zillabyte 0.0.6
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +15 -0
- data/ruby/README.md +2 -0
- data/ruby/lib/zillabyte.rb +20 -0
- data/ruby/lib/zillabyte/common/progress.rb +17 -0
- data/ruby/lib/zillabyte/harness.rb +16 -0
- data/ruby/lib/zillabyte/harness/aggregate.rb +29 -0
- data/ruby/lib/zillabyte/harness/counter.rb +11 -0
- data/ruby/lib/zillabyte/harness/each.rb +29 -0
- data/ruby/lib/zillabyte/harness/groupby.rb +9 -0
- data/ruby/lib/zillabyte/harness/helper.rb +326 -0
- data/ruby/lib/zillabyte/harness/live_delegator.rb +369 -0
- data/ruby/lib/zillabyte/harness/simple_function.rb +131 -0
- data/ruby/lib/zillabyte/harness/simple_spout.rb +90 -0
- data/ruby/lib/zillabyte/harness/sink.rb +23 -0
- data/ruby/lib/zillabyte/harness/spout.rb +48 -0
- data/ruby/lib/zillabyte/harness/topology.rb +132 -0
- data/ruby/lib/zillabyte/harness/tuple.rb +32 -0
- data/ruby/lib/zillabyte/version.rb +3 -0
- metadata +103 -0
@@ -0,0 +1,369 @@
|
|
1
|
+
require "rubygems"
|
2
|
+
require "json"
|
3
|
+
|
4
|
+
module Zillabyte
|
5
|
+
|
6
|
+
module Storm
|
7
|
+
|
8
|
+
module Protocol
|
9
|
+
|
10
|
+
class << self
|
11
|
+
attr_accessor :mode, :pending_commands, :pipe_name, :pipe_to_java, :pipe_from_java
|
12
|
+
end
|
13
|
+
|
14
|
+
self.pending_commands = []
|
15
|
+
|
16
|
+
def setup_pipes
|
17
|
+
pipe_name = Storm::Protocol.pipe_name
|
18
|
+
if(pipe_name and File.exist?(pipe_name+".in"))
|
19
|
+
Storm::Protocol.pipe_to_java = File.open("#{pipe_name}.in","w+")
|
20
|
+
else
|
21
|
+
Storm::Protocol.pipe_to_java = $stdout
|
22
|
+
end
|
23
|
+
|
24
|
+
if(pipe_name and File.exist?(pipe_name+".out"))
|
25
|
+
Storm::Protocol.pipe_from_java = File.open("#{pipe_name}.out","r+")
|
26
|
+
else
|
27
|
+
Storm::Protocol.pipe_from_java = $stdin
|
28
|
+
end
|
29
|
+
end
|
30
|
+
|
31
|
+
def read_message
|
32
|
+
msg = ""
|
33
|
+
loop do
|
34
|
+
line = Storm::Protocol.pipe_from_java.readline.chomp#STDIN.readline.chomp
|
35
|
+
break if line == "end"
|
36
|
+
msg << line
|
37
|
+
msg << "\n"
|
38
|
+
end
|
39
|
+
|
40
|
+
encoding_options = {
|
41
|
+
:invalid => :replace, # Replace invalid byte sequences
|
42
|
+
:undef => :replace, # Replace anything not defined in ASCII
|
43
|
+
:replace => '' # Use a blank for those replacements
|
44
|
+
}
|
45
|
+
msg.encode! Encoding.find('ASCII'), encoding_options
|
46
|
+
|
47
|
+
JSON.parse msg.chomp
|
48
|
+
end
|
49
|
+
|
50
|
+
def read_command
|
51
|
+
Storm::Protocol.pending_commands.shift ||
|
52
|
+
begin
|
53
|
+
msg = read_message
|
54
|
+
while msg.is_a? Array
|
55
|
+
Storm::Protocol.pending_taskids.push(msg)
|
56
|
+
msg = read_message
|
57
|
+
end
|
58
|
+
msg
|
59
|
+
rescue JSON::ParserError => e
|
60
|
+
fail "error: #{e}"
|
61
|
+
{}
|
62
|
+
end
|
63
|
+
end
|
64
|
+
|
65
|
+
def send_msg_to_parent(msg)
|
66
|
+
# log msg
|
67
|
+
# log msg.to_json
|
68
|
+
Storm::Protocol.pipe_to_java.write(msg.to_json+"\n")
|
69
