A Python library for the Datomic REST API.
While nothing will ever match the speed or power of Datomic's Clojure interface, this library will work well for the basics.
pip install datomicTo connect to database "test" in store "mem" on localhost at port 8888:
from datomic import *
db = DB('localhost', 8888, 'mem', 'test')
# create the database
db.create()
True
# get the state
db.info()
{'basis-t': 62, 'db/alias': 'mem/test'}You do not need to define a schema. If you want to, it is just a bunch of nested tuples.
Unless specified otherwise, each attribute is assumed to be a string with a cardinality of one.
S=(
('person',
('name', FULL, "A Persons name"),
('email', FULL, "A Person's email"),
('age', LONG, "A Person's age"),
('likes', REF, MANY, ISCOMP, "A Persons likes"),
),
('item',
('name', FULL),
('sku', UNIQ),
('active', BOOL),
('cat', MANY, ENUM('cat','dog','pony','horse','gerbil','sloth')),
),
('review',
('person', REF),
('item', REF),
('stars', LONG),
),
)You can pass the schema to DB when connecting.
db = DB(host, port, store, dbname, schema=S)
# get the schema data
db.schema.schema
['{:db/id #db/id[:db.part/db]
:db/ident :person/name
:db/fulltext true
:db/doc "A Persons name"
:db/valueType :db.type/string
:db/cardinality :db.cardinality/one
:db.install/_attribute :db.part/db}',
'...',
]
# transact the schema
db.tx_schema()
For a more comprehensive schema example, see datomic/datomic_test.py
# send raw edn string to db.tx() to transact
resp = db.tx('{:db/id #db/id[:db.part/user] :person/name "Bob"}')
{'db-before': {}, 'db-after': {}, 'tempids': [] }
# or, start a new transaction to accumulate many datums by calling db.tx() with no args
tx = db.tx()
# `person` will hold a tempid and resolve to an entity after the tx is executed
person = tx.add("person/", {
'name': "John Doe" ,
'age': 25,
})
# using "ns/" followed by saves some typing
item = tx.add("item/", {
'name': 'Item 1',
'sku': 'item-1-sku',
'active': True,
'cat': ['cat','dog'],
})
# another new entity, with a ref to our `person`
review = tx.add("review/", {
'item': item,
'stars': 4,
'person': person,
})
# we can nest a new entity in another entity
review2 = tx.add("review/", {
'item': item,
'stars': 5,
'person': tx.add("person/", {
'name': 'Nested Person',
'age': 22,
}),
})
# add another datum to `person`
tx.add(person, 'person/likes', item)
# does exactly the same thing as the previous example
person.add('person/likes', item)
# see our tempids so far
print person, item, review, review2
{'db/id': -1} {'db/id': -2} {'db/id': -4} {'db/id': -6}
# send the tx to datomic
tx.execute()
{'db-after': {'basis-t': 1042, 'db/alias': 'mem/test'},
'db-before': {'basis-t': 1040, 'db/alias': 'mem/test'},
'tx-data': [{'a': 50, 'added': True, 'e': 13194139534354, 'tx': 13194139534354, 'v': datetime.datetime(2013, 11, 9, 18, 55, 56, 657000, tzinfo=<UTC>)}, {'....'}]
}
# all entity ids are automatically resolved
print person, item, review, review2
{'db/id': 17592186045459} {'db/id': 17592186045460} {'db/id': 17592186045462} {'db/id': 17592186045464}
# access to the entity ids
print person.eid, item.eid, review.eid, review2.eid
17592186045459 17592186045460 17592186045462 17592186045464
# edn format
print unicode(person)
#db/id[:db.part/user 17592186045459] # fetch an entity
db.e(person)
{'person/age': 25, 'person/likes': ({'item/name': 'Item 1', 'item/sku': 'item-1-sku', 'item/cat': set(['dog', 'cat']), 'item/active': True, 'db/id': 17592186045460},), 'db/id': 17592186045459, 'person/name': 'John Doe'}
db.e(item.eid)
{'item/name': 'Item 1', 'item/sku': 'item-1-sku', 'item/cat': set(['dog', 'cat']), 'item/active': True, 'db/id': 17592186045460}
db.e(17592186045462)
{'review/person': {'db/id': 17592186045459}, 'review/stars': 4, 'db/id': 17592186045462, 'review/item': {'db/id': 17592186045460}}
# add datums to an entity
tx2 = db.tx()
person2 = tx2.add(person, 'person/email', '[email protected]')
tx2.execute()
person == person2
False
db.e(person2)
{'person/age': 25, 'person/email': '[email protected]', 'db/id': 17592186045440, 'person/name': 'John Doe', 'person/likes': ({'item/name': 'Item 1', 'item/sku': 'item-1-sku', 'item/cat': set(['dog', 'cat']), 'item/active': True, 'db/id': 17592186045441},)}
# send a edn string to db.q()
db.q('[...]')
# or, use db.find() to build a query in a more pythonic way
p_name = '?e :person/name ?n'
p_age = '?e :person/age ?a'
p_email = '?e :person/email ?m'
# get one
db.find('?e ?n').where(p_name).one()
[17592186045457, 'John Doe']
# one to dict
p = db.find('?e ?n').where(p_name,p_age).hashone()
p.items()
[('e', 17592186045457), ('n', 'John Doe')]
# OR input param
qa = db.find('?e ?n ?a').where(p_name, p_age)\
.param('?n', ['Nested Person', 'John Doe'])
qa.all()
[[17592186045463, 'Nested Person', 22], [17592186045459, 'John Doe', 25]]
qa.limit(1).all()
[[17592186045463, 'Nested Person', 22]]
# AND input param
qb = db.find('?e ?n ?a').where(p_name, p_age)\
.param('?n ?a', ('John Doe', 25))
qb.all()
[17592186045459, 'John Doe', 25]]
# unify external data
qc = db.find('?e ?n ?external').where(p_name, p_age)\
.param('?n ?external',
[ ['John Doe', 123.23], ['Nested Person', 456.00]])
qc.all()
[[17592186045459, 'John Doe', 123.23], [17592186045463, 'Nested Person', 456.0]]db.e(review2).get('review/stars')
5
db.retract(review2, 'review/stars', 5)
db.e(review2)
{'review/person': {'db/id': 17592186045463}, 'db/id': 17592186045464, 'review/item': {'db/id': 17592186045460}}db.datums() lazily fetches datums in the chunk size you specify.
for r in db.datoms('aevt', a='person/name', limit=100, chunk=100):
print r
{'a': 62, 'added': True, 'e': 17592186045459, 'tx': 13194139534354, 'v': 'John Doe'}
{'a': 62, 'added': True, 'e': 17592186045463, 'tx': 13194139534354, 'v': 'Nested Person'}for r in db.datoms('avet', a='item/sku', v='item-1-sku', limit=100):
print r
{'a': 67, 'added': True, 'e': 17592186045460, 'tx': 13194139534354, 'v': 'item-1-sku'}-
A python library for the C++ edn parser is in progress and should be more performant is in the works.
-
More test coverage
-
Better support for traversing the graph
-
Eager loading of entities
-
Materialized Views