scout-gear 9.1.0 → 10.0.1

Sign up to get free protection for your applications and to get access to all the features.
Files changed (116) hide show
  1. checksums.yaml +4 -4
  2. data/.vimproject +3 -3
  3. data/Rakefile +1 -2
  4. data/VERSION +1 -1
  5. data/lib/scout/offsite/exceptions.rb +9 -0
  6. data/lib/scout/offsite/ssh.rb +1 -0
  7. data/lib/scout/semaphore.rb +2 -0
  8. data/lib/scout/tsv/open.rb +1 -1
  9. data/lib/scout/tsv/parser.rb +1 -1
  10. data/lib/scout/tsv/persist/adapter.rb +2 -2
  11. data/lib/scout/tsv/persist.rb +1 -0
  12. data/lib/scout/tsv.rb +1 -1
  13. data/lib/scout/work_queue/exceptions.rb +18 -0
  14. data/lib/scout/work_queue.rb +1 -0
  15. data/lib/scout/workflow/definition.rb +1 -1
  16. data/lib/scout/workflow/step/config.rb +1 -1
  17. data/lib/scout/workflow/step.rb +2 -2
  18. data/lib/scout/workflow/task/inputs.rb +1 -1
  19. data/lib/scout/workflow/task.rb +2 -2
  20. data/lib/scout/workflow.rb +2 -2
  21. data/lib/scout-gear.rb +1 -10
  22. data/scout-gear.gemspec +8 -98
  23. data/test/scout/workflow/task/test_inputs.rb +0 -4
  24. metadata +7 -110
  25. data/lib/scout/cmd.rb +0 -347
  26. data/lib/scout/concurrent_stream.rb +0 -284
  27. data/lib/scout/config.rb +0 -168
  28. data/lib/scout/exceptions.rb +0 -151
  29. data/lib/scout/indiferent_hash/case_insensitive.rb +0 -30
  30. data/lib/scout/indiferent_hash/options.rb +0 -115
  31. data/lib/scout/indiferent_hash.rb +0 -96
  32. data/lib/scout/log/color.rb +0 -224
  33. data/lib/scout/log/color_class.rb +0 -269
  34. data/lib/scout/log/fingerprint.rb +0 -69
  35. data/lib/scout/log/progress/report.rb +0 -244
  36. data/lib/scout/log/progress/util.rb +0 -173
  37. data/lib/scout/log/progress.rb +0 -106
  38. data/lib/scout/log/trap.rb +0 -107
  39. data/lib/scout/log.rb +0 -441
  40. data/lib/scout/meta_extension.rb +0 -100
  41. data/lib/scout/misc/digest.rb +0 -63
  42. data/lib/scout/misc/filesystem.rb +0 -25
  43. data/lib/scout/misc/format.rb +0 -255
  44. data/lib/scout/misc/helper.rb +0 -31
  45. data/lib/scout/misc/insist.rb +0 -56
  46. data/lib/scout/misc/monitor.rb +0 -66
  47. data/lib/scout/misc/system.rb +0 -73
  48. data/lib/scout/misc.rb +0 -10
  49. data/lib/scout/named_array.rb +0 -138
  50. data/lib/scout/open/lock/lockfile.rb +0 -587
  51. data/lib/scout/open/lock.rb +0 -68
  52. data/lib/scout/open/remote.rb +0 -135
  53. data/lib/scout/open/stream.rb +0 -491
  54. data/lib/scout/open/util.rb +0 -244
  55. data/lib/scout/open.rb +0 -170
  56. data/lib/scout/path/find.rb +0 -204
  57. data/lib/scout/path/tmpfile.rb +0 -8
  58. data/lib/scout/path/util.rb +0 -127
  59. data/lib/scout/path.rb +0 -51
  60. data/lib/scout/persist/open.rb +0 -17
  61. data/lib/scout/persist/path.rb +0 -15
  62. data/lib/scout/persist/serialize.rb +0 -157
  63. data/lib/scout/persist.rb +0 -104
  64. data/lib/scout/resource/open.rb +0 -8
  65. data/lib/scout/resource/path.rb +0 -80
  66. data/lib/scout/resource/produce/rake.rb +0 -69
  67. data/lib/scout/resource/produce.rb +0 -151
  68. data/lib/scout/resource/scout.rb +0 -3
  69. data/lib/scout/resource/software.rb +0 -178
  70. data/lib/scout/resource/util.rb +0 -59
  71. data/lib/scout/resource.rb +0 -41
  72. data/lib/scout/simple_opt/accessor.rb +0 -54
  73. data/lib/scout/simple_opt/doc.rb +0 -126
  74. data/lib/scout/simple_opt/get.rb +0 -57
  75. data/lib/scout/simple_opt/parse.rb +0 -67
  76. data/lib/scout/simple_opt/setup.rb +0 -26
