编辑输出（根据检索到的数量和分号设置括号）

Question

每次我运行我的代码都会以 JSON 格式获得一行、两行或几行。我正在使用 Oracle 11g.

这是我在 SQL Fiddle 上的 CODE，您将在那里找到所有数据。

例如，在这种情况下，我得到了这样的东西：

{"sku":"99342435","PRICE":"9999",PRICES_FOR_CLIENTS:[{"group":"A","PRICE":"29223"},{"group":"B","PRICE":"33223"},{"group":"SUPERMARKET","PRICE":"48343"},{"group":"WALMART","PRICE":"40340"}]};

但我可以根据存储在表中的数据获得此输出：

{"sku":"99342435","PRICE":"9999",PRICES_FOR_CLIENTS:[{"group":"A","PRICE":"29223"},{"group":"B","PRICE":"33223"},{"group":"SUPERMARKET","PRICE":"48343"},{"group":"WALMART","PRICE":"40340"}]};

{"sku":"95453343","PRICE":"8778",PRICES_FOR_CLIENTS:[{"group":"A","PRICE":"29223"},{"group":"B","PRICE":"33223"},{"group":"SUPERMARKET","PRICE":"48343"},{"group":"WALMART","PRICE":"40340"}]};

我想让我的查询能够在 第一行 中放置一个 "open bracket"，检查是否存在其他行，如果没有显示新行，在末尾插入一个 右方括号 和一个分号；像这样：

[{"sku":"99342435","PRICE":"9999",PRICES_FOR_CLIENTS:[{"group":"A","PRICE":"29223"},{"group":"B","PRICE":"33223"},{"group":"SUPERMARKET","PRICE":"48343"},{"group":"WALMART","PRICE":"40340"}]}];

但另一种情况可能是存在 多于 1 行；在那种情况下，我想在第一行放置一个 左括号 而不是右括号。我只想打开第一行的括号并在最后一行和分号内关闭它。不同的行必须由逗号分隔。请按照这个例子：

[{"sku":"99342435","PRICE":"9999",PRICES_FOR_CLIENTS:[{"group":"A","PRICE":"29223"},{"group":"B","PRICE":"33223"},{"group":"SUPERMARKET","PRICE":"48343"},{"group":"WALMART","PRICE":"40340"}]},

{"sku":"95453343","PRICE":"8778",PRICES_FOR_CLIENTS:[{"group":"A","PRICE":"29223"},{"group":"B","PRICE":"33223"},{"group":"SUPERMARKET","PRICE":"48343"},{"group":"WALMART","PRICE":"40340"}]}];

试了很多次都做不到

你能帮帮我吗？

Answer 1

这开始属于仅仅因为您可以使它起作用的领域，并不意味着您应该这样做。但要回答您的问题，请查看此代码：

SELECT CASE WHEN sub2.TOTAL_ROW > 1 AND sub2.this_row = 1 THEN '[' ELSE NULL END||
       sub2.json||
       CASE WHEN sub2.total_row > 1 AND sub2.this_row = sub2.total_row THEN ']' ELSE NULL END AS JSON
FROM (SELECT '{"sku":"'||sub.item_code||'","PRICE":"'||sub.item_price||'",PRICES_FOR_CLIENTS:['||listagg('{"group":"'||sub.identifier||'","PRICE":"'||sub.price||'"}',',') WITHIN GROUP (ORDER BY sub.identifier)||']};' AS JSON, 
      COUNT(*) OVER () AS TOTAL_ROW,
      ROW_NUMBER() OVER (ORDER BY sub.item_code, sub.item_price) AS THIS_ROW                                         
      FROM (SELECT DECODE(ppc.customer_class_code, 'E', c.description, ppc.customer_class_code) AS IDENTIFIER, tpp.item_code, replace(tpp.price, ',', '.') AS ITEM_PRICE, REPLACE(avg(ppc.price), ',', '.') AS PRICE, 
            tpl.request_id, max(tpl.request_id) over (partition by tpp.item_code) as max_request
            FROM table_price_list tpl
            INNER JOIN table_price_product tpp ON tpp.list_header_id = tpl.list_header_id AND tpp.request_id = tpl.request_id
            INNER JOIN prices_per_client ppc ON tpp.item_code = ppc.item_code
            LEFT JOIN clients c ON ppc.customer_number = c.account_number
            WHERE SYSDATE BETWEEN NVL(tpp.start_date_active, SYSDATE) AND NVL(tpp.end_date_active, SYSDATE+1)
            GROUP BY DECODE(ppc.customer_class_code, 'E', c.description, ppc.customer_class_code), tpp.item_code, tpp.price, tpl.request_id) sub 
      WHERE sub.identifier IS NOT NULL
      and sub.request_id = sub.max_request
      GROUP BY sub.item_code, sub.item_price) sub2
ORDER BY sub2.this_row;

