BEFORE syntax support for presto engine.

[gupteaj]

Description

Presto issue #21971

Note - This PR has BEFORE syntax and engine side changes. For iceberg connector changes, test and doc, a new PR will be created.

Design description :
This feature will allow iceberg connector to query historical data using BEFORE syntax on a table.
Time travel version option will read bigint snapshot id value for the table. Time travel timestamp option will read
timestamp-with-time-zone value for the table.

examples

// Return table data for tab1 before snapshot id 8772871542276440693
select * from tab1 FOR SYSTEM_VERSION BEFORE 8772871542276440693;

// Return table data for tab1 before '2023-08-17 13:29:46.822 America/Los_Angeles'
select * from tab1 FOR SYSTEM_TIME BEFORE TIMESTAMP '2023-08-17 13:29:46.822 America/Los_Angeles';

Parser, Semantic Analyzer, Metadata

• support point in time query with BEFORE syntax on table reference

select * from tab1 FOR VERSION BEFORE  < bigint type >
select * from tab1 FOR TIMESTAMP BEFORE <timestamp with time zone type>

In parser tree, update TableVersionExpression class to hold time travel operator ( EQUAL/LESS_THAN) for AS OF/BEFORE state

• use table version operator in Semantic Analyzer code along with table version type and expression
• return a new error from Semantic Analyzer for BEFORE syntax
• rename AS OF code with generic table version state

Motivation and Context

Snowflake provides a BEFORE clause for time travel queries. When used in a query, this clause is specified in the FROM clause immediately after the table name. It determines the point in the past from which historical data is requested for the object. The AT keyword includes changes made by a statement or transaction with a timestamp equal to the specified parameter, while the BEFORE keyword refers to a point immediately preceding the specified parameter.

IBM Netezza supports BEFORE clause similar to Snowflake.

References -
Snowflake - https://docs.snowflake.com/en/sql-reference/constructs/at-before
IBM Netezza - https://cloud.ibm.com/docs/netezza?topic=netezza-runningqueries_tt#before_tt

Impact

A new table level option to return specific snapshot for iceberg connector

Test Plan

• Add new parser tests for BEFORE syntax
• Update TestIcebergTableVersion test for BEFORE syntax
• New iceberg PR will have time travel tests

Contributor checklist

• Please make sure your submission complies with our development, formatting, commit message, and attribution guidelines.
• PR description addresses the issue accurately and concisely. If the change is non-trivial, a GitHub Issue is referenced.
• Documented new properties (with its default value), SQL syntax, functions, or other functionality.
• If release notes are required, they follow the release notes guidelines.
• Adequate tests were added if applicable.
• CI passed.

Release Notes

Please follow release notes guidelines and fill in the release notes below.

Note - Iceberg PR will have release notes and doc section

[github-actions]

Codenotify: Notifying subscribers in CODENOTIFY files for diff 12a6e44...153bc1f.

Notify	File(s)
@aditi-pandit	presto-parser/src/main/antlr4/com/facebook/presto/sql/parser/SqlBase.g4
@elharo	presto-parser/src/main/antlr4/com/facebook/presto/sql/parser/SqlBase.g4
@kaikalur	presto-parser/src/main/antlr4/com/facebook/presto/sql/parser/SqlBase.g4
@rschlussel	presto-parser/src/main/antlr4/com/facebook/presto/sql/parser/SqlBase.g4

[ClarenceThreepwood]

Please include parser/formatter tests in TestSqlParser and TestStatementBuilder

[hantangwangd]

Overall looks good to me. Only some little nits.

[hantangwangd]

LGTM. Thanks for the fix.

[ClarenceThreepwood]

lgtm. Thanks!

[zacw7]

Hey @gupteaj , this change has broken a bunch of tests.

Stderr:
presto/presto-parser/src/main/java/com/facebook/presto/sql/parser/AstBuilder.java:1607: error: cannot find symbol
                return timestampExpression(getLocation(context), getTableVersionOperator((Token) context.tableVersionState().getChild(0).getPayload()), child);
                                                                                                        ^
  symbol:   method tableVersionState()
  location: variable context of type TableVersionContext
presto/presto-parser/src/main/java/com/facebook/presto/sql/parser/AstBuilder.java:1610: error: cannot find symbol
                return versionExpression(getLocation(context), getTableVersionOperator((Token) context.tableVersionState().getChild(0).getPayload()), child);

Do we need to upgrade the antlr version?

[tdcmeehan]

@zacw7 can you add a link to a test run?

[zacw7]

@zacw7 can you add a link to a test run?

They are meta internal tests. Let me check the build tool.

[ClarenceThreepwood]

I suspect that the Meta build tool chain needs to recompile the parser module more aggressively. This is the second such instance I have seen in recent times