  77. data/lib/scout/simple_opt.rb +0 -5
  78. data/lib/scout/tmpfile.rb +0 -129
  79. data/test/scout/indiferent_hash/test_case_insensitive.rb +0 -16
  80. data/test/scout/indiferent_hash/test_options.rb +0 -46
  81. data/test/scout/log/test_progress.rb +0 -108
  82. data/test/scout/misc/test_digest.rb +0 -30
  83. data/test/scout/misc/test_filesystem.rb +0 -30
  84. data/test/scout/misc/test_insist.rb +0 -13
  85. data/test/scout/misc/test_system.rb +0 -21
  86. data/test/scout/open/test_lock.rb +0 -52
  87. data/test/scout/open/test_remote.rb +0 -25
  88. data/test/scout/open/test_stream.rb +0 -676
  89. data/test/scout/open/test_util.rb +0 -73
  90. data/test/scout/path/test_find.rb +0 -119
  91. data/test/scout/path/test_util.rb +0 -22
  92. data/test/scout/persist/test_open.rb +0 -37
  93. data/test/scout/persist/test_path.rb +0 -37
  94. data/test/scout/persist/test_serialize.rb +0 -114
  95. data/test/scout/resource/test_path.rb +0 -46
  96. data/test/scout/resource/test_produce.rb +0 -92
  97. data/test/scout/resource/test_software.rb +0 -24
  98. data/test/scout/resource/test_util.rb +0 -36
  99. data/test/scout/simple_opt/test_doc.rb +0 -16
  100. data/test/scout/simple_opt/test_get.rb +0 -11
  101. data/test/scout/simple_opt/test_parse.rb +0 -10
  102. data/test/scout/simple_opt/test_setup.rb +0 -77
  103. data/test/scout/test_cmd.rb +0 -85
  104. data/test/scout/test_concurrent_stream.rb +0 -29
  105. data/test/scout/test_config.rb +0 -66
  106. data/test/scout/test_indiferent_hash.rb +0 -26
  107. data/test/scout/test_log.rb +0 -32
  108. data/test/scout/test_meta_extension.rb +0 -80
  109. data/test/scout/test_misc.rb +0 -6
  110. data/test/scout/test_named_array.rb +0 -43
  111. data/test/scout/test_open.rb +0 -146
  112. data/test/scout/test_path.rb +0 -54
  113. data/test/scout/test_persist.rb +0 -186
  114. data/test/scout/test_resource.rb +0 -26
  115. data/test/scout/test_tmpfile.rb +0 -53
  116. /data/test/scout/{log/test_color.rb → test_offsite.rb} +0 -0
@@ -1,135 +0,0 @@
1
- require_relative '../misc'
2
- require_relative '../path'
3
- require_relative '../cmd'
4
-
5
- module Open
6
- class << self
7
- attr_accessor :remote_cache_dir
8
-
9
- def remote_cache_dir
10
- @remote_cache_dir ||= Path.setup("var/cache/open-remote/").find
11
- end
12
- end
13
-
14
- def self.remote?(file)
15
- !! (file =~ /^(?:https?|ftp|ssh):\/\//)
16
- end
17
-
18
- def self.ssh?(file)
19
- !! (file =~ /^ssh:\/\//)
20
- end
21
-
22
- def self.ssh(file, options = {})
23
- m = file.match(/ssh:\/\/([^:]+):(.*)/)
24
- server = m[1]
25
- file = m[2]
26
- if server == 'localhost'
27
- Open.open(file)
28
- else
29
- CMD.cmd("ssh '#{server}' cat '#{file}'", :pipe => true, :autojoin => true)
30
- end
31
- end
32
-
33
- def self.wget(url, options = {})
34
- if ! (options[:force] || options[:nocache]) && cache_file = in_cache(url, options)
35
- return file_open(cache_file)
36
- end
37
-
38
- Log.low "WGET:\n -URL: #{ url }\n -OPTIONS: #{options.inspect}"
39
- options = IndiferentHash.add_defaults options, "--user-agent=" => 'rbbt', :pipe => true, :autojoin => true
40
-
41
- wait(options[:nice], options[:nice_key]) if options[:nice]
42
- options.delete(:nice)
43
- options.delete(:nice_key)
44
-
45
- pipe = options.delete(:pipe)
46
- quiet = options.delete(:quiet)
47
- post = options.delete(:post)
48
- cookies = options.delete(:cookies)
49
- nocache = options.delete(:nocache)
50
-
51
- options["--quiet"] = quiet if options["--quiet"].nil?
