rbbt-util 5.11.9 → 5.12.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/lib/rbbt/persist.rb +1 -1
- data/lib/rbbt/tsv.rb +1 -0
- data/lib/rbbt/tsv/dumper.rb +1 -2
- data/lib/rbbt/tsv/parallel/traverse.rb +2 -2
- data/lib/rbbt/tsv/parser.rb +6 -2
- data/lib/rbbt/tsv/stream.rb +55 -0
- data/lib/rbbt/tsv/util.rb +7 -1
- data/lib/rbbt/util/misc.rb +5 -762
- data/lib/rbbt/util/misc/concurrent_stream.rb +15 -0
- data/lib/rbbt/util/misc/development.rb +122 -0
- data/lib/rbbt/util/misc/inspect.rb +3 -3
- data/lib/rbbt/util/misc/manipulation.rb +136 -0
- data/lib/rbbt/util/misc/math.rb +50 -0
- data/lib/rbbt/util/misc/objects.rb +79 -0
- data/lib/rbbt/util/misc/omics.rb +10 -0
- data/lib/rbbt/util/misc/options.rb +280 -0
- data/lib/rbbt/util/misc/pipes.rb +140 -20
- data/lib/rbbt/util/misc/system.rb +90 -0
- data/lib/rbbt/util/tar.rb +0 -7
- data/lib/rbbt/workflow/accessor.rb +3 -3
- data/lib/rbbt/workflow/step/run.rb +69 -15
- data/lib/rbbt/workflow/task.rb +7 -5
- data/test/rbbt/tsv/test_stream.rb +92 -0
- data/test/rbbt/tsv/test_util.rb +1 -3
- data/test/rbbt/util/misc/test_pipes.rb +79 -0
- data/test/rbbt/workflow/test_task.rb +1 -0
- metadata +10 -2
@@ -0,0 +1,90 @@
|
|
1
|
+
require 'net/smtp'
|
2
|
+
|
3
|
+
module Misc
|
4
|
+
|
5
|
+
def self.hostname
|
6
|
+
@hostanem ||= `hostname`.strip
|
7
|
+
end
|
8
|
+
|
9
|
+
def self.pid_exists?(pid)
|
10
|
+
return false if pid.nil?
|
11
|
+
begin
|
12
|
+
Process.getpgid(pid.to_i)
|
13
|
+
true
|
14
|
+
rescue Errno::ESRCH
|
15
|
+
false
|
16
|
+
end
|
17
|
+
end
|
18
|
+
|
19
|
+
def self.send_email(from, to, subject, message, options = {})
|
20
|
+
IndiferentHash.setup(options)
|
21
|
+
options = Misc.add_defaults options, :from_alias => nil, :to_alias => nil, :server => 'localhost', :port => 25, :user => nil, :pass => nil, :auth => :login
|
22
|
+
|
23
|
+
server, port, user, pass, from_alias, to_alias, auth = Misc.process_options options, :server, :port, :user, :pass, :from_alias, :to_alias, :auth
|
24
|
+
|
25
|
+
msg = <<-END_OF_MESSAGE
|
26
|
+
From: #{from_alias} <#{from}>
|
27
|
+
To: #{to_alias} <#{to}>
|
28
|
+
Subject: #{subject}
|
29
|
+
|
30
|
+
#{message}
|
31
|
+
END_OF_MESSAGE
|
32
|
+
|
33
|
+
Net::SMTP.start(server, port, server, user, pass, auth) do |smtp|
|
34
|
+
smtp.send_message msg, from, to
|
35
|
+
end
|
36
|
+
end
|
37
|
+
|
38
|
+
def self.env_add(var, value, sep = ":", prepend = true)
|
39
|
+
ENV[var] ||= ""
|
40
|
+
return if ENV[var] =~ /(#{sep}|^)#{Regexp.quote value}(#{sep}|$)/
|
41
|
+
if prepend
|
42
|
+
ENV[var] = value + sep + ENV[var]
|
43
|
+
else
|
44
|
+
ENV[var] += sep + ENV[var]
|
45
|
+
end
|
46
|
+
end
|
47
|
+
|
48
|
+
def self.path_relative_to(basedir, path)
|
49
|
+
path = File.expand_path(path) unless path[0] == "/"
|
50
|
+
basedir = File.expand_path(basedir) unless basedir[0] == "/"
|
51
|
+
|
52
|
+
if path.index(basedir) == 0
|
53
|
+
if basedir[-1] == "/"
|
54
|
+
return path[basedir.length..-1]
|
55
|
+
else
|
56
|
+
return path[basedir.length+1..-1]
|
57
|
+
end
|
58
|
+
else
|
59
|
+
return nil
|
60
|
+
end
|
61
|
+
end
|
62
|
+
|
63
|
+
def self.common_path(dir, file)
|
64
|
+
file = File.expand_path file
|
65
|
+
dir = File.expand_path dir
|
66
|
+
|
67
|
+
return true if file == dir
|
68
|
+
while File.dirname(file) != file
|
69
|
+
file = File.dirname(file)
