hyperll 0.2.6 → 0.3.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.gitignore +4 -0
- data/Rakefile +7 -0
- data/ext/hyperll/delta_bytes.c +97 -0
- data/ext/hyperll/delta_bytes.h +24 -0
- data/ext/hyperll/extconf.rb +12 -0
- data/ext/hyperll/hyper_log_log_plus.c +473 -0
- data/ext/hyperll/hyperll.c +12 -0
- data/ext/hyperll/hyperll.h +17 -0
- data/ext/hyperll/register_set.c +196 -0
- data/ext/hyperll/register_set.h +19 -0
- data/ext/hyperll/sparse_set.c +76 -0
- data/ext/hyperll/sparse_set.h +24 -0
- data/ext/hyperll/varint.c +83 -0
- data/ext/hyperll/varint.h +21 -0
- data/hyperll.gemspec +2 -0
- data/lib/hyperll.rb +1 -0
- data/lib/hyperll/hyper_log_log.rb +0 -1
- data/lib/hyperll/hyper_log_log_plus.rb +10 -310
- data/lib/hyperll/version.rb +1 -1
- data/spec/fixtures/merge-many-sets.json +192 -0
- data/spec/hyperll/delta_bytes_spec.rb +5 -3
- data/spec/hyperll/hyper_log_log_plus_spec.rb +30 -1
- data/spec/hyperll/hyper_log_log_spec.rb +1 -1
- data/spec/hyperll/register_set_spec.rb +18 -1
- data/spec/hyperll/varint_spec.rb +5 -1
- metadata +32 -7
- data/lib/hyperll/delta_bytes.rb +0 -32
- data/lib/hyperll/register_set.rb +0 -90
- data/lib/hyperll/util.rb +0 -39
- data/lib/hyperll/varint.rb +0 -26
@@ -1,14 +1,16 @@
|
|
1
1
|
require 'spec_helper'
|
2
|
-
require 'hyperll
|
2
|
+
require 'hyperll'
|
3
3
|
|
4
4
|
module Hyperll
|
5
5
|
describe DeltaBytes do
|
6
6
|
it 'uncompresses bytes' do
|
7
|
-
expect(DeltaBytes.uncompress([
|
7
|
+
expect(DeltaBytes.uncompress([-46, 5, -64, 4])).to eq([722, 1298])
|
8
|
+
expect(DeltaBytes.uncompress([210, 5, 192, 4, 254, 67])).to eq([722, 1298, 10000])
|
8
9
|
end
|
9
10
|
|
10
11
|
it 'compresses bytes' do
|
11
|
-
expect(DeltaBytes.compress([722, 1298])).to eq([
|
12
|
+
expect(DeltaBytes.compress([722, 1298])).to eq([256 - 46, 5, 256 - 64, 4])
|
13
|
+
expect(DeltaBytes.compress([722, 1298, 10000])).to eq([210, 5, 192, 4, 254, 67])
|
12
14
|
end
|
13
15
|
end
|
14
16
|
end
|
@@ -1,6 +1,7 @@
|
|
1
1
|
require 'spec_helper'
|
2
2
|
require 'base64'
|
3
|
-
require 'hyperll
|
3
|
+
require 'hyperll'
|
4
|
+
require 'json'
|
4
5
|
|
5
6
|
module Hyperll
|
6
7
|
describe HyperLogLogPlus do
|
@@ -206,6 +207,22 @@ module Hyperll
|
|
206
207
|
expect(hllp.cardinality).to eq(6)
|
207
208
|
end
|
208
209
|
|
210
|
+
it 'merges and keeps the cardinality exact, handling elements that are common to both sets' do
|
211
|
+
# Serialization after offering [12, 13, 22, 34, 38, 40, 41, 46, 49]
|
212
|
+
hllp = HyperLogLogPlus.unserialize(Base64.decode64("/////gsQAQnMYsoMtgak9AGMiwK8VbKiAYmU0wPVwK38Dw=="))
|
213
|
+
|
214
|
+
# Serialization after offering [2, 6, 19, 29, 41, 48]
|
215
|
+
hllp2 = HyperLogLogPlus.unserialize(Base64.decode64("/////gsQAQbwdJz0Afq4AbSZAqxX4i4="))
|
216
|
+
|
217
|
+
expect(hllp.cardinality).to eq(9)
|
218
|
+
expect(hllp2.cardinality).to eq(6)