52
- options["--post-data="] ||= post if post
53
-
54
- if cookies
55
- options["--save-cookies"] = cookies
56
- options["--load-cookies"] = cookies
57
- options["--keep-session-cookies"] = true
58
- end
59
-
60
- stderr = case
61
- when options['stderr']
62
- options['stderr']
63
- when options['--quiet']
64
- false
65
- else
66
- nil
67
- end
68
-
69
- begin
70
- wget_options = options.dup
71
- wget_options = wget_options.merge( '-O' => '-') unless options.include?('--output-document')
72
- wget_options[:pipe] = pipe unless pipe.nil?
73
- wget_options[:stderr] = stderr unless stderr.nil?
74
-
75
- io = CMD.cmd("wget '#{ url }'", wget_options)
76
- if nocache && nocache.to_s != 'update'
77
- io
78
- else
79
- add_cache(url, io, options)
80
- open_cache(url, options)
81
- end
82
- rescue
83
- STDERR.puts $!.backtrace.inspect
84
- raise OpenURLError, "Error reading remote url: #{ url }.\n#{$!.message}"
85
- end
86
- end
87
-
88
- def self.download(url, file)
89
- CMD.cmd_log(:wget, "'#{url}' -O '#{file}'")
90
- end
91
-
92
- def self.digest_url(url, options = {})
93
- params = [url, options.values_at("--post-data", "--post-data="), (options.include?("--post-file")? Open.read(options["--post-file"]).split("\n").sort * "\n" : "")]
94
- Misc.digest([url, params])
95
- end
96
-
97
- def self.cache_file(url, options)
98
- File.join(self.remote_cache_dir, digest_url(url, options))
99
- end
100
-
101
- def self.in_cache(url, options = {})
102
- filename = cache_file(url, options)
103
- if File.exist? filename
104
- return filename
105
- else
106
- nil
107
- end
108
- end
109
-
110
- def self.remove_from_cache(url, options = {})
111
- filename = cache_file(url, options)
112
- if File.exist? filename
113
- FileUtils.rm filename
114
- else
115
- nil
116
- end
117
- end
118
-
119
- def self.add_cache(url, data, options = {})
120
- filename = cache_file(url, options)
121
- Open.sensible_write(filename, data, :force => true)
122
- end
123
-
124
- def self.open_cache(url, options = {})
125
- filename = cache_file(url, options)
126
- Open.open(filename)
127
- end
128
-
129
- def self.scp(source_file, target_file, target: nil, source: nil)
130
- CMD.cmd_log("ssh #{target} mkdir -p #{File.dirname(target_file)}")
131
- target_file = [target, target_file] * ":" if target && ! target_file.start_with?(target+":")
132
- source_file = [source, source_file] * ":" if source && ! source_file.start_with?(source+":")
133
- CMD.cmd_log("scp -r '#{ source_file }' #{target_file}")
134
- end
135
- end
@@ -1,491 +0,0 @@
1
- module Open
2
- BLOCK_SIZE = 1024 * 8
3
-
4
- class << self
5
- attr_accessor :sensible_write_lock_dir
6
-
7
- def sensible_write_lock_dir
8
- @sensible_write_lock_dir ||= Path.setup("tmp/sensible_write_locks").find
9
- end
10
- end
11
-
12
- class << self
13
- attr_accessor :sensible_write_dir
14
- def sensible_write_dir
15
- @sensible_write_dir ||= Path.setup("tmp/sensible_write").find
16
- end
17
- end
18
-
19
- def self.consume_stream(io, in_thread = false, into = nil, into_close = true, &block)
20
- return if Path === io
21
- return unless io.respond_to? :read
22
-
23
- if io.respond_to? :closed? and io.closed?
