Skip to main content

Tree Entities

TypeORM supports the Adjacency list and Closure table patterns for storing tree structures. To learn more about the hierarchy table take a look at this awesome presentation by Bill Karwin.

Adjacency list​

Adjacency list is a simple model with self-referencing. The benefit of this approach is simplicity, a drawback is that you can't load big trees all at once because of join limitations. To learn more about the benefits and use of Adjacency Lists look at this article by Matthew Schinckel. Example:

import {
    Entity,
    Column,
    PrimaryGeneratedColumn,
    ManyToOne,
    OneToMany,
} from "typeorm"

@Entity()
export class Category {
    @PrimaryGeneratedColumn()
    id: number

    @Column()
    name: string

    @Column()
    description: string

    @ManyToOne((type) => Category, (category) => category.children)
    parent: Category

    @OneToMany((type) => Category, (category) => category.parent)
    children: Category[]
}

Nested set​

Nested set is another pattern of storing tree structures in the database. It is very efficient for reads, but bad for writes. You cannot have multiple roots in the nested set. Example:

import {
    Entity,
    Tree,
    Column,
    PrimaryGeneratedColumn,
    TreeChildren,
    TreeParent,
    TreeLevelColumn,
} from "typeorm"

@Entity()
@Tree("nested-set")
export class Category {
    @PrimaryGeneratedColumn()
    id: number

    @Column()
    name: string

    @TreeChildren()
    children: Category[]

    @TreeParent()
    parent: Category
}

Materialized Path (aka Path Enumeration)​

Materialized Path (also called Path Enumeration) is another pattern of storing tree structures in the database. It is simple and effective. Example:

import {
    Entity,
    Tree,
    Column,
    PrimaryGeneratedColumn,
    TreeChildren,
    TreeParent,
    TreeLevelColumn,
} from "typeorm"

@Entity()
@Tree("materialized-path")
export class Category {
    @PrimaryGeneratedColumn()
    id: number

    @Column()
    name: string

    @TreeChildren()
    children: Category[]

    @TreeParent()
    parent: Category
}

Closure table​

Closure table stores relations between parent and child in a separate table in a special way. It's efficient in both reading and writing. Example:

import {
    Entity,
    Tree,
    Column,
    PrimaryGeneratedColumn,
    TreeChildren,
    TreeParent,
    TreeLevelColumn,
} from "typeorm"

@Entity()
@Tree("closure-table")
export class Category {
    @PrimaryGeneratedColumn()
    id: number

    @Column()
    name: string

    @TreeChildren()
    children: Category[]

    @TreeParent()
    parent: Category
}

You can specify the closure table name and/or closure table column names by setting optional parameter options into @Tree("closure-table", options). ancestorColumnName and descandantColumnName are callback functions, which receive the primary column's metadata and return the column's name.

@Tree("closure-table", {
    closureTableName: "category_closure",
    ancestorColumnName: (column) => "ancestor_" + column.propertyName,
    descendantColumnName: (column) => "descendant_" + column.propertyName,
})

Working with tree entities​

To bind tree entities to each other, it is required to set the parent in the child entity and then save them. for example:

const a1 = new Category()
a1.name = "a1"
await dataSource.manager.save(a1)

const a11 = new Category()
a11.name = "a11"
a11.parent = a1
await dataSource.manager.save(a11)

const a12 = new Category()
a12.name = "a12"
a12.parent = a1
await dataSource.manager.save(a12)

const a111 = new Category()
a111.name = "a111"
a111.parent = a11
await dataSource.manager.save(a111)

const a112 = new Category()
a112.name = "a112"
a112.parent = a11
await dataSource.manager.save(a112)

To load such a tree use TreeRepository:

const trees = await dataSource.manager.getTreeRepository(Category).findTrees()

trees will be the following:

[
    {
        "id": 1,
        "name": "a1",
        "children": [
            {
                "id": 2,
                "name": "a11",
                "children": [
                    {
                        "id": 4,
                        "name": "a111"
                    },
                    {
                        "id": 5,
                        "name": "a112"
                    }
                ]
            },
            {
                "id": 3,
                "name": "a12"
            }
        ]
    }
]

There are other special methods to work with tree entities through TreeRepository:

  • findTrees - Returns all trees in the database with all their children, children of children, etc.
const treeCategories = await dataSource.manager.getTreeRepository(Category).findTrees()
// returns root categories with sub categories inside

const treeCategoriesWithLimitedDepth = await dataSource.manager.getTreeRepository(Category).findTrees({ depth: 2 })
// returns root categories with sub categories inside, up to depth 2
  • findRoots - Roots are entities that have no ancestors. Finds them all. Does not load children's leaves.
const rootCategories = await dataSource.manager.getTreeRepository(Category).findRoots()
// returns root categories without sub categories inside
  • findDescendants - Gets all children (descendants) of the given entity. Returns them all in a flat array.
const children = await dataSource.manager.getTreeRepository(Category).findDescendants(parentCategory)
// returns all direct subcategories (without its nested categories) of a parentCategory
  • findDescendantsTree - Gets all children (descendants) of the given entity. Returns them in a tree - nested into each other.
const childrenTree = await repository.findDescendantsTree(parentCategory)
// returns all direct subcategories (with its nested categories) of a parentCategory
const childrenTreeWithLimitedDepth = await repository.findDescendantsTree(
    parentCategory,
    { depth: 2 },
)
// returns all direct subcategories (with its nested categories) of a parentCategory, up to depth 2
  • createDescendantsQueryBuilder - Creates a query builder used to get descendants of the entities in a tree.
const children = await repository
    .createDescendantsQueryBuilder(
        "category",
        "categoryClosure",
        parentCategory,
    )
    .andWhere("category.type = 'secondary'")
    .getMany()
  • countDescendants - Gets the number of descendants of the entity.
const childrenCount = await dataSource.manager.getTreeRepository(Category).countDescendants(parentCategory)
  • findAncestors - Gets all parents (ancestors) of the given entity. Returns them all in a flat array.
const parents = await repository.findAncestors(childCategory)
// returns all direct childCategory's parent categories (without "parent of parents")
  • findAncestorsTree - Gets all parents (ancestors) of the given entity. Returns them in a tree - nested into each other.
const parentsTree = await dataSource.manager.getTreeRepository(Category).findAncestorsTree(childCategory)
// returns all direct childCategory's parent categories (with "parent of parents")
  • createAncestorsQueryBuilder - Creates a query builder used to get the ancestors of the entities in a tree.
const parents = await repository
    .createAncestorsQueryBuilder("category", "categoryClosure", childCategory)
    .andWhere("category.type = 'secondary'")
    .getMany()
  • countAncestors - Gets the number of ancestors of the entity.
const parentsCount = await dataSource.manager.getTreeRepository(Category).countAncestors(childCategory)

For the following methods, options can be passed:

  • findTrees
  • findRoots
  • findDescendants
  • findDescendantsTree
  • findAncestors
  • findAncestorsTree

The following options are available:

  • relations - Indicates what relations of entity should be loaded (simplified left join form).

Examples:

const treeCategoriesWithRelations = await dataSource.manager.getTreeRepository(Category).findTrees({
    relations: ["sites"],
})
// automatically joins the sites relation

const parentsWithRelations = await dataSource.manager.getTreeRepository(Category).findAncestors(childCategory, {
    relations: ["members"],
})
// returns all direct childCategory's parent categories (without "parent of parents") and joins the 'members' relation