Mono-core python dict based script

extract-network-from-log.py

import sys,codecs
'''
format emiter receiver contacts  

example: 5767599 242682935 4

'''
links={}
i=0
f=file("/dev/stdin")


for line in f:
        i+=1
        #print line
        line=line.split(";");
        emiter=int(line[0])
        reciever=int(line[1])
        contacts=int(line[2])
        l=sorted([emiter,reciever])
        #rel_key='-'.join(map(str,l))
        rel_key=(l[0],l[1])
        di=0 if caller<callee else 1
        if not rel_key in links:
                links[rel_key]=[0,0]
        links[rel_key][di]+=contacts
        if i%1000000==0:
                print "read %i million contacts"%(i/1000000)
                print len(links),"relationships in dict"

f.close()
f=codecs.open("network-mutual-graph","w")
f2=codecs.open("network-directed-graph","w")

for link,contacs in links.iteritems():
        link=link.split('-')
        if contacts[0]==0:
                f2.write("%s %s %i \n"%(link[1],link[0],contacts[1]))
        elif contacts[1]==0:
                f2.write("%s %s %i \n"%(link[0],link[1],contacts[0]))
        else:
                f.write("%s %s %i %i \n"%(link[0],link[1],contacts[0],contacts[1]))