min_max 0.1.3 → 0.1.5
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/Cargo.lock +2 -2
- data/Cargo.toml +2 -1
- data/README.md +18 -3
- data/benchmarks/benchmarks.rb +85 -0
- data/ext/min_max/src/lib.rs +13 -38
- data/lib/min_max/version.rb +1 -1
- data/lib/min_max.rb +38 -49
- metadata +2 -1
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 6d739946dcb0939e2e05893222c927f4dea0eda65da5c0a00156402de3e49c29
|
4
|
+
data.tar.gz: 61b4453a94b93bb2ccefd0787a751cef5d3ef380ac0ed591c73c3b0afddc394b
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 70dd4402ca82a60cffbc451ea134ca9fc34391f28311d23726d22055bfc01db9ba62002f0d54c76282888993613f2bf5618dcfa7a964da39c6995747401f3a3f
|
7
|
+
data.tar.gz: 31ada4f8ed50624701e389c6a1d50b83b75a585ef703d2f18c6511498dc9373647fa1f88d959ccdcbf56336d38dc550e1ce96710644f65970f2341549f843d53
|
data/Cargo.lock
CHANGED
@@ -270,9 +270,9 @@ checksum = "24188a676b6ae68c3b2cb3a01be17fbf7240ce009799bb56d5b1409051e78fde"
|
|
270
270
|
|
271
271
|
[[package]]
|
272
272
|
name = "shlex"
|
273
|
-
version = "1.
|
273
|
+
version = "1.3.0"
|
274
274
|
source = "registry+https://github.com/rust-lang/crates.io-index"
|
275
|
-
checksum = "
|
275
|
+
checksum = "0fda2ff0d084019ba4d7c6f371c95d8fd75ce3524c3cb8fb653a3023f6323e64"
|
276
276
|
|
277
277
|
[[package]]
|
278
278
|
name = "syn"
|
data/Cargo.toml
CHANGED
data/README.md
CHANGED
@@ -1,7 +1,8 @@
|
|
1
|
-
# MinMax
|
1
|
+
# MinMax Heap
|
2
2
|
|
3
|
-
The MinMax gem provides a high-performance minmax heap implementation for Ruby, written in Rust.
|
4
|
-
|
3
|
+
The MinMax Heap gem provides a high-performance minmax heap implementation for Ruby, written in Rust.
|
4
|
+
The gem wraps the excellent [min-max-heap-rs](https://github.com/tov/min-max-heap-rs) Rust library.
|
5
|
+
It allows for the creation of a min-max-heap and supporting operations like pushing and popping multiple items, iterating over heap items, and converting heaps to arrays.
|
5
6
|
|
6
7
|
## Features
|
7
8
|
|
@@ -25,6 +26,7 @@ Add this line to your application's Gemfile:
|
|
25
26
|
gem 'min_max'
|
26
27
|
```
|
27
28
|
|
29
|
+
|
28
30
|
And then execute:
|
29
31
|
|
30
32
|
```bash
|
@@ -34,6 +36,10 @@ bundle install
|
|
34
36
|
Or install it yourself as:
