使用 Python 打乱 csv 文件的所有行

Question

I have an input csv file with data:

a   15
b   14
c   20
d   45

I want to generate a different csv file which will contain complete data rows from input file but rows should be shuffled.

like output file may contain values-

b 14
a 15
c 20
d 45 

I have tried this code:

import random
import sys
op=open('random.csv','w+')
ip=open(sys.argv[1],'r')
data=ip.read()
data1=str(random.choices(data))
op.write(data1)
op.close()

Answer 1

Another shot using pandas . You can read your .csv file with:

df = pd.read_csv('yourfile.csv', header=None)

and then using df.sample to shuffle your rows. This will return a random sample of your dataframe with rows shuffled. Using frac=1 you consider the whole set as sample:

In [18]: df
Out[18]: 
   0   1
0  a  15
1  b  14
2  c  20
3  d  45

In [19]: ds = df.sample(frac=1)

In [20]: ds
Out[20]: 
   0   1
1  b  14
3  d  45
0  a  15
2  c  20

If you need to save out again the new shuffled file you can just:

ds.to_csv('newfile.csv')

Answer 2

You can use the shuffle function from Python random module. Like this:

import random
fid = open("example.txt", "r")
li = fid.readlines()
fid.close()
print(li)

random.shuffle(li)
print(li)

fid = open("shuffled_example.txt", "w")
fid.writelines(li)
fid.close()

The print commands result in this:

['b 14\n', 'a 15\n', 'c 20\n', 'd 45\n']
['d 45\n', 'a 15\n', 'b 14\n', 'c 20\n']

And the new file is this:

d 45
a 15
b 14
c 20

Just make sure you have a newline at the end of each of your original lines.

Answer 3

You can use shuf .

Once you installed shuf , run

shuf -o shuffled-file.csv < file-to-shuffle.csv

Answer 4

There is a shuffle function in the random module. Also, you can you readlines() in order to have a list:

>>> ip=open('random.csv','r')
>>> data=ip.readlines()
>>> data
['a   15\n', 'b   14\n', 'c   20\n', 'd   45\n']
>>> from random import shuffle
>>> shuffle(data)
>>> data
['c   20\n', 'd   45\n', 'a   15\n', 'b   14\n']

If you have an header, just split the data, and shuffle the rows:

>>> ip=open('random.csv','r')
>>> data=ip.readlines()
>>> header, rest=data[0], data[1:]
>>> header
'h1  h2\n'
>>> rest
['a   15\n', 'b   14\n', 'c   20\n', 'd   45\n']
>>> shuffle(rest)
>>> rest
['c   20\n', 'd   45\n', 'a   15\n', 'b   14\n']
>>> [header]+rest
['h1  h2\n', 'c   20\n', 'd   45\n', 'a   15\n', 'b   14\n']

Using with statement:

>>> with open('random.csv','r') as ip:
...   data=ip.readlines()
...
>>> header, rest=data[0], data[1:]
>>> shuffle(rest)
>>> with open('output.csv','w') as out:
...   out.write(''.join([header]+rest))
...
>>>
~$ cat output.csv
h1  h2
d   45
b   14
a   15
c   20

Answer 5

I think you should read the actual lines of the file.

ip.readlines()

And random.shuffle() should be used to swap around the lines.

At the moment, you read an entire string and I think only randomly get a single character from the entire file.

Answer 6

If your CSV contains headers then you can shuffle it using pandas like this.

df = pd.read_csv(file_name) # avoid header=None. 
shuffled_df = df.sample(frac=1)
shuffled_df.to_csv(new_file_name, index=False)

This way you can avoid shuffling headers and remove index from your new CSV.

Answer 7

I'm using this code according to @cricket answer:

from random import shuffle

with open('input.csv','r') as f1:
    data=f1.readlines()

shuffle(data[0])

with open('output.csv','w') as f2:
    f2.write(''.join([data[1:]] + data[0]))

Answer 8

I follow this way.

import numpy as np
import pandas as pd

df = pd.read_csv("your_csv_file.csv", header=0)    
df.reindex(np.random.permutation(df.index))