|
219
|
+
|
220
|
+
# The set intersection of hllp and hllp2 has one element, 41, so after
|
221
|
+
# merging the cardinality should be 14.
|
222
|
+
hllp.merge(hllp2)
|
223
|
+
expect(hllp.cardinality).to eq(14)
|
224
|
+
end
|
225
|
+
|
209
226
|
it 'merges and keeps the cardinality exact' do
|
210
227
|
hllp = HyperLogLogPlus.unserialize(Base64.decode64("/////gsQAQOwX+yBA7TzAw=="))
|
211
228
|
hllp2 = HyperLogLogPlus.unserialize(Base64.decode64("/////gsQAQ7SKbociFqGigLUL9oagCWmC+IdlBqkE8g7jFiCnwE="))
|
@@ -270,5 +287,17 @@ module Hyperll
|
|
270
287
|
expect(hllp.cardinality).to eq(8) # 3 + 3 = 8; that's how it goes with hll
|
271
288
|
end
|
272
289
|
end
|
290
|
+
|
291
|
+
context 'merging multiple at a time' do
|
292
|
+
it 'merges' do
|
293
|
+
hllp = HyperLogLogPlus.unserialize([-1, -1, -1, -2, 4, 10, 1, 1, -110, 10].pack("C*"))
|
294
|
+
hllp2 = HyperLogLogPlus.unserialize([-1, -1, -1, -2, 4, 10, 1, 1, -46, 5].pack("C*"))
|
295
|
+
hllp3 = HyperLogLogPlus.unserialize([-1, -1, -1, -2, 4, 10, 1, 1, -124, 6].pack("C*"))
|
296
|
+
|
297
|
+
hllp.merge(hllp2, hllp3)
|
298
|
+
expect(hllp.format).to eq(:sparse)
|
299
|
+
expect(hllp.cardinality).to eq(3) # 1 + 1 + 1 = 3
|
300
|
+
end
|
301
|
+
end
|
273
302
|
end
|
274
303
|
end
|
@@ -1,5 +1,5 @@
|
|
1
1
|
require 'spec_helper'
|
2
|
-
require 'hyperll
|
2
|
+
require 'hyperll'
|
3
3
|
|
4
4
|
module Hyperll
|
5
5
|
describe RegisterSet do
|
@@ -68,5 +68,22 @@ module Hyperll
|
|
68
68
|
expect(value).to eq(merged[index])
|
69
69
|
end
|
70
70
|
end
|
71
|
+
|
72
|
+
it "serializes to a string" do
|
73
|
+
rs = RegisterSet.new(10)
|
74
|
+
rs[0] = 2
|
75
|
+
rs[1] = 3
|
76
|
+
rs[2] = 4
|
77
|
+
|
78
|
+
expect(rs.serialize).to eq("\x00\x00\x10b\x00\x00\x00\x00".force_encoding("ASCII-8BIT"))
|
79
|
+
end
|
80
|
+
|
81
|
+
it "unserializes from a string" do
|
82
|
+
rs = RegisterSet.new(10, "\x00\x00\x10b\x00\x00\x00\x00".unpack("N*"))
|
83
|
+
|
84
|
+
expect(rs[0]).to eq(2)
|
85
|
+
expect(rs[1]).to eq(3)
|
86
|
+
expect(rs[2]).to eq(4)
|
87
|
+
end
|
71
88
|
end
|
72
89
|
end
|
data/spec/hyperll/varint_spec.rb
CHANGED
@@ -1,5 +1,5 @@
|
|
1
1
|
require 'spec_helper'
|
2
|
-
require 'hyperll
|
2
|
+
require 'hyperll'
|
3
3
|
|
4
4
|
module Hyperll
|
5
5
|
describe Varint do
|
@@ -14,5 +14,9 @@ module Hyperll
|
|
14
14
|
expect(Varint.write_unsigned_var_int(0x81)).to eq([0x81, 0x01])
|
15
15
|
expect(Varint.write_unsigned_var_int(0x4081)).to eq([0x81, 0x81, 0x01])
|
16
16
|
end
|
17
|
+
|
18
|
+
it 'treats negative integers as their twos complement unsigned representation' do
|
19
|
+
expect(Varint.write_unsigned_var_int(-4)).to eq([252, 255, 255, 255, 15])
|
20
|
+
end
|
17
21
|
end
|
18
22
|
end
|
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: hyperll
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.
