def test_signModification_1(self): x = hamDb.explicitUnsignCast(5) x = hamDb.explicitSignCast(x) self.assertEqual(x, 5) tgtHash = -6076574518398440533 for hashVal in [data[1] for data in TEST_DATA]: x = hamDb.explicitUnsignCast(hashVal) x = hamDb.explicitSignCast(x) self.assertEqual(hashVal, x)
def test_signModification_1(self):
x = hamDb.explicitUnsignCast(5)
x = hamDb.explicitSignCast(x)
self.assertEqual(x, 5)
tgtHash = -6076574518398440533
for hashVal in [data[1] for data in TEST_DATA]:
x = hamDb.explicitUnsignCast(hashVal)
x = hamDb.explicitSignCast(x)
self.assertEqual(hashVal, x)
def test_signModification_2(self): for val in TEST_DATA_FLAT: tgtHash = b2i(val) x = hamDb.explicitUnsignCast(tgtHash) self.assertEqual(x, b2u(val)) x = hamDb.explicitSignCast(x) self.assertEqual(x, tgtHash)
def test_signModification_2(self):
for val in TEST_DATA_FLAT:
tgtHash = b2i(val)
x = hamDb.explicitUnsignCast(tgtHash)
self.assertEqual(x, b2u(val))
x = hamDb.explicitSignCast(x)
self.assertEqual(x, tgtHash)
def do_test_queries(self):
print("Doing test queries")
for x in range(TEST_QUERIES):
fk_hash = random.getrandbits(64)
fk_hash = hamDb.explicitSignCast(fk_hash)
self.tree.getWithinDistance(fk_hash, SEARCH_DIST)
print("Test queries complete")
def do_test_queries(self):
print("Doing test queries")
for x in range(TEST_QUERIES):
fk_hash = random.getrandbits(64)
fk_hash = hamDb.explicitSignCast(fk_hash)
self.tree.getWithinDistance(fk_hash, SEARCH_DIST)
print("Test queries complete")
def test_signModification_1(self):
self.log.info("test_signModification_1")
for val in TEST_DATA_Narrow:
tgtHash = b2i(val)
x = hamDb.explicitUnsignCast(tgtHash)
self.assertEqual(x, b2u(val))
x = hamDb.explicitSignCast(x)
self.assertEqual(x, tgtHash)
def test_signModification_3(self):
random.seed(500000)
for x in range(50000):
val = "".join([random.choice(("1", "0")) for z in range(64)])
tgtHash = b2i(val)
x = hamDb.explicitUnsignCast(tgtHash)
self.assertEqual(x, b2u(val))
x = hamDb.explicitSignCast(x)
self.assertEqual(x, tgtHash)
def test_signModification_3(self):
random.seed(500000)
for x in range(50000):
val = "".join([random.choice(("1", "0")) for z in range(64)])
tgtHash = b2i(val)
x = hamDb.explicitUnsignCast(tgtHash)
self.assertEqual(x, b2u(val))
x = hamDb.explicitSignCast(x)
self.assertEqual(x, tgtHash)
def buildOntoTree(self, tree):
# Fix the random ordering so profile results
# are sensible
random.seed(65413654911)
print("Building test-tree with %s nodes" % TREE_ENTRIES)
for x in range(TREE_ENTRIES):
fk_hash = random.getrandbits(64)
fk_hash = hamDb.explicitSignCast(fk_hash)
tree.insert(fk_hash, x)
if x % 10000 == 0:
print("Inserted %s items, %s%% complete" % (x, (x / TREE_ENTRIES * 100)))
print("Tree built")
def buildOntoTree(self, tree):
# Fix the random ordering so profile results
# are sensible
random.seed(65413654911)
print("Building test-tree with %s nodes" % TREE_ENTRIES)
for x in range(TREE_ENTRIES):
fk_hash = random.getrandbits(64)
fk_hash = hamDb.explicitSignCast(fk_hash)
tree.insert(fk_hash, x)
if x % 10000 == 0:
print("Inserted %s items, %s%% complete" %
(x, (x / TREE_ENTRIES * 100)))
print("Tree built")
# from cbktree import BkHammingTree, explicitSignCast
import unittest
import logging
import scanner.logSetup as logSetup
from bitstring import Bits
import pyximport
print("Have Cython")
pyximport.install()
import Tests.baseDbBkTree
import deduplicator.cyHamDb as hamDb
int_bits = lambda b: hamDb.explicitSignCast(int(b, 2))
TEST_DATA = {
# Format: id -> bitstring
1: int_bits('1011010010010110110111111000001000001000100011110001010110111011'),
2: int_bits('1011010010010110110111111000001000000001100011110001010110111011'),
3: int_bits('1101011110100100001011001101001110010011100010011101001000110101'),
}
SEARCH_DIST = 2 # 2 out of 64 bits
class TestBkTreeIssue2(unittest.TestCase):
def __init__(self, *args, **kwargs):
logSetup.initLogging()
# from cbktree import BkHammingTree, explicitSignCast
import pyximport
print("Have Cython")
pyximport.install()
import deduplicator.cyHamDb as hamDb
int_bits = lambda b: hamDb.explicitSignCast(int(b, 2))
TEST_DATA = {
# Format: id -> bitstring
1:
int_bits(
'1011010010010110110111111000001000001000100011110001010110111011'),
2:
int_bits(
'1011010010010110110111111000001000000001100011110001010110111011'),
3:
int_bits(
'1101011110100100001011001101001110010011100010011101001000110101'),
}
SEARCH_DIST = 2 # 2 out of 64 bits
import unittest
import scanner.logSetup as logSetup
from bitstring import Bits
import pyximport
print("Have Cython")
def random_q(self): fk_hash = random.getrandbits(64) fk_hash = hamDb.explicitSignCast(fk_hash) self.tree.getWithinDistance(fk_hash, SEARCH_DIST)
def random_q(self):
fk_hash = random.getrandbits(64)
fk_hash = hamDb.explicitSignCast(fk_hash)
self.tree.getWithinDistance(fk_hash, SEARCH_DIST)
