[Pig-dev] [jira] Created: (PIG-1703) Pig get confused if map value is not bytearray

Pig get confused if map value is not bytearray
----------------------------------------------

Key: PIG-1703
URL: https://issues.apache.org/jira/browse/PIG-1703
Project: Pig
Issue Type: Sub-task
Affects Versions: 0.8.0
Reporter: Daniel Dai
Assignee: Daniel Dai
Fix For: 0.9.0


It is the same nature of [PIG-999|https://issues.apache.org/jira/browse/PIG-999]. Just adding another test case:
{code}
a = load ':INPATH:/singlefile/studenttab10k' as (name: chararray, age: int, gpa: float);
sds = load ':INPATH:/somefile' using SomeLoader() as (s:map[], m:map[],
l:map[]);
views = FOREACH sds GENERATE s#'srcpvid' as srcpvid, flatten(l#'viewinfo') as viewinfo;
views1 = FILTER views BY srcpvid == '1234';
views2 = FILTER views1 BY (viewinfo#'it' EQ '25');
map_scalar = limit views2 1;
z = foreach a generate name, age+(double)map_scalar.viewinfo#'it' as some_sum;
store z into ':OUTPATH:.2';
{code}
Here l is a map of bags of maps. flatten(l#'viewinfo') suppose to get maps. However, internally Pig track all map key as bytearray. In the scalar case, ReadScalar will give bytearray as outputschema, but it is actually a map. What Pig does is to Stringize map, and then convert string back into map, which end up with nulls.

Pig get confused if map value is not bytearray
----------------------------------------------

Key: PIG-1703
URL: https://issues.apache.org/jira/browse/PIG-1703
Project: Pig
Issue Type: Sub-task
Affects Versions: 0.8.0
Reporter: Daniel Dai
Assignee: Daniel Dai
Fix For: 0.9.0


It is the same nature of [PIG-999|https://issues.apache.org/jira/browse/PIG-999]. Just adding another test case:
{code}
a = load ':INPATH:/singlefile/studenttab10k' as (name: chararray, age: int, gpa: float);
sds = load ':INPATH:/somefile' using SomeLoader() as (s:map[], m:map[],
l:map[]);
views = FOREACH sds GENERATE s#'srcpvid' as srcpvid, flatten(l#'viewinfo') as viewinfo;
views1 = FILTER views BY srcpvid == '1234';
views2 = FILTER views1 BY (viewinfo#'it' EQ '25');
map_scalar = limit views2 1;
z = foreach a generate name, age+(double)map_scalar.viewinfo#'it' as some_sum;
store z into ':OUTPATH:.2';
{code}
Here l is a map of bags of maps. flatten(l#'viewinfo') suppose to get maps. However, internally Pig track all map key as bytearray. In the scalar case, ReadScalar will give bytearray as outputschema, but it is actually a map. What Pig does is to Stringize map, and then convert string back into map, which end up with nulls.