針對這個問題，我先建張表，依據這個表再說明解決方案

表：

create table customer_table
(
id serial not null,
customer_id varchar(32),
identity_idvarchar(32),
customer_namevarchar(128),
)

comment on table customer_table is '客戶資訊表';
comment on column id is '主鍵';
comment on column customer_id is '客戶id';
comment on column identity_id, is '客戶身份id';
comment on column customer_name is '客戶名稱';

create index idx_customer_table_customer_id on customer_table (customer_id);
create unique index uk_customer_table_customer_id_identity_id on customer_table (customer_id,identity_id);

解決方案一、

我們可以直接通過設定唯一約束欄位或唯一約束索引，基於這個唯一約束，做衝突判斷，然後做更新處理，實現案例如下：

自動去重新增sql:

insert into customer_table (customer_id,identity_id,customer_name)
(#{customerId,jdbcType=VARCHAR},#{identityId,jdbcType=VARCHAR},#{customerName,jdbcType=VARCHAR})
on conflict (customer_id,identity_id) do update
set customer_name=#{customerName,jdbcType=VARCHAR}

解決方案二、

有些時候我們不能使用 on conflict 進行衝突判斷，那麼就需要我們自己判斷衝突點了，案例如下

insert into customer_table (customer_id,identity_id,customer_name)

(select

#{customerId,jdbcType=VARCHAR},#{identityId,jdbcType=VARCHAR},#{customerName,jdbcType=VARCHAR}

where not exists (

　　select id fromcustomer_table

　　wherecustomer_id =#{customerId,jdbcType=VARCHAR}

　　andidentity_id =#{identityId,jdbcType=VARCHAR}

　　)

)

總結：

方案二呢，使用更靈活，不依賴於唯一性約束欄位的建立,但是不好做更新，所以兩種方案各有利弊，可以現實場景自由選擇