Class: EventMachine::Iterator

Inherits:
Object
  • Object
show all
Defined in:
lib/em/iterator.rb

Overview

A simple iterator for concurrent asynchronous work.

Unlike ruby's built-in iterators, the end of the current iteration cycle is signaled manually, instead of happening automatically after the yielded block finishes executing. For example:

(0..10).each{ |num| }

becomes:

EM::Iterator.new(0..10).each{ |num,iter| iter.next }

This is especially useful when doing asynchronous work via reactor libraries and functions. For example, given a sync and async http api:

response = sync_http_get(url); ... async_http_get(url){ |response| ... }

a synchronous iterator such as:

responses = urls.map{ |url| sync_http_get(url) } ... puts 'all done!'

could be written as:

EM::Iterator.new(urls).map(proc{ |url,iter| async_http_get(url){ |res| iter.return(res) } }, proc{ |responses| ... puts 'all done!' })

Now, you can take advantage of the asynchronous api to issue requests in parallel. For example, to fetch 10 urls at a time, simply pass in a concurrency of 10:

EM::Iterator.new(urls, 10).each do |url,iter| async_http_get(url){ iter.next } end

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(list, concurrency = 1) ⇒ Iterator

Create a new parallel async iterator with specified concurrency.

i = EM::Iterator.new(1..100, 10)

will create an iterator over the range that processes 10 items at a time. Iteration is started via #each, #map or #inject

Raises:

  • (ArgumentError) 


51
52
53
54
55
56
57
58
59
 
# File 'lib/em/iterator.rb', line 51

def initialize(list, concurrency = 1)
  raise ArgumentError, 'argument must be an array' unless list.respond_to?(:to_a)
  raise ArgumentError, 'concurrency must be bigger than zero' unless (concurrency > 0)
  @list = list.to_a.dup
  @concurrency = concurrency

  @started = false
  @ended = false
end

Instance Attribute Details

#concurrency ⇒ Object

Returns the value of attribute concurrency.



70
71
72
 
# File 'lib/em/iterator.rb', line 70

def concurrency
  @concurrency
end

Instance Method Details

#each(foreach = nil, after = nil, &blk) ⇒ Object

Iterate over a set of items using the specified block or proc.

EM::Iterator.new(1..100).each do |num, iter| puts num iter.next end

An optional second proc is invoked after the iteration is complete.

EM::Iterator.new(1..100).each( proc{ |num,iter| iter.next }, proc{ puts 'all done' } )

Raises:

  • (ArgumentError) 


86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
 
# File 'lib/em/iterator.rb', line 86

def each(foreach=nil, after=nil, &blk)
  raise ArgumentError, 'proc or block required for iteration' unless foreach ||= blk
  raise RuntimeError, 'cannot iterate over an iterator more than once' if @started or @ended

  @started = true
  @pending = 0
  @workers = 0

  all_done = proc{
    after.call if after and @ended and @pending == 0
  }

  @process_next = proc{
    # p [:process_next, :pending=, @pending, :workers=, @workers, :ended=, @ended, :concurrency=, @concurrency, :list=, @list]
    unless @ended or @workers > @concurrency
      if @list.empty?
        @ended = true
        @workers -= 1
        all_done.call
      else
        item = @list.shift
        @pending += 1

        is_done = false
        on_done = proc{
          raise RuntimeError, 'already completed this iteration' if is_done
          is_done = true

          @pending -= 1

          if @ended
            all_done.call
          else
            EM.next_tick(@process_next)
          end
        }
        class << on_done
          alias :next :call
        end

        foreach.call(item, on_done)
      end
    else
      @workers -= 1
    end
  }

  spawn_workers

  self
end

#inject(obj, foreach, after) ⇒ Object

Inject the results of an asynchronous iteration onto a given object.

EM::Iterator.new(%w[ pwd uptime uname date ], 2).inject({}, proc{ |hash,cmd,iter| EM.system(cmd){ |output,status| hash[cmd] = status.exitstatus == 0 ? output.strip : nil iter.return(hash) } }, proc{ |results| p results })



187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
 
# File 'lib/em/iterator.rb', line 187

def inject(obj, foreach, after)
  each(proc{ |item,iter|
    is_done = false
    on_done = proc{ |res|
      raise RuntimeError, 'already returned a value for this iteration' if is_done
      is_done = true

      obj = res
      iter.next
    }
    class << on_done
      alias :return :call
      def next
        raise NoMethodError, 'must call #return on an inject iterator'
      end
    end

    foreach.call(obj, item, on_done)
  }, proc{
    after.call(obj)
  })
end

#map(foreach, after) ⇒ Object

Collect the results of an asynchronous iteration into an array.

EM::Iterator.new(%w[ pwd uptime uname date ], 2).map(proc{ |cmd,iter| EM.system(cmd){ |output,status| iter.return(output) } }, proc{ |results| p results })



148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
 
# File 'lib/em/iterator.rb', line 148

def map(foreach, after)
  index = 0

  inject([], proc{ |results,item,iter|
    i = index
    index += 1

    is_done = false
    on_done = proc{ |res|
      raise RuntimeError, 'already returned a value for this iteration' if is_done
      is_done = true

      results[i] = res
      iter.return(results)
    }
    class << on_done
      alias :return :call
      def next
        raise NoMethodError, 'must call #return on a map iterator'
      end
    end

    foreach.call(item, on_done)
  }, proc{ |results|
    after.call(results)
  })
end