Question

我有两张桌子

--
-- Table structure for table `d_statistics_report_temp`
--

CREATE TABLE IF NOT EXISTS `d_statistics_report_temp` (
  `srt_id` int(10) NOT NULL AUTO_INCREMENT ,
  `pub_name` varchar(100) DEFAULT NULL ,
  `sal_must_pay_date` date DEFAULT NULL ,
  `total_sales` float(7,0) DEFAULT NULL,
  `total_must_pay` float(7,0) NOT NULL,
  `pub_share` float(7,0) DEFAULT NULL ,
  `pub_id` int(11) DEFAULT NULL COMMENT ,
  `conversion_doc` varchar(255) NOT NULL,
  PRIMARY KEY (`srt_id`)
) ENGINE=MyISAM  DEFAULT CHARSET=utf8 AUTO_INCREMENT=6 ;

--
-- Dumping data for table `d_statistics_report_temp`
--

INSERT INTO `d_statistics_report_temp` 
(`srt_id`, `pub_name`, `sal_must_pay_date`, `total_sales`, `total_must_pay`, `pub_share`, `pub_id`, `conversion_doc`) 
VALUES
(1, 'sara', '2013-03-01', 50, 0, 15, 256, ''),
(2, 'sara', '2013-04-01', 20, 0, 15, 256, ''),
(3, 'sara', '2013-05-01', 80, 0, 15, 256, ''),
(4, 'sara', '2013-06-01', 10, 0, 15, 256, ''),
(5, 'sara', '2013-07-01', 20, 0, 15, 256, ''),
(6, 'sara', '2013-08-01', 70, 0, 15, 256, ''),
(7, 'sara', '2013-09-01', 90, 0, 15, 256, ''),
(8, 'adam', '2013-05-01', 10, 0, 15, 255, '')

和第二个表

--
-- Table structure for table `d_statistics_docs`
--

CREATE TABLE IF NOT EXISTS `d_statistics_docs` (
  `sd_id` int(11) NOT NULL AUTO_INCREMENT,
  `sd_pub` int(11) NOT NULL,
  `sd_date` date NOT NULL,
  `sd_doc` text NOT NULL,
  `price` int(11) NOT NULL,
  PRIMARY KEY (`sd_id`)
) ENGINE=MyISAM  DEFAULT CHARSET=utf8 AUTO_INCREMENT=16 ;

--
-- Dumping data for table `d_statistics_docs`
--

INSERT INTO `d_statistics_docs` (`sd_id`, `sd_pub`, `sd_date`, `sd_doc`, `price`) VALUES

(9,  256, '2013-03-01', 'scifiwallpaper1.jpg'),
(10, 256, '2013-04-01', 'scifiwallpaper1.jpg'),
(11, 256, '2013-05-01', ''),
(12, 256, '2013-06-01', 'scifiwallpaper1.jpg'),
(13, 256, '2013-07-01', ''),
(14, 256, '2013-08-01', ''),
(15, 256, '2013-09-01', 'scifiwallpaper1.jpg');

我想加入表格，为每total_sales

的每个pub_id（publisher_id）获取sd_date

但我想仅为此when sd_doc IS NULL条件total_sales获取sd_date

如果sd_doc NOT NULL本月返回total_sales且所有月份都在

之前

输出示例

9   256     2013-03-01  scifiwallpaper1.jpg 50

10  256     2013-04-01  scifiwallpaper1.jpg 20

11  256     2013-05-01         NULL         80

12  256    2013-06-01   scifiwallpaper1.jpg 90 // total 10 + 80

13  256     2013-07-01         NULL         20

14  256     2013-08-01         NULL        70

15  256     2013-09-01  scifiwallpaper1.jpg 180  // total of 90+70+20

* 注意：第二个表格中的sal_must_pay_date = sd_date *

Answer 1

您可以使用变量跟踪以前的行（只要您正确排序）。所以您的查询可以写成如下：

SET @i = 0.0;

SELECT  rt.srt_id,
        rt.pub_id,
        d.sd_date,
        CASE WHEN d.sd_doc != '' THEN d.sd_doc END sd_doc,
        rt.total_sales,
        @i + rt.total_sales AS Cumulative_Total,
        @i:= CASE WHEN d.sd_doc = '' THEN @i + rt.total_sales ELSE 0 END AS Increment
FROM    d_statistics_report_temp rt
        INNER JOIN d_statistics_docs d
            ON rt.pub_id = d.sd_pub
            AND rt.sal_must_pay_date = d.sd_date
ORDER BY rt.pub_id, d.sd_date;

在这种情况下，变量@i对每行进行更改，如果sd_doc不为空，则重置为0;如果sd_doc为空，则重置为先前的数量。

<强> Example on SQL Fiddle

如何仅按前一个日期连接两个表和总和行

1 个答案: