[英]How to find all connected subgraphs of an undirected graph
我需要一些幫助來解決我正在努力解決的問題。
示例表：
ID |Identifier1 | Identifier2
---------------------------------
1 | a | c
2 | b | f
3 | a | g
4 | c | h
5 | b | j
6 | d | f
7 | e | k
8 | i |
9 | l | h
我想對兩列之間相互關聯的標識符進行分組，並分配一個唯一的組 ID。
所需 Output：
Identifier | Gr_ID | Gr.Members
---------------------------------------------------
a | 1 | (a,c,g,h,l)
b | 2 | (b,d,f,j)
c | 1 | (a,c,g,h,l)
d | 2 | (b,d,f,j)
e | 3 | (e,k)
f | 2 | (b,d,f,j)
g | 1 | (a,c,g,h,l)
h | 1 | (a,c,g,h,l)
j | 2 | (b,d,f,j)
k | 3 | (e,k)
l | 1 | (a,c,g,h,l)
i | 4 | (i)
注意：Gr.Members 列不是必需的，主要用於更清晰的視圖。
所以組的定義是：如果一行與該組中的至少一行共享至少一個標識符，則該行屬於一個組
但是必須將組 ID 分配給每個標識符（由兩列的並集選擇）而不是行。
關於如何構建查詢以提供所需的 output 的任何幫助？
謝謝。
更新：下面是一些額外的樣本集，它們的預期值為 output。
給定表：
Identifier1 | Identifier2
----------------------------
a | f
a | g
a | NULL
b | c
b | a
b | h
b | j
b | NULL
b | NULL
b | g
c | k
c | b
d | l
d | f
d | g
d | m
d | a
d | NULL
d | a
e | c
e | b
e | NULL
預期 output：所有記錄應屬於同一組，組 ID = 1。
給定表：
Identifier1 | Identifier2
--------------------------
a | a
b | b
c | a
c | b
c | c
預期 output：記錄應在同一組中，組 ID = 1。
這是一個不使用游標但使用單個遞歸查詢的變體。
本質上，它將數據視為圖中的邊並遞歸遍歷圖的所有邊，在檢測到循環時停止。 然后它將所有找到的循環放在組中並給每個組一個編號。
請參閱下面有關其工作原理的詳細說明。 我建議您運行查詢 CTE-by-CTE 並檢查每個中間結果以了解它的作用。
示例 1
DECLARE @T TABLE (ID int, Ident1 char(1), Ident2 char(1));
INSERT INTO @T (ID, Ident1, Ident2) VALUES
(1, 'a', 'a'),
(2, 'b', 'b'),
(3, 'c', 'a'),
(4, 'c', 'b'),
(5, 'c', 'c');
樣本 2
我添加了一個帶有z值的行，以便有多個帶有不成對值的行。
DECLARE @T TABLE (ID int, Ident1 char(1), Ident2 char(1));
INSERT INTO @T (ID, Ident1, Ident2) VALUES
(1, 'a', 'a'),
(1, 'a', 'c'),
(2, 'b', 'f'),
(3, 'a', 'g'),
(4, 'c', 'h'),
(5, 'b', 'j'),
(6, 'd', 'f'),
(7, 'e', 'k'),
(8, 'i', NULL),
(88, 'z', 'z'),
(9, 'l', 'h');
示例 3
DECLARE @T TABLE (ID int, Ident1 char(1), Ident2 char(1));
INSERT INTO @T (ID, Ident1, Ident2) VALUES
(1, 'a', 'f'),
(2, 'a', 'g'),
(3, 'a', NULL),
(4, 'b', 'c'),
(5, 'b', 'a'),
(6, 'b', 'h'),
(7, 'b', 'j'),
(8, 'b', NULL),
(9, 'b', NULL),
(10, 'b', 'g'),
(11, 'c', 'k'),
(12, 'c', 'b'),
(13, 'd', 'l'),
(14, 'd', 'f'),
(15, 'd', 'g'),
(16, 'd', 'm'),
(17, 'd', 'a'),
(18, 'd', NULL),
(19, 'd', 'a'),
(20, 'e', 'c'),
(21, 'e', 'b'),
(22, 'e', NULL);
詢問
WITH
CTE_Idents
AS
(
SELECT Ident1 AS Ident
FROM @T
UNION
SELECT Ident2 AS Ident
FROM @T
)
,CTE_Pairs
AS
(
SELECT Ident1, Ident2
FROM @T
WHERE Ident1 <> Ident2
UNION
SELECT Ident2 AS Ident1, Ident1 AS Ident2
FROM @T
WHERE Ident1 <> Ident2
)
,CTE_Recursive
AS
(
SELECT
CAST(CTE_Idents.Ident AS varchar(8000)) AS AnchorIdent
, Ident1
, Ident2
, CAST(',' + Ident1 + ',' + Ident2 + ',' AS varchar(8000)) AS IdentPath
, 1 AS Lvl
FROM
CTE_Pairs
INNER JOIN CTE_Idents ON CTE_Idents.Ident = CTE_Pairs.Ident1
UNION ALL
SELECT
CTE_Recursive.AnchorIdent
, CTE_Pairs.Ident1
, CTE_Pairs.Ident2
