标签 pdo 下的文章

mysql批量更新多条记录的同一字段为不同值

mysql更新数据的某个字段，一般这样写：

UPDATE mytable SET myfield = 'value' WHERE other_field = 'other_value';

也可以这样用in指定要更新的记录：

UPDATE mytable SET myfield = 'value' WHERE other_field in ('other_values');

这里注意 ‘other_values’ 是一个逗号（，）分隔的字符串，如：1,2,3

如果更新多条数据而且每条记录要更新的值不同，可能很多人会这样写：

foreach ($values as $id => $myvalue) {
    $sql = "UPDATE mytable SET myfield = $myvalue WHERE id = $id";
    mysql_query($sql);
}

即是循环一条一条的更新记录。一条记录update一次，这样性能很差，也很容易造成阻塞。

那么能不能一条sql语句实现批量更新呢？mysql并没有提供直接的方法来实现批量更新，但是可以用点小技巧来实现。

UPDATE mytable
    SET myfield = CASE id
        WHEN 1 THEN 'myvalue1'
        WHEN 2 THEN 'myvalue2'
        WHEN 3 THEN 'myvalue3'
    END
WHERE other_field ('other_values')

如果where条件查询出记录的id不在CASE范围内，myfield将被设置为空。

如果更新多个值的话，只需要稍加修改：

UPDATE mytable
    SET myfield1 = CASE id
        WHEN 1 THEN 'myvalue11'
        WHEN 2 THEN 'myvalue12'
        WHEN 3 THEN 'myvalue13'
    END,
    myfield2 = CASE id
        WHEN 1 THEN 'myvalue21'
        WHEN 2 THEN 'myvalue22'
        WHEN 3 THEN 'myvalue23'
    END
WHERE id IN (1,2,3)

这里以php为例，构造这两条mysql语句：

  • 更新多条单个字段为不同值, mysql模式
$ids_values = array(
    1 => 11,
    2 => 22,
    3 => 33,
    4 => 44,
    5 => 55,
    6 => 66,
    7 => 77,
    8 => 88,
);
 
$ids = implode(',', array_keys($ids_values ));
$sql = "UPDATE mytable SET myfield = CASE id ";
foreach ($ids_values as $id=> $myvalue) {
    $sql .= sprintf("WHEN %d THEN %d ", $id, $myvalue);
}
$sql .= "END WHERE id IN ($ids)";
echo $sql.";<br/>";

输出：

UPDATE mytable SET myfield = CASE id WHEN 1 THEN 11 WHEN 2 THEN 22 WHEN 3 THEN 33 WHEN 4 THEN 44 WHEN 5 THEN 55 WHEN 6 THEN 66 WHEN 7 THEN 77 WHEN 8 THEN 88 END WHERE id IN (1,2,3,4,5,6,7,8);

  • 更新多个字段为不同值, PDO模式
$data = array(array('id' => 1, 'myfield1val' => 11, 'myfield2val' => 111), array('id' => 2, 'myfield1val' => 22, 'myfield2val' => 222));
$where_in_ids = implode(',', array_map(function($v) {return ":id_" . $v['id'];}, $data));
$update_sql = 'UPDATE mytable SET';
$params = array();

$update_sql .= ' myfield1 = CASE id';
foreach($data as $key => $item) {
    $update_sql .= " WHEN :id_" . $key . " THEN :myfield1val_" . $key . " ";
    $params[":id_" . $key] = $item['id'];
    $params[":myfield1val_" . $key] = $item['myfield1val'];
}
$update_sql .= " END";

$update_sql .= ',myfield2 = CASE id';
foreach($data as $key => $item) {
    $update_sql .= " WHEN :id_" . $key . " THEN :myfield2val_" . $key . " ";
    $params[":id_" . $key] = $item['id'];
    $params[":myfield1va2_" . $key] = $item['myfield2val'];
}
$update_sql .= " END";

$update_sql .= " WHERE id IN (" . $where_in_ids . ")";
echo $update_sql.";<br/>";
var_dump($params);

输出：

UPDATE mytable SET myfield1 = CASE id WHEN :id_0 THEN :myfield1val_0 WHEN :id_1 THEN :myfield1val_1 END,myfield2 = CASE id WHEN :id_0 THEN :myfield2val_0 WHEN :id_1 THEN :myfield2val_1 END WHERE id IN (:id_1,:id_2);

array (size=6)
 ':id_0' => int 1
 ':myfield1val_0' => int 11
 ':id_1' => int 2
 ':myfield1val_1' => int 22
 ':myfield1va2_0' => int 111
 ':myfield1va2_1' => int 222

另外三种批量更新方式

1. replace into 批量更新

replace into mytable(id, myfield) values (1,'value1'),(2,'value2'),(3,'value3');

2. insert into ...on duplicate key update 批量存在则更新

insert into mytable(id, myfield1, myfield2) values (1,'value11','value21'),(2,'value12','value22'),(3,'value13','value23') on duplicate key update myfield1=values(myfield1),myfield2=values(myfield2);

不需要以下语句就能批量更新

insert into mytable(id, myfield1, myfield2) values (1,'value11','value21'),(2,'value12','value22'),(3,'value13','value23') on duplicate key update myfield1=case id when values(id) then values(myfield1) end,myfield2=case id when values(id) then values(myfield2) end;

注意：即使没插入也会造成自增id的增加。

3. 临时表

DROP TABLE IF EXISTS `tmptable`;
create temporary table tmptable(id int(4) primary key,myfield varchar(50));
insert into tmptable values (1,'value1'),(2,'value2'),(3,'value3');
update mytable, tmptable set mytable.myfield = tmptable.myfield where mytable.id = tmptable.id;
  • 【replace into】和【insert into】更新都依赖于主键或唯一值，并都可能造成新增记录的操作的结构隐患
  • 【replace into】操作本质是对重复记录先delete然后insert，如果更新的字段不全缺失的字段将被设置成缺省值
  • 【insert into】则只是update重复的记录，更改的字段只能依循公式值
  • 【临时表】方式需要用户有temporary 表的create 权限
  • 数量较少时【replace into】和【insert into】性能最好，数量大时【临时表】最好，【CASE】则具有通用型也不具结构隐患