def getMergeTitle(): rows=tablemerge.getTitles(dbconfig.mergetable, limit=5000) if rows !=-1: f_title_ori=open(merge1_title_file_ori,'w') with open(merge1_title_file,'w') as fout: for row in rows: # id,title,summary,ctime,source mtid,title,summary,ctime=row[0],row[1].strip(),re.sub('\s+','',row[2]),row[3] msg_t_ori='%s %s'%(mtid,title) title=delpunc(' '.join(jieba.cut(title)).lower()).encode('utf-8') msg_t='%s %s'%(mtid,title) print msg_t_ori fout.write(msg_t+'\n') f_title_ori.write(msg_t_ori+'\n') f_title_ori.close()
def getTimeSpan(): rows=tablemerge.getTitles(dbconfig.mergetable, limit=5000) min_time,max_time=long(time.time()),0 if rows !=-1: for row in rows: # id,title,summary,ctime,source mtid,title,summary,ctime=row[0],row[1].strip(),re.sub('\s+','',row[2]),row[3] if ctime > max_time: max_time=ctime if ctime < min_time: min_time=ctime print time.time() print min_time print max_time print time.strftime('%Y-%m-%d %H:%M:%S',time.localtime(min_time)),'~',\ time.strftime('%Y-%m-%d %H:%M:%S',time.localtime(max_time))