def check_domain(self, obj, experiment):
"""
Either get or create assignment.
"""
from honest_ab.models import ExperimentDomainAllocation
from honest_ab.models import ExperimentDomain
try:
object_name = '.'.join([str(obj.__class__.__module__), str(obj.__class__.__name__)])
domain_allocation = ExperimentDomainAllocation.objects.get(
model_pk=obj.pk,
model=object_name
)
except ExperimentDomainAllocation.DoesNotExist:
num_domains = ExperimentDomain.objects.filter(active=1).count()
order = murmur.string_hash(str(obj.pk), domain_hash_key) % num_domains
domain = ExperimentDomain.objects.filter(active=1)[order]
try:
ExperimentDomainAllocation.objects.create(
experiment_domain=domain,
model=object_name,
model_pk=obj.pk,
)
except IntegrityError:
# This can occur in high traffic instances where two threads hit this method
# at the same time. Both will fail the get and both will try to create.
pass
return domain.pk == experiment.domain_id
else:
return domain_allocation.experiment_domain_id == experiment.domain_id
def check_fetched(self, bitmap, url):
mhash = murmur.string_hash(url)
if self.r.getbit(bitmap, mhash):
return True
else:
self.r.setbit(bitmap, mhash, 1)
return False
def check_fetched(url):
r = redis.Redis(connection_pool=POOL)
mhash = murmur.string_hash(url)
if r.getbit(config.BITMAP, mhash):
return True
else:
r.setbit(config.BITMAP, mhash, 1)
return False
def _make_decision(self, obj, experiment):
"""
Use a simple hash on the string value of the object pk.
"""
if random.random() * 100 > experiment.percentage_of_traffic:
result = HONEST_AB_SKIP_TYPE
else:
result = str(murmur.string_hash(str(obj.pk), experiment.pk) % experiment.number_of_classes)
return result
def extract_vector(features, bits):
vector = {}
for (fset, vals) in features.items():
for (f, v) in vals.items():
try:
h = murmur.string_hash("%s_%s" % (fset, f.encode("utf-8")))
h = h & (2 ** bits - 1)
vector[h + 1] = v
except UnicodeDecodeError, e:
print >> sys.stderr, "%s: %s" % (e, (f, v))
continue
def _make_decision(self, obj, experiment):
"""
Use a simple hash on the string value of the object pk.
"""
if random.random() * 100 > experiment.percentage_of_traffic:
result = HONEST_AB_SKIP_TYPE
else:
result = str(
murmur.string_hash(str(obj.pk), experiment.pk) %
experiment.number_of_classes)
return result
def minhash(features):
minhashes = []
sketches = []
#for feature in features:
#print("feature")
for num_minhash in range(NUM_MINHASHES):
minhashes.append(
min([murmur.string_hash(`feature`, num_minhash) for feature in features])
)
#print("this is the min hash array: ", minhashes)
#raw_input("\npress any key...")
#continue
for i in xrange(0,NUM_MINHASHES,SKETCH_RATIO):
sketch = murmur.string_hash('minhashes[i:i+SKETCH_RATIO]')
sketches.append(sketch)
return np.array(minhashes),sketches
def extract_features(sha, path_to_files_dir,
hash_dim=1024, split_regex=r"\s+"):
# first, read in the file as a big string:
file = read_file(sha=sha, dir=path_to_files_dir)
# next, split the big string into a bunch of different tokens ("words"):
tokens = re.split(pattern=split_regex, string=file)
# now take the module(hash of each token) so that each token is replaced
# by bucket (category) from 1:hash_dim.
token_hash_buckets = [
(murmur.string_hash(w) % (hash_dim - 1) + 1) for w in tokens
]
# Finally, we'll count how many hits each bucket got, so that our features
# always have length hash_dim, regardless of the size of the HTML file:
token_bucket_counts = np.zeros(hash_dim)
# this returns the frequency counts for each unique value in
# token_hash_buckets:
buckets, counts = np.unique(token_hash_buckets, return_counts=True)
# and now we insert these counts into our token_bucket_counts object:
for bucket, count in zip(buckets, counts):
token_bucket_counts[bucket] = count
return np.array(token_bucket_counts)
def haiku_nofriends(user):
"""haiku for someone with no friends."""
out = []
poem = Poem()
poem.seed_user = user
for i in [5, 7, 5]:
ln = ' '.join(generate_syllable_chain(i, user.corpus))
out.append(ln)
idnumber = murmur.string_hash((u''.join(out)).encode('utf-8'))
poem.lines = out
poem.poem_id = idnumber
percussion = [music.screen_to_track(user)]
melody = [music.words_to_track(w) for w in out]
composition = music.composificate(percussion, melody)
filename = "audio/{0}.wav".format(str(idnumber))
music.mp3ificate(composition, filename)
poem.save()
return poem
def haiku_nofriends(user):
"""haiku for someone with no friends."""
