def test_stress_wtrie(self):
ktrie = Trie()
strie = Trie()
etrie = Trie()
keywords = {}
search_terms = {}
exchange_ids = {}
with open(fixture) as f:
for data in f:
for word in data.split(' '):
vid = ktrie.add(word)
actual_vid = keywords.get(word)
if actual_vid is not None:
self.assertEqual(vid, actual_vid)
else:
keywords[word] = vid
vid = strie.add(data)
actual_vid = search_terms.get(data)
if actual_vid is not None:
self.assertEqual(vid, actual_vid)
else:
search_terms[data] = vid
nodes, kids, nodelen = etrie.serialize()
naddr, nlen = nodes.buffer_info()
kaddr, klen = kids.buffer_info()
#summarize(naddr, kaddr, nodelen)
#print_it(naddr, kaddr)
for dc, vid in exchange_ids.iteritems():
rvid = etrie.add(dc)
self.assertEqual(vid, rvid)
print dc, vid
value = value_for_vid(naddr, kaddr, vid)
self.assertEqual(dc, value)
if dc != value:
print " dc=%s adc=%s" % (dc, value)
avid = vid_for_value(naddr, kaddr, dc)
#print "vid=%s avid=%s" % (vid, avid)
self.assertEqual(vid, avid)
def test_stress_wtrie(self):
ktrie = Trie()
strie = Trie()
etrie = Trie()
keywords = {}
search_terms = {}
exchange_ids = {}
with open(fixture) as f:
for data in f:
for word in data.split(" "):
vid = ktrie.add(word)
actual_vid = keywords.get(word)
if actual_vid is not None:
self.assertEqual(vid, actual_vid)
else:
keywords[word] = vid
vid = strie.add(data)
actual_vid = search_terms.get(data)
if actual_vid is not None:
self.assertEqual(vid, actual_vid)
else:
search_terms[data] = vid
nodes, kids, nodelen = etrie.serialize()
naddr, nlen = nodes.buffer_info()
kaddr, klen = kids.buffer_info()
# summarize(naddr, kaddr, nodelen)
# print_it(naddr, kaddr)
for dc, vid in exchange_ids.iteritems():
rvid = etrie.add(dc)
self.assertEqual(vid, rvid)
print dc, vid
value = value_for_vid(naddr, kaddr, vid)
self.assertEqual(dc, value)
if dc != value:
print " dc=%s adc=%s" % (dc, value)
avid = vid_for_value(naddr, kaddr, dc)
# print "vid=%s avid=%s" % (vid, avid)
self.assertEqual(vid, avid)
def test_rtrie_in_memory(self):
s = unicode(u'séllsink').encode('utf-8')
#print "HELLSINK: %s" % s
t = Trie()
self.assertEqual(t.add('hello'), 1)
self.assertEqual(t.add('hell'), 2)
self.assertEqual(t.add('hello'), 1)
self.assertEqual(t.add('hellothere'), 3)
self.assertEqual(t.add('good'), 4)
self.assertEqual(t.add('goodbye'), 5)
self.assertEqual(t.add('hello'), 1)
self.assertEqual(t.add('hellsink'), 6)
self.assertEqual(t.add(s), 7)
t.print_it()
nodes, kids, _ = t.serialize()
nodeaddr, nodelen = nodes.buffer_info()
kidaddr, kidlen = kids.buffer_info()
print "LENS %s %s" % (nodelen, kidlen)
for i in range(8):
val = rtrie.value_for_vid(nodeaddr, kidaddr, i)
print "Value", i, val
self.assertEqual(rtrie.vid_for_value(nodeaddr, kidaddr, 'hello'), 1)
self.assertEqual(rtrie.vid_for_value(nodeaddr, kidaddr, 'hell'), 2)
self.assertEqual(rtrie.vid_for_value(nodeaddr, kidaddr, 'goodbye'), 5)
self.assertEqual(rtrie.vid_for_value(nodeaddr, kidaddr, 'hellsink'), 6)
self.assertEqual(rtrie.vid_for_value(nodeaddr, kidaddr, 'hellothere'),
3)
self.assertEqual(rtrie.vid_for_value(nodeaddr, kidaddr, 'good'), 4)
self.assertEqual(rtrie.vid_for_value(nodeaddr, kidaddr, s), 7)
self.assertIsNone(rtrie.vid_for_value(nodeaddr, kidaddr, 'notthere'))
self.assertIsNone(rtrie.vid_for_value(nodeaddr, kidaddr, 'h'))
self.assertIsNone(rtrie.vid_for_value(nodeaddr, kidaddr, 'he'))
self.assertIsNone(rtrie.vid_for_value(nodeaddr, kidaddr, 'hel'))
self.assertIsNone(rtrie.vid_for_value(nodeaddr, kidaddr, 'hells'))
def test_rtrie_in_memory(self):
s = unicode(u"séllsink").encode("utf-8")
# print "HELLSINK: %s" % s
t = Trie()
self.assertEqual(t.add("hello"), 1)
self.assertEqual(t.add("hell"), 2)
self.assertEqual(t.add("hello"), 1)
self.assertEqual(t.add("hellothere"), 3)
self.assertEqual(t.add("good"), 4)
self.assertEqual(t.add("goodbye"), 5)
self.assertEqual(t.add("hello"), 1)
self.assertEqual(t.add("hellsink"), 6)
self.assertEqual(t.add(s), 7)
t.print_it()
nodes, kids, _ = t.serialize()
nodeaddr, nodelen = nodes.buffer_info()
kidaddr, kidlen = kids.buffer_info()
print "LENS %s %s" % (nodelen, kidlen)
for i in range(8):
val = rtrie.value_for_vid(nodeaddr, kidaddr, i)
print "Value", i, val
self.assertEqual(rtrie.vid_for_value(nodeaddr, kidaddr, "hello"), 1)
self.assertEqual(rtrie.vid_for_value(nodeaddr, kidaddr, "hell"), 2)
self.assertEqual(rtrie.vid_for_value(nodeaddr, kidaddr, "goodbye"), 5)
self.assertEqual(rtrie.vid_for_value(nodeaddr, kidaddr, "hellsink"), 6)
self.assertEqual(rtrie.vid_for_value(nodeaddr, kidaddr, "hellothere"), 3)
self.assertEqual(rtrie.vid_for_value(nodeaddr, kidaddr, "good"), 4)
self.assertEqual(rtrie.vid_for_value(nodeaddr, kidaddr, s), 7)
self.assertIsNone(rtrie.vid_for_value(nodeaddr, kidaddr, "notthere"))
self.assertIsNone(rtrie.vid_for_value(nodeaddr, kidaddr, "h"))
self.assertIsNone(rtrie.vid_for_value(nodeaddr, kidaddr, "he"))
self.assertIsNone(rtrie.vid_for_value(nodeaddr, kidaddr, "hel"))
self.assertIsNone(rtrie.vid_for_value(nodeaddr, kidaddr, "hells"))
def _fetch_vid(vid, vid16_nodes, vid16_kids, vid_nodes, vid_kids):
return rtrie.value_for_vid(vid_nodes, vid_kids, vid)
def _fetch_vid(vid, vid16_nodes, vid16_kids, vid_nodes, vid_kids):
return rtrie.value_for_vid(vid_nodes, vid_kids, vid)