如何通过蜂巢中的collect_set（）操作使用订单

Question

在表1中，我有customer_id，item_id和item_rank（根据某些销售情况的项目等级）。我想为每个customer_id收集项目列表，并根据item_rank进行排列。

Customer_id  item_id rank_item
  23            2      3
  23            2      3
  23            4      2
  25            5      1
  25            4      2

我期望的输出是

Customer_id    item_list
  23             4,2
  25             5,4

我使用的代码是

 SELECT
    customer_id,
    concat_ws(',',collect_list (string(item_id))) AS item_list
FROM
    table1
GROUP BY
    customer_id
ORDER BY
    item_rank

Answer 1

您可以使用子查询来获取结果集（customer_id，item_id，item_rank），并按item_rank对其进行排序，然后collect_set在外部查询中使用。

询问

WITH table1 AS (
    SELECT 23 AS customer_id, 2 AS item_id, 3 AS item_rank UNION ALL
    SELECT 23 AS customer_id, 2 AS item_id, 3 AS item_rank UNION ALL
    SELECT 23 AS customer_id, 4 AS item_id, 2 AS item_rank UNION ALL
    SELECT 25 AS customer_id, 5 AS item_id, 1 AS item_rank UNION ALL
    SELECT 25 AS customer_id, 4 AS item_id, 2 AS item_rank
)
SELECT
    subquery.customer_id,
    collect_set(subquery.item_id) AS item_id_set
FROM (
    SELECT
        table1.customer_id,
        table1.item_id,
        table1.item_rank
    FROM table1
    DISTRIBUTE BY
        table1.customer_id
    SORT BY
        table1.customer_id,
        table1.item_rank
) subquery
GROUP BY
    subquery.customer_id
;

结果

    customer_id item_id_set
0   23  [4,2]
1   25  [5,4]

子查询用于DISTRIBUTE BY确保特定customer_id路由的所有行均到达同一reducer。然后，它使用SORT BY的排序customer_id和item_rank每个减速器内。我希望这足以满足要求，因为我没有注意到对最终结果集进行整体排序的要求。（如果需要按总顺序排序customer_id，那么我认为查询将必须使用ORDER BY，这将导致执行速度变慢。）

在内部，collect_setUDAF使用Java LinkedHashSet，它是一个保留顺序的集合，因此子查询中使用的排序顺序将在外部查询的集合中维护。这在Hive代码库中可见：

https://github.com/apache/hive/blob/release-2.0.0/ql/src/java/org/apache/hadoop/hive/ql/udf/generic/GenericUDAFMkCollectionEvaluator.java#L93