|
+
Storm::Protocol.pipe_to_java.write("end\n")
|
70
|
+
Storm::Protocol.pipe_to_java.flush
|
71
|
+
end
|
72
|
+
|
73
|
+
def send_pid(heartbeat_dir)
|
74
|
+
pid = Process.pid
|
75
|
+
send_msg_to_parent({'pid' => pid})
|
76
|
+
File.open("#{heartbeat_dir}/#{pid}", "w").close
|
77
|
+
end
|
78
|
+
|
79
|
+
def emit(*args)
|
80
|
+
|
81
|
+
stream = nil
|
82
|
+
tuple = nil
|
83
|
+
meta = {}
|
84
|
+
|
85
|
+
if (args.size == 1)
|
86
|
+
tuple = args[0]
|
87
|
+
elsif(args.size == 2)
|
88
|
+
stream = args[0]
|
89
|
+
tuple = args[1]
|
90
|
+
else
|
91
|
+
stream = args[0]
|
92
|
+
tuple = args[1]
|
93
|
+
meta = args[2]
|
94
|
+
end
|
95
|
+
|
96
|
+
m = {:command => :emit, :tuple => tuple, :stream => stream, :meta => meta}
|
97
|
+
send_msg_to_parent m
|
98
|
+
end
|
99
|
+
|
100
|
+
def ack(tup)
|
101
|
+
send_msg_to_parent :command => :ack, :id => tup.id
|
102
|
+
end
|
103
|
+
|
104
|
+
def done()
|
105
|
+
send_msg_to_parent :command => :done #, :id => tup.id
|
106
|
+
# read_task_ids
|
107
|
+
end
|
108
|
+
|
109
|
+
def fail(msg)
|
110
|
+
send_msg_to_parent :command => :fail, :msg => msg.to_s
|
111
|
+
end
|
112
|
+
|
113
|
+
def log(msg)
|
114
|
+
send_msg_to_parent :command => :log, :msg => msg.to_s
|
115
|
+
end
|
116
|
+
|
117
|
+
def handshake
|
118
|
+
setup_info = read_message
|
119
|
+
send_pid setup_info['pidDir']
|
120
|
+
[setup_info['conf'], setup_info['context']]
|
121
|
+
end
|
122
|
+
end
|
123
|
+
|
124
|
+
|
125
|
+
class Spout
|
126
|
+
include Storm::Protocol
|
127
|
+
|
128
|
+
def open(conf, context); end
|
129
|
+
|
130
|
+
def next_batch; end
|
131
|
+
|
132
|
+
def ack(id); end
|
133
|
+
|
134
|
+
def run(pipe_name)
|
135
|
+
Storm::Protocol.mode = 'spout'
|
136
|
+
Storm::Protocol.pipe_name = pipe_name
|
137
|
+
setup_pipes
|
138
|
+
open(*handshake)
|
139
|
+
|
140
|
+
begin
|
141
|
+
while true
|
142
|
+
msg = read_command
|
143
|
+
case msg['command']
|
144
|
+
when 'next'
|
145
|
+
next_batch
|
146
|
+
done
|
147
|
+
when 'ack'
|
148
|
+
ack(msg['id'])
|
149
|
+
when 'fail'
|
150
|
+
fail(msg['id'])
|
151
|
+
end
|
152
|
+
end
|
153
|
+
rescue Exception => e
|
154
|
+
fail 'Exception in spout: ' + e.message + ' - ' + e.backtrace.join('\n')
|
155
|
+
end
|
156
|
+
end
|
157
|
+
end
|
158
|
+
|
159
|
+
class Each
|
160
|
+
include Storm::Protocol
|
161
|
+
|
162
|
+
def prepare(conf, context); end
|
163
|
+
|
164
|
+
def execute(tuple); end
|
165
|
+
|
166
|
+
def run(pipe_name)
|
167
|
+
Storm::Protocol.mode = 'each'
|
168
|
+
Storm::Protocol.pipe_name = pipe_name
|
169
|
+
setup_pipes
|
170
|
+
prepare(*handshake)
|
171
|
+
|
172
|
+
begin
|
173
|
+
while true
|
174
|
+
t = nil
|
175
|
+
begin
|
176
|
+
t = Tuple.from_hash(read_command)
|
177
|
+
if(!t)
|
178
|
+
next
|
179
|
+
end
|
180
|
+
execute t
|
181
|
+
rescue Exception => e
|
182
|
+
fail 'Exception in bolt (A): ' + e.message + ' - ' + e.backtrace.join('\n')
|
183
|
+
ensure
