r75', 'r80', 'r85', 'r90', 'r95', 'r100', 'r105', 'r110', 'r115', 'r120', 'r125', 'r130', 'r135', 'r140', 'r145', 'r150', 'r155', 'r160', 'r165', 'r170', 'r175', 'r180', 'r185', 'r190', 'r195', 'r200', 'r205', 'r210', 'r215', 'r220', 'r225', 'r230', 'r235', 'r240', 'r245', 'r250', 'r255', 'r260', 'r265', 'r270', 'r275', 'r280', 'r285', 'r290', 'r295', 'r300', 'r305', 'r310', 'r315', 'r320', 'r325', 'r330', 'r335', 'r340', 'r345', 'r350', 'r355'])
print(len(actions))
for action in actions:
try:
os.makedirs(os.path.join(DATA_PATH, action))
except:
pass
您应该将 action 循环移到 directory 循环中。如果不这样做，action 循环仅使用 directory 循环的最后一个目录。
您的代码可以简化（除了 actions，您没有向我们展示）类似
import os
from collections import defaultdict
root = "/tmp/mp_data"
# Map labels (subdirectories of root) to data
data_per_label = defaultdict(list)
# Get all top-level directories within `root`
label_dirs = [
name for name in os.listdir(root) if os.path.isdir(os.path.join(root, name))
]
print(f"{label_dirs=}")
# Loop over each label directory
for label in label_dirs:
label_dir = os.path.join(root, label)
# Loop over each filename in the label directory
for filename in os.listdir(label_dir):
# Take care to only look at .npy files
if filename.endswith(".npy"):
filepath = os.path.join(label_dir, filename)
print(f"{label=} {filename=} {filepath=}")
data = filename # replace with np.load(filename)
data_per_label[label].append(data)
print(data_per_label)
给定一棵树
/tmp/mp_data
├── r1
│ └── a.npy
├── r2
│ └── b.npy
└── r3
└── c.npy
这会打印出来
label_dirs=['r1', 'r3', 'r2']
label='r1' filename='a.npy' filepath='/tmp/mp_data/r1/a.npy'
label='r3' filename='c.npy' filepath='/tmp/mp_data/r3/c.npy'
label='r2' filename='b.npy' filepath='/tmp/mp_data/r2/b.npy'
defaultdict(<class 'list'>, {'r1': ['/tmp/mp_data/r1/a.npy'], 'r3': ['/tmp/mp_data/r3/c.npy'], 'r2': ['/tmp/mp_data/r2/b.npy']})
这比你做的要简单得多。
have a directory that has 72 folders, and within each folder are 245 numpy files. The directory is called MP_Data. I would like to read all the 245 numpy files within each folder in any order. However, the loop through the folders should be in numerical order.
来自 os.walk 的文档：
When topdown is
True, the caller can modify the dirnames list in-place (perhaps usingdelor slice assignment), andwalk()will only recurse into the subdirectories whose names remain in dirnames; this can be used to prune the search, impose a specific order of visiting, or even to informwalk()about directories the caller creates or renames before it resumeswalk()again.
因此，我们使用该功能使 os.walk 以正确的顺序访问所有子目录和文件，并在遇到文件时读取它们。我们应该只需要os.walk一次。因此：
window = []
for current, subdirectories, filenames in sorted(os.walk(directory_1)):
subdirectories[:] = natsorted(subdirectories)
for filename in filenames:
window.append(np.load(os.path.join(current, filename)))
读取每个文件一次并构建数据列表后，我们就可以为每个文件克隆该数据 action in actions:
from copy import deepcopy
sequences = [deepcopy(window) for action in actions]
labels = [label_map[action] for action in actions]
假设您实际上需要为每个操作单独的数据，也就是说。这将取决于您程序的其余部分。