We can combine this with LowCardinality DataType above, e.g.

Adding support for LowCardinality and Tuples DataType in ClickHouse (Java) ([#29533](https://github.com/apache/beam/pull/29533), [#29715](https://github.com/apache/beam/pull/29715)).

Release cut is today so this may not get into 2.53.0. Could leave it as is and fix when the PR is finalized

Consider put the documentation Tuple to the table above. Broken sentence here currently.

Please check this sentence

The caller has a condition if (type.toLowerCase().trim().startsWith("tuple(")) { below, so here payload always (case insensitive) starts with "tuple(", so here the replace of "Tuple\(" won't work

The condition is on with lower case, but the actual parsing is on the original string

I mean below when " if (type.toLowerCase().trim().startsWith("tuple(")) {" the this preprocessing will be executed.

So Tuple( (with backslash) is considered here but not the condition in the caller there. Coild this cause issue?

You're right. Would you mind adding some comments about the proprocessing rule and example, for example, sth like // Tuple(a String, b Integer) -> ...

Is COLON and SEMI_COLON for tuple or another feature?

Using that since i have some ambiguity in parsing (looking for other options), but for now decided to leave it that way

If it's get added to SDK then there is expectation of backward compatibility. So I would prefer to keep consistent with ClickHouse's SQL syntax

It's a temporary solution since there is some ambiguity when using javacc having difficulty to detecting between tokens and strings

Removed COLON and SEMI_ COLON

Is :; a standard syntax for ClickHouse? https://clickhouse.com/docs/en/sql-reference/data-types/tuple seems not mentioning this pattern.

@Abacn Removed :;

LowCardinality will go to 2.53.0 but Tuples does not, as 2.53.0 is already cut

Please check this sentence

I mean below when " if (type.toLowerCase().trim().startsWith("tuple(")) {" the this preprocessing will be executed.

So Tuple( (with backslash) is considered here but not the condition in the caller there. Coild this cause issue?

Hi @mzitnik,
Shouldn't each key inside a tuple be mapped to their respective data type ? Any reason for being DATETIME?

Hi,

This quick adaptation seems to work:

case TUPLE:
     List<TableSchema.Column> columns = columnType.tupleTypes().entrySet().stream().map(
                              x -> TableSchema.Column.of(x.getKey(), x.getValue())
                         ).collect(Collectors.toList());
     TableSchema tupleSchema = TableSchema.of(columns.toArray(new TableSchema.Column[0]));
     Schema tupleAsRowSchema = getEquivalentSchema(tupleSchema);
     return Schema.FieldType.row(tupleAsRowSchema);

Hi,

Also, it seems that Array(Tuple(*)) is not supported when calling ClickHouseIO.getTableSchema.

This seems to fix the issue :
if (type.toLowerCase().trim().startsWith("tuple(")) { String content = tuplePreprocessing(type); columnType = TableSchema.ColumnType.parse(content); } else if (type.toLowerCase().trim().startsWith("array(tuple(")) { String content = tuplePreprocessing(type); columnType = TableSchema.ColumnType.parse(content); }

@erube and @wattache, thanks for the feedback i will fork and also add some tests to cover Array(Tuple(*))