我有一个剧集表，通过导入长CSV电子表格定期更新。通常，大多数记录保持不变，但无论如何它们都包含在CSV中。为了检测哪些行实际更新，我在更新CURRENT_TIMESTAMP列“mtime”上添加了TIMESTAMP默认CURRENT_TIMESTAMP。
不幸的是，每次都为每个导入的行重置每一行的时间戳值，即使它们的值都没有实际更改。
实际表的列数多于CSV文件中包含的列数，因此在UPDATE查询中。下面是表格布局和通过导入CSV文件触发的示例UPDATE查询。任何人都可以识别可能触发TIMESTAMP列重置的内容，即使这些值没有不同吗？非常感谢！
+---------------+-----------------------+------+-----+---------------------+-----------------------------+
| Field | Type | Null | Key | Default | Extra |
+---------------+-----------------------+------+-----+---------------------+-----------------------------+
| id | mediumint(7) unsigned | NO | PRI | NULL | auto_increment |
| ctime | timestamp | NO | | 0000-00-00 00:00:00 | |
| mtime | timestamp | NO | | CURRENT_TIMESTAMP | on update CURRENT_TIMESTAMP |
| network_id | mediumint(7) unsigned | NO | MUL | 0 | |
| series_id | mediumint(6) unsigned | NO | MUL | NULL | |
| season_id | mediumint(7) unsigned | NO | MUL | NULL | |
| num | mediumint(7) unsigned | NO | | NULL | |
| title | varchar(150) | NO | | NULL | |
| isactive | tinyint(1) | NO | MUL | 1 | |
| istve | tinyint(1) | NO | | 0 | |
| isest | tinyint(1) | NO | | 0 | |
| ismob | tinyint(1) | NO | | 0 | |
| isvod | tinyint(1) | NO | | 0 | |
| tve_start | date | NO | | 0000-00-00 | |
| est_start | date | NO | | 0000-00-00 | |
| mob_start | date | NO | | 0000-00-00 | |
| vod_start | date | NO | | 0000-00-00 | |
| tve_end | date | NO | | 0000-00-00 | |
| est_end | date | NO | | 0000-00-00 | |
| mob_end | date | NO | | 0000-00-00 | |
| vod_end | date | NO | | 0000-00-00 | |
| fposter | varchar(150) | NO | | | |
| iswebisode | tinyint(1) | NO | | 0 | |
| date_air | date | NO | | 0000-00-00 | |
| roll_start | date | NO | | 0000-00-00 | |
| roll_end | date | NO | | 0000-00-00 | |
| season_start | date | NO | | 0000-00-00 | |
| season_end | date | NO | | 0000-00-00 | |
| premier_start | date | NO | | 0000-00-00 | |
| premier_end | date | NO | | 0000-00-00 | |
| iscore | tinyint(1) | NO | | 1 | |
| note | varchar(150) | NO | | | |
| tvn_filename | varchar(150) | NO | | | |
| tve_filename | varchar(150) | NO | | | |
| repeat_start | date | NO | | 0000-00-00 | |
| repeat_end | date | NO | | 0000-00-00 | |
| hasedited | tinyint(1) | NO | | 1 | |
| import_key | char(32) | NO | | | |
+---------------+-----------------------+------+-----+---------------------+-----------------------------+
UPDATE `episode` SET `network_id`='25', `series_id`='321', `season_id`='6887', `num`='10', `title`='EPISODE 810', `isvod`='1', `vod_start`='2011-09-05', `vod_end`='2011-10-31', `ismob`='1', `mob_start`='2011-09-05', `isest`='1', `est_start`='2011-09-05', `note`='', `date_air`='2011-09-04', `roll_start`='0000-00-00', `roll_end`='0000-00-00', `premier_start`='2011-09-04', `premier_end`='2011-10-09', `tvn_filename`='', `tve_filename`='' WHERE `id`='2821'
