How can i update some entity with ManyToMany relation

user_13267506 0
Viewed 1256

help me please. I really don't understand how it's works I have an entity called Product, with a ManyToMany relation with another entity called Car with cascade true flag enabled.

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

  @ManyToMany(() => Car, { cascade: true })
  @JoinTable()
  cars: Car[]
}

The Car entity doesn't contain a reference to the Product entity. That means like i think that the relation is unidirectional. So, when i save a product, i set the cars for this product:

const cars = await this.carService.getByIds(createProductDto.cars)

const newProduct = this.productsRepository.create({
    ...createProductDto,
    cars
})

await this.productsRepository.save(newProduct)

I get the cars entities by ids and save them to the product. This works well. I'm not sure that it is the best way, but i don't know another.

The problems comes when i try to update some existing product in the db. I get the fallowing errors:

ERROR [ExceptionsHandler] duplicate key value violates unique constraint "PK_f7b49d8fd8d09ea2fc2e7542f3f"

or

Error: Cannot query across many-to-many for property cars

or other... Help me please. I want to understand what is the best way to work with such relations. What is the best way to create, update and remove it. Thanks

typeorm nestjs postgresql javascript
1 Answers

You claim to have a M:M between Cars and Products, but that is not what you describe. You describe a 1:M for Cars and Products. Referring to Cars in the Products table means a given Car can have many Products, but a given Product can only be used by 1 car. That is the result of referencing Cars from Products (I would not classify as unidirectional as given either one I can get the other, it just the number I can get changes.) However, I believe what you want a M:M. The typical implementation for a M:M requires a 3rd table. With this configuration neither Cars nor Products table references the other, instead the third table references both. Unfortunately, I do not know or recognize your obfuscation manager (ORM you are using). So I'll sketch a possible ddl definition for the tables.

create table cars ( car_id   integer generated always as identity primary key
                  , vin      text 
                  , manufacturer text
                  -- other Car attributes
                  );
                 
create table products( prod_id integer generated always as identity primary key
                     , manufacturer text 
                     -- other product 
                     ); 
                     
-- create the reference/intersection table for M:M Cars:Products
create table car_products( car_id    
                         , prod_id   
                         -- attributes that pertain only to the combination of 
                         -- of the other tables 
                         , model_code text 
                         , serial_number text
                         -- constraints  
                         , constraint car_products_pk
                                      primary key (car_id, prod_id)
                         , constraint car_products_2_cars_fk
                                      foreign key (car_id) 
                                      references cars(car_id) 
                                      on delete cascade
                         , constatint car_products_2_products_fk
                                      foreign key(prod_id)
                                      references products(prod_id) 
                                      on delete cascade                                      
                         );

Now with this setup a given Car can have any number of Products and a given product can be user by any number of cars. (Note: In this set up a Product is differently from a part.) I hope this helps you understand 1:M relationships vs M:M relationships.

user_76238560
Related