, CAST(CTE_Recursive.IdentPath + CTE_Pairs.Ident2 + ',' AS varchar(8000)) AS IdentPath
, CTE_Recursive.Lvl + 1 AS Lvl
FROM
CTE_Pairs
INNER JOIN CTE_Recursive ON CTE_Recursive.Ident2 = CTE_Pairs.Ident1
WHERE
CTE_Recursive.IdentPath NOT LIKE CAST('%,' + CTE_Pairs.Ident2 + ',%' AS varchar(8000))
)
,CTE_RecursionResult
AS
(
SELECT AnchorIdent, Ident1, Ident2
FROM CTE_Recursive
)
,CTE_CleanResult
AS
(
SELECT AnchorIdent, Ident1 AS Ident
FROM CTE_RecursionResult
UNION
SELECT AnchorIdent, Ident2 AS Ident
FROM CTE_RecursionResult
)
SELECT
CTE_Idents.Ident
,CASE WHEN CA_Data.XML_Value IS NULL
THEN CTE_Idents.Ident ELSE CA_Data.XML_Value END AS GroupMembers
,DENSE_RANK() OVER(ORDER BY
CASE WHEN CA_Data.XML_Value IS NULL
THEN CTE_Idents.Ident ELSE CA_Data.XML_Value END
) AS GroupID
FROM
CTE_Idents
CROSS APPLY
(
SELECT CTE_CleanResult.Ident+','
FROM CTE_CleanResult
WHERE CTE_CleanResult.AnchorIdent = CTE_Idents.Ident
ORDER BY CTE_CleanResult.Ident FOR XML PATH(''), TYPE
) AS CA_XML(XML_Value)
CROSS APPLY
(
SELECT CA_XML.XML_Value.value('.', 'NVARCHAR(MAX)')
) AS CA_Data(XML_Value)
WHERE
CTE_Idents.Ident IS NOT NULL
ORDER BY Ident;
結果 1
+-------+--------------+---------+
| Ident | GroupMembers | GroupID |
+-------+--------------+---------+
| a | a,b,c, | 1 |
| b | a,b,c, | 1 |
| c | a,b,c, | 1 |
+-------+--------------+---------+
結果 2
+-------+--------------+---------+
| Ident | GroupMembers | GroupID |
+-------+--------------+---------+
| a | a,c,g,h,l, | 1 |
| b | b,d,f,j, | 2 |
| c | a,c,g,h,l, | 1 |
| d | b,d,f,j, | 2 |
| e | e,k, | 3 |
| f | b,d,f,j, | 2 |
| g | a,c,g,h,l, | 1 |
| h | a,c,g,h,l, | 1 |
| i | i | 4 |
| j | b,d,f,j, | 2 |
| k | e,k, | 3 |
| l | a,c,g,h,l, | 1 |
| z | z | 5 |
+-------+--------------+---------+
結果 3
+-------+--------------------------+---------+
| Ident | GroupMembers | GroupID |
+-------+--------------------------+---------+
| a | a,b,c,d,e,f,g,h,j,k,l,m, | 1 |
| b | a,b,c,d,e,f,g,h,j,k,l,m, | 1 |
| c | a,b,c,d,e,f,g,h,j,k,l,m, | 1 |
| d | a,b,c,d,e,f,g,h,j,k,l,m, | 1 |
| e | a,b,c,d,e,f,g,h,j,k,l,m, | 1 |
| f | a,b,c,d,e,f,g,h,j,k,l,m, | 1 |
| g | a,b,c,d,e,f,g,h,j,k,l,m, | 1 |
| h | a,b,c,d,e,f,g,h,j,k,l,m, | 1 |
| j | a,b,c,d,e,f,g,h,j,k,l,m, | 1 |
| k | a,b,c,d,e,f,g,h,j,k,l,m, | 1 |
| l | a,b,c,d,e,f,g,h,j,k,l,m, | 1 |
| m | a,b,c,d,e,f,g,h,j,k,l,m, | 1 |
+-------+--------------------------+---------+
我將使用第二組示例數據進行說明。
CTE_Idents
CTE_Idents給出出現在Ident1和Ident2列中的所有標識符的列表。 因為它們可以以任意順序出現，我們UNION兩列在一起。 UNION還會刪除任何重復項。
+-------+
| Ident |
+-------+
| NULL |
| a |
| b |
| c |
| d |
| e |
| f |
| g |
| h |
| i |
| j |
| k |
| l |
| z |
+-------+
CTE_Pairs
CTE_Pairs給出了圖在兩個方向上的所有邊的列表。 同樣， UNION用於刪除任何重復項。
+--------+--------+
| Ident1 | Ident2 |
+--------+--------+
| a | c |
| a | g |
| b | f |
| b | j |
| c | a |
| c | h |
| d | f |
| e | k |
| f | b |
