Sorting Scientific Number With Unix Sort

问题

I tried to sort these number with Unix sort, but it doesn't seem to work:

This is my command:

sort -nr file.txt

What's the right way to do it?

回答1:

Use -g (long form --general-numeric-sort) instead of -n. The -g option passes the numbers through strtod to obtain their value, and it will recognize this format.

I'm not sure if this is available on all implementations of sort, or just the GNU one.

回答2:

if your sort doesn't have -g, another way.

$ printf "%.200f\n" $(<file) |sort -n |xargs printf "%g\n"
1e-104
9e-99
1e-91
3e-19
2e-13
2e-13

$ sort -g file
1e-104
9e-99
1e-91
3e-19
2e-13
2e-13

$ printf "%.200f\n" `cat file` |sort -n |xargs printf "%g\n"

回答3:

Just do two things:

Use -g (or --general-numeric-sort) to make sort treat Exp-numbers correctly.
Use LC_ALL=C. The sort is very sensible to locale if your data may contain some language-specific symbols except ASCII. So using LC_ALL=C is the universal approach for every case you use the sort, it makes sort to treat the input stream as binary, and you wouldn't have any problems.

So the universal solution is:

cat file.txt | LC_ALL=C sort -gr | less

Also I made an alias for sort in my .bashrc file:

alias csort="LC_ALL=C sort"

for much convinient use.

回答4:

Ok, here is a not fully tested version of Python script. Supposed usage:

sort_script.py file.txt

Unfortunately I developed this on Windows, and with 2 different versions of Python installed I cannot test it properly. Warning: requires newest Python (with, print functions added or changed). Note: back_to_file flag can be an optional parameter, although with Unix you can always redirect ... even in Windows you can.

#!/usr/bin/env python3.1
# Note: requires newer python

import sys

#Remove this line:
sys.argv = ('', 'file.txt')

assert(len(sys.argv) == 2)

with open(sys.argv[1], 'r') as fin:
    lines = fin.readlines()

lines_sorted = sorted(lines, key=lambda x: float(x))

back_to_file = False # Change this if needed

if back_to_file:
    with open(sys.argv[1], 'w') as fout:
        fout.writelines(lines_sorted)
else:
    for lns in lines_sorted:
        print(lns, end='') # Suppress new line

来源：https://stackoverflow.com/questions/2642349/sorting-scientific-number-with-unix-sort

标签

Linux

unix

sorting

numbers