out = []
poem = Poem()
poem.seed_user = user
for i in [5, 7, 5]:
ln = ' '.join(generate_syllable_chain(i, user.corpus))
out.append(ln)
idnumber = murmur.string_hash((u''.join(out)).encode('utf-8'))
poem.lines = out
poem.poem_id = idnumber
percussion = [music.screen_to_track(user)]
melody = [music.words_to_track(w) for w in out]
composition = music.composificate(percussion, melody)
filename = "audio/{0}.wav".format(str(idnumber))
music.mp3ificate(composition, filename)
poem.save()
return poem
def haiku(user):
#Can't do users until we're closed under corpus
out = []
poem = Poem()
poem.seed_user = user
users = [user]
for i in [5, 7, 5]:
ln = ' '.join(generate_syllable_chain(i, user.corpus))
out.append(ln)
user = get_friend(user)
users.append(user)
idnumber = murmur.string_hash((u''.join(out)).encode('utf-8'))
poem.lines = out
poem.poem_id = idnumber
percussion = [music.screen_to_track(u) for u in users]
melody = [music.words_to_track(w) for w in out]
composition = music.composificate(percussion, melody)
filename = "audio/{0}.wav".format(str(idnumber))
music.mp3ificate(composition, filename)
poem.save()
return poem
def stanza(user, max_length, lines):
out = []
poem = Poem()
poem.seed_user = user
for _ in range(lines):
ln = ' '.join(
generate_chain(length=max_length, markov=user.corpus,
strict=False))
out.append(ln)
user = get_friend(user)
idnumber = murmur.string_hash((u''.join(out)).encode('utf-8'))
poem.poem_id = idnumber
poem.lines = out
percussion = [music.screen_to_track(user)]
melody = [music.words_to_track(w) for w in out]
composition = music.composificate(percussion, melody)
filename = "audio/{0}.wav".format(str(idnumber))
music.mp3ificate(composition, filename)
poem.save()
return poem
def haiku(user):
#Can't do users until we're closed under corpus
out = []
poem = Poem()
poem.seed_user = user
users = [user]
for i in [5,7,5]:
ln = ' '.join(generate_syllable_chain(i, user.corpus))
out.append(ln)
user = get_friend(user)
users.append(user)
idnumber = murmur.string_hash((u''.join(out)).encode('utf-8'))
poem.lines = out
poem.poem_id = idnumber
percussion = [music.screen_to_track(u) for u in users]
melody = [music.words_to_track(w) for w in out]
composition = music.composificate(percussion, melody)
filename = "audio/{0}.wav".format(str(idnumber))
music.mp3ificate(composition, filename)
poem.save()
return poem
def stanza(user, max_length, lines):
out = []
poem = Poem()
poem.seed_user = user
for _ in range(lines):
ln = ' '.join(generate_chain(length = max_length,
markov = user.corpus,
strict = False))
out.append(ln)
user = get_friend(user)
idnumber = murmur.string_hash((u''.join(out)).encode('utf-8'))
poem.poem_id = idnumber
poem.lines = out
percussion = [music.screen_to_track(user)]
melody = [music.words_to_track(w) for w in out]
composition = music.composificate(percussion, melody)
filename = "audio/{0}.wav".format(str(idnumber))
music.mp3ificate(composition, filename)
poem.save()
return poem
continue
net_start = dottedQuadToNum(ip)
net_size = pow(2, 32 - ipfx)
#print net_start, net_size
for i in range(net_size):
ip_num = net_start + i
if not use_salt:
b = str(ip_num)
else:
b = str(ip_num) + str(salt)
if hash_function == Hashes.MURMUR:
h = murmur.string_hash(b)
if long(h) % mod == 0:
ip_str = numToDottedQuad(ip_num)
print ip_str
elif hash_function == Hashes.SHA1:
h = SHA.new(b)
if number.bytes_to_long(h.digest()) % mod == 0:
ip_str = numToDottedQuad(ip_num)
print ip_str
else:
h = MD5.new(b)
if number.bytes_to_long(h.digest()) % mod == 0:
ip_str = numToDottedQuad(ip_num)