|
184
|
+
done()
|
185
|
+
end
|
186
|
+
end
|
187
|
+
rescue Exception => e
|
188
|
+
fail 'Exception in bolt (B): ' + e.message + ' - ' + e.backtrace.join('\n')
|
189
|
+
end
|
190
|
+
end
|
191
|
+
end
|
192
|
+
|
193
|
+
class AggregateStart
|
194
|
+
include Storm::Protocol
|
195
|
+
|
196
|
+
def start(tuple); end
|
197
|
+
|
198
|
+
def run(pipe_name)
|
199
|
+
Storm::Protocol.mode = 'agg_start'
|
200
|
+
Storm::Protocol.pipe_name = pipe_name
|
201
|
+
setup_pipes
|
202
|
+
handshake
|
203
|
+
|
204
|
+
begin
|
205
|
+
while true
|
206
|
+
t = nil
|
207
|
+
begin
|
208
|
+
t = Tuple.from_hash(read_command)
|
209
|
+
if(!t)
|
210
|
+
next
|
211
|
+
end
|
212
|
+
start t
|
213
|
+
rescue Exception => e
|
214
|
+
fail 'Exception in bolt (A): ' + e.message + ' - ' + e.backtrace.join('\n')
|
215
|
+
ensure
|
216
|
+
done()
|
217
|
+
end
|
218
|
+
end
|
219
|
+
rescue Exception => e
|
220
|
+
fail 'Exception in bolt (B): ' + e.message + ' - ' + e.backtrace.join('\n')
|
221
|
+
end
|
222
|
+
end
|
223
|
+
end
|
224
|
+
|
225
|
+
class AggregateAggregate
|
226
|
+
include Storm::Protocol
|
227
|
+
|
228
|
+
def aggregate(tuple1, tuple2); end
|
229
|
+
|
230
|
+
def run(pipe_name)
|
231
|
+
Storm::Protocol.mode = 'agg_aggregate'
|
232
|
+
Storm::Protocol.pipe_name = pipe_name
|
233
|
+
setup_pipes
|
234
|
+
handshake
|
235
|
+
|
236
|
+
begin
|
237
|
+
while true
|
238
|
+
t = nil
|
239
|
+
begin
|
240
|
+
t1 = Tuple.from_hash(read_command)
|
241
|
+
if(!t1)
|
242
|
+
next
|
243
|
+
end
|
244
|
+
t2 = Tuple.from_hash(read_command)
|
245
|
+
if(!t2)
|
246
|
+
next
|
247
|
+
end
|
248
|
+
aggregate t1, t2
|
249
|
+
rescue Exception => e
|
250
|
+
fail 'Exception in bolt (A): ' + e.message + ' - ' + e.backtrace.join('\n')
|
251
|
+
ensure
|
252
|
+
done()
|
253
|
+
end
|
254
|
+
end
|
255
|
+
rescue Exception => e
|
256
|
+
fail 'Exception in bolt (B): ' + e.message + ' - ' + e.backtrace.join('\n')
|
257
|
+
end
|
258
|
+
end
|
259
|
+
end
|
260
|
+
|
261
|
+
class AggregateComplete
|
262
|
+
include Storm::Protocol
|
263
|
+
|
264
|
+
def complete(tuple); end
|
265
|
+
|
266
|
+
def run(pipe_name)
|
267
|
+
Storm::Protocol.mode = 'agg_complete'
|
268
|
+
Storm::Protocol.pipe_name = pipe_name
|
269
|
+
setup_pipes
|
270
|
+
handshake
|
271
|
+
|
272
|
+
begin
|
273
|
+
while true
|
274
|
+
t = nil
|
275
|
+
begin
|
276
|
+
t = Tuple.from_hash(read_command)
|
277
|
+
if(!t)
|
278
|
+
next
|
279
|
+
end
|
280
|
+
complete t
|
281
|
+
rescue Exception => e
|
282
|
+
fail 'Exception in bolt (A): ' + e.message + ' - ' + e.backtrace.join('\n')
|
283
|
+
ensure
|
284
|
+
done()
|
285
|
+
end
|
286
|
+
end
|
287
|
+
rescue Exception => e
|
288
|
+
fail 'Exception in bolt (B): ' + e.message + ' - ' + e.backtrace.join('\n')
|
289
|
+
end
|
290
|
+
end
|
291
|
+
end
|
292
|
+
end
|
293
|
+
|
294
|
+
module Harness
|
295
|
+
|
296
|
+
class SpoutController < Storm::Spout
|
297
|
+
|
298
|
+