|
35
37
|
```bash
|
36
38
|
gem install min_max
|
39
|
+
|
40
|
+
# or manually specify target . E.g.
|
41
|
+
|
42
|
+
CARGO_BUILD_TARGET=x86_64-apple-darwin gem install min_max
|
37
43
|
```
|
38
44
|
|
39
45
|
## Usage
|
@@ -84,4 +90,13 @@ heap.length # Alias for size
|
|
84
90
|
heap.clear
|
85
91
|
```
|
86
92
|
|
93
|
+
## Count number of times an item is in the heap
|
94
|
+
```ruby
|
95
|
+
heap.count(item)
|
96
|
+
```
|
97
|
+
|
98
|
+
## Check if item is contained in the heap
|
99
|
+
```ruby
|
100
|
+
heap.contains?(item)
|
101
|
+
```
|
87
102
|
|
@@ -0,0 +1,85 @@
|
|
1
|
+
require "bundler/setup"
|
2
|
+
require "min_max"
|
3
|
+
|
4
|
+
require 'bundler/inline'
|
5
|
+
|
6
|
+
gemfile do
|
7
|
+
source 'https://rubygems.org'
|
8
|
+
gem 'rb_heap'
|
9
|
+
gem 'algorithms'
|
10
|
+
gem 'ruby-heap'
|
11
|
+
gem 'pqueue'
|
12
|
+
end
|
13
|
+
|
14
|
+
require 'benchmark'
|
15
|
+
require 'rb_heap'
|
16
|
+
require 'min_max'
|
17
|
+
|
18
|
+
|
19
|
+
data = 100_000.times.map{ Random.rand(0...1000_000_000) }
|
20
|
+
rb_hp = Heap.new(:<)
|
21
|
+
mm_hp = MinMax[]
|
22
|
+
|
23
|
+
Benchmark.bm do |x|
|
24
|
+
x.report("push_rb_heap"){ data.each{|d| rb_hp.add(d) } }
|
25
|
+
x.report("push_mm_heap"){ data.each{|d| mm_hp.push(d) } }
|
26
|
+
x.report("push_mm_heap_batches"){ mm_hp = MinMax[]; data.each_slice(1000){|d| mm_hp.push(*d) } }
|
27
|
+
end
|
28
|
+
|
29
|
+
Benchmark.bm do |x|
|
30
|
+
x.report("pop_rb_heap"){ 100_000.times{|d| rb_hp.pop } }
|
31
|
+
x.report("pop_mm_heap"){ 100_000.times{|d| mm_hp.pop_min } }
|
32
|
+
x.report("pop_mm_heap_batches"){ 1000.times{|d| mm_hp.pop_min(100) } }
|
33
|
+
end
|
34
|
+
|
35
|
+
Object.send(:remove_const, :Heap)
|
36
|
+
|
37
|
+
require 'Heap'
|
38
|
+
puts "# ruby-heap vs min_max"
|
39
|
+
ruby_hp = Heap::BinaryHeap::MinHeap.new
|
40
|
+
mm_hp = MinMax[]
|
41
|
+
|
42
|
+
Benchmark.bm do |x|
|
43
|
+
x.report("push_ruby_heap"){ data.each{|d| ruby_hp.add(d) } }
|
44
|
+
x.report("push_mm_heap"){ data.each{|d| mm_hp.push(d) } }
|
45
|
+
x.report("push_mm_heap_batches"){ mm_hp = MinMax[]; data.each_slice(1000){|d| mm_hp.push(*d) } }
|
46
|
+
end
|
47
|
+
|
48
|
+
Benchmark.bm do |x|
|
49
|
+
x.report("pop_ruby_heap"){ 100_000.times{|d| ruby_hp.extract_min! } }
|
50
|
+
x.report("pop_mm_heap"){ 100_000.times{|d| mm_hp.pop_min } }
|
51
|
+
x.report("pop_mm_heap_batches"){ 1000.times{|d| mm_hp.pop_min(100) } }
|
52
|
+
end
|
53
|
+
|
54
|
+
|
55
|
+
require 'algorithms'
|
56
|
+