| f | d |
| g | a |
| h | c |
| h | l |
| j | b |
| k | e |
| l | h |
+--------+--------+
CTE_Recursive
CTE_Recursive是查詢的主要部分，它從每個唯一標識符開始遞歸遍歷圖。 這些起始行由UNION ALL的第一部分生成。 UNION ALL的第二部分遞歸地連接到自身，將Ident2鏈接到Ident1 。 由於我們預先制作的CTE_Pairs寫在兩個方向的所有邊緣，我們能夠始終只能鏈接Ident2到Ident1 ，我們將圖中的所有路徑。 同時，查詢構建IdentPath - 到目前為止已遍歷的以逗號分隔的標識符字符串。 它用於WHERE過濾器：
CTE_Recursive.IdentPath NOT LIKE CAST('%,' + CTE_Pairs.Ident2 + ',%' AS varchar(8000))
一旦我們遇到之前包含在路徑中的標識符，遞歸就會停止，因為連接的節點列表已用完。 AnchorIdent是遞歸的起始標識符，稍后將用於對結果進行分組。 Lvl並未真正使用，我將其包含在內是為了更好地了解正在發生的事情。
+-------------+--------+--------+-------------+-----+
| AnchorIdent | Ident1 | Ident2 | IdentPath | Lvl |
+-------------+--------+--------+-------------+-----+
| a | a | c | ,a,c, | 1 |
| a | a | g | ,a,g, | 1 |
| b | b | f | ,b,f, | 1 |
| b | b | j | ,b,j, | 1 |
| c | c | a | ,c,a, | 1 |
| c | c | h | ,c,h, | 1 |
| d | d | f | ,d,f, | 1 |
| e | e | k | ,e,k, | 1 |
| f | f | b | ,f,b, | 1 |
| f | f | d | ,f,d, | 1 |
| g | g | a | ,g,a, | 1 |
| h | h | c | ,h,c, | 1 |
| h | h | l | ,h,l, | 1 |
| j | j | b | ,j,b, | 1 |
| k | k | e | ,k,e, | 1 |
| l | l | h | ,l,h, | 1 |
| l | h | c | ,l,h,c, | 2 |
| l | c | a | ,l,h,c,a, | 3 |
| l | a | g | ,l,h,c,a,g, | 4 |
| j | b | f | ,j,b,f, | 2 |
| j | f | d | ,j,b,f,d, | 3 |
| h | c | a | ,h,c,a, | 2 |
| h | a | g | ,h,c,a,g, | 3 |
| g | a | c | ,g,a,c, | 2 |
| g | c | h | ,g,a,c,h, | 3 |
| g | h | l | ,g,a,c,h,l, | 4 |
| f | b | j | ,f,b,j, | 2 |
| d | f | b | ,d,f,b, | 2 |
| d | b | j | ,d,f,b,j, | 3 |
| c | h | l | ,c,h,l, | 2 |
| c | a | g | ,c,a,g, | 2 |
| b | f | d | ,b,f,d, | 2 |
| a | c | h | ,a,c,h, | 2 |
| a | h | l | ,a,c,h,l, | 3 |
+-------------+--------+--------+-------------+-----+
CTE_CleanResult
CTE_CleanResult葉僅從相關部分CTE_Recursive並再次合並兩個Ident1和Ident2使用UNION 。
+-------------+-------+
| AnchorIdent | Ident |
+-------------+-------+
| a | a |
| a | c |
| a | g |
| a | h |
| a | l |
| b | b |
| b | d |
| b | f |
| b | j |
| c | a |
| c | c |
| c | g |
| c | h |
| c | l |
| d | b |
| d | d |
| d | f |
| d | j |
| e | e |
| e | k |
| f | b |
| f | d |
| f | f |
| f | j |
| g | a |
| g | c |
| g | g |
| g | h |
| g | l |
| h | a |
| h | c |
| h | g |
| h | h |
| h | l |
| j | b |
| j | d |
| j | f |
| j | j |
| k | e |
| k | k |
| l | a |
| l | c |
| l | g |
| l | h |
| l | l |
+-------------+-------+
最終選擇
現在我們需要為每個AnchorIdent構建一串以逗號分隔的Ident值。 CROSS APPLY with FOR XML可以做到。 DENSE_RANK()計算每個AnchorIdent的GroupID編號。
此腳本根據需要生成測試集 1、2 和 3 的輸出。 算法注釋作為腳本中的注釋。
意識到：
#tree 。 所以使用這個腳本需要將源數據插入#treeNULL值。 替換NULL與值CHAR(0)插入時#tree使用ISNULL(source_col,CHAR(0))來規避這個缺點。 從最終結果中進行選擇時，使用NULLIF(node,CHAR(0))將CHAR(0)替換為NULL 。請注意，使用遞歸 CTE的答案更優雅，因為它是單個 SQL 語句，但對於使用遞歸 CTE 的大型輸入集，可能會產生極短的執行時間（請參閱有關該答案的評論）。 下面描述的解決方案雖然更復雜，但對於大型輸入集應該運行得更快。
SET NOCOUNT ON;
CREATE TABLE #tree(node_l CHAR(1),node_r CHAR(1));
