How to only print lines with unique fields?

For example... if I had a file like this:

A   16  chr11   36595888
A   0   chr1    155517200
B   16  chr1    43227072
C   0   chr20   55648508
D   0   chr2    52375454
D   16  chr2    73574214
D   0   chr3    93549403
E   16  chr3    3315671

I need to print only the lines which have a unique first column:

B   16开发者_高级运维  chr1    43227072
C   0   chr20   55648508
E   16  chr3    3315671

It's similar to awk '!_[$1]++', but I want to remove all lines which have non-unique fist field.

Bash and python solutions preferably.

---

in bash, assuming first column has fixed with (3):

sort input-file.txt | uniq -u -w 3

'-u' option prints only the unique lines and '-w 3' compares no more than the first 3 characters.

---

How about this:

#!/usr/bin/env python
from collections import defaultdict
data = defaultdict(list)
with open('file', 'rb') as f:
    for line in sorted(f.readlines()):
        data[line[0]].append(line)
for key in sorted(data.iterkeys()):
    if len(data[key]) == 1:
        print data[key]

---

awk '
  {count[$1]++; line[$1]=$0}
  END {for (val in count) if (count[val]==1) print line[val]}
' filename

That may alter the order of lines. If that's a problem, try this 2-pass approach:

awk '
  NR==FNR {count[$1]++; next}
  count[$1] == 1 {print}
' filename filename

---

sed one liner solution:

sed ':a;$bb;N;/^\(.\).*\n\1[^\n]*$/ba;:b;s/^\(.\).*\n\1[^\n]*\n*//;ta;/./P;D' file

---

in python, much easier to read and tweak:

d = dict()
for line in open('input-file.txt', 'r'):
  key = line.split(' ', 1)[0]
  d.setdefault(key, list()).append(line.rstrip())

for k, v in sorted(d.items()):
  if len(v) == 1:
     print v[0]

---

import sys
from collections import OrderedDict
lines = OrderedDict()
for line in sys.stdin:
    field0 = line.strip().split('\t')[0]
    lines[field0] = None if field0 in lines else line
for line in lines.values():
    if line is not None:
        sys.stdout.write(line)

If you don't care aout preserving order you could use a plain old dict ({}) instead of OrderedDict.

This implementation doesn't care if the duplicate fields are adjacent.