|
70
|
+
return true if file == dir
|
71
|
+
end
|
72
|
+
|
73
|
+
return false
|
74
|
+
end
|
75
|
+
|
76
|
+
# WARN: probably not thread safe...
|
77
|
+
def self.in_dir(dir)
|
78
|
+
old_pwd = FileUtils.pwd
|
79
|
+
res = nil
|
80
|
+
begin
|
81
|
+
FileUtils.mkdir_p dir unless File.exists? dir
|
82
|
+
FileUtils.cd dir
|
83
|
+
res = yield
|
84
|
+
ensure
|
85
|
+
FileUtils.cd old_pwd
|
86
|
+
end
|
87
|
+
res
|
88
|
+
end
|
89
|
+
|
90
|
+
end
|
data/lib/rbbt/util/tar.rb
CHANGED
@@ -43,13 +43,6 @@ module Misc
|
|
43
43
|
|
44
44
|
string = tar.string
|
45
45
|
|
46
|
-
#chunk_size = 1024
|
47
|
-
#Zlib::GzipWriter.wrap(gz) do |writer|
|
48
|
-
# while chunk = tar.read(chunk_size) do
|
49
|
-
# writer.write chunk
|
50
|
-
# end
|
51
|
-
# writer.close
|
52
|
-
#end
|
53
46
|
z = Zlib::GzipWriter.new(gz)
|
54
47
|
z.write string
|
55
48
|
z.close
|
@@ -49,9 +49,9 @@ class Step
|
|
49
49
|
end
|
50
50
|
|
51
51
|
begin
|
52
|
-
@info_cache = Misc.insist(2,
|
53
|
-
Misc.insist(2,
|
54
|
-
Misc.insist(3, 0.
|
52
|
+
@info_cache = Misc.insist(2, 3, info_file) do
|
53
|
+
Misc.insist(2, 1, info_file) do
|
54
|
+
Misc.insist(3, 0.2, info_file) do
|
55
55
|
Open.open(info_file) do |file|
|
56
56
|
INFO_SERIALIAZER.load(file) || {}
|
57
57
|
end
|
@@ -2,6 +2,47 @@ class Step
|
|
2
2
|
|
3
3
|
attr_reader :stream
|
4
4
|
|
5
|
+
STREAM_CACHE = {}
|
6
|
+
STREAM_CACHE_MUTEX = Mutex.new
|
7
|
+
def self.dup_stream(stream)
|
8
|
+
case stream
|
9
|
+
when IO, File
|
10
|
+
return stream if stream.closed?
|
11
|
+
STREAM_CACHE_MUTEX.synchronize do
|
12
|
+
if STREAM_CACHE[stream].nil?
|
13
|
+
STREAM_CACHE[stream] = stream
|
14
|
+
else
|
15
|
+
Misc.dup_stream(STREAM_CACHE[stream])
|
16
|
+
end
|
17
|
+
end
|
18
|
+
when TSV::Dumper, TSV::Parser
|
19
|
+
stream = stream.stream
|
20
|
+
return stream if stream.closed?
|
21
|
+
|
22
|
+
STREAM_CACHE_MUTEX.synchronize do
|
23
|
+
if STREAM_CACHE[stream].nil?
