def save(self): #get one .tar.gz file proxy001_tarfiles = LogFilesTarGzGlob.test_glob3() tar_num = len(proxy001_tarfiles) for i in range(tar_num): #get 1 gzfile. gzfile = proxy001_tarfiles[i] #gzfile open. gzfileinfo = GzFileInfo(gzfile) gzfile = ExtractFiles(gzfileinfo.gzfilename) print "gzfileinfo: " + gzfileinfo.gzfilename #log file extract one by one for extractedfileobj in gzfile.extract_files(): cur_file = gzfile.cur_extract_file print cur_file #db collection resetting. table = self.conn.set_cur_col(cur_file) log_data = LogDataLineSplit(extractedfileobj) for data in log_data.read_line(): #build structure item = LogDataStructure(data).log_data item.__setitem__("log_server", gzfileinfo.log_server) item.__setitem__("log_date", gzfileinfo.log_date) table.insert(item)
def test(): #get one .tar.gz file files = LogFilesTarGzGlob.test_glob() gzfile = files[0] gzfileinfo = GzFileInfo(gzfile) gzfile = ExtractFiles(gzfileinfo.gzfilename) extractedfileobj = None for i in gzfile.extract_files(): extractedfileobj = i print extractedfileobj.name print gzfile.cur_extract_file #print gzfile.cur_extract_file.split('.')[1].split('/')[1] print "gzfileinfo: " + gzfileinfo.log_server + gzfileinfo.log_date return extractedfileobj