如何使用Flink创建一个按STRUCT/ROW类型的内部字段分区的Iceberg表？

Question

我尝试像这样创建表格：

Flink SQL> create table test (nested ROW(id STRING, name STRING)) partitioned by (nested.id);

但我收到此错误：

[ERROR] Could not execute SQL statement. Reason:
org.apache.flink.sql.parser.impl.ParseException: Encountered "." at line 1, column 78.
Was expecting one of:
    ")" ...
    "," ...

我尝试了很多选项：

partitioned by ('nested.id')

，

partitioned by (nested.id)

，但它们都失败了。文档中也没有提到这一点。

Answer 1

几乎没有这方面的文档，但我想通了：

Schema schema = new Schema(
        optional(1, "id", Types.IntegerType.get()),
        optional(2, "name", Types.StringType.get()),
        optional(3, "address", Types.StructType.of(
                optional(4, "street", Types.StringType.get()),
                optional(5, "city", Types.StringType.get()),
                optional(6, "state", Types.StringType.get()),
                optional(7, "zip", Types.IntegerType.get())
        ))
);

Table table = new TableBuilder()
        .withSchema(schema)
        .withPartitionSpec(PartitionSpec.builderFor(schema)
                .identity("id")
                .bucket("address.city", 16)
                .build())
        .build();