|
24
|
+
STREAM_CACHE[stream] = stream
|
25
|
+
else
|
26
|
+
Misc.dup_stream(STREAM_CACHE[stream])
|
27
|
+
end
|
28
|
+
end
|
29
|
+
else
|
30
|
+
stream
|
31
|
+
end
|
32
|
+
end
|
33
|
+
|
34
|
+
def self.purge_stream_cache
|
35
|
+
return
|
36
|
+
STREAM_CACHE_MUTEX.synchronize do
|
37
|
+
STREAM_CACHE.collect{|k,s|
|
38
|
+
Thread.new do
|
39
|
+
Misc.consume_stream s
|
40
|
+
end
|
41
|
+
}
|
42
|
+
STREAM_CACHE.clear
|
43
|
+
end
|
44
|
+
end
|
45
|
+
|
5
46
|
def get_stream
|
6
47
|
@mutex.synchronize do
|
7
48
|
@stream = begin
|
@@ -12,9 +53,16 @@ class Step
|
|
12
53
|
end
|
13
54
|
end
|
14
55
|
|
56
|
+
def dup_inputs
|
57
|
+
@inputs.collect do |input|
|
58
|
+
Step.dup_stream input
|
59
|
+
end
|
60
|
+
end
|
61
|
+
|
15
62
|
def _exec
|
16
63
|
@exec = true if @exec.nil?
|
17
|
-
|
64
|
+
#@task.exec_in((bindings ? bindings : self), *@inputs)
|
65
|
+
@task.exec_in((bindings ? bindings : self), *dup_inputs)
|
18
66
|
end
|
19
67
|
|
20
68
|
def exec(no_load=false)
|
@@ -299,26 +347,32 @@ class Step
|
|
299
347
|
|
300
348
|
join_stream
|
301
349
|
|
302
|
-
return if not Open.exists? info_file
|
350
|
+
return self if not Open.exists? info_file
|
351
|
+
|
352
|
+
return self if info[:joined]
|
303
353
|
pid = @pid
|
304
354
|
|
305
355
|
Misc.insist [0.1, 0.2, 0.5, 1] do
|
306
356
|
pid ||= info[:pid]
|
307
357
|
end
|
308
358
|
|
309
|
-
|
310
|
-
|
311
|
-
|
312
|
-
|
313
|
-
|
314
|
-
|
315
|
-
|
316
|
-
|
317
|
-
|
318
|
-
|
319
|
-
|
320
|
-
|
321
|
-
|
359
|
+
begin
|
360
|
+
if pid.nil?
|
361
|
+
dependencies.each{|dep| dep.join }
|
362
|
+
self
|
363
|
+
else
|
364
|
+
begin
|
365
|
+
Log.debug{"Waiting for pid: #{pid}"}
|
366
|
+
Process.waitpid pid
|
367
|
+
rescue Errno::ECHILD
|
368
|
+
Log.debug{"Process #{ pid } already finished: #{ path }"}
|
369
|
+
end if Misc.pid_exists? pid
|
370
|
+
pid = nil
|
371
|
+
dependencies.each{|dep| dep.join }
|
372
|
+
self
|
373
|
+
end
|
374
|
+
ensure
|
375
|
+
set_info :joined, true
|
322
376
|
end
|
323
377
|
self
|
324
378
|
end
|
data/lib/rbbt/workflow/task.rb
CHANGED
@@ -29,15 +29,13 @@ module Task
|
|
29
29
|
end
|
30
30
|
end
|
31
31
|
|
32
|
-
def param_options
|
33
|
-
end
|
34
|
-
|
35
32
|
def take_input_values(input_values)
|
36
33
|
return [] if @inputs.nil?
|
37
34
|
values = []
|
35
|
+
defaults = IndiferentHash.setup(@input_defaults || {})
|
38
36
|
@inputs.each do |input|
|
39
37
|
value = input_values[input]
|
40
|
-
value =
|
38
|
+
value = defaults[input] if value.nil?
