fly on exercises

data/3.PostgreSQL高阶/3.SQL高级技巧/1.递归查询/config.json

@@ -2,5 +2,5 @@
   "node_id": "pg-2e691ed3a847424eb887b40aca750c4e",
   "keywords": [],
   "children": [],
-  "export": []
+  "export": ["continuous"]
 }
\ No newline at end of file

data/3.PostgreSQL高阶/3.SQL高级技巧/1.递归查询/continuous.json

+{
+  "type": "code_options",
+  "author": "刘鑫",
+  "source": "continuous.md",
+  "notebook_enable": false
+}
\ No newline at end of file

data/3.PostgreSQL高阶/3.SQL高级技巧/1.递归查询/continuous.md

+# 获取连续区间
+
+SmartMarket 交易所的系统中，所有订单在生成时，都从一个 PostgreSQL Sequence 中获取唯一的 序列号，完成交易计算后各个撮合程序将其插入如下的 orders 表中：
+
+```postgresql
+create table orders
+(
+    id      integer primary key,
+    meta    jsonb default '{}'::jsonb,
+    content jsonb default '{}'::jsonb
+    -- ignore other definitions...
+);
+```
+
+后续的结算系统需要连续的获取订单，以便处理一些顺序敏感的业务。但是撮合发生在很多个异步节点上，它们只能保证最终会将 所有订单都保存到 orders 表，确保 id 列最终是连续的，但是最新插入的一段记录集有可能不连续。而我们希望结算系统成批
+的读取数据，以优化性能，那么在结算系统有它最后处理的订单id的前提下，下面哪一个查询可以确保从这个id向前读取 id 连续 的一批订单？
+
+## 答案
+
+```postgresql
+with recursive r(id) as (select id
+                         from orders
+                         where id = $1
+                         union
+                         select d.id
+                         from orders as d
+                                  join r on d.id = r.id + 1)
+select orders.id, meta, content
+from orders
+         join r on orders.id = r.id;
+```
+
+## 选项
+
+### A
+
+没有办法构造一个简单查询实现这个功能，因为它的数据过滤条件递归的依赖查询结果。
+
+### B
+
+```postgresql
+select data.id, meta, content
+from orders
+         join orders as r on orders.id = r.id - 1
+where id = $1;
+```
+
+### C
+
+```postgresql
+select id, meta, content
+from orders
+where id in (select id from orders where id = id + 1);
+```
+
+### D
+
+```postgresql
+with r as (select id
+           from orders
+           where id = $1
+           union
+           select d.id
+           from orders as d
+                    join r on d.id = r.id + 1)
+select orders.id, meta, content
+from orders
+         join r on orders.id = r.id;
+```
\ No newline at end of file