minheap = Containers::MinHeap.new()
|
57
|
+
mm_hp = MinMax[]
|
58
|
+
|
59
|
+
Benchmark.bm do |x|
|
60
|
+
x.report("push_algos_heap"){ data.each{|d| minheap.push(d) } }
|
61
|
+
x.report("push_mm_heap"){ data.each{|d| mm_hp.push(d) } }
|
62
|
+
x.report("push_mm_heap_batches"){ mm_hp = MinMax[]; data.each_slice(1000){|d| mm_hp.push(*d) } }
|
63
|
+
end
|
64
|
+
|
65
|
+
Benchmark.bm do |x|
|
66
|
+
x.report("pop_algos_heap"){ 100_000.times{|d| minheap.pop } }
|
67
|
+
x.report("pop_mm_heap"){ 100_000.times{|d| mm_hp.pop_min } }
|
68
|
+
x.report("pop_mm_heap_batches"){ 1000.times{|d| mm_hp.pop_min(100) } }
|
69
|
+
end
|
70
|
+
|
71
|
+
require 'pqueue'
|
72
|
+
pqueue = PQueue.new()
|
73
|
+
mm_hp = MinMax[]
|
74
|
+
|
75
|
+
Benchmark.bm do |x|
|
76
|
+
x.report("push_algos_heap"){ data.each{|d| pqueue.push(d) } }
|
77
|
+
x.report("push_mm_heap"){ data.each{|d| mm_hp.push(d) } }
|
78
|
+
x.report("push_mm_heap_batches"){ mm_hp = MinMax[]; data.each_slice(1000){|d| mm_hp.push(*d) } }
|
79
|
+
end
|
80
|
+
|
81
|
+
Benchmark.bm do |x|
|
82
|
+
x.report("pop_algos_heap"){ 100_000.times{|d| minheap.pop } }
|
83
|
+
x.report("pop_mm_heap"){ 100_000.times{|d| mm_hp.pop_min } }
|
84
|
+
x.report("pop_mm_heap_batches"){ 1000.times{|d| mm_hp.pop_min(100) } }
|
85
|
+
end
|
data/ext/min_max/src/lib.rs
CHANGED
@@ -1,5 +1,5 @@
|
|
1
1
|
use magnus::scan_args::scan_args;
|
2
|
-
use magnus::value::
|
2
|
+
use magnus::value::ReprValue;
|
3
3
|
use magnus::{
|
4
4
|
block::{block_given, Yield},
|
5
5
|
define_class, function, method,
|
@@ -9,32 +9,11 @@ use magnus::{
|
|
9
9
|
use magnus::{DataTypeFunctions, Integer, TypedData};
|
10
10
|
use min_max_heap::MinMaxHeap;
|
11
11
|
use std::cell::RefCell;
|
12
|
-
use std::
|
13
|
-
use std::sync::{Arc, RwLock};
|
12
|
+
use std::rc::Rc;
|
14
13
|
|
15
14
|
#[derive(Debug, Clone)]
|
16
15
|
struct PriorityOrderableValue(i64, i64);
|
17
16
|
|
18
|
-
#[macro_use]
|
19
|
-
extern crate lazy_static;
|
20
|
-
|
21
|
-
struct MEM(
|
22
|
-
RwLock<HashMap<i64, BoxValue<Value>>>,
|
23
|
-
RwLock<HashMap<i64, RefCell<usize>>>,
|
24
|
-
);
|
25
|
-
|
26
|
-
impl MEM {
|
27
|
-
fn new() -> Self {
|
28
|
-
MEM(RwLock::new(HashMap::new()), RwLock::new(HashMap::new()))
|
29
|
-
}
|
30
|
-
}
|
31
|
-
|
32
|
-
unsafe impl Sync for MEM {}
|
33
|
-
|
34
|
-
lazy_static! {
|
35
|
-
static ref MEMORY: MEM = MEM::new();
|
36
|
-
}
|
37
|
-
|
38
17
|
impl PartialOrd for PriorityOrderableValue {
|
39
18
|
fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
|
40
19
|
Some(self.cmp(other))
|
@@ -43,9 +22,8 @@ impl PartialOrd for PriorityOrderableValue {
|
|
43
22
|
|
44
23
|
impl Ord for PriorityOrderableValue {
|
45
24
|
fn cmp(&self, other: &Self) -> std::cmp::Ordering {
|
46
|
-
|
47
|
-
|
48
|
-
}
|
25
|
+
let (PriorityOrderableValue(p1, _), PriorityOrderableValue(p2, _)) = (self, other);
|
26
|
+
p1.cmp(p2)
|
49
27
|
}
|
50
28
|
}
|
51
29
|
|
@@ -53,11 +31,8 @@ impl Eq for PriorityOrderableValue {}
|
|
53
31
|
|
54
32
|
impl PartialEq for PriorityOrderableValue {
|
55
33
|
fn eq(&self, other: &Self) -> bool {
|
56
|
-
|
57
|
-
|
58
|
-
p1.eq(p2) && (v1).eq(v2)
|
59
|
-
}
|
60
|
-
}
|
34
|
+
let (PriorityOrderableValue(p1, v1), PriorityOrderableValue(p2, v2)) = (self, other);
|
35
|
+
p1.eq(p2) && (v1).eq(v2)
|
61
36
|
}
|
62
37
|
}
|
63
38
|
|
@@ -65,21 +40,21 @@ unsafe impl Send for RubyMinMaxHeap {}
|
|
65
40
|
#[derive(DataTypeFunctions, TypedData, Clone)]
|
66
41
|
#[magnus(class = "MinMax", size, free_immediately, mark)]
|
67
42
|
struct RubyMinMaxHeap {
|
68
|
-
heap:
|
43
|
+
heap: Rc<RefCell<MinMaxHeap<PriorityOrderableValue>>>,
|
69
44
|
}
|
70
45
|
|
71
46
|
impl RubyMinMaxHeap {
|
72
47
|
fn new() -> Result<Self, Error> {
|
73
48
|
Ok(RubyMinMaxHeap {
|
74
|
-
heap:
|
49
|
+
heap: Rc::new(RefCell::new(MinMaxHeap::new())),
|
75
50
|
})
|
76
51
|
}
|
77
52
|
|
78
53
|
fn push(&self, values: RArray) -> Result<(), Error> {
|
79
54
|
let mut hp = self.heap.borrow_mut();
|
80
|
-
let values_vec = values.to_vec::<(i64, i64)>()
|
81
|
-
values_vec.
|
82
|
-
hp.push(PriorityOrderableValue(priority, key));
|
55
|
+
let values_vec = values.to_vec::<(i64, i64)>()?;
|
56
|
+
values_vec.iter().for_each(|(priority, key)| {
|
57
|
+
hp.push(PriorityOrderableValue(*priority, *key));
|
83
58
|
});
|
84
59
|
Ok(())
|
85
60
|
}
|
@@ -213,8 +188,8 @@ impl RubyMinMaxHeap {
|
|
213
188
|
Ok(ary)
|
214
189
|
}
|
215
190
|
|
216
|
-
fn clear(&self)
|
217
|
-
|
191
|
+
fn clear(&self) {
|
192
|
+
self.heap.borrow_mut().clear()
|
218
193
|
}
|
219
194
|
}
|
220
195
|
|
data/lib/min_max/version.rb
CHANGED
data/lib/min_max.rb
CHANGED
@@ -6,46 +6,35 @@ require_relative "min_max/min_max"
|
|
6
6
|
class MinMax
|
7
7
|
class Error < StandardError; end
|
8
8
|
|
9
|
+
attr_reader :priority_blk, :storage, :mtx
|
10
|
+
|
9
11
|
def self.[](*args, &blk)
|
10
12
|
new(*args, &blk)
|
11
13
|
end
|
12
14
|
|
13
15
|
def self.new(*args, &blk)
|
14
16
|
self._new.tap{|s|
|
15
|
-
s.