|
41
39
|
values << value
|
42
40
|
end
|
43
41
|
values
|
@@ -46,7 +44,11 @@ module Task
|
|
46
44
|
def exec(*args)
|
47
45
|
case
|
48
46
|
when (args.length == 1 and not inputs.nil? and inputs.length > 1 and Hash === args.first)
|
49
|
-
|
47
|
+
begin
|
48
|
+
self.call *take_input_values(IndiferentHash.setup(args.first))
|
49
|
+
ensure
|
50
|
+
purge_stream_cache
|
51
|
+
end
|
50
52
|
else
|
51
53
|
self.call *args
|
52
54
|
end
|
@@ -0,0 +1,92 @@
|
|
1
|
+
require File.join(File.expand_path(File.dirname(__FILE__)), '../..', 'test_helper.rb')
|
2
|
+
require 'rbbt/tsv'
|
3
|
+
require 'rbbt/tsv/stream'
|
4
|
+
require 'rbbt'
|
5
|
+
|
6
|
+
class TestStream < Test::Unit::TestCase
|
7
|
+
def test_collapse_stream
|
8
|
+
text=<<-EOF
|
9
|
+
#: :sep=" "
|
10
|
+
#Row LabelA LabelB LabelC
|
11
|
+
row1 A B C
|
12
|
+
row1 a b c
|
13
|
+
row2 AA BB CC
|
14
|
+
row2 aa bb cc
|
15
|
+
EOF
|
16
|
+
|
17
|
+
s = StringIO.new text
|
18
|
+
tsv = TSV.open TSV.collapse_stream(s)
|
19
|
+
assert_equal ["A", "a"], tsv["row1"][0]
|
20
|
+
assert_equal ["BB", "bb"], tsv["row2"][1]
|
21
|
+
end
|
22
|
+
|
23
|
+
def test_paste_stream
|
24
|
+
text1=<<-EOF
|
25
|
+
#: :sep=" "
|
26
|
+
#Row LabelA LabelB LabelC
|
27
|
+
row1 A B C
|
28
|
+
row2 AA BB CC
|
29
|
+
row3 AAA BBB CCC
|
30
|
+
EOF
|
31
|
+
|
32
|
+
text2=<<-EOF
|
33
|
+
#: :sep=" "
|
34
|
+
#Row Labela Labelb
|
35
|
+
row1 a b
|
36
|
+
row2 aa bb
|
37
|
+
row3 aaa bbb
|
38
|
+
EOF
|
39
|
+
|
40
|
+
text3=<<-EOF
|
41
|
+
#: :sep=" "
|
42
|
+
#Row LabelC
|
43
|
+
row1 c
|
44
|
+
row2 cc
|
45
|
+
row3 ccc
|
46
|
+
EOF
|
47
|
+
|
48
|
+
s1 = StringIO.new text1
|
49
|
+
s2 = StringIO.new text2
|
50
|
+
s3 = StringIO.new text3
|
51
|
+
tsv = TSV.open TSV.paste_streams([s1,s2,s3], :sep => " ", :type => :list)
|
52
|
+
assert_equal ["A", "B", "C", "a", "b", "c"], tsv["row1"]
|
53
|
+
assert_equal ["AA", "BB", "CC", "aa", "bb", "cc"], tsv["row2"]
|
54
|
+
assert_equal ["AAA", "BBB", "CCC", "aaa", "bbb", "ccc"], tsv["row3"]
|
55
|
+
end
|
56
|
+
|
57
|
+
def test_paste_stream_sort
|
58
|
+
text1=<<-EOF
|
59
|
+
#: :sep=" "
|
60
|
+
#Row LabelA LabelB LabelC
|
61
|
+
row2 AA BB CC
|
62
|
+
row1 A B C
|
63
|
+
row3 AAA BBB CCC
|
64
|
+
EOF
|
65
|
+
|
66
|
+
text2=<<-EOF
|
67
|
+
#: :sep=" "
|
68
|
+
#Row Labela Labelb
|
69
|
+
row1 a b
|
70
|
+
row3 aaa bbb
|
71
|
+
row2 aa bb
|
72
|
+
EOF
|
73
|
+
|
74
|
+
text3=<<-EOF
|
75
|
+
#: :sep=" "
|
76
|
+
#Row Labelc
|
77
|
+
row3 ccc
|
78
|
+
row1 c
|
79
|
+
row2 cc
|
80
|
+
EOF
|
81
|
+
|
82
|
+
s1 = StringIO.new text1
|
83
|
+
s2 = StringIO.new text2
|
84
|
+
s3 = StringIO.new text3
|
85
|
+
tsv = TSV.open TSV.paste_streams([s1,s2,s3], :sep => " ", :type => :list, :sort => true)
|
86
|
+
assert_equal "Row", tsv.key_field
|
87
|
+
assert_equal %w(LabelA LabelB LabelC Labela Labelb Labelc), tsv.fields
|
88
|
+
assert_equal ["A", "B", "C", "a", "b", "c"], tsv["row1"]
|
89
