How to apply several QualifierFilter to a row in HBase

问题

we would like to filter a scan on a HBase table with two QualifierFilters. Means we would like to only get the rows of the table which do have a certain column 'col_A' AND (!) a certain other column 'col_B'.

Our current approach looks like this:

FilterList filterList = new FilterList(FilterList.Operator.MUST_PASS_ALL);
Filter filter1 = new QualifierFilter(CompareOp.EQUAL, new BinaryComparator("col_A".getBytes()));
filterList.addFilter(filter1);
Filter filter2 = new QualifierFilter(CompareOp.EQUAL, new BinaryComparator("col_B".getBytes()));
filterList.addFilter(filter2);

Scan scan = new Scan();
scan.setFilter(filterList);
...

The ResultScanner does not return any results from this scan although there are several rows in the HBase table which do have both columns 'col_A' and 'col_B'.

If we only apply filter1 to the scan everything works fine an we do get all the rows which have 'col_A'. If we only apply filter2 to the scan it is the same. We do get all rows which have 'col_B'.

Only if we combine these two filters we do not get any results.

What would be the right way to get only the rows from the table which do have col_A AND col_B?

回答1:

You can achieve this by defining the following filters:

List<Filter> filters = new ArrayList<Filter>(2);
byte[] colfam = Bytes.toBytes("c");
byte[] fakeValue = Bytes.toBytes("DOESNOTEXIST");
byte[] colA = Bytes.toBytes("col_A");
byte[] colB = Bytes.toBytes("col_B");

SingleColumnValueFilter filter1 = 
    new SingleColumnValueFilter(colfam, colA , CompareOp.NOT_EQUAL, fakeValue);  
filter1.setFilterIfMissing(true);
filters.add(filter1);

SingleColumnValueFilter filter2 = 
    new SingleColumnValueFilter(colfam, colB, CompareOp.NOT_EQUAL, fakeValue);          
filter2.setFilterIfMissing(true);
filters.add(filter2);

FilterList filterList = new FilterList(FilterList.Operator.MUST_PASS_ALL, filters);
Scan scan = new Scan();
scan.setFilter(filterList);

The idea here is to define one SingleColumnValueFilter per column you are looking for, each with a fake value and a CompareOp.NOT_EQUAL operator. I.e: such a SingleColumnValueFilter will return all columns for a given name.

Source: http://mapredit.blogspot.com/2012/05/using-filters-in-hbase-to-match-two.html

回答2:

I think this line is the issue -

FilterList filterList = new FilterList(FilterList.Operator.MUST_PASS_ALL);

You want it to be -

FilterList filterList = new FilterList(FilterList.Operator.MUST_PASS_ONE);

The filter will try to find a column that has both the column qualifier and there is no such column

来源：https://stackoverflow.com/questions/13379350/how-to-apply-several-qualifierfilter-to-a-row-in-hbase

标签

filter

hbase

database-scan