|
17
|
+
s.instance_eval{
|
18
|
+
@priority_blk = (blk || proc{|x| x.respond_to?(:priority) ? x.priority : x.to_i })
|
19
|
+
@storage = Hash.new{|h,k| h[k] = [0, nil] }
|
20
|
+
@mtx ||= Mutex.new
|
21
|
+
}
|
16
22
|
s.push(*args)
|
17
23
|
}
|
18
24
|
end
|
19
25
|
|
20
|
-
def priority_blk=(bll)
|
21
|
-
@priority_blk = bll
|
22
|
-
end
|
23
|
-
|
24
|
-
def storage
|
25
|
-
@storage ||= Hash.new
|
26
|
-
end
|
27
|
-
|
28
|
-
def counts
|
29
|
-
@counts ||= Hash.new(0)
|
30
|
-
end
|
31
|
-
|
32
|
-
def mtx
|
33
|
-
@_mtx ||= Mutex.new
|
34
|
-
end
|
35
|
-
|
36
26
|
def push(*args)
|
37
|
-
|
38
|
-
|
39
|
-
|
40
|
-
|
41
|
-
|
42
|
-
|
43
|
-
|
44
|
-
|
45
|
-
|
46
|
-
_push(mapped)
|
27
|
+
mapped = args.map do |a|
|
28
|
+
hash = a.hash
|
29
|
+
entry = self.storage[hash]
|
30
|
+
entry[0] += 1
|
31
|
+
entry[1] ||= a
|
32
|
+
[
|
33
|
+
(self.priority_blk.call(a) rescue 0),
|
34
|
+
hash
|
35
|
+
]
|
47
36
|
end
|
48
|
-
|
37
|
+
_push(mapped)
|
49
38
|
end
|
50
39
|
|
51
40
|
def add(*args)
|
@@ -53,29 +42,13 @@ class MinMax
|
|
53
42
|
end
|
54
43
|
|
55
44
|
def pop_max(*args)
|
56
|
-
|
57
|
-
|
58
|
-
popped.kind_of?(Array) ? popped.map{|p| retrieve(p) } : retrieve(popped)
|
59
|
-
}
|
45
|
+
popped = _pop_max(*args)
|
46
|
+
popped.kind_of?(Array) ? popped.map{|p| retrieve(p) } : retrieve(popped)
|
60
47
|
end
|
61
48
|
|
62
49
|
def pop_min(*args)
|
63
|
-
|
64
|
-
|
65
|
-
popped.kind_of?(Array) ? popped.map{|p| retrieve(p) } : retrieve(popped)
|
66
|
-
}
|
67
|
-
end
|
68
|
-
|
69
|
-
def retrieve(hash, remove=true)
|
70
|
-
if remove
|
71
|
-
if (counts[hash] -= 1) == 0
|
72
|
-
storage.delete(hash)
|
73
|
-
else
|
74
|
-
storage[hash]
|
75
|
-
end
|
76
|
-
else
|
77
|
-
storage[hash]
|
78
|
-
end
|
50
|
+
popped = _pop_min(*args)
|
51
|
+
popped.kind_of?(Array) ? popped.map{|p| retrieve(p) } : retrieve(popped)
|
79
52
|
end
|
80
53
|
|
81
54
|
def peek_min(*args)
|
@@ -100,6 +73,14 @@ class MinMax
|
|
100
73
|
each.to_a
|
101
74
|
end
|
102
75
|
|
76
|
+
def count(val)
|
77
|
+
counts.has_key?(val.hash) ? counts[val.hash] : 0
|
78
|
+
end
|
79
|
+
|
80
|
+
def contains?(val)
|
81
|
+
counts.has_key?(val.hash) && counts[val.hash] > 0
|
82
|
+
end
|
83
|
+
|
103
84
|
def to_a_asc
|
104
85
|
mtx.synchronize { _to_a_asc.map{|p| retrieve(p, false) } }
|
105
86
|
end
|
@@ -109,7 +90,15 @@ class MinMax
|
|
109
90
|
end
|
110
91
|
|
111
92
|
def inspect
|
112
|
-
"MinMax[#{
|
93
|
+
"MinMax[#{_each.first(10).map{|v| retrieve(v, false).to_s }.join(", ")}#{size > 10 ? ", ..." : ""}]"
|
94
|
+
end
|
95
|
+
|
96
|
+
private
|
97
|
+
def retrieve(hash, remove=true)
|
98
|
+
entry = self.storage[hash]
|
99
|
+
self.storage.delete(hash) if remove && (entry[0] -= 1) == 0
|
100
|
+
entry[1]
|
113
101
|
end
|
102
|
+
|
114
103
|
end
|
115
104
|
|
metadata
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: min_max
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.1.
|
4
|
+
version: 0.1.5
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Wouter Coppieters
|
@@ -66,6 +66,7 @@ files:
|
|
66
66
|
- LICENSE.txt
|
67
67
|
- README.md
|
68
68
|
- Rakefile
|
69
|
+
- benchmarks/benchmarks.rb
|
69
70
|
- ext/min_max/Cargo.toml
|
70
71
|
- ext/min_max/extconf.rb
|
71
72
|
- ext/min_max/src/lib.rs
|