One way to do this is just parse the first of every four lines and leave the others untouched:

#!/usr/bin/env python                                                                                                                              

import sys

l = 0
for e in sys.stdin:
    e = e.rstrip()
    if l % 4 == 0:
        p = e[1:].split(':')
        h = ':'.join(p[0:-1])
        t = p[-1].split('+')
        e = []
        e.append('@')
        e.append(t[0][0:8])
        e.append(t[0][-8:])
        e.append(t[1][0:8])
        e.append(t[1][-8:])
        e.append(':')
        e.append(h)
        e = ''.join(e)
    sys.stdout.write('%s\n' % (e))
    l += 1

Example input:

$ cat input.fq
@M01581:1209:000000000-D3YJT:1:1102:16163:1449 2:N:0:TCTCGCGCGGACAGGGACAGCCGCGCCCACGCTACTTACTCCT+TTAAGGAGTCGTCGGCAGCGTCTCCACGCTGCTCTGA
TTTTCTTCTTTTTTTTTTTTTTTTTTTTTTTTTTCTTTCTTTCC
+
1111113331311100A0////////////////0111221121

Example output:

$ ./rewrite_fq.py < input.fq
@TCTCGCGCTTACTCCTTTAAGGAGTGCTCTGA:M01581:1209:000000000-D3YJT:1:1102:16163:1449 2:N:0
TTTTCTTCTTTTTTTTTTTTTTTTTTTTTTTTTTCTTTCTTTCC
+
1111113331311100A0////////////////0111221121

Or if you have compressed FASTQ, use a bash process substitution to feed uncompressed standard input to the script:

$ ./rewrite_fq.py < <(gunzip -c input.fq.gz)

To write the result to a compressed file:

$ ./rewrite_fq.py < <(gunzip -c input.fq.gz) | gzip -c > output.fq.gz

Etc.

sed solution: output:

$ sed '/@/ s/\@\(.*\)\:\([A-Z]\{8\}\).*\([A-Z]\{8\}\)+\([A-Z]\{8\}\).*\([A-Z]\{8\}\)/\@\2\3\4\5\:\1/g' test.fastq 

@TCTCGCGCTTACTCCTTTAAGGAGTGCTCTGA:M01581:1209:000000000-D3YJT:1:1102:16163:1449 2:N:0
TTTTCTTCTTTTTTTTTTTTTTTTTTTTTTTTTTCTTTCTTTCC
+
1111113331311100A0////////////////0111221121

input:

$ cat test.fastq 

@M01581:1209:000000000-D3YJT:1:1102:16163:1449 2:N:0:TCTCGCGCGGACAGGGACAGCCGCGCCCACGCTACTTACTCCT+TTAAGGAGTCGTCGGCAGCGTCTCCACGCTGCTCTGA
TTTTCTTCTTTTTTTTTTTTTTTTTTTTTTTTTTCTTTCTTTCC
+
1111113331311100A0////////////////0111221121