|
+
assert_equal ["AA", "BB", "CC", "aa", "bb", "cc"], tsv["row2"]
|
90
|
+
assert_equal ["AAA", "BBB", "CCC", "aaa", "bbb", "ccc"], tsv["row3"]
|
91
|
+
end
|
92
|
+
end
|
data/test/rbbt/tsv/test_util.rb
CHANGED
@@ -18,12 +18,10 @@ row2 A B Id3
|
|
18
18
|
|
19
19
|
assert_equal 2, TSV.field_match_counts(tsv, ["a","A","a","b","Id3"])["ValueA"]
|
20
20
|
assert_equal nil, TSV.field_match_counts(tsv, ["ValueA"])["ValueA"]
|
21
|
-
|
22
21
|
end
|
23
|
-
|
24
22
|
end
|
25
23
|
|
26
|
-
def
|
24
|
+
def _test_marshal
|
27
25
|
content =<<-EOF
|
28
26
|
#Id ValueA ValueB OtherID
|
29
27
|
row1 a|aa|aaa b Id1|Id2
|
@@ -0,0 +1,79 @@
|
|
1
|
+
require File.expand_path(File.dirname(__FILE__) + '/../../../test_helper')
|
2
|
+
require 'test/unit'
|
3
|
+
require 'rbbt/tsv'
|
4
|
+
require 'rbbt/util/misc'
|
5
|
+
|
6
|
+
class TestMiscPipes < Test::Unit::TestCase
|
7
|
+
|
8
|
+
def test_collapse_stream
|
9
|
+
text=<<-EOF
|
10
|
+
row1 A B C
|
11
|
+
row1 a b c
|
12
|
+
row2 AA BB CC
|
13
|
+
row2 aa bb cc
|
14
|
+
EOF
|
15
|
+
|
16
|
+
s = StringIO.new text
|
17
|
+
tsv = TSV.open Misc.collapse_stream(s,nil, " "), :sep => " "
|
18
|
+
assert_equal ["A", "a"], tsv["row1"][0]
|
19
|
+
assert_equal ["BB", "bb"], tsv["row2"][1]
|
20
|
+
end
|
21
|
+
|
22
|
+
def test_paste_stream
|
23
|
+
text1=<<-EOF
|
24
|
+
row1 A B C
|
25
|
+
row2 AA BB CC
|
26
|
+
row3 AAA BBB CCC
|
27
|
+
EOF
|
28
|
+
|
29
|
+
text2=<<-EOF
|
30
|
+
row1 a b
|
31
|
+
row2 aa bb
|
32
|
+
EOF
|
33
|
+
|
34
|
+
text3=<<-EOF
|
35
|
+
row1 c
|
36
|
+
row2 cc
|
37
|
+
row3 ccc
|
38
|
+
EOF
|
39
|
+
|
40
|
+
s1 = StringIO.new text1
|
41
|
+
s2 = StringIO.new text2
|
42
|
+
s3 = StringIO.new text3
|
43
|
+
tsv = TSV.open Misc.paste_streams([s1,s2,s3],nil, " "), :sep => " ", :type => :list
|
44
|
+
assert_equal ["A", "B", "C", "a", "b", "c"], tsv["row1"]
|
45
|
+
assert_equal ["AA", "BB", "CC", "aa", "bb", "cc"], tsv["row2"]
|
46
|
+
assert_equal ["AAA", "BBB", "CCC", "", "", "ccc"], tsv["row3"]
|
47
|
+
end
|
48
|
+
|
49
|
+
def test_sort_stream
|
50
|
+
text =<<-EOF
|
51
|
+
#: :sep=" "
|
52
|
+
#Row LabelA LabelB LabelC
|
53
|
+
row2 AA BB CC
|
54
|
+
row3 AAA BBB CCC
|
55
|
+
row1 A B C
|
56
|
+
EOF
|
57
|
+
s = StringIO.new text
|
58
|
+
sorted = Misc.sort_stream(s)
|
59
|
+
assert_equal %w(#: #Row row2 row3 row1), text.split("\n").collect{|l| l.split(" ").first}
|
60
|
+
assert_equal %w(#: #Row row1 row2 row3), sorted.read.split("\n").collect{|l| l.split(" ").first}
|
61
|
+
end
|
62
|
+
|
63
|
+
def test_dup_stream
|
64
|
+
text =<<-EOF
|
65
|
+
#: :sep=" "
|
66
|
+
#Row LabelA LabelB LabelC
|
67
|
+
row2 AA BB CC
|
68
|
+
row3 AAA BBB CCC
|
69
|
+
row1 A B C
|
70
|
+
EOF
|
71
|
+
|
72
|
+
TmpFile.with_file(text) do |tmp|
|
73
|
+
io = Open.open(tmp)
|
74
|
+
dup = Misc.dup_stream(io)
|
75
|
+
assert_equal text, dup.read
|
76
|
+
assert_equal text, io.read
|
77
|
+
end
|
78
|
+
end
|
79
|
+
end
|