WikiElection network construction

wikielections.py

import re
import urllib2
import gzip
import StringIO
import numpy as np

def readDataset(save=True):
    content = ''
    outFilePath = 'wikielect.txt'
    try:
        with open(outFilePath, 'r') as f:
            content = f.read()
    except:
        response = urllib2.urlopen('http://snap.stanford.edu/data/wikiElec.ElecBs3.txt.gz')
        compressedFile = StringIO.StringIO()
        compressedFile.write(response.read())
        compressedFile.seek(0)
        with gzip.GzipFile(fileobj=compressedFile, mode='rb') as f:
            content = f.read()
            if save:
                with open(outFilePath, 'w') as outfile:
                    outfile.write(content)

    return content

content = readDataset()
content = content.split('\n')

def formAdj(records,size_estimate=10000):
    targetPattern = re.compile('N\s+([0-9]+)\s+[a-zA-Z]+')
    sourcePattern = re.compile('V\s+(0|1|\-1)\s+([0-9]+)')
    target = -1
    max_index = 0
    G = np.zeros([size_estimate,size_estimate])
    for record in records:
        match = targetPattern.match(record)
        if match:
            target = int(match.groups()[0])
        else:
            match = sourcePattern.match(record)
            if match:
                source = int(match.groups()[1])
                sign = 1 if (int(match.groups()[0])>-1) else -1
                G[source,target] = sign
                max_index = np.max([source,target,max_index])
    return G

G = formAdj(content)