CREATE NONCLUSTERED INDEX NIX_tree_node_l ON #tree(node_l)INCLUDE(node_r); -- covering indices to speed up lookup
CREATE NONCLUSTERED INDEX NIX_tree_node_r ON #tree(node_r)INCLUDE(node_l);
INSERT INTO #tree(node_l,node_r) VALUES
('a','c'),('b','f'),('a','g'),('c','h'),('b','j'),('d','f'),('e','k'),('i','i'),('l','h'); -- test set 1
--('a','f'),('a','g'),(CHAR(0),'a'),('b','c'),('b','a'),('b','h'),('b','j'),('b',CHAR(0)),('b',CHAR(0)),('b','g'),('c','k'),('c','b'),('d','l'),('d','f'),('d','g'),('d','m'),('d','a'),('d',CHAR(0)),('d','a'),('e','c'),('e','b'),('e',CHAR(0)); -- test set 2
--('a','a'),('b','b'),('c','a'),('c','b'),('c','c'); -- test set 3
CREATE TABLE #sets(node CHAR(1) PRIMARY KEY,group_id INT); -- nodes with group id assigned
CREATE TABLE #visitor_queue(node CHAR(1)); -- contains nodes to visit
CREATE TABLE #visited_nodes(node CHAR(1) PRIMARY KEY CLUSTERED WITH(IGNORE_DUP_KEY=ON)); -- nodes visited for nodes on the queue; ignore duplicate nodes when inserted
CREATE TABLE #visitor_ctx(node_l CHAR(1),node_r CHAR(1)); -- context table, contains deleted nodes as they are visited from #tree
DECLARE @last_created_group_id INT=0;
-- Notes:
-- 1. This algorithm is destructive in its input set, ie #tree will be empty at the end of this procedure
-- 2. This algorithm does not accept NULL values. Populate #tree with CHAR(0) for NULL values (using ISNULL(source_col,CHAR(0)), or COALESCE(source_col,CHAR(0)))
-- 3. When selecting from #sets, to regain the original NULL values use NULLIF(node,CHAR(0))
WHILE EXISTS(SELECT*FROM #tree)
BEGIN
TRUNCATE TABLE #visited_nodes;
TRUNCATE TABLE #visitor_ctx;
-- push first nodes onto the queue (via #visitor_ctx -> #visitor_queue)
DELETE TOP (1) t
OUTPUT deleted.node_l,deleted.node_r INTO #visitor_ctx(node_l,node_r)
FROM #tree AS t;
INSERT INTO #visitor_queue(node) SELECT node_l FROM #visitor_ctx UNION SELECT node_r FROM #visitor_ctx; -- UNION to filter when node_l equals node_r
INSERT INTO #visited_nodes(node) SELECT node FROM #visitor_queue; -- keep track of nodes visited
-- work down the queue by visiting linked nodes in #tree; nodes are deleted as they are visited
WHILE EXISTS(SELECT*FROM #visitor_queue)
BEGIN
TRUNCATE TABLE #visitor_ctx;
-- pop_front for node on the stack (via #visitor_ctx -> @node)
DELETE TOP (1) s
OUTPUT deleted.node INTO #visitor_ctx(node_l)
FROM #visitor_queue AS s;
DECLARE @node CHAR(1)=(SELECT node_l FROM #visitor_ctx);
TRUNCATE TABLE #visitor_ctx;
-- visit nodes in #tree where node_l or node_r equal target @node;
-- delete visited nodes from #tree, output to #visitor_ctx
DELETE t
OUTPUT deleted.node_l,deleted.node_r INTO #visitor_ctx(node_l,node_r)
FROM #tree AS t
WHERE t.node_l=@node OR t.node_r=@node;
-- insert visited nodes in the queue that haven't been visited before
