How to Implement a Many-To-Many Relationship in Postgresql

How to implement a many-to-many relationship in PostgreSQL?

The SQL DDL (data definition language) statements could look like this:

CREATE TABLE product (
  product_id serial PRIMARY KEY  -- implicit primary key constraint
, product    text NOT NULL
, price      numeric NOT NULL DEFAULT 0
);

CREATE TABLE bill (
  bill_id  serial PRIMARY KEY
, bill     text NOT NULL
, billdate date NOT NULL DEFAULT CURRENT_DATE
);

CREATE TABLE bill_product (
  bill_id    int REFERENCES bill (bill_id) ON UPDATE CASCADE ON DELETE CASCADE
, product_id int REFERENCES product (product_id) ON UPDATE CASCADE
, amount     numeric NOT NULL DEFAULT 1
, CONSTRAINT bill_product_pkey PRIMARY KEY (bill_id, product_id)  -- explicit pk
);

I made a few adjustments:

• The n:m relationship is normally implemented by a separate table - bill_product in this case.

• I added serial columns as surrogate primary keys. In Postgres 10 or later consider an IDENTITY column instead. See:

  • Safely rename tables using serial primary key columns
  • Auto increment table column
  • https://www.2ndquadrant.com/en/blog/postgresql-10-identity-columns/

  I highly recommend that, because the name of a product is hardly unique (not a good "natural key"). Also, enforcing uniqueness and referencing the column in foreign keys is typically cheaper with a 4-byte integer (or even an 8-byte bigint) than with a string stored as text or varchar.

• Don't use names of basic data types like date as identifiers. While this is possible, it is bad style and leads to confusing errors and error messages. Use legal, lower case, unquoted identifiers. Never use reserved words and avoid double-quoted mixed case identifiers if you can.

• "name" is not a good name. I renamed the column of the table product to be product (or product_name or similar). That is a better naming convention. Otherwise, when you join a couple of tables in a query - which you do a lot in a relational database - you end up with multiple columns named "name" and have to use column aliases to sort out the mess. That's not helpful. Another widespread anti-pattern would be just "id" as column name.
  
  I am not sure what the name of a bill would be. bill_id will probably suffice in this case.

• price is of data type numeric to store fractional numbers precisely as entered (arbitrary precision type instead of floating point type). If you deal with whole numbers exclusively, make that integer. For example, you could save prices as Cents.

• The amount ("Products" in your question) goes into the linking table bill_product and is of type numeric as well. Again, integer if you deal with whole numbers exclusively.

• You see the foreign keys in bill_product? I created both to cascade changes: ON UPDATE CASCADE. If a product_id or bill_id should change, the change is cascaded to all depending entries in bill_product and nothing breaks. Those are just references without significance of their own.
  
  I also used ON DELETE CASCADE for bill_id: If a bill gets deleted, its details die with it.
  
  Not so for products: You don't want to delete a product that's used in a bill. Postgres will throw an error if you attempt this. You would add another column to product to mark obsolete rows ("soft-delete") instead.

• All columns in this basic example end up to be NOT NULL, so NULL values are not allowed. (Yes, all columns - primary key columns are defined UNIQUE NOT NULL automatically.) That's because NULL values wouldn't make sense in any of the columns. It makes a beginner's life easier. But you won't get away so easily, you need to understand NULL handling anyway. Additional columns might allow NULL values, functions and joins can introduce NULL values in queries etc.

• Read the chapter on CREATE TABLE in the manual.

• Primary keys are implemented with a unique index on the key columns, that makes queries with conditions on the PK column(s) fast. However, the sequence of key columns is relevant in multicolumn keys. Since the PK on bill_product is on (bill_id, product_id) in my example, you may want to add another index on just product_id or (product_id, bill_id) if you have queries looking for a given product_id and no bill_id. See:

  • PostgreSQL composite primary key
  • Is a composite index also good for queries on the first field?
  • Working of indexes in PostgreSQL

• Read the chapter on indexes in the manual.

What is the best way to implement a many-to-many relationship in PostgreSQL?

You can get what you want just inserting the Product Name and the Order name letting Postgres handle duplication. You elevate the indexes on them to unique constraints (no hashing needed) then let the ON CONFLICT clause handle duplication. That is what select, if not found insert logic is attempting anyway. The only issue becomes Postgres not returning the ID on duplicates. To overcome that the insert for Product_Orders retrieves the ids from the appropriate names. This can be done in a single statement with a couple CTEs but then the returning clause is necessary and whatever is returned is null where the name already exists. So it just gets ignored anyway. However a SQL function/procedure can have multiple statements, so (imho) 3 statements are clearer. (see example here)

create or replace 
procedure gen_product_order(
          myproduct varchar
        , myorder   varchar
        )
  language sql
as $$
  insert into products(product_name) 
       values (myproduct) 
           on conflict (product_name) do nothing; 
 
  insert into orders (order_name)  
      values (myorder)
          on conflict (order_name) do nothing;      
 
    insert into product_orders (product_id, order_id) -- insert both fk ids into the product_order table
       select product_id, order_id 
         from (select product_id 
                 from products 
                where product_name = myproduct
              ) prd
            , (select order_id 
                 from orders
                where order_name = myorder
              ) ord  
        on conflict (product_id, order_id) do nothing; 
$$; 

Note: I've used a procedure rather that a function returning void but if you Postgres version does not support a function would work.