|
4
|
+
version: 0.3.0
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Andy Lindeman
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date: 2013-10-
|
11
|
+
date: 2013-10-31 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: bundler
|
@@ -52,11 +52,26 @@ dependencies:
|
|
52
52
|
- - ~>
|
53
53
|
- !ruby/object:Gem::Version
|
54
54
|
version: '2.14'
|
55
|
+
- !ruby/object:Gem::Dependency
|
56
|
+
name: rake-compiler
|
57
|
+
requirement: !ruby/object:Gem::Requirement
|
58
|
+
requirements:
|
59
|
+
- - ~>
|
60
|
+
- !ruby/object:Gem::Version
|
61
|
+
version: 0.9.1
|
62
|
+
type: :development
|
63
|
+
prerelease: false
|
64
|
+
version_requirements: !ruby/object:Gem::Requirement
|
65
|
+
requirements:
|
66
|
+
- - ~>
|
67
|
+
- !ruby/object:Gem::Version
|
68
|
+
version: 0.9.1
|
55
69
|
description: HyperLogLog implementation in pure Ruby
|
56
70
|
email:
|
57
71
|
- andy@andylindeman.com
|
58
72
|
executables: []
|
59
|
-
extensions:
|
73
|
+
extensions:
|
74
|
+
- ext/hyperll/extconf.rb
|
60
75
|
extra_rdoc_files: []
|
61
76
|
files:
|
62
77
|
- .gitignore
|
@@ -66,18 +81,27 @@ files:
|
|
66
81
|
- LICENSE.txt
|
67
82
|
- README.md
|
68
83
|
- Rakefile
|
84
|
+
- ext/hyperll/delta_bytes.c
|
85
|
+
- ext/hyperll/delta_bytes.h
|
86
|
+
- ext/hyperll/extconf.rb
|
87
|
+
- ext/hyperll/hyper_log_log_plus.c
|
88
|
+
- ext/hyperll/hyperll.c
|
89
|
+
- ext/hyperll/hyperll.h
|
90
|
+
- ext/hyperll/register_set.c
|
91
|
+
- ext/hyperll/register_set.h
|
92
|
+
- ext/hyperll/sparse_set.c
|
93
|
+
- ext/hyperll/sparse_set.h
|
94
|
+
- ext/hyperll/varint.c
|
95
|
+
- ext/hyperll/varint.h
|
69
96
|
- hyperll.gemspec
|
70
97
|
- lib/hyperll.rb
|
71
|
-
- lib/hyperll/delta_bytes.rb
|
72
98
|
- lib/hyperll/hyper_log_log.rb
|
73
99
|
- lib/hyperll/hyper_log_log_plus.rb
|
74
100
|
- lib/hyperll/murmur_hash.rb
|
75
|
-
- lib/hyperll/register_set.rb
|
76
|
-
- lib/hyperll/util.rb
|
77
|
-
- lib/hyperll/varint.rb
|
78
101
|
- lib/hyperll/version.rb
|
79
102
|
- spec/fixtures/10000.txt
|
80
103
|
- spec/fixtures/mega.json
|
104
|
+
- spec/fixtures/merge-many-sets.json
|
81
105
|
- spec/hyperll/delta_bytes_spec.rb
|
82
106
|
- spec/hyperll/hyper_log_log_plus_spec.rb
|
83
107
|
- spec/hyperll/hyper_log_log_spec.rb
|
@@ -114,6 +138,7 @@ summary: HyperLogLog implementation in pure Ruby
|
|
114
138
|
test_files:
|
115
139
|
- spec/fixtures/10000.txt
|
116
140
|
- spec/fixtures/mega.json
|
141
|
+
- spec/fixtures/merge-many-sets.json
