我通过cygwin命令行在windows10工作站上使用sqlcmd将sqlserver数据库中的大量字符串数据作为插入内容转储到文本文件中。
它可以很好地加载到mysql中，但在mariadb中我得到以下错误：

ERROR 1366 (22007) at line 384: 
    Incorrect string value: '\x87ao Su...' for column 'description' at row 1995

我的sql server db具有以下设置：

SELECT DISTINCT C.collation_name 
FROM   sys.tables AS T 
       INNER JOIN sys.columns C 
               ON T.object_id = C.object_id 
WHERE  collation_name IS NOT NULL; 

collation_name: Latin1_General_CI_AS

SELECT DISTINCT C.collation_name 
FROM   sys.tables AS T 
       INNER JOIN sys.columns C 
               ON T.object_id = C.object_id 
WHERE  collation_name IS NOT NULL; 
collation_name: Latin1_General_CI_AS

这是mysql（它在这里插入）：

SHOW FULL COLUMNS FROM forecast
--------------

+--------------------+------------------+-----------------+------+-----+-------------------+----------------+---------------------------------+---------+
| Field              | Type             | Collation       | Null | Key | Default           | Extra          | Privileges                      | Comment |
+--------------------+------------------+-----------------+------+-----+-------------------+----------------+---------------------------------+---------+
| description        | varchar(255)     | utf8_unicode_ci | YES  |     | NULL              |                | select,insert,update,references |         |

SELECT @@character_set_database, @@collation_database, @@collation_connection
--------------

+--------------------------+----------------------+------------------------+
| @@character_set_database | @@collation_database | @@collation_connection |
+--------------------------+----------------------+------------------------+
| utf8                     | utf8_unicode_ci      | utf8_general_ci        |
+--------------------------+----------------------+------------------------+
1 row in set (0.01 sec)

这是mariadb给出的上面的错误：

SHOW FULL COLUMNS FROM forecast
--------------

+--------------------+------------------+-----------------+------+-----+---------------------+----------------+---------------------------------+---------+
| Field              | Type             | Collation       | Null | Key | Default             | Extra          | Privileges                      | Comment |
+--------------------+------------------+-----------------+------+-----+---------------------+----------------+---------------------------------+---------+
| description        | varchar(255)     | utf8_unicode_ci | YES  |     | NULL                |                | select,insert,update,references |         |
+--------------------+------------------+-----------------+------+-----+---------------------+----------------+---------------------------------+---------+
10 rows in set (0.01 sec)

MariaDB [tardis]> SELECT @@character_set_database, @@collation_database, @@collation_connection;
--------------
SELECT @@character_set_database, @@collation_database, @@collation_connection
--------------

+--------------------------+----------------------+------------------------+
| @@character_set_database | @@collation_database | @@collation_connection |
+--------------------------+----------------------+------------------------+
| utf8                     | utf8_unicode_ci      | utf8_general_ci        |
+--------------------------+----------------------+------------------------+
1 row in set (0.01 sec)

据我所知，mariadb的行为应该和mysql一样，但事实并非如此。
我该怎么办？
我可以用sql server函数转换文本-目前我只是转义'字符-但是如何转换呢？
[更新]刚刚发现，当我从文件运行import sql命令时，mariadb对数据没有任何问题。
只有当我在命令行上将它直接从sqlcmd管道传输到mysql.exe时，我才会得到错误：

$ SQLCMD.EXE -i "c:\dev\workspace\sql\generate-inserts-for-migration.sql" \
       -h-1 -S "myhost" -d MyDBName -r1 -W \
       -b -f o:65001 | mysql -h myhost.amazonaws.com -u adam -p

我想这样做，这样我就不必在硬盘上创建10gb的sql文件，甚至可以在云上同一网络上的docker容器（空间有限）中运行它。