我更新了你的SQLFiddle

这个查询所做的是在前一个查询的基础上进行扩展。它使用分析函数 COUNT 和 ROW_NUMBER 计算总行数和这一行数。我根本没有对它们进行分区，因为我们希望它们考虑所有返回的行。如果总行数 > 1 并且这是第一行，则以“[”开头。在中间附加 JSON。如果这是最后一行（总计 > 1 且 ROW_NUMBER = COUNT），则将“]”附加到末尾。

编辑 1：更改为始终有括号

SELECT DECODE(sub2.this_row, 1, '[', NULL)||
       sub2.json||
       DECODE(sub2.this_row, sub2.total_row, ']', NULL) AS JSON
FROM (SELECT '{"sku":"'||sub.item_code||'","PRICE":"'||sub.item_price||'",PRICES_FOR_CLIENTS:['||listagg('{"group":"'||sub.identifier||'","PRICE":"'||sub.price||'"}',',') WITHIN GROUP (ORDER BY sub.identifier)||']};' AS JSON, 
      COUNT(*) OVER () AS TOTAL_ROW,
      ROW_NUMBER() OVER (ORDER BY sub.item_code, sub.item_price) AS THIS_ROW                                         
      FROM (SELECT DECODE(ppc.customer_class_code, 'E', c.description, ppc.customer_class_code) AS IDENTIFIER, tpp.item_code, replace(tpp.price, ',', '.') AS ITEM_PRICE, REPLACE(avg(ppc.price), ',', '.') AS PRICE, 
            tpl.request_id, max(tpl.request_id) over (partition by tpp.item_code) as max_request
            FROM table_price_list tpl
            INNER JOIN table_price_product tpp ON tpp.list_header_id = tpl.list_header_id AND tpp.request_id = tpl.request_id
            INNER JOIN prices_per_client ppc ON tpp.item_code = ppc.item_code
            LEFT JOIN clients c ON ppc.customer_number = c.account_number
            WHERE SYSDATE BETWEEN NVL(tpp.start_date_active, SYSDATE) AND NVL(tpp.end_date_active, SYSDATE+1)
            GROUP BY DECODE(ppc.customer_class_code, 'E', c.description, ppc.customer_class_code), tpp.item_code, tpp.price, tpl.request_id) sub 
      WHERE sub.identifier IS NOT NULL
      and sub.request_id = sub.max_request
      GROUP BY sub.item_code, sub.item_price) sub2
ORDER BY sub2.this_row;

这是更新后的 SQLFiddle (Link)。

鉴于更简单的逻辑，我将 CASE 换成了 DECODE。如果该行是第一行，它会得到左括号。如果它是最后一行，它会得到右括号。如果两者都是，则两者都得到。

Answer 2

处理此问题的最简单方法是对现有查询的结果执行另一个 LISTAGG，并将其包含在 [ 和 ]:

中

SELECT '[' || LISTAGG(JSON, ',') WITHIN GROUP(ORDER BY NULL) || ']' AS JSON
FROM (
  SELECT '{"sku":"'||sub.item_code||'","PRICE":"'||sub.item_price||'","PRICES_FOR_CLIENTS":['||listagg('{"group":"'||sub.identifier||'","PRICE":"'||sub.price||'"}',',') WITHIN GROUP (ORDER BY sub.identifier)||']}' AS JSON                                              
  FROM (SELECT DECODE(ppc.customer_class_code, 'E', c.description, ppc.customer_class_code) AS IDENTIFIER, tpp.item_code, replace(tpp.price, ',', '.') AS ITEM_PRICE, REPLACE(avg(ppc.price), ',', '.') AS PRICE, 
        tpl.request_id, max(tpl.request_id) over (partition by tpp.item_code) as max_request
        FROM table_price_list tpl
        INNER JOIN table_price_product tpp ON tpp.list_header_id = tpl.list_header_id AND tpp.request_id = tpl.request_id
        INNER JOIN prices_per_client ppc ON tpp.item_code = ppc.item_code
        LEFT JOIN clients c ON ppc.customer_number = c.account_number
        WHERE SYSDATE BETWEEN NVL(tpp.start_date_active, SYSDATE) AND NVL(tpp.end_date_active, SYSDATE+1)
        GROUP BY DECODE(ppc.customer_class_code, 'E', c.description, ppc.customer_class_code), tpp.item_code, tpp.price, tpl.request_id) sub 
  WHERE sub.identifier IS NOT NULL
  and sub.request_id = sub.max_request
  GROUP BY sub.item_code, sub.item_price
) j

Demo on SQLFiddle

注意需要用双引号将PRICES_FOR_CLIENTS括起来，否则查询的输出是无效的JSON。您还需要从原始查询输出的字符串末尾删除 ;。

编辑输出（根据检索到的数量和分号设置括号）

Editing the output (set brackets depending on the quantity retrieved and the semicolon)

sql

oracle11g