请考虑以下示例数据集：
SELECT ID, patID, OP1Datum, OP1OPVerfahren, revision FROM table LIMIT 50;
+----+-------+------------+----------------+----------+
| ID | patID | OP1Datum | OP1OPVerfahren | revision |
+----+-------+------------+----------------+----------+
| 1 | 649 | 0000-00-00 | 0 | 0 |
| 2 | 649 | 2010-09-23 | 0 | 1 |
| 3 | 649 | 2010-09-23 | 0 | 2 |
| 4 | 649 | 2010-09-23 | 0 | 3 |
| 5 | 649 | 2010-09-23 | 0 | 4 |
| 6 | 649 | 2010-09-23 | 0 | 5 |
| 7 | 649 | 2010-09-23 | 0 | 6 |
| 8 | 649 | 2010-09-23 | 0 | 7 |
| 9 | 649 | 2010-09-23 | 0 | 8 |
| 10 | 649 | 2010-09-23 | 0 | 9 |
| 12 | 673 | 2010-08-16 | 0 | 0 |
| 13 | 676 | 2010-07-05 | 0 | 0 |
| 14 | 649 | 2010-02-17 | 0 | 10 |
| 15 | 649 | 2010-02-17 | 0 | 11 |
| 16 | 676 | 2010-07-05 | 0 | 1 |
| 17 | 718 | 2010-12-01 | 0 | 0 |
| 18 | 720 | 0000-00-00 | 0 | 0 |
| 19 | 720 | 0000-00-00 | 0 | 1 |
| 20 | 649 | 2010-02-17 | 0 | 12 |
| 21 | 649 | 2010-02-17 | 0 | 13 |
| 22 | 649 | 2010-02-17 | 0 | 14 |
| 23 | 649 | 2010-02-17 | 0 | 15 |
| 24 | 649 | 2010-02-17 | 0 | 16 |
| 25 | 719 | 2008-12-03 | 0 | 0 |
| 26 | 719 | 2008-12-03 | 0 | 1 |
| 27 | 721 | 2008-12-03 | 0 | 1 |
| 28 | 649 | 2010-02-16 | 0 | 17 |
| 29 | 649 | 2010-02-16 | 0 | 18 |
| 30 | 720 | 2011-02-03 | 0 | 2 |
| 31 | 720 | 2011-02-03 | 0 | 3 |
| 32 | 720 | 2011-02-03 | 0 | 4 |
| 33 | 720 | 2011-02-03 | 0 | 5 |
| 34 | 720 | 2011-02-03 | 0 | 6 |
| 35 | 676 | 2010-07-05 | 0 | 2 |
| 36 | 718 | 2010-12-01 | 0 | 1 |
| 37 | 719 | 2008-12-03 | 0 | 2 |
| 38 | 721 | 2008-12-03 | 0 | 2 |
| 39 | 649 | 2010-02-16 | 0 | 19 |
| 40 | 673 | 2010-08-16 | 0 | 1 |
| 41 | 694 | 2011-02-16 | 0 | 0 |
| 42 | 649 | 2010-02-16 | 0 | 20 |
| 43 | 649 | 2010-10-23 | 0 | 21 |
| 44 | 649 | 2010-02-16 | 0 | 22 |
| 45 | 649 | 2010-02-16 | 0 | 23 |
| 46 | 649 | 2010-02-16 | 0 | 24 |
| 47 | 724 | 2011-03-02 | 0 | 0 |
| 48 | 724 | 2011-03-02 | 0 | 1 |
| 49 | 694 | 2011-02-16 | 0 | 1 |
| 50 | 694 | 2011-02-16 | 0 | 2 |
| 51 | 649 | 2010-02-16 | 0 | 25 |
+----+-------+------------+----------------+----------+
例如，patID 649显示了数据记录的26个修订版，即。即此记录已保存26次。
我想只显示具有单独最高版本号的行。所以理想的数据集是这样的：
+----+-------+------------+----------------+----------+
| ID | patID | OP1Datum | OP1OPVerfahren | revision |
+----+-------+------------+----------------+----------+
