The following SQL queries describe two tables and their contents, a small and simplified part of a big inventory system.
The first is a transaction log listing bought and sold quantities of various items.
CREATE TABLE `Inventar_actions` (
`PartID` char(8) NOT NULL,
`DateSeq` smallint(4) unsigned NOT NULL,
`ActionType` enum('Bought','Sold') NOT NULL,
`Units` bigint(20) unsigned NOT NULL,
PRIMARY KEY (`PartID`,`DateSeq`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;
INSERT INTO `test`.`Inventar_actions` (`PartID`, `DateSeq`, `ActionType`, `Units`) VALUES
('0001', '1020', 'Bought', '100'),('0001', '1025', 'Sold', '20'),('0001', '1028', 'Sold', '30'),
('0001', '1031', 'Bought', '100'),('0001', '1035', 'Sold', '50');
The second table is a calendar, listing business days and their sequential numbers from an arbitrary epoch:
CREATE TABLE `BusinessDates` (
`DateSeq` int(10) unsigned NOT NULL,
`BusinessDate` date NOT NULL,
PRIMARY KEY (`DateSeq`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;
INSERT INTO `test`.`BusinessDates` (`DateSeq`, `BusinessDate`) VALUES
('1015', '2020-02-03'),('1016', '2020-02-04'),('1017', '2020-02-05'),('1018', '2020-02-06'),('1019', '2020-02-07'),
('1020', '2020-02-10'),('1021', '2020-02-11'),('1022', '2020-02-12'),('1023', '2020-02-13'),('1024', '2020-02-14'),
('1025', '2020-02-17'),('1026', '2020-02-18'),('1027', '2020-02-19'),('1028', '2020-02-20'),('1029', '2020-02-21'),
('1030', '2020-02-24'),('1031', '2020-02-25'),('1032', '2020-02-26'),('1033', '2020-02-27'),('1034', '2020-02-28'),
('1035', '2020-03-02'),('1036', '2020-03-03'),('1037', '2020-03-04'),('1038', '2020-03-05'),('1039', '2020-03-06')
;
What I need is a daily inventory, one that looks something like this:
| 1020 | 2020-02-10 | 100 | Bought | 100 |
| 1021 | 2020-02-11 | | | 100 |
| 1022 | 2020-02-12 | | | 100 |
| 1023 | 2020-02-13 | | | 100 |
| 1024 | 2020-02-14 | | | 100 |
| 1025 | 2020-02-17 | 20 | Sold | 80 |
| 1026 | 2020-02-18 | | | 80 |
| 1027 | 2020-02-19 | | | 80 |
| 1028 | 2020-02-20 | 30 | Sold | 50 |
| 1029 | 2020-02-21 | | | 50 |
| 1030 | 2020-02-24 | | | 50 |
| 1031 | 2020-02-25 | 100 | Bought | 150 |
| 1032 | 2020-02-26 | | | 150 |
| 1033 | 2020-02-27 | | | 150 |
| 1034 | 2020-02-28 | | | 150 |
| 1035 | 2020-03-02 | 50 | Sold | 100 |
| 1036 | 2020-03-03 | | | 100 |
| 1037 | 2020-03-04 | | | 100 |
| 1038 | 2020-03-05 | | | 100 |
| 1039 | 2020-03-06 | | | 100 |
That is, for each business date what is the balance of each PartID, given parts bought and sold.
I used the following query to line up dates, date-sequence and transactions:
SELECT d.DateSeq, d.BusinessDate, a.`Units`, a.`ActionType`
FROM `BusinessDates` AS d LEFT JOIN
(SELECT DateSeq, ActionType, Units FROM `Inventar_actions` WHERE PartID = '0001') AS a USING (DateSeq)
WHERE d.DateSeq BETWEEN 1020 AND 1039
ORDER BY d.DateSeq;
This produces the first four columns of the desired output.
Unfortunately, I couldn't find a query that will also produce the last column, the balance inventory of parts.
Please note that there are a lot of PartID's and thousands of business days.
