sc._jsc.hadoopConfiguration().set("fs.s3n.awsAccessKeyId", AWS_ACCESS_KEY_ID)
sc._jsc.hadoopConfiguration().set("fs.s3n.awsSecretAccessKey", AWS_SECRET_ACCESS_KEY)
for item in d:
name = item
fp = "s3n://bucket-data/"+name
p = d[item]
data = read_song_text(sc, fp, p)
# sparse vector from data
data_sparse_list = data.map(lambda x:(x[0][1], x[1]))
rdd_1 = data.map(lambda x:(x[0][0], x[1]))
# initialize the parameters
m, n, b, c = 1000,1000,25,2
# create the model for LSH
model = lsh.run(data_sparse_list, p, m, n, b, c)
print ("start printing filename %s" % name)
# Get similar buckets
cnt = model.buckets.count()
# print result time taken
timetaken = (time.time() - start_time)
print 'Found %s clusters.' % cnt
print("--- %s seconds ---" % timetaken)
tup = cnt, timetaken
d[name] = cnt , (time.time() - start_time)
# write log file
f = open('output.txt', 'a')
mystr = str(time.time())
help = "Number of times to hash the elements. Larger numbers diversify " +
"signatures, increasing likelihood similar vectors will be hashed together. " +
"This is also the length of the signature. [DEFAULT: 1000]")
parser.add_argument("-b", "--bands", type = int, default = 25,
help = "Number of bands. Each band will have (n / b) elements. Larger " +
"numbers of elements increase confidence in element similarity. [DEFAULT: 25]")
parser.add_argument("-c", "--minbucketsize", type = int, default = 2,
help = "Minimum bucket size (0 to disable). Buckets with fewer than this " +
"number of elements will be dropped. [DEFAULT: 2]")
args = vars(parser.parse_args())
sc = SparkContext(conf = SparkConf())
# Read the input data.
print now(), 'Starting'
raw_lines, data = read_text(sc, args['input'])
p = 65537
m, n, b, c = args['bins'], args['numrows'], args['bands'], args['minbucketsize']
vector_buckets = lsh.run(data, p, m, n, b, c)
bucket_ids = vector_buckets.map(lambda (vector, bucket): bucket).distinct()
print now(), 'Found %s clusters.' % bucket_ids.count()
bucket_vectors = vector_buckets.map(lambda (vector, bucket): (bucket, vector)).groupByKey()
for (bucket, vectors) in bucket_vectors.collect():
print 'Bucket %s' % bucket
for vector in vectors:
print '\tDocument %s: %s ...' % (vector, raw_lines[vector][:100])
print '*' * 40
parser = argparse.ArgumentParser(description = 'Spark LSH',
epilog = 'lol lsh', add_help = 'How to use',
prog = 'python driver.py <arguments>')
parser.add_argument("-i", "--input", required = True,
help = "Input directory of text files.")
# Optional parameters.
parser.add_argument("-m", "--bins", type = int, default = 1000,
help = "Number of bins into which to hash the data. Smaller numbers " +
"increase collisions, producing larger clusters. [DEFAULT: 1000]")
parser.add_argument("-n", "--numrows", type = int, default = 1000,
help = "Number of times to hash the elements. Larger numbers diversify " +
"signatures, increasing likelihood similar vectors will be hashed together. " +
"This is also the length of the signature. [DEFAULT: 1000]")
parser.add_argument("-b", "--bands", type = int, default = 25,
help = "Number of bands. Each band will have (n / b) elements. Larger " +
"numbers of elements increase confidence in element similarity. [DEFAULT: 25]")
parser.add_argument("-c", "--minbucketsize", type = int, default = 2,
help = "Minimum bucket size (0 to disable). Buckets with fewer than this " +
"number of elements will be dropped. [DEFAULT: 2]")
args = vars(parser.parse_args())
sc = SparkContext(conf = SparkConf())
# Read the input data.
data = read_text(sc, args['input'])
p = 65537
m, n, b, c = args['bins'], args['numrows'], args['bands'], args['minbucketsize']
model = lsh.run(data, p, m, n, b, c)
print 'Found %s clusters.' % model.buckets.count()
help="Number of times to hash the elements. Larger numbers diversify "
+
"signatures, increasing likelihood similar vectors will be hashed together. "
+ "This is also the length of the signature. [DEFAULT: 1000]")
parser.add_argument(
"-b",
"--bands",
type=int,
default=25,
help="Number of bands. Each band will have (n / b) elements. Larger " +
"numbers of elements increase confidence in element similarity. [DEFAULT: 25]"
)
parser.add_argument(
"-c",
"--minbucketsize",
type=int,
default=2,
help="Minimum bucket size (0 to disable). Buckets with fewer than this "
+ "number of elements will be dropped. [DEFAULT: 2]")
args = vars(parser.parse_args())
sc = SparkContext(conf=SparkConf())
# Read the input data.
data = read_text(sc, args['input'])
p = 65537
m, n, b, c = args['bins'], args['numrows'], args['bands'], args[
'minbucketsize']
model = lsh.run(data, p, m, n, b, c)
print 'Found %s clusters.' % model.buckets.count()