def initialize(harness, progress)
|
299
|
+
@harness = harness
|
300
|
+
@progress = progress
|
301
|
+
end
|
302
|
+
|
303
|
+
def open(*args)
|
304
|
+
@harness._prepare.call(self, *args) if @harness._prepare
|
305
|
+
end
|
306
|
+
|
307
|
+
def next_batch
|
308
|
+
@harness._next_batch.call(self)
|
309
|
+
end
|
310
|
+
|
311
|
+
end
|
312
|
+
|
313
|
+
class EachController < Storm::Each
|
314
|
+
|
315
|
+
def initialize(harness, progress)
|
316
|
+
@harness = harness
|
317
|
+
@progress = progress
|
318
|
+
end
|
319
|
+
|
320
|
+
def execute(*args)
|
321
|
+
@harness._execute.call(self, *args)
|
322
|
+
end
|
323
|
+
|
324
|
+
def prepare(*args)
|
325
|
+
@harness._prepare.call(self, *args) if @harness._prepare
|
326
|
+
end
|
327
|
+
|
328
|
+
end
|
329
|
+
|
330
|
+
class AggregateStartController < Storm::AggregateStart
|
331
|
+
|
332
|
+
def initialize(harness, progress)
|
333
|
+
@harness = harness
|
334
|
+
@progress = progress
|
335
|
+
end
|
336
|
+
|
337
|
+
def start(*args)
|
338
|
+
@harness._start.call(self, *args)
|
339
|
+
end
|
340
|
+
|
341
|
+
end
|
342
|
+
|
343
|
+
class AggregateAggregateController < Storm::AggregateAggregate
|
344
|
+
|
345
|
+
def initialize(harness, progress)
|
346
|
+
@harness = harness
|
347
|
+
@progress = progress
|
348
|
+
end
|
349
|
+
|
350
|
+
def aggregate(*args)
|
351
|
+
@harness._aggregate.call(self, *args)
|
352
|
+
end
|
353
|
+
|
354
|
+
end
|
355
|
+
|
356
|
+
class AggregateCompleteController < Storm::AggregateComplete
|
357
|
+
|
358
|
+
def initialize(harness, progress)
|
359
|
+
@harness = harness
|
360
|
+
@progress = progress
|
361
|
+
end
|
362
|
+
|
363
|
+
def complete(*args)
|
364
|
+
@harness._complete.call(self, *args)
|
365
|
+
end
|
366
|
+
|
367
|
+
end
|
368
|
+
end
|
369
|
+
end
|
@@ -0,0 +1,131 @@
|
|
1
|
+
require 'optparse'
|
2
|
+
|
3
|
+
class Zillabyte::Harness::SimpleFunction
|
4
|
+
attr_accessor :_nodes, :_relation, :_matches, :_emits, :_prepare, :_execute, :_name, :_info_file, :_options
|
5
|
+
|
6
|
+
def self.build()
|
7
|
+
h = Zillabyte::Harness::SimpleFunction.new()
|
8
|
+
yield(h)
|
9
|
+
h._name = h._name || Dir.pwd.split("/")[-1]
|
10
|
+
Zillabyte::Harness::Helper.check_name("simple_function", h._name, {})
|
11
|
+
Zillabyte::Harness::Helper.check_emits("simple_function", h._emits, {})
|
12
|
+
generic_emits = h.get_generic_emits
|
13
|
+
|
14
|
+
h._nodes = []
|
15
|
+
h._options = Zillabyte::Harness::Helper.opt_parser()
|
16
|
+
|
17
|
+
if(h._options[:command] == :info)
|
18
|
+
h._info_file = File.open(h._options[:file],"w+")
|
19
|
+
hash = {"language" => "ruby", "name" => h._name}
|
20
|
+
Zillabyte::Harness::Helper.write_hash_to_file(hash, h._info_file)
|
21
|
+
end
|
22
|
+
|
23
|
+
h.build_spout()
|
24
|
+
fn = h.build_each(generic_emits)
|
25
|
+
h.build_sink()
|
26
|
+
|
27
|
+
if(h._options[:command] == :execute and h._options[:name] == h._name)
|
28
|
+