24
- io.join if io.respond_to? :join
25
- return
26
- end
27
-
28
- if in_thread
29
- consumer_thread = Thread.new(Thread.current) do |parent|
30
- Thread.current["name"] = "Consumer #{Log.fingerprint io}"
31
- Thread.current.report_on_exception = false
32
- consume_stream(io, false, into, into_close)
33
- end
34
-
35
- io.threads.push(consumer_thread) if io.respond_to?(:threads)
36
- Thread.pass until consumer_thread["name"]
37
-
38
- consumer_thread
39
- else
40
- if into
41
- Log.low "Consuming stream #{Log.fingerprint io} -> #{Log.fingerprint into}"
42
- else
43
- Log.low "Consuming stream #{Log.fingerprint io}"
44
- end
45
-
46
- begin
47
- into = into.find if Path === into
48
-
49
- if String === into
50
- dir = File.dirname(into)
51
- Open.mkdir dir unless File.exist?(dir)
52
- into_path, into = into, File.open(into, 'w')
53
- end
54
-
55
- into_close = false unless into.respond_to? :close
56
-
57
- while c = io.read(BLOCK_SIZE)
58
- into << c if into
59
- last_c = c if c
60
- break if io.closed?
61
- end
62
-
63
- io.join if io.respond_to? :join
64
- io.close unless io.closed?
65
- into.join if into and into_close and into.respond_to?(:joined?) and not into.joined?
66
- into.close if into and into_close and not into.closed?
67
- block.call if block_given?
68
-
69
- last_c
70
- rescue Aborted
71
- Thread.current["exception"] = true
72
- Log.low "Consume stream Aborted #{Log.fingerprint io} into #{into_path || into}"
73
- io.abort $! if io.respond_to? :abort
74
- into.close if into.respond_to?(:closed?) && ! into.closed?
75
- FileUtils.rm into_path if into_path and File.exist?(into_path)
76
- rescue Exception
77
- Thread.current["exception"] = true
78
- Log.low "Consume stream Exception reading #{Log.fingerprint io} into #{into_path || into} - #{$!.message}"
79
- exception = (io.respond_to?(:stream_exception) && io.stream_exception) ? io.stream_exception : $!
80
- io.abort exception if io.respond_to? :abort
81
- into.close if into.respond_to?(:closed?) && ! into.closed?
82
- into_path = into if into_path.nil? && String === into
83
- if into_path and File.exist?(into_path)
84
- FileUtils.rm into_path
85
- end
86
- raise exception
87
- end
88
- end
89
- end
90
-
91
- def self.sensible_write(path, content = nil, options = {}, &block)
92
- force = IndiferentHash.process_options options, :force
93
-
94
- if File.exist?(path) and not force
95
- Open.consume_stream content
96
- return
97
- end
98
-
99
- lock_options = IndiferentHash.pull_keys options.dup, :lock
100
- lock_options = lock_options[:lock] if Hash === lock_options[:lock]
101
- tmp_path = TmpFile.tmp_for_file(path, {:dir => Open.sensible_write_dir})
102
- tmp_path_lock = TmpFile.tmp_for_file(path, {:dir => Open.sensible_write_lock_dir})
103
-
104
- tmp_path_lock = nil if FalseClass === options[:lock]
105
-
106
- Open.lock tmp_path_lock, lock_options do
107
-
108
- if File.exist? path and not force
109
- Log.warn "Path exists in sensible_write, not forcing update: #{ path }"
110
- Open.consume_stream content
111
- else
112
- FileUtils.mkdir_p File.dirname(tmp_path) unless File.directory?(File.dirname(tmp_path))
113
- FileUtils.rm_f tmp_path if File.exist? tmp_path
114
- Log.low "Sensible write stream #{Log.fingerprint content} -> #{Log.fingerprint path}" if IO === content
115
- begin
116
- case
117
- when block_given?
