iepy-develop/iepy/selfpreprocess/BiddingKG/dl/fingerprint/documentFingerprint.py

import hashlib
import codecs

def getMD5(sourceHtml):
    if sourceHtml is not None and len(sourceHtml)>0:
        if isinstance(sourceHtml,str):
            bs = sourceHtml.encode()
        elif isinstance(sourceHtml,bytes):
            bs = sourceHtml
        else:
            return ""
        md5 = hashlib.md5()
        md5.update(bs)
        return md5.hexdigest()
    return ""

def getFingerprint(sourceHtml):
    md5 = getMD5(sourceHtml)
    if md5!="":
        _fingerprint = "md5=%s"%(md5)
    else:
        _fingerprint = ""
    return _fingerprint

if __name__=="__main__":
    sourceHtml = text = codecs.open("C:\\Users\\User\\Desktop\\2.html","rb",encoding="utf8").read()
    print(getFingerprint(sourceHtml))