private void generateVisitData() throws Exception {
long counter = 0;
Integer[] appIds = new Integer[] {1, 2, 3, 4, 5, 6};
Long[] userIds =
new Long[] {
100011L, 100010L, 10090L, 10008L, 10007L, 10006L, 10005L, 10004L, 10003L, 10002L
};
String[] visitTime =
new String[] {
"2014-07-05 15:15:15",
"2014-07-10 15:15:15",
"2014-07-15 15:15:15",
"2014-07-20 15:15:15",
"2014-07-25 15:15:15",
"2014-07-30 15:15:15",
"2014-08-04 15:15:15",
"2014-08-09 15:15:15",
"2014-08-14 15:15:15"
};
Random rnd = new Random();
// 添加模拟的数据
for (int i = 0; i < 100; i++) {
counter++;
int order = rnd.nextInt(5);
int uOrder = rnd.nextInt(9);
String orgText =
String.format(
"auc\t%d\t%d\t1001\t3264456\trefer.page\tvisit.page\tvisit.param\t%s",
userIds[uOrder], appIds[order].intValue(), visitTime[uOrder]);
VisitLogEntry entry = new VisitLogEntry();
entry.parse(orgText);
mrPageVisitor.addInput(new LongWritable(counter), entry);
}
}
/**
* @param key
* @param value
* @param context
* @throws IOException
* @throws InterruptedException
* @should write to context
*/
@Override
protected void map(LongWritable key, VisitLogEntry value, Context context)
throws IOException, InterruptedException {
// 如果用户的ID小于等于0, 直接忽略掉
if (value.getUserId() <= 0) {
return;
}
// 去掉注册日期中的时间部分
Calendar cal = Calendar.getInstance();
cal.setTimeInMillis(value.getVisitTime());
GregorianCalendar gCal =
new GregorianCalendar(
cal.get(Calendar.YEAR), cal.get(Calendar.MONTH), cal.get(Calendar.DATE));
// 设置键中的各个字段信息
writableKey.setStatDate(gCal.getTimeInMillis());
writableKey.setAppId(value.getAppId());
writableKey.setTerminalCode(value.getTerminalCode());
writableKey.setUserId(value.getUserId());
context.write(writableKey, writableValue);
}