118
- File.open(tmp_path, 'wb', &block)
119
- when String === content
120
- File.open(tmp_path, 'wb') do |f| f.write content end
121
- when (IO === content or StringIO === content or File === content)
122
- Open.write(tmp_path) do |f|
123
- while block = content.read(BLOCK_SIZE)
124
- f.write block
125
- break if content.closed?
126
- end
127
- end
128
- else
129
- File.open(tmp_path, 'wb') do |f| end
130
- end
131
-
132
- begin
133
- Misc.insist do
134
- Open.mv tmp_path, path, lock_options
135
- end
136
- rescue Exception
137
- raise $! unless File.exist? path
138
- end
139
-
140
- Open.touch path if File.exist? path
141
- content.join if content.respond_to?(:join) and not Path === content and not (content.respond_to?(:joined?) && content.joined?)
142
-
143
- Open.notify_write(path)
144
- rescue Aborted
145
- Log.low "Aborted sensible_write -- #{ Log.reset << path }"
146
- content.abort if content.respond_to? :abort
147
- Open.rm path if File.exist? path
148
- rescue Exception
149
- exception = (AbortedStream === content and content.exception) ? content.exception : $!
150
- Log.low "Exception in sensible_write: [#{Process.pid}] #{exception.message} -- #{ path }"
151
- content.abort(exception) if content.respond_to? :abort
152
- Open.rm path if File.exist? path
153
- raise exception
154
- rescue
155
- raise $!
156
- ensure
157
- FileUtils.rm_f tmp_path if File.exist? tmp_path
158
- if Lockfile === lock_options[:lock] and lock_options[:lock].locked?
159
- lock_options[:lock].unlock
160
- end
161
- end
162
- end
163
- end
164
- end
165
-
166
- PIPE_MUTEX = Mutex.new
167
-
168
- OPEN_PIPE_IN = []
169
- def self.pipe
170
- OPEN_PIPE_IN.delete_if{|pipe| pipe.closed? }
171
- res = PIPE_MUTEX.synchronize do
172
- sout, sin = IO.pipe
173
- OPEN_PIPE_IN << sin
174
-
175
- [sout, sin]
176
- end
177
- Log.low{"Creating pipe #{[Log.fingerprint(res.last), Log.fingerprint(res.first)] * " -> "}"}
178
- res
179
- end
180
-
181
- def self.with_fifo(path = nil, clean = true, &block)
182
- begin
183
- erase = path.nil?
184
- path = TmpFile.tmp_file if path.nil?
185
- File.rm path if clean && File.exist?(path)
186
- File.mkfifo path
187
- yield path
188
- ensure
189
- FileUtils.rm path if erase && File.exist?(path)
190
- end
191
- end
192
-
193
- def self.release_pipes(*pipes)
194
- PIPE_MUTEX.synchronize do
195
- pipes.flatten.each do |pipe|
196
- pipe.close unless pipe.closed?
197
- end
198
- end
199
- end
200
-
201
- def self.purge_pipes(*save)
202
- PIPE_MUTEX.synchronize do
203
- OPEN_PIPE_IN.each do |pipe|
204
- next if save.include? pipe
205
- pipe.close unless pipe.closed?
206
- end
207
- end
208
- end
209
-
210
- def self.open_pipe(do_fork = false, close = true)
211
- raise "No block given" unless block_given?
212
-
213
- sout, sin = Open.pipe
214
-
215
- if do_fork
216
-
217
- pid = Process.fork {
218
- begin
219
- purge_pipes(sin)
220
- sout.close
221
-
222
- yield sin
223
- sin.close if close and not sin.closed?
224
-
225
- rescue Exception
226
- Log.exception $!