|
117
142
|
- spec/hyperll/delta_bytes_spec.rb
|
118
143
|
- spec/hyperll/hyper_log_log_plus_spec.rb
|
119
144
|
- spec/hyperll/hyper_log_log_spec.rb
|
data/lib/hyperll/delta_bytes.rb
DELETED
@@ -1,32 +0,0 @@
|
|
1
|
-
require_relative 'varint'
|
2
|
-
|
3
|
-
module Hyperll
|
4
|
-
class DeltaBytes
|
5
|
-
def self.compress(bytes)
|
6
|
-
compressed = Varint.write_unsigned_var_int(bytes.length)
|
7
|
-
previous_value = 0
|
8
|
-
|
9
|
-
bytes.each do |b|
|
10
|
-
compressed.concat(Varint.write_unsigned_var_int(b - previous_value))
|
11
|
-
previous_value = b
|
12
|
-
end
|
13
|
-
|
14
|
-
compressed
|
15
|
-
end
|
16
|
-
|
17
|
-
def self.uncompress(bytes)
|
18
|
-
uncompressed = []
|
19
|
-
previous_value = 0
|
20
|
-
|
21
|
-
length = Varint.read_unsigned_var_int(bytes)
|
22
|
-
length.times do
|
23
|
-
next_value = Varint.read_unsigned_var_int(bytes)
|
24
|
-
|
25
|
-
uncompressed << next_value + previous_value
|
26
|
-
previous_value = uncompressed.last
|
27
|
-
end
|
28
|
-
|
29
|
-
uncompressed
|
30
|
-
end
|
31
|
-
end
|
32
|
-
end
|
data/lib/hyperll/register_set.rb
DELETED
@@ -1,90 +0,0 @@
|
|
1
|
-
require_relative 'util'
|
2
|
-
|
3
|
-
module Hyperll
|
4
|
-
class RegisterSet
|
5
|
-
include Enumerable
|
6
|
-
include Util
|
7
|
-
|
8
|
-
LOG2_BITS_PER_WORD = 6
|
9
|
-
REGISTER_SIZE = 5
|
10
|
-
|
11
|
-
attr_reader :count, :size
|
12
|
-
|
13
|
-
def initialize(count, values = nil)
|
14
|
-
@count = count
|
15
|
-
|
16
|
-
@bits = count / LOG2_BITS_PER_WORD
|
17
|
-
if @bits.zero?
|
18
|
-
@size = 1
|
19
|
-
elsif (@bits % INTEGER_SIZE).zero?
|
20
|
-
@size = @bits
|
21
|
-
else
|
22
|
-
@size = @bits + 1
|
23
|
-
end
|
24
|
-
|
25
|
-
@values = values || Array.new(@size, 0)
|
26
|
-
end
|
27
|
-
|
28
|
-
def []=(position, value)
|
29
|
-
bucket = position / LOG2_BITS_PER_WORD
|
30
|
-
shift = REGISTER_SIZE * (position - (bucket * LOG2_BITS_PER_WORD))
|
31
|
-
|
32
|
-
@values[bucket] = ((@values[bucket] & ~(0x1f * POWERS_OF_TWO[shift])) | (value * POWERS_OF_TWO[shift]))
|
33
|
-
@values[bucket] &= INT_MASK if @values[bucket] > INT_MASK
|
34
|
-
end
|
35
|
-
|
36
|
-
def [](position)
|
37
|
-
bucket = position / LOG2_BITS_PER_WORD
|
38
|
-
shift = REGISTER_SIZE * (position - (bucket * LOG2_BITS_PER_WORD))
|
39
|
-
|
40
|
-
return (@values[bucket] & (0x1f * POWERS_OF_TWO[shift])) / POWERS_OF_TWO[shift]
|
41
|
-
end
|
42
|
-
|
43
|
-
def each
|
44
|
-
return enum_for(:each) unless block_given?