What the best way to model a many to many relationship

A proper model that allows everything you need while enforcing referential integrity could look like this:

CREATE TABLE contact (
  contact_id serial PRIMARY KEY
, name text
, phone text
, ...
);

CREATE TABLE product (
  product_id serial PRIMARY KEY
, ...
);

CREATE TABLE product_role (
  role_id int PRIMARY KEY
, role text UNIQUE
);

CREATE TABLE product_contact (
  product_id int REFERENCES product
, contact_id int REFERENCES contact
, role_id    int REFERENCES product_role
, PRIMARY KEY (product_id, contact_id, role_id)
);

If the same contact can never act in more than one role for the same product, don't include the role in the PK:

, PRIMARY KEY (product_id, contact_id)

This allows to simply add a row to product_role to allow and additional type of contact.

If there is only a hand full of distinct roles, the data type "char" might be convenient for role_id.

• Any downsides of using data type "text" for storing strings?

Basics:

• How to implement a many-to-many relationship in PostgreSQL?

Modeling Many-to-Many Relationships in Postgresql

Since you're using Postgres, you want to normalize your tables. What normalization means, in a sentence, is that your columns are represented by the key, the whole key, and nothing but the key. Normalization can be a three-step process, but I'm just going to show the final result.

Generally, in a database, table names are singular. User, Collection, Item. Also, it's less confusing to name your ID fields with the table name. It makes join SQL easier to read.

So let's start with the User table.

User
----
User ID
User Name
User Email
...

So far, so good. All of the columns have to do with a user.

Next, let's look at Collection

Collection
----------
Collection ID
Collection Name
Owner ID

So far, so good. The Owner ID is the User ID of the owner of the collection.

Next, let's look at Item.

Item
----
Item ID
Item Name
Item Description
...

So far, so good. All of the columns have to do with an item.

Now, let's look at the relationship between Collection and Item you described in problem 1. A Collection can have one or more items. An Item can be in one or more collections.

When you have a many-to-many relationship, you create a junction table. So let's create a CollectionItem junction table.

CollectionItem
--------------
CollectionItem ID
Collection ID
Item ID
CollectionItem timestamp
CollectionItem contributor ID

Where CollectionItem ID is the primary clustering key. You also have a unique index on (Collection ID, Item ID), so you can pull the collection together. You can also have a unique index on (Item ID, Collection ID), so you can see which items are in multiple collections.

You also have a unique index on (Collection ID, CollectionItem contributor ID). This allows you to see which contributor (user) contributed the item to the collection. This user ID could be the owner ID from the collections row or a different contributor.

Looking at problem 2, we need a Vote table. The Vote table is another junction table connecting a collection, item, and voter (user).

Vote
----
Vote ID
Collection ID
Item ID
Voter ID
Vote timestamp

Where Vote ID is the primary clustering key, and you have a unique index on (Collection ID, Item ID, Voter ID). You may also have two other unique indexes, depending on whether or not you want to group the votes by voter or item.

Edited to add:

You also need some kind of Permission table.

Permission
----------
Permission ID
Owner ID
Contributor ID

Where Permission ID is the primary clustering key and you have a unique index on (Owner ID, Contributor ID). You can also have a Permission Type, which you haven't defined, so I can't add it to the Permission table. The Permission Type would also be a part of the unique index.

I hope this explanation has been helpful.

PostgreSQL many-to-many relationship with some particular constrains

Create a unique constraint:

Alter table UserReaction
      add constraint one_reaction_per_customer_per_post
          unique ( customer_id, post_id); 
      

Postgresql: One-To-Many Relationship With Multiple Conditions On The Many Side

Does this solve your problem? I count all occurrences for your expected courses and the sum must be 2.

demo:db<>fiddle

SELECT
    s.id
FROM students s
JOIN courses c
ON c.student_id = s.id
GROUP BY s.id
HAVING SUM(
    ((c.course_name = 'math' AND c.credits = 5)
    OR
    (c.course_name = 'history' AND c.credits = 3))::int
) = 2

Alternative to the SUM(condition::int) you could use the COUNT() with a FILTER clause:

demo:db<>fiddle

HAVING COUNT(*) FILTER (WHERE
    (c.course_name = 'math' AND c.credits = 5)
    OR
    (c.course_name = 'history' AND c.credits = 3)
) = 2

---

Related Topics