Algen
Algen generates opionated ORM classes for sqlalchemy given a simple schema
either as a commandline string or as a yaml file. It is designed to have
minimal dependencies and is trivially extensible. A command line tool
is bundled along to help generate the models. For DB specific types,
only postgres is currently supported. The tool currently assumes that
sqlalchemy's declarative base object is to be imported like
from .alchemy_base import Base
The library prefers making the code verbose rather than concise for the
sake of having better auto-completion and help from your editor/IDE
e.g. first style is preferred to the second one:
def update(self, a=None, b=None):
if a is not None:
self.a = a
if b is not None:
self.b = bdef update(self, **kwargs):
for k,v in kwargs.items():
if hasattr(self, k) and v is not None:
setattr(self, k, v)CLI
$ algen --help
Usage: algen [OPTIONS]
Options:
-n, --name TEXT Name of model
-c, --columns TEXT Column definition. e.g. col_name:col_type Can be
used multiple times hence named columns. e.g. -c
foo:Int -c bar:Unicode(20)
-d, --destination PATH Destination directory. Default will assume 'models'
directory inside the current working directory
-y, --yaml PATH Yaml file describing the Model. This supersedes the
column definition provided through --columns option.
--help Show this message and exit.Given a file as follows:
Person:
columns:
- name: id
type: BigInteger
primary_key: True
auto_increment: True
- name: name
type: Unicode(255)
- name: is_vip
type: Boolean
- name: created_at
type: DateTime(timezone=True)
server_default: now() at time zone 'utc'
- name: updated_at
type: DateTime(timezone=True)
server_default: now() at time zone 'utc'
relationships:
- name: addresses
class: Address
back_populates: 'person'
Address:
columns:
- name: id
type: BigInteger
primary_key: True
auto_increment: True
- name: line1
type: Unicode()
- name: line2
type: Unicode()
- name: line3
type: Unicode()
- name: postcode
type: Unicode(10)
index: True
- name: created_at
type: DateTime(timezone=True)
server_default: now() at time zone 'utc'
- name: updated_at
type: DateTime(timezone=True)
server_default: now() at time zone 'utc'
foreign_keys:
- name: person_id
type: BigInteger
reference:
table: persons
column: id
nullable: False
relationships:
- name: person
class: Person
back_populates: 'addresses'
The cli tool will create two the following two files person.py and address.py.
from __future__ import unicode_literals, absolute_import, print_function
from collections import namedtuple
from sqlalchemy import Column, Unicode, BigInteger, DateTime, ForeignKey
from sqlalchemy.orm import relationship
from .alchemy_base import Base
__author__ = 'danishabdullah'
class Address(Base):
__tablename__ = 'addresses'
id = Column(BigInteger, auto_increment=True, primary_key=True)
line1 = Column(Unicode(), )
line2 = Column(Unicode(), )
line3 = Column(Unicode(), )
postcode = Column(Unicode(10), index=True)
created_at = Column(DateTime(timezone=True), server_default="now() at time zone 'utc'")
updated_at = Column(DateTime(timezone=True), server_default="now() at time zone 'utc'")
# --- Foreign Keys ---
person_id = Column(BigInteger, ForeignKey('persons.id'), nullable=False)
# --- Relationships ---
person = relationship('Person', back_populates='addresses')
def __init__(self, id=None, line1=None, line2=None, line3=None, postcode=None, created_at=None, updated_at=None, person_id=None):
self.id = id
self.line1 = line1
self.line2 = line2
self.line3 = line3
self.postcode = postcode
self.created_at = created_at
self.updated_at = updated_at
self.person_id = person_id
def add(self, session):
session.add(self)
def update(self, line1=None, line2=None, line3=None, postcode=None, created_at=None, updated_at=None, person_id=None):
