Clickhouse 计算自动增量

Question

存在带有整数标识符列的旧表。

尝试重现自动增量：

INSERT legacy_table (id, name)
SELECT
    (SELECT max(id) FROM legacy_table) + rowNumberInAllBlocks() id,
    name
FROM other_table

问题是SELECT max(id) FROM legacy_table 只计算一次，会产生重复。

是否可以每行执行SELECT max(id) 或其他自动增量解决方案？

Answer 1

我会依靠数组函数来使行为可预测。

备注：

• 不应将任何外部插入到目标表和源表中

• 一次应该只运行一个脚本（所以不要将它应用于分布式表，而是在每个分片上顺序执行）

• 对于大表，嵌套查询可能会失败，所以添加 WHERE 子句来限制处理数据的计数是有意义的

INSERT INTO target
SELECT result.1 AS id, result.2 AS name
FROM (
  WITH (SELECT max(id) FROM target) AS target_table_max_id
  SELECT 
    groupArray((id, name)) origin_data,
    arrayMap((origin_row, row_index) -> (target_table_max_id + row_index, origin_row.2), 
      origin_data, 
      arrayEnumerate(origin_data)) origin_data_updated,
    arrayJoin(origin_data_updated) result
  FROM origin);

准备测试环境的脚本：

# prepare test tables
CREATE TABLE target (`id` Int32, `name` String) ENGINE = Memory;
CREATE TABLE origin (`id` Int32, `name` String) ENGINE = Memory; 

# fill test tables 
INSERT INTO target 
SELECT  number AS id, toString(id) AS name
FROM numbers(8);

INSERT INTO origin 
SELECT number + 2048 AS id, toString(id) AS name
FROM numbers(4);