Commit Graph

2 Commits

Author SHA1 Message Date
Nathan Salmon
34353218ce IMPALA-4675: Case-insensitive matching of Parquet fields.
The query option PARQUET_FALLBACK_SCHEMA_RESOLUTION
allows matching of Parquet fields by name instead of by
index (the default).

Parquet column names are case sensitive, but Impala treats
db/table/column/field names as case-insensitive. Today,
there is no way today to select Parquet columns with mixed
casing via SQL using the name-based field resolution policy.

This patch changes the matching of Parquet fields to be
case-insensitive.

Testing:
- Modified the data files backing complextypestbl
  to contain fields with mixed casing.
- Several existing tests run against this table,
  including the test for name-based resolution.
- I confirmed that without this fix, the existing
  name-based resolution tests fail on the modified
  data files.
- I locally ran test_scanners.py and test_nested_types.py
  on exhaustive with this fix.

Change-Id: I87395f84ba29b4c3d8e41be1ea4e89e500b8a9f4
Reviewed-on: http://gerrit.cloudera.org:8080/5891
Reviewed-by: Alex Behm <alex.behm@cloudera.com>
Tested-by: Impala Public Jenkins
2017-03-03 10:20:07 +00:00
Skye Wanderman-Milne
bcc73a36da Nested types: read and materialize nested types in Parquet scanner
This patch modifies the Parquet scanner to resolve nested schemas, and
read and materialize collection types. The high-level modification is
to create a CollectionColumnReader that recursively materializes map-
and array-type slots.

This patch also adds many tests, most of which query a new table
called complextypestbl. This table contains hand-generated data that
is meant to expose edge cases in the scanner. The tests mostly test
the scanner, with a few tests of other functionality (e.g. array
serialization).

I ran a local benchmark comparing this scanner code to the original
scanner code on an expanded version of tpch_parquet.lineitem with
48009720 rows. My benchmark involved selecting different numbers of
columns with a single scanner thread, and I looked at the HDFS scan
node time in the query profiles. This code introduces a 10%-20%
regression in single-threaded scan time.

Change-Id: Id27fb728934e8346444f61752c9278d8010e5f3a
Reviewed-on: http://gerrit.cloudera.org:8080/576
Reviewed-by: Alex Behm <alex.behm@cloudera.com>
Tested-by: Internal Jenkins
2015-09-02 19:23:54 +00:00