sequelize/sequelize-typescript

virtual getter using pk is null on update

willmac321 opened this issue · 1 comments

When using a virtual column that modifies a primary key, if the primary key is not supplied on an update, the pk is null and the virtual column will try to operate on that null value. The value should always have a value?

Given a model like

UserModel {
   @PrimaryKey
   @Column({ type: Sequelize.BIGINT, autoIncrement: true })
   public id!: number;

   @Column({type:Sequelize.STRING})
    public comment!: string;

    @Column({
    type: Sequelize.VIRTUAL,
    get() {
      console.log(this);
      return thisFuncCantHaveNull(this.getDataValue('id'));
    },
}
///
const [affectedCount, rows] = UserModel.update({comment:'i like Mondays'}, {
        where: { id: suppliedId },
        transaction,
        returning: true,
});

then when the update is called on a record that exists already, the console.log of this will print something like:

UserModel {
  dataValues: {
    id: null,
    comment: 'i like Mondays',
  },
  _previousDataValues: {
   comment: null,
  },
  uniqno: 1,
  _changed: Set(1) { 'comment' },
  _options: { isNewRecord: true, _schema: null, _schemaDelimiter: '' },
  isNewRecord: true
}

I think it makes sense that under the hood a new record is made and then updated with the existing record, however the null id is problematic since our virtual getter is trying to calculate a value on something that shouldn't be null... :(