2012-08-22
author:周海汉 2012.8.22
一、shell 查询
hbase 查询相当简单，提供了get和scan两种方式，也不存在多表联合查询的问题。复杂查询需通过hive创建相应外部表，用sql语句自动生成mapreduce进行。 但是这种简单，有时为了达到目的，也不是那么顺手。至少和sql查询方式相差较大。
hbase 提供了很多过滤器，可对行键，列，值进行过滤。过滤方式可以是子串，二进制，前缀，正则比较等。条件可以是AND,OR等 组合。所以通过过滤，还是能满足需求，找到正确的结果的。
1.1 过滤器类型
HBase 最新官方文档中文版(http://abloz.com/hbase/book.html)中有对过滤器的描述。过滤器分为5种类型：
-
构造型过滤器：用于包含其他一组过滤器的过滤器。包括：FilterList
-
列值型过滤器：对每列的值进行过滤的. 相当于sql查询中的=和like 包括：
SingleColumnValueFilter
比较器，包括：
<a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/RegexStringComparator.html" target="_top">RegexStringComparator</a> 支持值比较的正则表达式
<a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/SubstringComparator.html" target="_top">SubstringComparator</a> 用于检测一个子串是否存在于值中。大小写不敏感。
<a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/BinaryPrefixComparator.html" target="_top">BinaryPrefixComparator</a> 二进制前缀比较
<a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/BinaryComparator.html" target="_top">BinaryComparator</a> 二进制比较
- 键值元数据过滤器：用于对列进行过滤的。包括：
<a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/FamilyFilter.html" target="_top">FamilyFilter</a> 用于过滤列族。 通常，在Scan中选择ColumnFamilie优于在过滤器中做。
<a href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/QualifierFilter.html" target="_top">QualifierFilter</a> 用于基于列名(即 Qualifier)过滤.
ColumnPrefixFilter 可基于列名(即Qualifier)前缀过滤。
MultipleColumnPrefixFilter 和 ColumnPrefixFilter 行为差不多，但可以指定多个前缀。
<a style="font-family: 'Segoe UI', Calibri, 'Myriad Pro', Myriad, 'Trebuchet MS', Helvetica, Arial, sans-serif; font-size: 13px; line-height: 19px; white-space: normal;" href="http://hbase.apache.org/apidocs/org/apache/hadoop/hbase/filter/ColumnRangeFilter.html" target="_top">ColumnRangeFilter</a><span style="font-family: 'Segoe UI', Calibri, 'Myriad Pro', Myriad, 'Trebuchet MS', Helvetica, Arial, sans-serif; font-size: 13px; line-height: 19px; white-space: normal;"> 可以进行高效内部扫描。 </span>
-
Rowkey：对行键进行过滤。通常认为行选择时Scan采用 startRow/stopRow 方法比较好。然而 RowFilter 也可以用。
-
工具：如FirstKeyOnlyFilter用于统计行数。
二、示例
1.FirstKeyOnlyFilter,一种方便的计算行数的过滤器
hbase(main):002:0> scan 'toplist_ware_ios_1009_201231',{COLUMNS=>'info',FILTER=>"(FirstKeyOnlyFilter())"}
0000000001 column=info:loginid, timestamp=1343625459713, value=jjm168131013
0000000002 column=info:loginid, timestamp=1343625459713, value=loveswh
...
21 row(s) in 0.5480 seconds
2.列名子串进行过滤
hbase(main):006:0> scan 'toplist_ware_ios_1009_201231',{COLUMNS=>['info:'],FILTER=>"(QualifierFilter(=,'substring:id'))"}
ROW COLUMN+CELL
0000000001 column=info:loginid, timestamp=1343625459713, value=jjm168131013
0000000001 column=info:userid, timestamp=1343625459713, value=168131013
0000000002 column=info:loginid, timestamp=1343625459713, value=loveswh
0000000002 column=info:userid, timestamp=1343625459713, value=100898152
hbase(main):005:0> scan 'toplist_ware_ios_1009_201231',{COLUMNS=>['info:loginid'],FILTER=>"(QualifierFilter(=,'substring:id'))"}
ROW COLUMN+CELL
0000000001 column=info:loginid, timestamp=1343625459713, value=jjm168131013
0000000002 column=info:loginid, timestamp=1343625459713, value=loveswh
hbase(main):007:0> scan 'toplist_ware_ios_1009_201231',{COLUMNS=>['info:'],FILTER=>"(QualifierFilter(=,'substring:nid'))"}
ROW COLUMN+CELL
0000000001 column=info:loginid, timestamp=1343625459713, value=jjm168131013
0000000002 column=info:loginid, timestamp=1343625459713, value=loveswh
hbase(main):008:0> scan 'toplist_ware_ios_1009_201231',{COLUMNS=>['info:'],FILTER=>"(QualifierFilter(=,'substring:nick'))"}
ROW COLUMN+CELL
0000000001 column=info:nick, timestamp=1343625459713, value=xE5xAExB6xE6x9Cx89xE8x99x8ExE5xAEx9
D
0000000002 column=info:nick, timestamp=1343625459713, value=loveswh08
3.Value 过滤
3.1 正则过滤
hbase(main):004:0> scan 'toplist_ware_ios_1009_201231',{COLUMNS=>'info',FILTER=>"(SingleColumnValueFilter('info','nick',=,'regexstring:.*99',true,true))"}
ROW COLUMN+CELL
0000000009 column=info:loginid, timestamp=1343625459713, value=zgh1968