CREATE TABLE IF NOT EXISTS `episode` (
`id` mediumint(7) unsigned NOT NULL AUTO_INCREMENT,
`ctime` timestamp NOT NULL DEFAULT '0000-00-00 00:00:00',
`mtime` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
`network_id` mediumint(7) unsigned NOT NULL DEFAULT '0',
`series_id` mediumint(6) unsigned NOT NULL,
`season_id` mediumint(7) unsigned NOT NULL,
`num` mediumint(7) unsigned NOT NULL,
`title` varchar(150) COLLATE latin1_general_ci NOT NULL,
`isactive` tinyint(1) NOT NULL DEFAULT '1',
`istve` tinyint(1) NOT NULL DEFAULT '0',
`isest` tinyint(1) NOT NULL DEFAULT '0',
`ismob` tinyint(1) NOT NULL DEFAULT '0',
`isvod` tinyint(1) NOT NULL DEFAULT '0',
`tve_start` date NOT NULL DEFAULT '0000-00-00',
`est_start` date NOT NULL DEFAULT '0000-00-00',
`mob_start` date NOT NULL DEFAULT '0000-00-00',
`vod_start` date NOT NULL DEFAULT '0000-00-00',
`tve_end` date NOT NULL DEFAULT '0000-00-00',
`est_end` date NOT NULL DEFAULT '0000-00-00',
`mob_end` date NOT NULL DEFAULT '0000-00-00',
`vod_end` date NOT NULL DEFAULT '0000-00-00',
`fposter` varchar(150) COLLATE latin1_general_ci NOT NULL DEFAULT '',
`iswebisode` tinyint(1) NOT NULL DEFAULT '0',
`date_air` date NOT NULL DEFAULT '0000-00-00',
`roll_start` date NOT NULL DEFAULT '0000-00-00',
`roll_end` date NOT NULL DEFAULT '0000-00-00',
`season_start` date NOT NULL DEFAULT '0000-00-00',
`season_end` date NOT NULL DEFAULT '0000-00-00',
`premier_start` date NOT NULL DEFAULT '0000-00-00',
`premier_end` date NOT NULL DEFAULT '0000-00-00',
`iscore` tinyint(1) NOT NULL DEFAULT '1',
`note` varchar(150) COLLATE latin1_general_ci NOT NULL DEFAULT '',
`tvn_filename` varchar(150) COLLATE latin1_general_ci NOT NULL DEFAULT '',
`tve_filename` varchar(150) COLLATE latin1_general_ci NOT NULL DEFAULT '',
`repeat_start` date NOT NULL DEFAULT '0000-00-00',
`repeat_end` date NOT NULL DEFAULT '0000-00-00',
`hasedited` tinyint(1) NOT NULL DEFAULT '1',
`import_key` char(32) COLLATE latin1_general_ci NOT NULL DEFAULT '',
PRIMARY KEY (`id`),
KEY `season_id` (`season_id`),
KEY `idx_series` (`isactive`,`season_id`),
KEY `idx_master_cron` (`isactive`,`hasedited`),
KEY `network_id` (`network_id`),
KEY `series_id` (`series_id`),
KEY `season_id_2` (`season_id`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1 COLLATE=latin1_general_ci ;
答案 0 :(得分：0)
这是预期的行为。 MySQL文档中没有任何内容表明只有在提供新值时才会发生更新，从而触发ON UPDATE处理程序。
您没有说明您的导入方法是什么，因此这可能不适用。您确实显示了工具生成的更新语句。如果您对它有任何控制权，或者它可以选择自己编写：通过在where子句中添加更多检查来进行可选更新：
UPDATE `episode` SET `network_id`='25', `series_id`='321', ...
WHERE `id`='2821' AND `network_id`<>'25' AND `series_id`<>'321' ...
这样，如果值相同，则WHERE子句不会与现有行匹配，不会发生更新，也不会触发时间戳ON UPDATE。