如何在Oracle SQL中對唯一標記進行分組？

Question

我有這樣一個表：

GROUPID | TAG 
------ | ------ 
1  | Tag1 Tag2 
1  | Tag1 Tag3 
1  | Tag1 Tag4 
2  | Tag5 Tag6 
2  | Tag4 Tag3 

而且我想要得到的結果如下：

GROUP | TAG 
------- | ------- 
1  | Tag1 Tag2 Tag3 Tag4 
2  | Tag3 Tag4 Tag5 Tag6 

而且想法如何篩選在Oracle SQL複製的標籤？

謝謝

Answer 1

正如評論中提到的mathguy，這樣的查詢可以通過更改設計來簡化。
但是，考慮到這種格式的數據，仍然可以提取一組獨特的TAG/GROUPID。以下是一個示例方法：

在第一階段中，我們將使用REGEXP_COUNT來查看每行有多少個TAG。然後，我們將生成位置標記號爲每個行中的每個TAG。最後，我們將在每行中的每個集合式TAG的給定位置提取標籤。

首先，創建測試表：

CREATE TABLE GROUPID_TAG(
    GROUPID NUMBER, 
    "TAG" VARCHAR2(256) 
); 

INSERT INTO GROUPID_TAG VALUES (1,'Tag1 Tag2'); 
INSERT INTO GROUPID_TAG VALUES (1,'Tag1 Tag3'); 
INSERT INTO GROUPID_TAG VALUES (1,'Tag1 Tag4'); 
INSERT INTO GROUPID_TAG VALUES (2,'Tag5 Tag6'); 
INSERT INTO GROUPID_TAG VALUES (2,'Tag4 Tag3'); 

的下面查詢將產生兩列，與單個TAG在ONLY_ONE_TAG柱（但更多的行每GROUPID）

WITH COUNTED_TAG AS (
    SELECT GROUPID, "TAG", REGEXP_COUNT("TAG",'(^|)[^ ]{1,}') AS TAG_COUNT FROM GROUPID_TAG), 
    KEYED_COUNTED_TAG AS (
    SELECT GROUPID, "TAG", TAG_COUNT, TAG_KEG_GENERATOR.TAG_KEY FROM COUNTED_TAG 
     INNER JOIN (SELECT LEVEL AS TAG_KEY FROM DUAL CONNECT BY LEVEL <= 999) TAG_KEG_GENERATOR 
     ON TAG_KEG_GENERATOR.TAG_KEY <= COUNTED_TAG.TAG_COUNT) 
SELECT DISTINCT GROUPID, REPLACE(REGEXP_SUBSTR("TAG",'(^|)[^ ]{1,}',1,TAG_KEY),' ','') AS ONLY_ONE_TAG 
FROM KEYED_COUNTED_TAG 
ORDER BY 1 ASC, 2 ASC; 

運行它給出：

GROUPID ONLY_ONE_TAG 
1  Tag1   
1  Tag2   
1  Tag3   
1  Tag4   
2  Tag3   
2  Tag4   
2  Tag5   
2  Tag6  

此時的數據可能比原始狀態更容易使用。但是，如果你想每個GROUPID重新聚合成一行，這裏就是一個例子。與我們過去的查詢開始，我們將添加一個LISTAGG到聚合的事情：

WITH COUNTED_TAG AS (
    SELECT GROUPID, "TAG", REGEXP_COUNT("TAG",'(^|)[^ ]{1,}') AS TAG_COUNT FROM GROUPID_TAG), 
    KEYED_COUNTED_TAG AS (
    SELECT GROUPID, "TAG", TAG_COUNT, TAG_KEG_GENERATOR.TAG_KEY FROM COUNTED_TAG 
     INNER JOIN (SELECT LEVEL AS TAG_KEY FROM DUAL CONNECT BY LEVEL <= 999) TAG_KEG_GENERATOR 
     ON TAG_KEG_GENERATOR.TAG_KEY <= COUNTED_TAG.TAG_COUNT), 
    DISTINCT_TAG AS(SELECT DISTINCT GROUPID, REPLACE(REGEXP_SUBSTR("TAG",'(^|)[^ ]{1,}',1,TAG_KEY),' ','') AS ONLY_ONE_TAG 
        FROM KEYED_COUNTED_TAG) 
SELECT GROUPID, LISTAGG(ONLY_ONE_TAG,' ') WITHIN GROUP (ORDER BY ONLY_ONE_TAG ASC) AS AGGREGATED_TAG 
FROM DISTINCT_TAG 
GROUP BY GROUPID 
ORDER BY 1 ASC; 

結果：

GROUPID AGGREGATED_TAG  
1  Tag1 Tag2 Tag3 Tag4 
2  Tag3 Tag4 Tag5 Tag6 

然後，添加一些額外的標籤，來進行測試：

INSERT INTO GROUPID_TAG VALUES (1,'Wookie Hobbit @_@'); 
INSERT INTO GROUPID_TAG VALUES (2,'HAL-9000 Thor'); 

而且再次查詢：

WITH COUNTED_TAG AS (
    SELECT GROUPID, "TAG", REGEXP_COUNT("TAG",'(^|)[^ ]{1,}') AS TAG_COUNT FROM GROUPID_TAG), 
    KEYED_COUNTED_TAG AS (
    SELECT GROUPID, "TAG", TAG_COUNT, TAG_KEG_GENERATOR.TAG_KEY FROM COUNTED_TAG 
     INNER JOIN (SELECT LEVEL AS TAG_KEY FROM DUAL CONNECT BY LEVEL <= 999) TAG_KEG_GENERATOR 
     ON TAG_KEG_GENERATOR.TAG_KEY <= COUNTED_TAG.TAG_COUNT), 
    DISTINCT_TAG AS(SELECT DISTINCT GROUPID, REPLACE(REGEXP_SUBSTR("TAG",'(^|)[^ ]{1,}',1,TAG_KEY),' ','') AS ONLY_ONE_TAG 
        FROM KEYED_COUNTED_TAG) 
SELECT GROUPID, LISTAGG(ONLY_ONE_TAG,' ') WITHIN GROUP (ORDER BY ONLY_ONE_TAG ASC) AS AGGREGATED_TAG 
FROM DISTINCT_TAG 
GROUP BY GROUPID 
ORDER BY 1 ASC; 

結果：

GROUPID AGGREGATED_TAG       
1  @_@ Hobbit Tag1 Tag2 Tag3 Tag4 Wookie 
2  HAL-9000 Tag3 Tag4 Tag5 Tag6 Thor