0000000009 column=info:nick, timestamp=1343625459713, value=zwy99
0000000009 column=info:score, timestamp=1343625459713, value=5
0000000009 column=info:userid, timestamp=1343625459713, value=100366262
1 row(s) in 0.2520 seconds
3.2 子串
需导入
import org.apache.hadoop.hbase.filter.CompareFilter
import org.apache.hadoop.hbase.filter.SingleColumnValueFilter
import org.apache.hadoop.hbase.filter.SubstringComparator
import org.apache.hadoop.hbase.util.Bytes
hbase(main):028:0> scan 'toplist_ware_ios_1001_201231',{COLUMNS =>'info:nick', FILTER=>SingleColumnValueFilter.new(Bytes.toBytes('info'),Bytes.toBytes('nick'),CompareFilter::CompareOp.valueOf('EQUAL'),SubstringComparator.new('8888'))}
ROW COLUMN+CELL
0000000002 column=info:nick, timestamp=1343625446556, value=xE7x81x8F????xE3x81x8A??8888
1 row(s) in 0.0330 seconds
3.3 二进制
子串等不支持多字节文字，所以用二进制来进行比较
hbase(main):010:0> scan 'toplist_ware_ios_1009_201231',{COLUMNS=>['info:'],FILTER=>"(QualifierFilter(=,'substring:nick') AND ValueFilter(=,'binary:7789xE6xB4x81') )"}
ROW COLUMN+CELL
0000000016 column=info:nick, timestamp=1343625459713, value=7789xE6xB4x81
1 row(s) in 0.1710 seconds
4 综合列名子串和值二进制比较
hbase(main):012:0> scan 'toplist_ware_ios_1009_201231',{COLUMNS=>['info:'],FILTER=>"(QualifierFilter(=,'substring:nick') AND ValueFilter(=,'binary:7789xE6xB4x81') )"}
ROW COLUMN+CELL
0000000016 column=info:nick, timestamp=1343625459713, value=7789xE6xB4x81
1 row(s) in 0.0120 seconds
hbase(main):014:0> scan 'toplist_ware_ios_1009_201231',{COLUMNS=>"info:",FILTER=>"(PrefixFilter('000000002')) AND (QualifierFilter(=,'substring:nick')"}
ROW COLUMN+CELL
0000000020 column=info:nick, timestamp=1343625459713, value=Denny_feng
0000000021 column=info:nick, timestamp=1343625459713, value=xE5xB0x8FxE7xBDx97xE6x95x99xE7xBBx8
31
2 row(s) in 0.0440 seconds
5. 行查询
hbase(main):005:0> get 'toplist_ware_ios_1009_201231','0000000009'
COLUMN CELL
info:loginid timestamp=1343625459713, value=zgh1968
info:nick timestamp=1343625459713, value=zwy99
info:score timestamp=1343625459713, value=5
info:userid timestamp=1343625459713, value=100366262
4 row(s) in 0.1000 seconds
hbase(main):006:0> get 'toplist_ware_ios_1009_201231','0000000009','info:nick'
COLUMN CELL
info:nick timestamp=1343625459713, value=zwy99
1 row(s) in 0.0100 seconds
hbase(main):009:0> scan 'toplist_ware_ios_1009_201231',FILTER=>"PrefixFilter('000000002')"
ROW COLUMN+CELL
0000000020 column=info:loginid, timestamp=1343625459713, value=jjm169212318
0000000020 column=info:nick, timestamp=1343625459713, value=Denny_feng
0000000020 column=info:score, timestamp=1343625459713, value=1
0000000020 column=info:userid, timestamp=1343625459713, value=169212318
0000000021 column=info:loginid, timestamp=1343625459713, value=jjm169371841
0000000021 column=info:nick, timestamp=1343625459713, value=xE5xB0x8FxE7xBDx97xE6x95x99xE7xBBx8
31
0000000021 column=info:score, timestamp=1343625459713, value=1
0000000021 column=info:userid, timestamp=1343625459713, value=169371841
2 row(s) in 0.0180 seconds
hbase(main):010:0> scan 'toplist_ware_ios_1009_201231',FILTER=>"PrefixFilter('000000002')",LIMIT=>1
ROW COLUMN+CELL
0000000020 column=info:loginid, timestamp=1343625459713, value=jjm169212318
0000000020 column=info:nick, timestamp=1343625459713, value=Denny_feng
0000000020 column=info:score, timestamp=1343625459713, value=1
0000000020 column=info:userid, timestamp=1343625459713, value=169212318
1 row(s) in 0.0170 seconds
hbase(main):011:0> scan 'toplist_ware_ios_1009_201231',{COLUMNS=>"info:nick",FILTER=>"PrefixFilter('000000002')",LIMIT=>1}
ROW COLUMN+CELL
0000000020 column=info:nick, timestamp=1343625459713, value=Denny_feng
1 row(s) in 0.0160 seconds
查询MPID和GameID同时等于某个值的记录：
hbase(main):014:0> scan 'award_1211',{FILTER=>"(PrefixFilter('2012-11-26')) AND (SingleColumnValueFilter('info','MPID',=,'regexstring:8639',true,true)) AND (SingleColumnValueFilter('info','gameID',=,'regexstring:1001',true,true))",LIMIT=>2}
如非注明转载, 均为原创. 本站遵循知识共享CC协议,转载请注明来源
