spark-instrumented-optimizer/sql/core
gatorsmile 9dd5399d78 [SPARK-12723][SQL] Comprehensive Verification and Fixing of SQL Generation Support for Expressions
#### What changes were proposed in this pull request?

Ensure that all built-in expressions can be mapped to its SQL representation if there is one (e.g. ScalaUDF doesn't have a SQL representation). The function lists are from the expression list in `FunctionRegistry`.

window functions, grouping sets functions (`cube`, `rollup`, `grouping`, `grouping_id`), generator functions (`explode` and `json_tuple`) are covered by separate JIRA and PRs. Thus, this PR does not cover them. Except these functions, all the built-in expressions are covered. For details, see the list in `ExpressionToSQLSuite`.

Fixed a few issues. For example, the `prettyName` of `approx_count_distinct` is not right. The `sql` of `hash` function is not right, since the `hash` function does not accept `seed`.

Additionally, also correct the order of expressions in `FunctionRegistry` so that people are easier to find which functions are missing.

cc liancheng

#### How was the this patch tested?
Added two test cases in LogicalPlanToSQLSuite for covering `not like` and `not in`.

Added a new test suite `ExpressionToSQLSuite` to cover the functions:

1. misc non-aggregate functions + complex type creators + null expressions
2. math functions
3. aggregate functions
4. string functions
5. date time functions + calendar interval
6. collection functions
7. misc functions

Author: gatorsmile <gatorsmile@gmail.com>

Closes #11314 from gatorsmile/expressionToSQL.
2016-02-22 22:17:56 -08:00
..
src [SPARK-12723][SQL] Comprehensive Verification and Fixing of SQL Generation Support for Expressions 2016-02-22 22:17:56 -08:00
pom.xml [SPARK-6363][BUILD] Make Scala 2.11 the default Scala version 2016-01-30 00:20:28 -08:00