|
45
|
-
@count.times do |i|
|
46
|
-
yield self[i]
|
47
|
-
end
|
48
|
-
end
|
49
|
-
|
50
|
-
def update_if_greater(position, value)
|
51
|
-
bucket = position / LOG2_BITS_PER_WORD
|
52
|
-
shift = REGISTER_SIZE * (position - (bucket * LOG2_BITS_PER_WORD));
|
53
|
-
mask = (0x1f * POWERS_OF_TWO[shift])
|
54
|
-
|
55
|
-
current_value = @values[bucket] & mask
|
56
|
-
new_value = value * POWERS_OF_TWO[shift]
|
57
|
-
if current_value < new_value
|
58
|
-
@values[bucket] = ((@values[bucket] & ~mask) | new_value)
|
59
|
-
@values[bucket] &= INT_MASK if @values[bucket] > INT_MASK
|
60
|
-
true
|
61
|
-
else
|
62
|
-
false
|
63
|
-
end
|
64
|
-
end
|
65
|
-
|
66
|
-
def merge(other)
|
67
|
-
@size.times do |bucket|
|
68
|
-
word = 0
|
69
|
-
LOG2_BITS_PER_WORD.times do |j|
|
70
|
-
mask = 0x1f << (REGISTER_SIZE * j);
|
71
|
-
|
72
|
-
this_val = self.values[bucket] & mask
|
73
|
-
other_val = other.values[bucket] & mask
|
74
|
-
word |= [this_val, other_val].max
|
75
|
-
end
|
76
|
-
|
77
|
-
@values[bucket] = word
|
78
|
-
end
|
79
|
-
end
|
80
|
-
|
81
|
-
def serialize
|
82
|
-
@values.pack("N*")
|
83
|
-
end
|
84
|
-
|
85
|
-
protected
|
86
|
-
def values
|
87
|
-
@values
|
88
|
-
end
|
89
|
-
end
|
90
|
-
end
|
data/lib/hyperll/util.rb
DELETED
@@ -1,39 +0,0 @@
|
|
1
|
-
module Hyperll
|
2
|
-
module Util
|
3
|
-
INT_MASK = 0xFFFFFFFF
|
4
|
-
INTEGER_SIZE = 32
|
5
|
-
|
6
|
-
POWERS_OF_TWO = 0.upto(32).map { |i| 2**i }.freeze
|
7
|
-
|
8
|
-
def number_of_leading_zeros(x)
|
9
|
-
return 32 if x == 0
|
10
|
-
|
11
|
-
n = 0
|
12
|
-
if x <= 0x0000FFFF
|
13
|
-
n += 16
|
14
|
-
x *= POWERS_OF_TWO[16]
|
15
|
-
end
|
16
|
-
|
17
|
-
if x <= 0x00FFFFFF
|
18
|
-
n += 8;
|
19
|
-
x *= POWERS_OF_TWO[8]
|
20
|
-
end
|
21
|
-
|
22
|
-
if x <= 0x0FFFFFFF
|
23
|
-
n += 4
|
24
|
-
x *= POWERS_OF_TWO[4]
|
25
|
-
end
|
26
|
-
|
27
|
-
if x <= 0x3FFFFFFF
|
28
|
-
n += 2
|
29
|
-
x *= POWERS_OF_TWO[2]
|
30
|
-
end
|
31
|
-
|
32
|
-
if x <= 0x7FFFFFFF
|
33
|
-
n += 1
|
34
|
-
end
|
35
|
-
|
36
|
-
n
|
37
|
-
end
|
38
|
-
end
|
39
|
-
end
|
data/lib/hyperll/varint.rb
DELETED
@@ -1,26 +0,0 @@
|
|
1
|
-
module Hyperll
|
2
|
-
class Varint
|
3
|
-
def self.read_unsigned_var_int(bytes)
|
4
|
-
value, i, b = 0, 0, 0
|
5
|
-
while (b = bytes.shift) & 0x80 != 0
|
6
|
-
value |= (b & 0x7F) << i
|
7
|
-
|
8
|
-
i += 7
|
9
|
-
raise "Variable length quantity is too long" if i > 35
|
10
|
-
end
|
11
|
-
|
12
|
-
value | (b << i)
|
13
|
-
end
|
14
|
-
|
15
|
-
def self.write_unsigned_var_int(value)
|
16
|
-
bytes = []
|
17
|
-
while (value & 0xFFFFFF80) != 0
|
18
|
-
bytes << ((value & 0x7F) | 0x80)
|
19
|
-
value >>= 7
|
20
|
-
end
|
21
|
-
|
22
|
-
bytes << (value & 0x7F)
|
23
|
-
bytes
|
24
|
-
end
|
25
|
-
end
|
26
|
-
end
|