INSERT INTO #visitor_queue(node)
(SELECT node_l FROM #visitor_ctx UNION SELECT node_r FROM #visitor_ctx) EXCEPT (SELECT node FROM #visited_nodes);
-- keep track of visited nodes (duplicates are ignored by the IGNORE_DUP_KEY option for the PK)
INSERT INTO #visited_nodes(node)
SELECT node_l FROM #visitor_ctx UNION SELECT node_r FROM #visitor_ctx;
END
SET @last_created_group_id+=1; -- create new group id
-- insert group into #sets
INSERT INTO #sets(group_id,node)
SELECT group_id=@last_created_group_id,node
FROM #visited_nodes;
END
SELECT node=NULLIF(node,CHAR(0)),group_id FROM #sets ORDER BY node; -- nodes with their assigned group id
SELECT g.group_id,m.members -- groups with their members
FROM
(SELECT DISTINCT group_id FROM #sets) AS g
CROSS APPLY (
SELECT members=STUFF((
SELECT ','+ISNULL(CAST(NULLIF(si.node,CHAR(0)) AS VARCHAR(4)),'NULL')
FROM #sets AS si
WHERE si.group_id=g.group_id
FOR XML PATH('')
),1,1,'')
) AS m
ORDER BY g.group_id;
DROP TABLE #visitor_queue;
DROP TABLE #visited_nodes;
DROP TABLE #visitor_ctx;
DROP TABLE #sets;
DROP TABLE #tree;
第 1 組的輸出：
+------+----------+
| node | group_id |
+------+----------+
| a | 1 |
| b | 2 |
| c | 1 |
| d | 2 |
| e | 4 |
| f | 2 |
| g | 1 |
| h | 1 |
| i | 3 |
| j | 2 |
| k | 4 |
| l | 1 |
+------+----------+
第 2 組的輸出：
+------+----------+
| node | group_id |
+------+----------+
| NULL | 1 |
| a | 1 |
| b | 1 |
| c | 1 |
| d | 1 |
| e | 1 |
| f | 1 |
| g | 1 |
| h | 1 |
| j | 1 |
| k | 1 |
| l | 1 |
| m | 1 |
+------+----------+
第 3 組的輸出：
+------+----------+
| node | group_id |
+------+----------+
| a | 1 |
| b | 1 |
| c | 1 |
+------+----------+
我的建議是使用帶有游標的存儲過程。 它易於實施且相對較快。 只需兩步：
詢問：
CREATE TABLE #PairIds
(
Ident1 VARCHAR(10),
Ident2 VARCHAR(10)
)
INSERT INTO #PairIds
VALUES ('a', 'c'),
('b', 'f'),
('a', 'g'),
('c', 'h'),
('b', 'j'),
('d', 'f'),
('e', 'k'),
('l', 'h')
exec [dbo].[sp_GetIdentByGroup]
結果：
Ident | GroupID --------------------------------------------------- a | 1 | b | 2 | c | 1 | d | 2 | e | 3 | f | 2 | g | 1 | h | 1 | j | 2 | k | 3 | l | 1 |
創建存儲過程的代碼：
CREATE PROCEDURE [dbo].[sp_GetIdentByGroup]
AS
BEGIN
DECLARE @message VARCHAR(70);
DECLARE @IdentInput1 varchar(20)
DECLARE @IdentInput2 varchar(20)
DECLARE @Counter INT
DECLARE @Group1 INT
DECLARE @Group2 INT
DECLARE @Ident varchar(20)
DECLARE @IdentCheck1 varchar(20)
DECLARE @IdentCheck2 varchar(20)
SET @Counter = 1
DECLARE @IdentByGroupCursor TABLE (
Ident varchar(20) UNIQUE CLUSTERED,
GroupID INT
);
-- Use a cursor to select your data, which enables SQL Server to extract
-- the data from your local table to the variables.
declare ins_cursor cursor for
select Ident1, Ident2 from #PairIds
open ins_cursor
fetch next from ins_cursor into @IdentInput1, @IdentInput2 -- At this point, the data from the first row
-- is in your local variables.
