-
Notifications
You must be signed in to change notification settings - Fork 332
SQL:Recursive calculation of tree like hierarchical structure
SQL SERVER has a task table where the id field is the primary key and the parent field is the foreign key pointing to the id, used to describe how multiple nodes form a tree. The root node of a tree is null, and the records with a null parent are the child nodes of the root, that is, the secondary nodes. The weight field only has a value when it is a leaf node.
id | name | weight | parent |
---|---|---|---|
1 | t1 | null | null |
22 | t22 | null | 1 |
3 | t3 | 0.03 | 1 |
4 | t4 | 0.1 | 22 |
55 | t55 | null | 22 |
6 | t6 | null | null |
7 | t7 | 0.01 | 6 |
11 | t11 | 1 | 55 |
12 | t12 | 2 | 55 |
Now we need to recursively calculate the hierarchical structure of the tree. Parallel nodes are arranged in order of their IDs, and the new name field=level * underline + original name, which is used to visually represent the hierarchical relationship; The new weight field is the sum of the weights of all subordinate nodes, and the level field represents the hierarchy.
name | weight | level |
---|---|---|
t1 | 3.13 | 1 |
_t3 | 0.03 | 2 |
_t22 | 3.1 | 2 |
__t4 | 0.1 | 3 |
__t55 | 3 | 3 |
___t11 | 1 | 4 |
___t12 | 2 | 4 |
t6 | 0.01 | 1 |
_t7 | 0.01 | 2 |
SPL code:
A | B | |
---|---|---|
1 | =data=MSSQL.query("select * from task order by id") | |
2 | =res=create(name,weight,level) | |
3 | =data.select(!parent).(tree( id, name, weight,1)) | |
4 | func tree(I,N,W,L) | =r=res.insert@n(0, fill("_",L-1)+N, 0, L ) |
5 | =data.select(parent==I).sum(tree(id, name, weight, L+1) ) | |
6 | return r.weight = W + B5 | |
7 | return res |
A1: Query the database through JDBC.
A2: Create an empty result set.
A3: Filter out the child nodes of the root and loop through these nodes using recursive function.
A4-B6: Function tree, which adds new records to the result set, recursively processes directly subordinate nodes, and returns the weight of the new record. It has four parameters: id, name, weight, and level.
B4: Add new records to the result set and spell out the new name and level. The new weight will be calculated later, so let's set it to 0 for now.
B5: Filter out the direct subordinate nodes of this node, recursively calculate the weights of these nodes, and sum them up.
B6: Calculate the new weight and modify this record to return the new weight.
A7: Return the result set.
Problem source:https://stackoverflow.com/questions/78409018/create-a-task-sheet-by-cte
SPL Resource: SPL Official Website | SPL Blog | Download esProc SPL | SPL Source Code