Please help...
This works with mysql 5.x and 8.0
CREATE TABLE `Inventar_actions` ( `PartID` char(8) NOT NULL, `DateSeq` smallint(4) unsigned NOT NULL, `ActionType` enum('Bought','Sold') NOT NULL, `Units` bigint(20) unsigned NOT NULL, PRIMARY KEY (`PartID`,`DateSeq`) ) ENGINE=InnoDB DEFAULT CHARSET=utf8; INSERT INTO `Inventar_actions` (`PartID`, `DateSeq`, `ActionType`, `Units`) VALUES ('0001', '1020', 'Bought', '100'),('0001', '1025', 'Sold', '20'),('0001', '1028', 'Sold', '30'), ('0001', '1031', 'Bought', '100'),('0001', '1035', 'Sold', '50');✓ ✓
CREATE TABLE `BusinessDates` ( `DateSeq` int(10) unsigned NOT NULL, `BusinessDate` date NOT NULL, PRIMARY KEY (`DateSeq`) ) ENGINE=InnoDB DEFAULT CHARSET=utf8; INSERT INTO `BusinessDates` (`DateSeq`, `BusinessDate`) VALUES ('1015', '2020-02-03'),('1016', '2020-02-04'),('1017', '2020-02-05'),('1018', '2020-02-06'),('1019', '2020-02-07'), ('1020', '2020-02-10'),('1021', '2020-02-11'),('1022', '2020-02-12'),('1023', '2020-02-13'),('1024', '2020-02-14'), ('1025', '2020-02-17'),('1026', '2020-02-18'),('1027', '2020-02-19'),('1028', '2020-02-20'),('1029', '2020-02-21'), ('1030', '2020-02-24'),('1031', '2020-02-25'),('1032', '2020-02-26'),('1033', '2020-02-27'),('1034', '2020-02-28'), ('1035', '2020-03-02'),('1036', '2020-03-03'),('1037', '2020-03-04'),('1038', '2020-03-05'),('1039', '2020-03-06') ;✓ ✓
SELECT d.DateSeq , d.BusinessDate , a.`Units` , a.`ActionType` , CASE WHEN a.`ActionType` = 'Bought' THEN @inventory := @inventory + a.`Units` WHEN a.`ActionType` = 'Sold' THEN @inventory := @inventory - a.`Units` ELSE @inventory := @inventory END inventory FROM (SELECT * FROM `BusinessDates` ORDER BY BusinessDate ASC) AS d LEFT JOIN (SELECT DateSeq, ActionType, Units FROM `Inventar_actions` WHERE PartID = '0001') AS a USING (DateSeq) ,(SELECT @inventory := 0) b WHERE d.DateSeq BETWEEN 1020 AND 1039 ORDER BY d.DateSeq;DateSeq | BusinessDate | Units | ActionType | inventory ------: | :----------- | ----: | :--------- | :-------- 1020 | 2020-02-10 | 100 | Bought | 100 1021 | 2020-02-11 | null | null | 100 1022 | 2020-02-12 | null | null | 100 1023 | 2020-02-13 | null | null | 100 1024 | 2020-02-14 | null | null | 100 1025 | 2020-02-17 | 20 | Sold | 80 1026 | 2020-02-18 | null | null | 80 1027 | 2020-02-19 | null | null | 80 1028 | 2020-02-20 | 30 | Sold | 50 1029 | 2020-02-21 | null | null | 50 1030 | 2020-02-24 | null | null | 50 1031 | 2020-02-25 | 100 | Bought | 150 1032 | 2020-02-26 | null | null | 150 1033 | 2020-02-27 | null | null | 150 1034 | 2020-02-28 | null | null | 150 1035 | 2020-03-02 | 50 | Sold | 100 1036 | 2020-03-03 | null | null | 100 1037 | 2020-03-04 | null | null | 100 1038 | 2020-03-05 | null | null | 100 1039 | 2020-03-06 | null | null | 100
db<>fiddle here