pipe_name = h._options[:pipe]
|
29
|
+
c = Zillabyte::Harness::EachController.new(fn, progress = Zillabyte::Common::Progress.new)
|
30
|
+
c.run(pipe_name)
|
31
|
+
end
|
32
|
+
|
33
|
+
h
|
34
|
+
end
|
35
|
+
|
36
|
+
def name(v)
|
37
|
+
@_name = v
|
38
|
+
end
|
39
|
+
|
40
|
+
def matches(v, options = {})
|
41
|
+
case v
|
42
|
+
when String
|
43
|
+
@_relation = { :query => v, :options => options.is_a?(Hash) ? options : {} }
|
44
|
+
when Array
|
45
|
+
@_matches = v
|
46
|
+
end
|
47
|
+
end
|
48
|
+
|
49
|
+
def emits(v)
|
50
|
+
@_emits = v
|
51
|
+
end
|
52
|
+
|
53
|
+
def prepare(&block)
|
54
|
+
@_prepare = block
|
55
|
+
end
|
56
|
+
|
57
|
+
def execute(&block)
|
58
|
+
@_execute = block
|
59
|
+
end
|
60
|
+
|
61
|
+
def get_generic_emits()
|
62
|
+
generic_emits = []
|
63
|
+
@_emits.each do |relation|
|
64
|
+
temit = []
|
65
|
+
relation[1].each do |column|
|
66
|
+
column.each do |col, type|
|
67
|
+
temit << col
|
68
|
+
end
|
69
|
+
end
|
70
|
+
generic_emits << [relation[0], temit]
|
71
|
+
end
|
72
|
+
generic_emits
|
73
|
+
end
|
74
|
+
|
75
|
+
def build_spout()
|
76
|
+
h = Zillabyte::Harness::Spout.new(false)
|
77
|
+
h._matches = @_matches if @_matches
|
78
|
+
h._relation = @_relation if @_relation
|
79
|
+
@_nodes << h
|
80
|
+
if(@_options[:command] == :info)
|
81
|
+
info_hash = {"name" => h._name, "type" => h._type}
|
82
|
+
if(h._relation)
|
83
|
+
info_hash["relation"] = h._relation
|
84
|
+
elsif(h._matches)
|
85
|
+
info_hash["matches"] = h._matches
|
86
|
+
end
|
87
|
+
Zillabyte::Harness::Helper.write_hash_to_file(info_hash, @_info_file)
|
88
|
+
end
|
89
|
+
end
|
90
|
+
|
91
|
+
def build_each(generic_emits)
|
92
|
+
h = Zillabyte::Harness::Each.new()
|
93
|
+
h._name = @_name
|
94
|
+
h._emits = generic_emits
|
95
|
+
h._prepare = @_prepare
|
96
|
+
h._execute = @_execute
|
97
|
+
@_nodes << h
|
98
|
+
if(@_options[:command] == :info)
|
99
|
+
info_hash = {"name" => h._name, "type" => h._type, "emits" => h._emits}
|
100
|
+
Zillabyte::Harness::Helper.write_hash_to_file(info_hash, @_info_file)
|
101
|
+
end
|
102
|
+
h
|
103
|
+
end
|
104
|
+
|
105
|
+
def build_sink()
|
106
|
+
# Construct the sink...
|
107
|
+
n_sinks = @_emits.length
|
108
|
+
@_emits.each do |emit|
|
109
|
+
h = Zillabyte::Harness::Sink.new()
|
110
|
+
h._name = emit[0]
|
111
|
+
columns = emit[1]
|
112
|
+
columns.each do |col|
|
113
|
+
col.each do |cname, ctype|
|
114
|
+
h.column(cname, ctype)
|
115
|
+
end
|
116
|
+
end
|
117
|
+
if(n_sinks > 1)
|
118
|
+
h._consumes = h._name
|
119
|
+
end
|
120
|
+
@_nodes << h
|
121
|
+
if(@_options[:command] == :info)
|
122
|
+
info_hash = {"name" => h._name, "type" => h._type, "columns" => h._columns}
|
123
|
+
if(h._consumes)
|
124
|
+
info_hash["consumes"] = h._consumes
|
125
|
+
end
|
126
|
+
Zillabyte::Harness::Helper.write_hash_to_file(info_hash, @_info_file)
|
127
|
+
end
|
128
|
+
end
|
129
|
+
end
|
130
|
+
|
131
|
+
end
|