# This function only updates a value if it is not None.
# Falsy values go through in the normal way.
# To set things to None use the usual syntax:
# Address.column_name = None
if line1 is not None:
self.line1 = line1
if line2 is not None:
self.line2 = line2
if line3 is not None:
self.line3 = line3
if postcode is not None:
self.postcode = postcode
if created_at is not None:
self.created_at = created_at
if updated_at is not None:
self.updated_at = updated_at
if person_id is not None:
self.person_id = person_id
def delete(self, session):
session.delete(self)
def to_dict(self):
return {x: y for x, y in self.__dict__.items() if not x.startswith("_sa")}
def get_proxy_cls(self):
# AddressProxy is useful when you want to persist data
# independent of the sqlalchemy session. It's just a namedtuple
# that has very low memory/cpu footprint compared the regular
# orm class instances.
keys = self.to_dict().keys()
name = "AddressProxy"
return namedtuple(name, keys)
def to_proxy(self):
# Proxy-ing is useful when you want to persist data
# independent of the sqlalchemy session. It's just a namedtuple
# that has very low memory/cpu footprint compared the regular
# orm class instances.
cls = self._get_proxy_cls()
return cls(**self.to_dict())
@classmethod
def from_proxy(cls, proxy):
return cls(**proxy._asdict())
def __hash__(self):
return hash(str(self.id))
def __eq__(self, other):
return (self.id == other.id)
def __neq__(self, other):
return not (self.id == other.id)
def __str__(self):
return "<Address: {id}>".format(id=self.id)
def __unicode__(self):
return "<Address: {id}>".format(id=self.id)
def __repr__(self):
return "<Address: {id}>".format(id=self.id)
from __future__ import unicode_literals, absolute_import, print_function
from collections import namedtuple
from sqlalchemy import Column, Unicode, BigInteger, Boolean, DateTime
from sqlalchemy.orm import relationship
from .alchemy_base import Base
__author__ = 'danishabdullah'
class Person(Base):
__tablename__ = 'persons'
id = Column(BigInteger, auto_increment=True, primary_key=True)
name = Column(Unicode(255), )
is_vip = Column(Boolean, )
created_at = Column(DateTime(timezone=True), server_default="now() at time zone 'utc'")
updated_at = Column(DateTime(timezone=True), server_default="now() at time zone 'utc'")
# --- Foreign Keys ---
# --- Relationships ---
addresses = relationship('Address', back_populates='person')
def __init__(self, id=None, name=None, is_vip=None, created_at=None, updated_at=None):
self.id = id
self.name = name
self.is_vip = is_vip
self.created_at = created_at
self.updated_at = updated_at
def add(self, session):
session.add(self)
def update(self, name=None, is_vip=None, created_at=None, updated_at=None):
# This function only updates a value if it is not None.
# Falsy values go through in the normal way.
# To set things to None use the usual syntax:
# Person.column_name = None
if name is not None:
self.name = name
if is_vip is not None:
self.is_vip = is_vip
if created_at is not None:
self.created_at = created_at
if updated_at is not None:
self.updated_at = updated_at
def delete(self, session):
session.delete(self)
def to_dict(self):
return {x: y for x, y in self.__dict__.items() if not x.startswith("_sa")}
def get_proxy_cls(self):
# PersonProxy is useful when you want to persist data
# independent of the sqlalchemy session. It's just a namedtuple
# that has very low memory/cpu footprint compared the regular
# orm class instances.
keys = self.to_dict().keys()
name = "PersonProxy"
return namedtuple(name, keys)
def to_proxy(self):
# Proxy-ing is useful when you want to persist data
# independent of the sqlalchemy session. It's just a namedtuple
# that has very low memory/cpu footprint compared the regular
# orm class instances.
cls = self._get_proxy_cls()
return cls(**self.to_dict())
@classmethod
def from_proxy(cls, proxy):
return cls(**proxy._asdict())
def __hash__(self):
return hash(str(self.id))
def __eq__(self, other):
return (self.id == other.id)
def __neq__(self, other):
return not (self.id == other.id)
def __str__(self):
return "<Person: {id}>".format(id=self.id)
def __unicode__(self):
return "<Person: {id}>".format(id=self.id)
def __repr__(self):
return "<Person: {id}>".format(id=self.id)
