Relations & grouping

[CrispenGari]

===
Dataloom 2.0.0

Release Notes - dataloom

We have release the new dataloom Version 2.0.0 (2024-02-12)

Features

• Renaming the class Dataloom to Loom.

  from dataloom import Loom

• Eager data fetching in relationships

  • Now you can fetch your child relationship together in your query

  user = mysql_loom.find_one(
    instance=User,
    filters=[Filter(column="id", value=userId)],
    include=[Include(model=Profile, select=["id", "avatar"], has="one")],
  )
  print(user)

  • You can apply limits, offsets, filters and orders to your child associations during queries

    post = mysql_loom.find_one(
      instance=Post,
      filters=[Filter(column="userId", value=userId)],
      select=["title", "id"],
      include=[
          Include(
              model=User,
              select=["id", "username"],
              has="one",
              include=[Include(model=Profile, select=["avatar", "id"], has="one")],
          ),
          Include(
              model=Category,
              select=["id", "type"],
              has="many",
              order=[Order(column="id", order="DESC")],
              limit=2,
          ),
      ],
    )

• Now return_dict has bee removed as an option in dataloom in the query functions like find_by_pk, find_one, find_many and find_all now works starting from this version. If you enjoy working with python objects you have to maneuver them manually using experimental features.

  from dataloom.decorators import initialize
  
  @initialize(repr=True, to_dict=True, init=True, repr_identifier="id")
  class Profile(Model):
      __tablename__: Optional[TableColumn] = TableColumn(name="profiles")
      id = PrimaryKeyColumn(type="int", auto_increment=True)
      avatar = Column(type="text", nullable=False)
      userId = ForeignKeyColumn(
          User,
          maps_to="1-1",
          type="int",
          required=True,
          onDelete="CASCADE",
          onUpdate="CASCADE",
      )
  
  # now you can do this
  
  profile = mysql_loom.find_many(
      instance=Profile,
  )
  print([Profile(**p) for p in profile]) # ? = [<Profile:id=1>]
  print([Profile(**p) for p in profile][0].id) # ? = 1

  • These are experimental decorators they are little bit slow and they work perfect in a single instance, you can not nest relationships on them.
  • You can use them if you know how your data is structured and also if you know how to manipulate dictionaries

• Deprecated join_next_filter_with to the use of join_next_with

• Values that was required as list e.g, select, include etc can now be passed as a single value.

  • Before

  res = mysql_loom.find_by_pk(Profile, pk=profileId, select={"id", "avatar"}) # invalid
  res = mysql_loom.find_by_pk(Profile, pk=profileId, select=("id", "avatar")) # invalid
  res = mysql_loom.find_by_pk(Profile, pk=profileId, select="id") # invalid
  res = mysql_loom.find_by_pk(Profile, pk=profileId, select=["id"]) # valid

  • Now

  res = mysql_loom.find_by_pk(Profile, pk=profileId, select={"id", "avatar"}) # valid
  res = mysql_loom.find_by_pk(Profile, pk=profileId, select=("id", "avatar")) # valid
  res = mysql_loom.find_by_pk(Profile, pk=profileId, select="id") # valid
  res = mysql_loom.find_by_pk(Profile, pk=profileId, select=["id"]) # valid

• Updated the documentation.

• Grouping data in queries will also be part of this release, using the class Group and Having.

posts = pg_loom.find_many(
          Post,
          select="id",
          filters=Filter(column="id", operator="gt", value=1),
          group=Group(
              column="id",
              function="MAX",
              having=Having(column="id", operator="in", value=(2, 3, 4)),
              return_aggregation_column=True,
          ),
)