将文本文件读入结构化二维 numpy 数组

Question

I have a structured 2D numpy array of shape: [2,2]

filled example [2,2] array:

Main_Clt_data_array:

[ [ ((10, 10), 20, 300.) ((20, 20), 20, 300.) ]
  [((30, 30), 30, 300.) ((40, 40), 40, 300.)] ] 

---

It is of the following datatype

Cord_dtype = np.dtype([('X', np.float64), ('Y', np.float64)])
Clt_data_dtype=np.dtype([('Coord', Cord_dtype),('Angle', np.float64), ('Length', np.float64)])

---

Using np.savetxt i was able to save the array into a txt file maintaing same shape and order.

'Trial.txt': enter image description here

---

but when i try to read into python using genfromtxt,It does not work.: example:

read_array = np.genfromtxt('Trial.txt',dtype=Clt_data_dtype)

OUTPUT :

('READ_ARRAY_shape :', (2L,))

('READ_ARRAY :', array([((nan, nan), nan, nan), ((nan, nan), nan, nan)], 'V32'))

---

Hope some one could guide me and Thanks in advance for any input

also note i cannot use pandas as i am running it only numpy compatable program

Answer 1

In [17]: Cord_dtype = np.dtype([('X', np.float64), ('Y', np.float64)])
    ...: Clt_data_dtype=np.dtype([('Coord', Cord_dtype),('Angle', np.float64), ('Length', np.float64)])

Creating your array:

In [18]: arr = np.array([ [ ((10, 10), 20, 300.), ((20, 20), 20, 300.) ],
    ...:   [((30, 30), 30, 300.), ((40, 40), 40, 300.)] ], dtype=Clt_data_dtype)
    ...: 
    ...: 
In [19]: arr
Out[19]: 
array([[((10., 10.), 20., 300.), ((20., 20.), 20., 300.)],
       [((30., 30.), 30., 300.), ((40., 40.), 40., 300.)]],
      dtype=[('Coord', [('X', '<f8'), ('Y', '<f8')]), ('Angle', '<f8'), ('Length', '<f8')])
In [20]: arr.shape
Out[20]: (2, 2)

genfromtxt can recreate it from a csv style file like:

In [21]: txt = """10 10 20 300.
    ...: 20 20 20 300.
    ...: 30 30 30 300.
    ...: 40 40 40 300.
    ...: """
     
In [22]: np.genfromtxt(txt.splitlines(), dtype=Clt_data_dtype)
Out[22]: 
array([((10., 10.), 20., 300.), ((20., 20.), 20., 300.),
       ((30., 30.), 30., 300.), ((40., 40.), 40., 300.)],
      dtype=[('Coord', [('X', '<f8'), ('Y', '<f8')]), ('Angle', '<f8'), ('Length', '<f8')])

In true csv fashion this just contains rows and columns of numbers. It does not have the delimiters and () that your img shows. It may be possible to parse your image and create the file, but a csv reader can't handle.