def murmur2_hash_c(bytes, seed=0x9747b28c):
"""murmur2_hash_c
Use the murmur c-extension's string_hash routine
"""
return murmur.string_hash(str(bytes), seed)
continue
net_start = dottedQuadToNum(ip)
net_size = pow(2, 32 - ipfx)
#print net_start, net_size
for i in range(net_size):
ip_num = net_start + i
if not use_salt:
b = str(ip_num)
else:
b = str(ip_num) + str(salt)
if hash_function == Hashes.MURMUR:
h = murmur.string_hash(b)
if long(h) % mod == 0 :
ip_str = numToDottedQuad(ip_num)
print ip_str
elif hash_function == Hashes.SHA1:
h = SHA.new(b)
if number.bytes_to_long(h.digest()) % mod == 0 :
ip_str = numToDottedQuad(ip_num)
print ip_str
else:
h = MD5.new(b)
if number.bytes_to_long(h.digest()) % mod == 0 :
ip_str = numToDottedQuad(ip_num)
'f': 4226522672,
'i': 3451942824,
'h': 1069002520,
'k': 3288208012,
'j': 3131388162,
'm': 3020367812,
'l': 2169669117,
'o': 1720432690,
'n': 1785613168,
'q': 2083633015,
'p': 834694889,
's': 389143345,
'r': 744399309,
'u': 1479000828,
't': 2418444476,
'w': 1340422676,
'v': 3414904798,
'y': 3657681515,
'x': 372604132,
'z': 2195360465
}
actual_dict = {
letter: murmur.string_hash(letter)
for letter in string.ascii_lowercase
}
unmatched_items = set(expected_dict.items()) ^ set(actual_dict.items())
assert len(unmatched_items) == 0
analyzer = StandardAnalyzer(Version.LUCENE_30)
queryparser = QueryParser(Version.LUCENE_30, "text", analyzer)
searcher = IndexSearcher(dir)
nonzeros = 0
for i, l in enumerate(sys.stdin):
if i % 100 == 0:
print >> sys.stderr, "Read %d lines from sys.stdin (bloom filter has %s nonzeros)..." % (
i, percent(nonzeros, BLOOM_FILTER_SIZE))
print >> sys.stderr, stats()
l = string.strip(l)
added_this_sentence = 0
for newl in retrieve(l, searcher, queryparser):
# Iterate until we have added DESIRED_NEW_DOCUMENTS_PER_ORIGINAL_DOCUMENT documents
if added_this_sentence >= DESIRED_NEW_DOCUMENTS_PER_ORIGINAL_DOCUMENT:
break
newl = string.strip(newl)
# Hash the sentence
idx = murmur.string_hash(newl.encode("utf-8")) % BLOOM_FILTER_SIZE
# Don't use duplicate sentences
if usedsentences[idx]: continue
usedsentences[idx] = True
nonzeros += 1
added_this_sentence += 1
print newl.encode("utf-8")
def murmur2_hash_c(bytes, seed=0x9747b28c):
"""murmur2_hash_c
Use the murmur c-extension's string_hash routine
"""
return murmur.string_hash(str(bytes), seed)
def is_validation_example(e):
import common.hyperparameters
HYPERPARAMETERS = common.hyperparameters.read("language-model")
examples_per_validation = int(1/HYPERPARAMETERS["PERCENT_OF_TRAINING_EXAMPLES_FOR_VALIDATION"])
return murmur.string_hash(`e`) % examples_per_validation == 0
indexDir = "lucene.ukwac"
dir = SimpleFSDirectory(File(indexDir))
analyzer = StandardAnalyzer(Version.LUCENE_30)
queryparser = QueryParser(Version.LUCENE_30, "text", analyzer)
searcher = IndexSearcher(dir)
nonzeros = 0
for i, l in enumerate(sys.stdin):
if i % 100 == 0:
print >> sys.stderr, "Read %d lines from sys.stdin (bloom filter has %s nonzeros)..." % (i, percent(nonzeros, BLOOM_FILTER_SIZE))
print >> sys.stderr, stats()
l = string.strip(l)
added_this_sentence = 0
for newl in retrieve(l, searcher, queryparser):
# Iterate until we have added DESIRED_NEW_DOCUMENTS_PER_ORIGINAL_DOCUMENT documents
if added_this_sentence >= DESIRED_NEW_DOCUMENTS_PER_ORIGINAL_DOCUMENT: break
newl = string.strip(newl)
# Hash the sentence
idx = murmur.string_hash(newl.encode("utf-8")) % BLOOM_FILTER_SIZE
# Don't use duplicate sentences
if usedsentences[idx]: continue
usedsentences[idx] = True
nonzeros += 1
added_this_sentence += 1
print newl.encode("utf-8")
