[SPARK-17166] [SQL] Store Table Properties in CTAS that is Converted to Data Source Tables

[gatorsmile]

What changes were proposed in this pull request?

CTAS lost table properties after conversion to data source tables. For example,

CREATETABLEt TBLPROPERTIES('prop1'='c', 'prop2'='d') ASSELECT1as a, 1as b

The output of DESC FORMATTED t does not have the related properties.

|Table Parameters: | | |
| rawDataSize |-1 | |
| numFiles |1 | |
| transient_lastDdlTime |1471670983 | |
| totalSize |496 | |
| spark.sql.sources.provider|parquet | |
| EXTERNAL |FALSE | |
| COLUMN_STATS_ACCURATE |false | |
| numRows |-1 | |
| | | |
|# Storage Information | | |
|SerDe Library: |org.apache.hadoop.hive.ql.io.parquet.serde.ParquetHiveSerDe | |
|InputFormat: |org.apache.hadoop.hive.ql.io.parquet.MapredParquetInputFormat | |
|OutputFormat: |org.apache.hadoop.hive.ql.io.parquet.MapredParquetOutputFormat | |
|Compressed: |No | |
|Storage Desc Parameters: | | |
| serialization.format |1 | |
| path |file:/private/var/folders/4b/sgmfldk15js406vk7lw5llzw0000gn/T/warehouse-f3aa2927-6464-4a35-a715-1300dde6c614/t| |

After the fix, the properties specified by users are stored as serde properties, since the table properties are used for storing table schemas and system generated properties.

|Table Parameters: | | |
| rawDataSize |-1 | |
| numFiles |1 | |
| transient_lastDdlTime |1471672182 | |
| totalSize |496 | |
| spark.sql.sources.provider|parquet | |
| EXTERNAL |FALSE | |
| COLUMN_STATS_ACCURATE |false | |
| numRows |-1 | |
| | | |
|# Storage Information | | |
|SerDe Library: |org.apache.hadoop.hive.ql.io.parquet.serde.ParquetHiveSerDe | |
|InputFormat: |org.apache.hadoop.hive.ql.io.parquet.MapredParquetInputFormat | |
|OutputFormat: |org.apache.hadoop.hive.ql.io.parquet.MapredParquetOutputFormat | |
|Compressed: |No | |
|Storage Desc Parameters: | | |
| prop2 |d | |
| prop1 |c | |
| serialization.format |1 | |
| path |file:/private/var/folders/4b/sgmfldk15js406vk7lw5llzw0000gn/T/warehouse-78c38cea-02c9-40aa-9b20-9803686069ae/t| |
+----------------------------+--------------------------------------------------------------------------------------------------------------+-------+

How was this patch tested?

Added a test case.

[gatorsmile]

cc @cloud-fan@yhuai This is what we discussed in another PR. Could you please review whether this is a right fix? Thanks!

[SparkQA]

Test build #64125 has finished for PR 14727 at commit bffc412.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[cloud-fan]

is this what we want? Why do the table properties in Hive serde table should go to storage properties in data source table?

Ideally data source table should have data source options(storage properties) and table properties. Currently we don't support specifying table properties for data source tables, but it doesn't mean we will never do it. I think we can do it when unify the CREATE TABLE syntax.

[gatorsmile]

uh, agree! Let me close this PR. Thanks!