def test_termindex():
terms = [("a", "alfa"), ("a", "bravo"), ("a", "charlie"), ("a", "delta"),
("b", "able"), ("b", "baker"), ("b", "dog"), ("b", "easy")]
st = RamStorage()
tw = TermIndexWriter(st.create_file("test.trm"))
for i, t in enumerate(terms):
tw.add(t, FileTermInfo(1.0, i))
tw.close()
tr = TermIndexReader(st.open_file("test.trm"))
for i, (t1, t2) in enumerate(zip(tr.keys(), terms)):
assert_equal(t1, t2)
ti = tr.get(t1)
assert_equal(ti.weight(), 1.0)
assert_equal(ti.doc_frequency(), i)
def test_termindex():
terms = [("a", "alfa"), ("a", "bravo"), ("a", "charlie"), ("a", "delta"),
("b", "able"), ("b", "baker"), ("b", "dog"), ("b", "easy")]
st = RamStorage()
tw = TermIndexWriter(st.create_file("test.trm"))
for i, t in enumerate(terms):
tw.add(t, FileTermInfo(1.0, i))
tw.close()
tr = TermIndexReader(st.open_file("test.trm"))
for i, (t1, t2) in enumerate(zip(tr.keys(), terms)):
assert_equal(t1, t2)
ti = tr.get(t1)
assert_equal(ti.weight(), 1.0)
assert_equal(ti.doc_frequency(), i)
def test_random_termkeys():
def random_fieldname():
return "".join(chr(random.randint(65, 90)) for _ in xrange(1, 20))
def random_token():
return "".join(unichr(random.randint(0, 0xd7ff)) for _ in xrange(1, 20))
domain = sorted([(random_fieldname(), random_token()) for _ in xrange(1000)])
st = RamStorage()
tw = TermIndexWriter(st.create_file("test.trm"))
for term in domain:
tw.add(term, FileTermInfo(1.0, 1))
tw.close()
tr = TermIndexReader(st.open_file("test.trm"))
for term in domain:
assert term in tr
def test_random_termkeys():
def random_fieldname():
return "".join(chr(random.randint(65, 90)) for _ in xrange(1, 20))
def random_token():
return "".join(
unichr(random.randint(0, 0xd7ff)) for _ in xrange(1, 20))
domain = sorted([(random_fieldname(), random_token())
for _ in xrange(1000)])
st = RamStorage()
tw = TermIndexWriter(st.create_file("test.trm"))
for term in domain:
tw.add(term, FileTermInfo(1.0, 1))
tw.close()
tr = TermIndexReader(st.open_file("test.trm"))
for term in domain:
assert term in tr
def test_termkey():
with TempStorage("termkey") as st:
tw = TermIndexWriter(st.create_file("test.trm"))
tw.add(("alfa", u("bravo")), FileTermInfo(1.0, 3))
tw.add(("alfa", u('\xc3\xa6\xc3\xaf\xc5\ufffd\xc3\xba')), FileTermInfo(4.0, 6))
tw.add(("text", u('\xe6\u2014\xa5\xe6\u0153\xac\xe8\xaa\u017e')), FileTermInfo(7.0, 9))
tw.close()
tr = TermIndexReader(st.open_file("test.trm"))
assert ("alfa", u("bravo")) in tr
assert ("alfa", u('\xc3\xa6\xc3\xaf\xc5\ufffd\xc3\xba')) in tr
assert ("text", u('\xe6\u2014\xa5\xe6\u0153\xac\xe8\xaa\u017e')) in tr
tr.close()
def test_termkey():
with TempStorage("termkey") as st:
tw = TermIndexWriter(st.create_file("test.trm"))
tw.add(("alfa", u("bravo")), FileTermInfo(1.0, 3))
tw.add(("alfa", u('\xc3\xa6\xc3\xaf\xc5\ufffd\xc3\xba')),
FileTermInfo(4.0, 6))
tw.add(("text", u('\xe6\u2014\xa5\xe6\u0153\xac\xe8\xaa\u017e')),
FileTermInfo(7.0, 9))
tw.close()
tr = TermIndexReader(st.open_file("test.trm"))
assert ("alfa", u("bravo")) in tr
assert ("alfa", u('\xc3\xa6\xc3\xaf\xc5\ufffd\xc3\xba')) in tr
assert ("text", u('\xe6\u2014\xa5\xe6\u0153\xac\xe8\xaa\u017e')) in tr
tr.close()