227
- Kernel.exit!(-1)
228
- end
229
- Kernel.exit! 0
230
- }
231
- sin.close
232
-
233
- ConcurrentStream.setup sout, :pids => [pid]
234
- else
235
-
236
- ConcurrentStream.setup sin, :pair => sout
237
- ConcurrentStream.setup sout, :pair => sin
238
-
239
- thread = Thread.new do
240
- begin
241
- ConcurrentStream.process_stream(sin, :message => "Open pipe") do
242
- Thread.current.report_on_exception = false
243
- Thread.current["name"] = "Pipe input #{Log.fingerprint sin} => #{Log.fingerprint sout}"
244
-
245
- yield sin
246
- end
247
- end
248
- end
249
-
250
- sin.threads = [thread]
251
- sout.threads = [thread]
252
-
253
- Thread.pass until thread["name"]
254
- end
255
-
256
- sout
257
- end
258
-
259
- def self.tee_stream_thread_multiple(stream, num = 2)
260
- in_pipes = []
261
- out_pipes = []
262
- num.times do
263
- sout, sin = Open.pipe
264
- in_pipes << sin
265
- out_pipes << sout
266
- end
267
-
268
- Log.low("Tee stream #{Log.fingerprint stream} -> #{Log.fingerprint out_pipes}")
269
-
270
- filename = stream.filename if stream.respond_to? :filename
271
-
272
- splitter_thread = Thread.new(Thread.current) do |parent|
273
- begin
274
- Thread.current.report_on_exception = false
275
- Thread.current["name"] = "Splitter #{Log.fingerprint stream}"
276
-
277
- skip = [false] * num
278
- while block = stream.read(BLOCK_SIZE)
279
-
280
- in_pipes.each_with_index do |sin,i|
281
- begin
282
- sin.write block
283
- rescue IOError
284
- Log.warn("Tee stream #{i} #{Log.fingerprint stream} IOError: #{$!.message} (#{Log.fingerprint sin})");
285
- skip[i] = true
286
- rescue
287
- Log.warn("Tee stream #{i} #{Log.fingerprint stream} Exception: #{$!.message} (#{Log.fingerprint sin})");
288
- raise $!
289
- end unless skip[i]
290
- end
291
- break if stream.closed?
292
- end
293
-
294
- stream.join if stream.respond_to? :join
295
- in_pipes.first.close unless in_pipes.first.closed?
296
- rescue Aborted, Interrupt
297
- stream.abort if stream.respond_to?(:abort) && ! stream.aborted?
298
- out_pipes.reverse.each do |sout|
299
- sout.threads.delete(Thread.current)
300
- begin
301
- sout.abort($!) if sout.respond_to?(:abort) && ! sout.aborted?
302
- rescue
303
- end
304
- end
305
- in_pipes.each do |sin|
306
- sin.close unless sin.closed?
307
- end
308
- Log.low "Tee aborting #{Log.fingerprint stream}"
309
- raise $!
310
- rescue Exception
311
- begin
312
- stream.abort($!) if stream.respond_to?(:abort) && ! stream.aborted?
313
- out_pipes.reverse.each do |sout|
314
- sout.threads.delete(Thread.current)
315
- begin
316
- sout.abort($!) if sout.respond_to?(:abort) && ! sout.aborted?
317
- rescue
318
- end
319
- end
320
- in_pipes.each do |sin|
321
- sin.close unless sin.closed?
322
- end
323
- Log.low "Tee exception #{Log.fingerprint stream}"
324
- rescue
325
- ensure
326
- begin
327
- in_pipes.each do |sin|
328
- sin.close unless sin.closed?
329
- end
330
- ensure
331
- raise $!
332
- end
333
- end
334
- end
335
- end
336
-
337
- Thread.pass until splitter_thread["name"]
338
-
339
- main_pipe = out_pipes.first
340
-
341
- ConcurrentStream.setup(main_pipe, :threads => [splitter_thread], :filename => filename, :autojoin => true)
342
-
343
- out_pipes[1..-1].each do |sout|
344
- ConcurrentStream.setup sout, :filename => filename, :threads => [splitter_thread]
345
- end
346
-
347
- main_pipe.callback = proc do
348
- begin
349
- stream.join if stream.respond_to?(:join) && ! stream.joined?
350
- in_pipes[1..-1].each do |sin|
351
- sin.close unless sin.closed?
352
- end
353
- rescue
354
- main_pipe.abort_callback.call($!)
355
- raise $!
