| 1 |
fmoscato |
1.1 |
########################################
|
| 2 |
|
|
## author: fmoscato
|
| 3 |
|
|
## script that take an input file containing a list of
|
| 4 |
|
|
## site:dataset
|
| 5 |
|
|
## and group by site
|
| 6 |
|
|
## input are input_file out_dir
|
| 7 |
|
|
##############################################
|
| 8 |
|
|
|
| 9 |
|
|
|
| 10 |
|
|
#!/bin/python
|
| 11 |
|
|
import sys
|
| 12 |
|
|
|
| 13 |
|
|
Usage = """
|
| 14 |
|
|
loadData.py
|
| 15 |
|
|
Options:
|
| 16 |
|
|
--input_file file that contains list of the form site:dataset
|
| 17 |
|
|
--out_dir output directory where print the group by result
|
| 18 |
|
|
"""
|
| 19 |
|
|
|
| 20 |
|
|
|
| 21 |
|
|
|
| 22 |
|
|
def loadData(input_file,out_dir):
|
| 23 |
|
|
|
| 24 |
|
|
fs=open(input_file,"r")
|
| 25 |
|
|
sites={}
|
| 26 |
|
|
lines =fs.readlines()
|
| 27 |
|
|
fs.close()
|
| 28 |
|
|
|
| 29 |
|
|
for f in lines:
|
| 30 |
|
|
ar=f.split(":")
|
| 31 |
|
|
if len(ar)<2:
|
| 32 |
|
|
print f
|
| 33 |
|
|
sys.exit(1)
|
| 34 |
|
|
if ar[0]=="":
|
| 35 |
|
|
ar[0]="EMPTY_MSS"
|
| 36 |
|
|
if ar[0] not in sites:
|
| 37 |
|
|
sites[ar[0]]=[]
|
| 38 |
|
|
sites[ar[0]].append(ar[1])
|
| 39 |
|
|
|
| 40 |
|
|
for i,lst in sites.iteritems():
|
| 41 |
|
|
t=i.replace(" ","")
|
| 42 |
|
|
if t=="": continue
|
| 43 |
|
|
f=open("%s/%s"%(out_dir,t),"w")
|
| 44 |
|
|
for item in lst:
|
| 45 |
|
|
f.write(item)
|
| 46 |
|
|
|
| 47 |
|
|
|
| 48 |
|
|
|
| 49 |
|
|
def main(opts, args):
|
| 50 |
|
|
|
| 51 |
|
|
input_file=''
|
| 52 |
|
|
out_dir=''
|
| 53 |
|
|
|
| 54 |
|
|
for opt, val in opts:
|
| 55 |
|
|
if opt=='--input_file':
|
| 56 |
|
|
input_file = val
|
| 57 |
|
|
if opt=='--out_dir':
|
| 58 |
|
|
out_dir = val
|
| 59 |
|
|
|
| 60 |
|
|
if not input_file or not out_dir:
|
| 61 |
|
|
print Usage
|
| 62 |
|
|
sys.exit(1)
|
| 63 |
|
|
|
| 64 |
|
|
loadData(input_file,out_dir)
|
| 65 |
|
|
|
| 66 |
|
|
if __name__ == '__main__':
|
| 67 |
|
|
import sys
|
| 68 |
|
|
import getopt
|
| 69 |
|
|
|
| 70 |
|
|
try:
|
| 71 |
|
|
opts, args = getopt.getopt(sys.argv[1:], "", ["input_file=","out_dir="])
|
| 72 |
|
|
except getopt.GetoptError, msg:
|
| 73 |
|
|
print
|
| 74 |
|
|
print msg
|
| 75 |
|
|
print
|
| 76 |
|
|
print Usage
|
| 77 |
|
|
sys.exit(1)
|
| 78 |
|
|
|
| 79 |
|
|
if len(opts) <1 :
|
| 80 |
|
|
print
|
| 81 |
|
|
print Usage
|
| 82 |
|
|
sys.exit(1)
|
| 83 |
|
|
|
| 84 |
|
|
|
| 85 |
|
|
main( opts, args)
|
| 86 |
|
|
|
