目前我有一个初始系统在工作,它读取一个文件,每行格式如下:
REVISION 12 30364918 Anarchism 2005-12-06T17:44:47Z RJII 141644
使用此代码:
CREATE EXTERNAL TABLE mytable(type STRING, aid BIGINT, rid BIGINT, title STRING, ts STRING, uname STRING, uid STRING) ROW FORMAT DELIMITED FIELDS TERMINATED BY ' ' STORED AS TEXTFILE LOCATION '/my/local/path/to/file';
但现在我有一个文件,每行都有这样的格式:
2001-04-29T15:43:48Z [10656,251129]
(注意时间戳后面有一个标签)
我不知道如何从中提取数据。我在网上寻找建议,但似乎什么都找不到。
如果理论上我想从这个
ts STRING -> 2001-04-29T15:43:48Z
aid BIGINT -> 10656
rid BIGINT -> 251129
看起来第二个字段是一个JSON数组。您可以创建映射到实际底层结构的表,并定义一个视图来提取值。Brickhouse中的"from_json"one_answers"json_split"UDF(http://github.com/klout/brickhouse)可以为您解析JSON
CREATE EXTERNAL TABLE mytable(datestr STRING, jsonArray STRING)
ROW FORMAT FIELDS DELIMITED BY 't';
CREATE VIEW myview AS
SELECT datestr, numArr[0] as aid, numArr[1] as rid
FROM ( SELECT datestr, from_json( jsonArray, array( cast(1 as bigint) ) ;