356
- end
357
- end
358
-
359
- main_pipe.abort_callback = proc do |exception|
360
- stream.abort(exception)
361
- out_pipes[1..-1].each do |sout|
362
- sout.abort(exception)
363
- end
364
- end
365
-
366
- out_pipes
367
- end
368
-
369
- def self.tee_stream_thread(stream)
370
- tee_stream_thread_multiple(stream, 2)
371
- end
372
-
373
- def self.tee_stream(stream)
374
- tee_stream_thread(stream)
375
- end
376
-
377
- def self.read_stream(stream, size)
378
- str = nil
379
- Thread.pass while IO.select([stream],nil,nil,1).nil?
380
- while not str = stream.read(size)
381
- IO.select([stream],nil,nil,1)
382
- Thread.pass
383
- raise ClosedStream if stream.eof?
384
- end
385
-
386
- while str.length < size
387
- raise ClosedStream if stream.eof?
388
- IO.select([stream],nil,nil,1)
389
- if new = stream.read(size-str.length)
390
- str << new
391
- end
392
- end
393
- str
394
- end
395
-
396
- def self.read_stream(stream, size)
397
- str = ""
398
- while str.length < size
399
- missing = size - str.length
400
- more = stream.read(missing)
401
- str << more
402
- end
403
- str
404
- end
405
-
406
- def self.sort_stream(stream, header_hash: "#", cmd_args: "-u", memory: false)
407
- sout = Open.open_pipe do |sin|
408
- ConcurrentStream.process_stream(stream) do
409
- line = stream.gets
410
- while line && line.start_with?(header_hash) do
411
- sin.puts line
412
- line = stream.gets
413
- end
414
-
415
- line_stream = Open.open_pipe do |line_stream_in|
416
- line_stream_in.puts line if line
417
- Open.consume_stream(stream, false, line_stream_in)
418
- end
419
- Log.low "Sub-sort stream #{Log.fingerprint stream} -> #{Log.fingerprint line_stream}"
420
-
421
- if memory
422
- line_stream.read.split("\n").sort.each do |line|
423
- sin.puts line
424
- end
425
- else
426
- io = CMD.cmd("env LC_ALL=C sort #{cmd_args || ""}", :in => line_stream, :pipe => true)
427
- Open.consume_stream(io, false, sin)
428
- end
429
- end
430
- end
431
- Log.low "Sort #{Log.fingerprint stream} -> #{Log.fingerprint sout}"
432
- sout
433
- end
434
-
435
- #def self.sort_stream(stream, header_hash = "#", cmd_args = "-u")
436
- # StringIO.new stream.read.split("\n").sort.uniq * "\n"
437
- #end
438
-
439
- def self.collapse_stream(s, line: nil, sep: "\t", header: nil, &block)
440
- sep ||= "\t"
441
- Open.open_pipe do |sin|
442
-
443
- sin.puts header if header
444
-
445
- line ||= s.gets
446
-
447
- current_parts = []
448
- while line
449
- key, *parts = line.chomp.split(sep, -1)
450
- case
451
- when key.nil?
452
- when current_parts.nil?
453
- current_parts = parts
454
- current_key = key
455
- when current_key == key
456
- parts.each_with_index do |part,i|
457
- if current_parts[i].nil?
458
- current_parts[i] = "|" << part
459
- else
460
- current_parts[i] = current_parts[i] << "|" << part
461
- end
462
- end
463
-
464
- (parts.length..current_parts.length-1).to_a.each do |pos|
465
- current_parts[pos] = current_parts[pos] << "|" << ""
466
- end
467
- when current_key.nil?
468
- current_key = key
469
- current_parts = parts
470
- when current_key != key
471
- if block_given?
472
- res = block.call(current_parts)
473
- sin.puts [current_key, res] * sep
474
- else
475
- sin.puts [current_key, current_parts].flatten * sep
476
- end
477
- current_key = key
478
- current_parts = parts
479
- end
480
- line = s.gets
481
- end
482
-
483
- if block_given?
484
- res = block.call(current_parts)
485
- sin.puts [current_key, res] * sep
486
- else
487
- sin.puts [current_key, current_parts].flatten * sep
488
- end unless current_key.nil?
489
- end
490
- end
491
- end