def test_hdt_simple_scan(subj, pred, obj, expected):
with HDTFileConnector('tests/data/watdiv.10M.hdt') as backend:
iterator, c = backend.search(subj, pred, obj)
assert iterator.has_next()
while iterator.has_next() and len(expected) > 0:
assert_next_triple(iterator, expected)
assert not iterator.has_next()
assert len(expected) == 0
def test_hdt_resume_scan(subj, pred, obj, expected):
# don't test for scan that yield one matching RDF triple
if len(expected) > 1:
with HDTFileConnector('tests/data/watdiv.10M.hdt') as backend:
iterator, c = backend.search(subj, pred, obj)
assert iterator.has_next()
# read first triple, then stop and reload a new iterator
assert_next_triple(iterator, expected)
last_read = iterator.last_read()
iterator, c = backend.search(subj, pred, obj, last_read=last_read)
while iterator.has_next() and len(expected) > 0:
assert_next_triple(iterator, expected)
assert not iterator.has_next()
assert len(expected) == 0
# scan_test.py
# Author: Thomas MINIER - MIT License 2017-2020
import pytest
from sage.query_engine.sage_engine import SageEngine
from sage.query_engine.iterators.scan import ScanIterator
from sage.query_engine.iterators.projection import ProjectionIterator
from sage.database.hdt.connector import HDTFileConnector
hdtDoc = HDTFileConnector('tests/data/test.hdt')
engine = SageEngine()
triple = {
'subject': '?s1',
'predicate': 'http://example.org/p1',
'object': '?common',
'graph': 'watdiv100'
}
@pytest.mark.asyncio
async def test_projection_read():
iterator, card = hdtDoc.search(triple['subject'], triple['predicate'],
triple['object'])
scan = ScanIterator(iterator, triple, card)
proj = ProjectionIterator(scan, ['?common'])
(results, saved, done, _) = await engine.execute(proj, 10e7)
assert len(results) == card
for res in results:
assert '?common' in res and '?s1' not in res
assert done
def test_hdt_scan_unknown_pattern():
with HDTFileConnector('tests/data/watdiv.10M.hdt') as backend:
iterator, c = backend.search('http://example.org#toto', None, None)
assert next(iterator) is None
# filter_iterator_test.py
# Author: Thomas MINIER - MIT License 2017-2020
import pytest
from sage.query_engine.sage_engine import SageEngine
from sage.query_engine.iterators.scan import ScanIterator
from sage.query_engine.iterators.filter import FilterIterator
from sage.query_engine.iterators.projection import ProjectionIterator
from sage.query_engine.iterators.loader import load
from sage.database.hdt.connector import HDTFileConnector
from tests.utils import DummyDataset
import math
hdtDoc = HDTFileConnector('tests/data/watdiv.10M.hdt')
engine = SageEngine()
triple = {
'subject': 'http://db.uwaterloo.ca/~galuc/wsdbm/Offer1000',
'predicate': '?p',
'object': '?o',
'graph': 'watdiv100'
}
@pytest.mark.asyncio
async def test_simple_filter_iterator():
expression = "?p = <http://schema.org/eligibleRegion>"
iterator, card = hdtDoc.search(triple['subject'], triple['predicate'],
triple['object'])
scan = ProjectionIterator(ScanIterator(iterator, triple, card))
iterator = FilterIterator(scan, expression)
(results, saved, done, _) = await engine.execute(iterator, math.inf)
assert len(results) == 4
def test_hdt_scan_unknown_pattern():
with HDTFileConnector('tests/data/watdiv.10M.hdt') as backend:
with pytest.raises(StopIteration):
iterator, c = backend.search('http://example.org#toto', None, None)
next(iterator)
# scan_test.py
# Author: Thomas MINIER - MIT License 2017-2020
import pytest
import asyncio
from rdflib.plugins.sparql.algebra import translateQuery, translateUpdate
from rdflib.plugins.sparql.parser import parseQuery, parseUpdate
from sage.query_engine.sage_engine import SageEngine
from sage.database.hdt.connector import HDTFileConnector
from sage.query_engine.optimizer.query_parser import parse_query
from tests.utils import DummyDataset
hdtDoc = HDTFileConnector('tests/data/context.hdt')
dataset = DummyDataset(hdtDoc, 'context')
engine = SageEngine()
queries = [("""
select ?md5 where {
?s <http://isa> ?o
BIND(URI(CONCAT("http://",MD5(CONCAT(STR(?s),STR(<http://isa>),STR(?o))))) as ?md5)
}
""", 5),
("""
select ?o1 where {
?s <http://isa> ?o
BIND(URI(CONCAT("http://",MD5(CONCAT(STR(?s),STR(<http://isa>),STR(?o))))) as ?md5)
?md5 <http://source> ?o1
}
""", 5),
("""
select ?md5 where {