| 1 | 649 | 0000-00-00 | 0 | 25 |
| 2 | 673 | 2010-08-16 | 0 | 1 |
| 3 | 676 | 2010-07-05 | 0 | 2 |
| 4 | 718 | 2010-12-01 | 0 | 1 |
| 5 | 719 | 2008-12-03 | 0 | 2 |
| 6 | 720 | 2011-02-03 | 0 | 6 |
| 7 | 721 | 2008-12-03 | 0 | 2 |
| 8 | 724 | 2011-03-02 | 0 | 1 |
| 9 | 694 | 2011-02-16 | 0 | 2 |
+----+-------+------------+----------------+----------+
有谁知道如何处理这个？
考虑以下：
DROP TABLE IF EXISTS my_table;
CREATE TABLE my_table
(id SERIAL PRIMARY KEY
,patient_id INT NOT NULL
,dt DATE
);
INSERT INTO my_table VALUES
( 1,649,'0000-00-00'),
( 2,649,'2010-09-23'),
( 3,649,'2010-09-23'),
( 4,649,'2010-09-23'),
( 5,649,'2010-09-23'),
( 6,649,'2010-09-23'),
( 7,649,'2010-09-23'),
( 8,649,'2010-09-23'),
( 9,649,'2010-09-23'),
(10,649,'2010-09-23'),
(12,673,'2010-08-16'),
(13,676,'2010-07-05'),
(14,649,'2010-02-17'),
(15,649,'2010-02-17'),
(16,676,'2010-07-05'),
(17,718,'2010-12-01'),
(18,720,'0000-00-00'),
(19,720,'0000-00-00'),
(20,649,'2010-02-17'),
(21,649,'2010-02-17'),
(22,649,'2010-02-17'),
(23,649,'2010-02-17'),
(24,649,'2010-02-17'),
(25,719,'2008-12-03'),
(26,719,'2008-12-03'),
(27,721,'2008-12-03'),
(28,649,'2010-02-16'),
(29,649,'2010-02-16'),
(30,720,'2011-02-03'),
(31,720,'2011-02-03'),
(32,720,'2011-02-03'),
(33,720,'2011-02-03'),
(34,720,'2011-02-03'),
(35,676,'2010-07-05'),
(36,718,'2010-12-01'),
(37,719,'2008-12-03'),
(38,721,'2008-12-03'),
(39,649,'2010-02-16'),
(40,673,'2010-08-16'),
(41,694,'2011-02-16'),
(42,649,'2010-02-16'),
(43,649,'2010-10-23'),
(44,649,'2010-02-16'),
(45,649,'2010-02-16'),
(46,649,'2010-02-16'),
(47,724,'2011-03-02'),
(48,724,'2011-03-02'),
(49,694,'2011-02-16'),
(50,694,'2011-02-16'),
(51,649,'2010-02-16');
一个问题：
SELECT x.*
, y.rev
FROM my_table x
JOIN
( SELECT MIN(id) id, COUNT(*)-1 rev FROM my_table GROUP BY patient_id ) y
ON y.id = x.id
ORDER
BY id;
+----+------------+------------+-----+
| id | patient_id | dt | rev |
+----+------------+------------+-----+
| 1 | 649 | 0000-00-00 | 25 |
| 12 | 673 | 2010-08-16 | 1 |
| 13 | 676 | 2010-07-05 | 2 |
| 17 | 718 | 2010-12-01 | 1 |
| 18 | 720 | 0000-00-00 | 6 |
| 25 | 719 | 2008-12-03 | 2 |
| 27 | 721 | 2008-12-03 | 1 |
| 41 | 694 | 2011-02-16 | 2 |
| 47 | 724 | 2011-03-02 | 1 |
+----+------------+------------+-----+
不是答案;评论太久了......
这是我认为对问题的第一部分有效的方法......
