Add support for TABLE command in Postgresql by yanghykevin · Pull Request #2364 · apache/datafusion-sqlparser-rs

yanghykevin · 2026-06-04T20:07:04Z

Hi all. We're an AWS team using sqlparser-rs for MySQL/PG parsing and plan to contribute a series of dialect compatibility improvements. Happy to coordinate on approach.

Change Description:

This closes Issue: #742

Renamed Table struct → ExplicitTable. The original Table struct was created for only one specific use case: CREATE TABLE x AS TABLE y.
The new name matches is more precise and matches SQL:2016 terminology.

Grammar: https://www.postgresql.org/docs/current/sql-select.html#SQL-TABLE

c2main · 2026-06-06T11:51:16Z

+    );
+
+    for sql in [
+        "TABLE films WHERE x > 0",


might be interesting to test: TABLE db.schema.table is rejected.

Currently it is accepted. Based on the existing design philosophy I thought the parser should be permissive on this and relies on downstream semantic analyzer to do verification. But I want to get your opinion on this and will be happy to make the change rejecting this with a trait method

so this is fixed by your second commit 👍

c2main · 2026-06-06T11:53:41Z

+    dialects.verified_stmt("TABLE films");
+    // Schema-qualified.
+    dialects.verified_stmt("TABLE myschema.films");
+    // Database-qualified (three-part name).


this is for MySQL, not PostgreSQL, right ? Maybe a subtle overlap with another PR to come for MySQL ?

Yes, actually I will enable this command and add mysql testing in the next revision as well since this implementation works for both

c2main · 2026-06-06T12:03:09Z

Hi all. We're an AWS team using sqlparser-rs for MySQL/PG parsing and plan to contribute a series of dialect compatibility improvements. Happy to coordinate on approach.

Hi. FYI, we have similar plan at Data Bene for Oracle/PG.

…SQL. MySQL and PostgreSQL limit to 2 (schema.table); the Generic and ANSI dialects remain permissive. Also enables supports_table_command() for MySQL with tests.

docteurklein · 2026-06-24T06:17:00Z

Eager to see this landing as I was about to fix it myself. Thanks!

yanghykevin · 2026-07-21T17:54:56Z

Hi @iffyio 👋 — this PR has a community approval and is passing, but the CI workflow needs maintainer approval to run (and it needs a committer review to merge). Could you take a look when you have a chance? Thanks!

iffyio · 2026-07-26T07:38:23Z

+    /// Returns true if the dialect supports the `TABLE` command
+    /// (SQL:2016 `<explicit table>`). See [`ExplicitTable`].
+    fn supports_table_command(&self) -> bool {
+        false
+    }
+
+    /// Returns true if the dialect supports Postgres inheritance modifiers
+    /// (`ONLY` prefix and trailing `*`) on the `TABLE` command.
+    /// See [`InheritanceModifier`].
+    fn supports_explicit_table_inheritance_modifiers(&self) -> bool {
+        false
+    }
+
+    /// Returns the maximum number of dot-separated parts allowed in a
+    /// table name for the `TABLE` command. For example, `2` means only
+    /// `schema.table` is accepted; `3` would allow `catalog.schema.table`.
+    ///
+    /// Returns `None` if the dialect does not restrict the number of parts.
+    fn table_command_max_name_parts(&self) -> Option<usize> {
+        None


I think we should be able to drop these dialect methods, the latter two are a bit too specific I think, and in general since we're introducing an entirely new statement it should be fine to have the parser always accept the statement without it conflicting with other dialects

iffyio · 2026-07-26T07:40:13Z

-    pub table_name: Option<String>,
-    /// Optional schema/catalog name qualifying the table.
-    pub schema_name: Option<String>,
+pub struct ExplicitTable {


hmm not sure if there's a need to rename Table (its not clear to me what we gain by doing so)?

iffyio · 2026-07-26T07:41:48Z

+#[derive(Debug, Copy, Clone, PartialEq, PartialOrd, Eq, Ord, Hash)]
+#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
+#[cfg_attr(feature = "visitor", derive(Visit, VisitMut))]
+pub enum InheritanceModifier {


Suggested change

pub enum InheritanceModifier {

pub enum TableInheritanceModifier {

iffyio · 2026-07-26T07:43:12Z

+    /// The (possibly schema-qualified) table name.
+    pub name: ObjectName,
+    /// Postgres inheritance modifier (`ONLY` or trailing `*`), if present.
+    pub inheritance: InheritanceModifier,


this should be an Option instead of introducing a custom None variant in the enum

iffyio · 2026-07-26T07:44:06Z

+pub struct ExplicitTable {
+    /// The (possibly schema-qualified) table name.
+    pub name: ObjectName,
+    /// Postgres inheritance modifier (`ONLY` or trailing `*`), if present.


Suggested change

/// Postgres inheritance modifier (`ONLY` or trailing `*`), if present.

/// Inheritance modifier.

the rest of the comment is duplicated, the pg part we can drop since other dialects can support it either now or in the future

iffyio · 2026-07-26T07:45:18Z

+/// Postgres inheritance-hierarchy modifier for table references.
+///
+/// Controls whether a query against a table includes rows from descendant
+/// tables (inheritance children or partitions). See
+/// <https://www.postgresql.org/docs/current/ddl-inherit.html>.


Suggested change

/// Postgres inheritance-hierarchy modifier for table references.

///

/// Controls whether a query against a table includes rows from descendant

/// tables (inheritance children or partitions). See

/// <https://www.postgresql.org/docs/current/ddl-inherit.html>.

/// Inheritance-hierarchy modifier for table references.

///

/// Controls whether a query against a table includes rows from descendant

/// tables (inheritance children or partitions).

/// [Postgres]: https://www.postgresql.org/docs/current/ddl-inherit.html.

iffyio · 2026-07-26T07:46:32Z

                }
+                Keyword::TABLE if self.dialect.supports_table_command() => {
+                    self.prev_token();
+                    self.parse_query().map(Into::into)


which statement type is being returned?

Add support for TABLE command in Postgresql

7cae107

c2main reviewed Jun 6, 2026

View reviewed changes

Reject catalog-qualified names in TABLE command for MySQL and Postgre…

e499266

…SQL. MySQL and PostgreSQL limit to 2 (schema.table); the Generic and ANSI dialects remain permissive. Also enables supports_table_command() for MySQL with tests.

docteurklein approved these changes Jun 24, 2026

View reviewed changes

hongyi yang and others added 2 commits June 29, 2026 11:04

Merge remote-tracking branch 'origin/main' into table-command

db1fe42

Merge branch 'main' into table-command

faae354

iffyio reviewed Jul 26, 2026

View reviewed changes

	pub enum InheritanceModifier {
	pub enum TableInheritanceModifier {

	/// Postgres inheritance modifier (`ONLY` or trailing `*`), if present.
	/// Inheritance modifier.

Uh oh!

Conversation

yanghykevin commented Jun 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Change Description:

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

c2main commented Jun 6, 2026

Uh oh!

docteurklein commented Jun 24, 2026

Uh oh!

yanghykevin commented Jul 21, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

yanghykevin commented Jun 4, 2026 •

edited

Loading