Parse MySQL and PostgreSQL CREATE and ALTER TABLESPACE

[minleejae]

Purpose

Parse MySQL and PostgreSQL CREATE TABLESPACE and ALTER TABLESPACE. Today CREATE TABLESPACE fails in every configuration, for example CREATE TABLESPACE ts1 ADD DATAFILE 'ts1.ibd' ENGINE=InnoDB and CREATE TABLESPACE fast OWNER u1 LOCATION '/ssd1/pg'. ALTER TABLESPACE fast RENAME TO faster parses only as an UnsupportedStatement token list. Tools that read schema scripts therefore cannot see a tablespace's name, data file, location or options.

Current design (As-is)

• Create() has no tablespace alternative. CREATE TABLESPACE reaches CreateIndex() and fails there, at ADD (MySQL) or at the end of the statement (PostgreSQL).
• Alter() captures every ALTER TABLESPACE statement as an UnsupportedStatement. Only the PostgreSQL bulk move ALTER TABLE|INDEX|MATERIALIZED VIEW ALL IN TABLESPACE … SET TABLESPACE … is modeled, as AlterTablespaceMove.
• Option models for these clauses already exist:
  • MySQL ENGINE, ENCRYPTION, AUTOEXTEND_SIZE and ENGINE_ATTRIBUTE are TableOption kinds, parsed by the shared CREATE/ALTER TABLE option productions.
  • PostgreSQL WITH (…), SET (…) and RESET (…) lists are IndexOption lists or names, as in CREATE TABLE … WITH and ALTER INDEX … SET/RESET.

Target design (To-be)

• Add CreateTablespace. It records:
  • the name, and the MySQL UNDO keyword;
  • PostgreSQL: the owner, the LOCATION (StringValue) and the WITH parameters (List<IndexOption>);
  • MySQL: the ADD DATAFILE file (StringValue) and the options (List<TableOption>) in source order.
• Add AlterTablespace, with an explicit Action:
  • RENAME (RENAME TO);
  • PostgreSQL OWNER (OWNER TO), SET_PARAMETERS (SET (…)) and RESET_PARAMETERS (RESET (…));
  • MySQL SET_STATE (ALTER UNDO TABLESPACE … SET ACTIVE|INACTIVE) and OPTIONS.
• Reuse the existing option models instead of adding parallel ones:
  • MySQL options go through the existing table option productions and become TableOption values. The only new kind is TableOption.Kind.FILE_BLOCK_SIZE, which only tablespaces use.
  • PostgreSQL parameters reuse PostgreSqlIndexOptions(); RESET keeps names, as ALTER INDEX … RESET does.
• Accept only what the server grammars define (MySQL 8.4 sql_yacc.yy, PostgreSQL 18 gram.y):
  • CREATE UNDO TABLESPACE requires ADD DATAFILE, and undo tablespaces accept only ENGINE. ALTER rejects FILE_BLOCK_SIZE.
  • MySQL options may be separated by commas; the commas are not kept.
  • PostgreSQL requires LOCATION. A statement cannot mix PostgreSQL and MySQL clauses.
  • The clause words (UNDO, DATAFILE, LOCATION, OWNER, INACTIVE, FILE_BLOCK_SIZE) are matched in context. No new keywords are reserved.
• Choose the statement as follows:
  • CREATE [UNDO] TABLESPACE always parses to CreateTablespace. It failed before, so invalid forms still raise a parse error.
  • ALTER [UNDO] TABLESPACE becomes AlterTablespace only when its grammar matches the whole statement. A known option that the clause does not define, such as FILE_BLOCK_SIZE after ALTER, raises a parse error.
  • Other ALTER TABLESPACE forms still parse to UnsupportedStatement, as before. These include Oracle ADD DATAFILE … SIZE, OFFLINE and ENCRYPTION ONLINE …, and NDB ADD DATAFILE and WAIT.
  • AlterTablespaceMove is unchanged.
• Out of scope:
  • the NDB-only clauses (USE LOGFILE GROUP, INITIAL_SIZE, MAX_SIZE, EXTENT_SIZE, NODEGROUP, WAIT, COMMENT) and the STORAGE ENGINE spelling;
  • DROP TABLESPACE.
• Wire both statements into:
  • the StatementVisitor defaults, StatementVisitorAdapter, StatementDeParser and TablesNamesFinder (no tables; parameter values are visited);
  • validation, through the new Feature.createTablespace and Feature.alterTablespace. The MySQL 8.0 and PostgreSQL capabilities and FeaturesAllowed.CREATE/ALTER enable them;
  • statement feature analysis: MODIFIES_SCHEMA.
• Document the API in the usage guide and list the statements in README and index.

Validation

• 46 new tests cover:
  • every MySQL and PostgreSQL form: parse, deparse and reparse with the dialect, plus the AST fields;
  • option order, optional commas, and the equality of tablespace and CREATE TABLE option tokens;
  • rejected forms, including UNDO restrictions, FILE_BLOCK_SIZE after ALTER, mixed PostgreSQL/MySQL clauses and PostgreSQL values without =;
  • Oracle and NDB ALTER TABLESPACE forms that stay UnsupportedStatement, also inside scripts;
  • unchanged AlterTablespaceMove, and clause words used as identifiers or tablespace names;
  • builder deparsing, expression visiting, script boundaries, statement features and validation capabilities.
• Java 17 offline Gradle spotlessApply check: 10,412 tests, 10,387 passed, 25 skipped, zero failures/errors. The grammar choice conflict check passes.
• Compared with master (bdf8ac31) on a corpus of 635 MySQL 8.4 and PostgreSQL 18 DDL/DCL statements, each accepted by the server, plus variants:
  • The six CREATE/ALTER TABLESPACE statements (two MySQL, four PostgreSQL) now parse to the new statements; four of them also without a dialect.
  • No case regressed, and no other case changed.
• MySQL 8.4 and PostgreSQL 18.3: every corpus statement this branch parses ran as written on one server and as rendered by toString() on a second, identical server, in the same order. All 207 MySQL and 329 PostgreSQL statements give the same result (success, or the same error).

Integration validation

Applied #2806–#2819 together on master (bdf8ac31), with the stacked PRs on top of their bases (#2806 → #2807 → #2808, #2806 → #2809 → #2810 and #2811 → #2812).

• Each branch merges cleanly into master. Between the branches, conflicts are limited to adjacent additions (README/index rows, usage sections, imports, visitor and feature lists, one enum line and one alias guard).
• The integrated build passes Java 17 offline Gradle spotlessApply check: 11,494 tests, 11,469 passed, 25 skipped, zero failures/errors. The grammar choice conflict check passes.
• On a corpus of 635 MySQL 8.4 and PostgreSQL 18 DDL/DCL statements, each accepted by the server, the integrated build parses 265 of 271 MySQL statements (master: 206) and 363 of 364 PostgreSQL statements (master: 328). Every MySQL 8.4 non-reserved word and PostgreSQL 18 unreserved or column-name keyword parses as a column and table name.
• On paired MySQL 8.4 and PostgreSQL 18.3 servers, all 263 MySQL and 362 PostgreSQL statements rendered by the integrated build give the same results as the originals.