def wordlist_docrequest(docpath, request_list):
#filemd5 = md5(docpath)
gettext = docsimlarity.fileio(docpath)
gettext = docsimlarity.textread(gettext)
request_list = docsimlarity.tf_text(gettext)
#request_list["FileMD5"] = filemd5
return json.dumps(request_list)
def wordlist_train(dirpath, listall):
list_scripts = os.listdir(dirpath)
dictall_md5 = {}
for sample in list_scripts:
filemd5 = md5(dirpath + sample)
gettext = docsimlarity.fileio(dirpath + sample)
gettext = docsimlarity.textread(gettext)
listall = docsimlarity.tf_text(gettext)
dictall_md5[filemd5] = listall
return json.dumps(dictall_md5)