从文件中读取每一行作为 python 中的 numpy 数组

Question

我有一个包含 4 行的文件 'test.txt'，如下所示：

[35,51,3701],[10,6,7700.75],[42,31,4700.5],[31,41,1700.25],[6,25,3901.25],[10,12,3301.5]
[4,13,3709.25],[13,24,3761.5]
[11,18,3101.25],[26,7,3671],[10,69,7800.75]
[5,5,3701.25]

我以第一行手动为例。这是我想做的：

arr = np.array([[35,51,3701],[10,6,7700.75],[42,31,4700.5],[31,41,1700.25],[6,25,3901.25],[10,12,3301.5]])

p, b, a = [], [], []
for i in range(0, len(arr)):
        b = b + [arr[i][0]]
        a = a + [arr[i][1]]
        p = p + [arr[i][2]]

print (p)
print (b)
print (a)

结果是（这正是我想要的）：

[3701.0, 7700.75, 4700.5, 1700.25, 3901.25, 3301.5]
[35.0, 10.0, 42.0, 31.0, 6.0, 10.0]
[51.0, 6.0, 31.0, 41.0, 25.0, 12.0]

但是，当我尝试逐行读取文件时，它不起作用。这是我使用的代码：

f = open('test.txt')

for line in f:
  print(line)
  arr = np.array([line.rstrip()])
  p, b, a = [], [], []
  for i in range(0, len(arr)):
          b = b + [arr[i][0]]
          a = a + [arr[i][1]]
          p = p + [arr[i][2]]

  print (p)
  print (b)
  print (a)
  print ("\n\n")

和结果（不是预期的）：

[35,51,3701],[10,6,7700.75],[42,31,4700.5],[31,41,1700.25],[6,25,3901.25],[10,12,3301.5]

['5']
['[']
['3']



[4,13,3709.25],[13,24,3761.5]

[',']
['[']
['4']



[11,18,3101.25],[26,7,3671],[10,69,7800.75]

['1']
['[']
['1']



[5,5,3701.25]
[',']
['[']
['5']

你能告诉我如何得到手动示例中的结果吗？谢谢

Answer 1

问题在行arr = np.array([line.rstrip()])。 line.rstrip() returns 字符串，不是数值。可以用ast.literal_eval来解析，例如：

from ast import literal_eval

f = open("a.txt", "r")

for line in f:
    print(line)
    arr = np.array(literal_eval(line.rstrip() + ","))
    p, b, a = [], [], []
    for i in range(len(arr)):
        b = b + [arr[i][0]]
        a = a + [arr[i][1]]
        p = p + [arr[i][2]]

    print(p)
    print(b)
    print(a)
    print("\n\n")

打印：

[35,51,3701],[10,6,7700.75],[42,31,4700.5],[31,41,1700.25],[6,25,3901.25],[10,12,3301.5]

[3701.0, 7700.75, 4700.5, 1700.25, 3901.25, 3301.5]
[35.0, 10.0, 42.0, 31.0, 6.0, 10.0]
[51.0, 6.0, 31.0, 41.0, 25.0, 12.0]



[4,13,3709.25],[13,24,3761.5]

[3709.25, 3761.5]
[4.0, 13.0]
[13.0, 24.0]



[11,18,3101.25],[26,7,3671],[10,69,7800.75]

[3101.25, 3671.0, 7800.75]
[11.0, 26.0, 10.0]
[18.0, 7.0, 69.0]



[5,5,3701.25]
[3701.25]
[5.0]
[5.0]

Answer 2

IIUC，你可以试试ast.literal_eval:

from ast import literal_eval
for line in f: # use your file here
    a = np.array(literal_eval(f'[{line}]'))
    p,b,a = a.T.tolist()
    print(p)
    print(b)
    print(a)
    print('---')

或者，如果您不想使用 numpy，请使用：

p,b,a = zip(*literal_eval(f'[{line}]'))

输出：

[35.0, 10.0, 42.0, 31.0, 6.0, 10.0]
[51.0, 6.0, 31.0, 41.0, 25.0, 12.0]
[3701.0, 7700.75, 4700.5, 1700.25, 3901.25, 3301.5]
---
[4.0, 13.0]
[13.0, 24.0]
[3709.25, 3761.5]
---
[11.0, 26.0, 10.0]
[18.0, 7.0, 69.0]
[3101.25, 3671.0, 7800.75]
---
[5.0]
[5.0]
[3701.25]
---

从文件中读取每一行作为 python 中的 numpy 数组

Read each line from file as numpy arrays in python

python

arrays

numpy