Benchmarking the new Memory2 rdflib triplestore implementation

benchmark.py

import random
from string import ascii_uppercase, ascii_lowercase, digits
from rdflib import Graph, ConjunctiveGraph, URIRef, BNode, Literal
import timeit

ALPHABET = ascii_uppercase + ascii_lowercase + digits

def make_word(length=8):
    return ''.join(random.choice(ALPHABET) for _ in range(length))

def make_random_uriref():
    return URIRef("http://"
                  + make_word()
                  + (".com/" if random.choice("CO") == "C" else ".org/")
                  + make_word()
                  + ("#" if random.choice("HS") == "H" else "/")
                  + make_word())

def make_random_literal():
    l_type = random.choice("SIB")
    if l_type == "S":
        return Literal(make_word())
    elif l_type == "I":
        return Literal(random.randint(1, 1000000))
    else:
        return Literal(random.choice([True, False]))

def make_triple():
    # subject is a bnode or uri
    if random.choice('BU') == 'B':
        s = BNode()
    else:
        s = make_random_uriref()
    # predicate is a uri
    p = make_random_uriref()
    # object is a bnode or uri or literal
    p_choice = random.choice('BUL')
    if p_choice == "B":
        o = BNode()
    elif p_choice == "U":
        o = make_random_uriref()
    else:
        o = make_random_literal()

    return (s, p, o)

global_triples = [make_triple() for _ in range(100000)]

_setup = r"""

triples = global_triples

"""

_run1 = r"""
g = Graph("Memory1")
for t in triples:
    g.add(t)

_ = list(g.triples((None, None, None)))

g.remove((None, None, None))
del g
"""

_run2 = r"""
g = Graph("Memory2")
for t in triples:
    g.add(t)

_ = list(g.triples((None, None, None)))

g.remove((None, None, None))
del g
"""

_run3 = r"""
g = Graph("IOMemory")
for t in triples:
    g.add(t)

_ = list(g.triples((None, None, None)))

g.remove((None, None, None))
del g
"""

a = timeit.timeit(_run1, setup=_setup, number=10, globals={"Graph": Graph, "global_triples": global_triples})
b = timeit.timeit(_run2, setup=_setup, number=10, globals={"Graph": Graph, "global_triples": global_triples})
c = timeit.timeit(_run3, setup=_setup, number=10, globals={"Graph": Graph, "global_triples": global_triples})
print(a/10.0)
print(b/10.0)
print(c/10.0)