DROP TABLE IF EXISTS method;
CREATE TABLE method
(method_id SERIAL PRIMARY KEY
,method VARCHAR(255) NOT NULL UNIQUE
);
INSERT INTO method VALUES
( 1, "Keine Operation"),
( 2, "Bioenterics Intragastric Ballon (BIB)"),
( 3, "Gastric Banding"),
( 4, "Roux-en-Y Gastric Bypass"),
( 5, "Roux-en-Y Gastric Bypass banded"),
( 6, "Scopinaro"),
( 7, "Duodenal Switch (DS)"),
( 8, "Sleeve Resection"),
( 9, "Gastric Pacemaker"),
( 10, "Billroth II"),
( 12, "Gastroplasty"),
( 13, "Fobi / Capella Bypass"),
( 14, "Larrad"),
( 15, "Santoro"),
( 16, "DJB"),
( 17, "TOGA"),
( 18, "Endobarrier"),
( 19, "Gastric Plication"),
( 20, "Stomaphyx"),
( 21, "Omega Loop Bypass"),
( 22, "Omega Loop Bypass banded"),
( 23, "Long Limb Bypass"),
( 24, "Distal Very Long Gastric Bypass (Thurnheer)"),
( 25, "Endoscopic Sclerosation"),
( 26, "Swedish Adjustable Gastric Bypass (SAGB)"),
( 27, "Vertical Banded Gastroplasty (VBG)"),
( 28, "Plastic Abdominal Wall Reconstruction (PAWR)"),
( 29, "Inner Hernia Repair"),
( 30, "Single Anastomosis Duodeno-Ileal Bypass with Sleeve Gastrectomy (SADI-S)"),
(100, "Anderes OP-Verfahren");
DROP TABLE IF EXISTS therapy;
CREATE TABLE therapy
(therapy_id SERIAL PRIMARY KEY
,method_id INT NOT NULL
,date DATE NOT NULL
);
INSERT INTO therapy (SELECT NULL,1+(RAND()*30),'2015-01-01' + INTERVAL RAND()*100 DAY);
INSERT INTO therapy SELECT NULL,1+(RAND()*30),'2015-01-01' + INTERVAL RAND()*100 DAY FROM therapy; -- Repeat this line a few times...
SELECT m.*
, COUNT(t.therapy_id) total
FROM method m
LEFT
JOIN therapy t
ON t.method_id = m.method_id
GROUP
BY m.method_id;
+-----------+--------------------------------------------------------------------------+-------+
| method_id | method | total |
+-----------+--------------------------------------------------------------------------+-------+
| 1 | Keine Operation | 38 |
| 2 | Bioenterics Intragastric Ballon (BIB) | 79 |
| 3 | Gastric Banding | 72 |
| 4 | Roux-en-Y Gastric Bypass | 73 |
| 5 | Roux-en-Y Gastric Bypass banded | 59 |
| 6 | Scopinaro | 66 |
| 7 | Duodenal Switch (DS) | 58 |
| 8 | Sleeve Resection | 67 |
| 9 | Gastric Pacemaker | 64 |
| 10 | Billroth II | 67 |
| 12 | Gastroplasty | 55 |
| 13 | Fobi / Capella Bypass | 68 |
| 14 | Larrad | 61 |
| 15 | Santoro | 69 |
| 16 | DJB | 69 |
| 17 | TOGA | 60 |
| 18 | Endobarrier | 67 |
| 19 | Gastric Plication | 77 |
| 20 | Stomaphyx | 65 |
| 21 | Omega Loop Bypass | 86 |
| 22 | Omega Loop Bypass banded | 83 |
| 23 | Long Limb Bypass | 69 |
| 24 | Distal Very Long Gastric Bypass (Thurnheer) | 58 |
| 25 | Endoscopic Sclerosation | 67 |
| 26 | Swedish Adjustable Gastric Bypass (SAGB) | 69 |
| 27 | Vertical Banded Gastroplasty (VBG) | 71 |
| 28 | Plastic Abdominal Wall Reconstruction (PAWR) | 69 |
| 29 | Inner Hernia Repair | 72 |
| 30 | Single Anastomosis Duodeno-Ileal Bypass with Sleeve Gastrectomy (SADI-S) | 77 |
| 100 | Anderes OP-Verfahren | 0 |
+-----------+--------------------------------------------------------------------------+-------+
这部分问题的其余部分是显示问题，在应用程序代码中处理得更为恰当。
对于问题的下一部分，我们需要看到MCVE以及期望的结果（参见Why should I provide an MCVE for what seems to me to be a very simple SQL query）。显然，不要使用真实的患者数据！