-- Move through the table with the @@FETCH_STATUS=0
WHILE @@FETCH_STATUS=0
BEGIN
SET @Group1 = null
SET @Group2 = null
SELECT TOP 1 @Group1 = GroupID, @IdentCheck1 = Ident
FROM @IdentByGroupCursor
WHERE Ident in (@IdentInput1)
SELECT TOP 1 @Group2 = GroupID, @IdentCheck2 = Ident
FROM @IdentByGroupCursor
WHERE Ident in (@IdentInput2)
IF (@Group1 IS NOT NULL AND @Group2 IS NOT NULL)
BEGIN
IF @Group1 > @Group2
BEGIN
UPDATE @IdentByGroupCursor
SET GroupID = @Group2
WHERE
GroupID = @Group1
END
IF @Group2 > @Group1
BEGIN
UPDATE @IdentByGroupCursor
SET GroupID = @Group1
WHERE
GroupID = @Group2
END
END
ELSE IF @Group1 IS NOT NULL
BEGIN
UPDATE @IdentByGroupCursor
SET GroupID = @Group1
WHERE
Ident IN (@IdentInput1)
END
ELSE IF @Group2 IS NOT NULL
BEGIN
UPDATE @IdentByGroupCursor
SET GroupID = @Group2
WHERE
Ident IN (@IdentInput2)
END
IF (@Group1 IS NOT NULL AND @Group2 IS NOT NULL)
BEGIN
IF @Group1 > @Group2
BEGIN
UPDATE @IdentByGroupCursor
SET GroupID = @Group2
WHERE
GroupID = @Group1
END
IF @Group2 > @Group1
BEGIN
UPDATE @IdentByGroupCursor
SET GroupID = @Group1
WHERE
GroupID = @Group2
END
END
IF @Group1 IS NULL
BEGIN
INSERT INTO @IdentByGroupCursor (Ident, GroupID)
VALUES (@IdentInput1, ISNULL(@Group2, @Counter))
END
IF @Group2 IS NULL
BEGIN
INSERT INTO @IdentByGroupCursor (Ident, GroupID)
VALUES (@IdentInput2, ISNULL(@Group1, @COunter))
END
IF (@Group1 IS NULL OR @Group2 IS NULL)
BEGIN
SET @COunter = @COunter +1
END
-- Once the execution has taken place, you fetch the next row of data from your local table.
fetch next from ins_cursor into @IdentInput1, @IdentInput2
End
-- When all the rows have inserted you must close and deallocate the cursor.
-- Failure to do this will not let you re-use the cursor.
close ins_cursor
deallocate ins_cursor
SELECT Ident ,DENSE_RANK() OVER( ORDER BY GroupID ASC) AS GroupID
FROM @IdentByGroupCursor
ORDER BY Ident
END
GO
Sp_GetIdentByGroup 有一個速度索引，並使用游標准備所需的結果集。 存儲過程需要#PairIds 表存在。
有關SQL 如何在特定組中對彼此相關的標識符進行分組的更多信息。
sp_GetIdentByGroup 是很棒的方法，我正在使用 CTE 函數搜索類似的解決方案幾天，但它們太慢了。例如，我們在 5 秒內完成了 20 條記錄，在 1:40 分鍾內完成了 30 條記錄，而 35 條記錄運行了很長時間。
你的程序設法排名 150 項目是一組與秒的分裂。 謝謝！
如何使用Teradata SQL分組/列出無向圖的所有節點
[英]How to group/List all nodes of a undirected graph using teradata sql
如何將連接的子圖 ID 分配給 Oracle SQL 中無向圖中的每個節點？
[英]How to assign connected subgraph ID to each node in an undirected graph in Oracle SQL?
一個簡單的SQL Select查詢，用於抓取社交圖中所有已連接的人員？
[英]A simple SQL Select query to crawl all connected people in a social graph?
聲明:本站的技術帖子網頁，遵循CC BY-SA 4.0協議，如果您需要轉載，請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.