td 0.10.85 → 0.10.86
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/ChangeLog +6 -0
- data/lib/td/command/bulk_import.rb +5 -345
- data/lib/td/command/import.rb +9 -1
- data/lib/td/command/list.rb +3 -3
- data/lib/td/command/runner.rb +2 -1
- data/lib/td/version.rb +1 -1
- metadata +19 -47
checksums.yaml
ADDED
@@ -0,0 +1,7 @@
|
|
1
|
+
---
|
2
|
+
SHA1:
|
3
|
+
metadata.gz: e3a7c671d16731ca4b03f3063b539b9d88165483
|
4
|
+
data.tar.gz: 348c5b2ca638734cfc6f47a5f47500a9e9c188d0
|
5
|
+
SHA512:
|
6
|
+
metadata.gz: 960cae5f7a35493c5b68d5cca143015947a0b438ace3856c6b922f3779e52fafbba0908b8ef50de7eda17770d42f760fd208695101e7213f558045c7f7fc8c0c
|
7
|
+
data.tar.gz: 9c1438ba072e135c4fdf96ff5fe3d50f49a6617ceb93ecda9a14b9b580cae61ccb5e03e6d721852f046c5f6ac6826efef8dc339815012e747e1eba0b3785b506
|
data/ChangeLog
CHANGED
@@ -197,6 +197,10 @@ module Command
|
|
197
197
|
end
|
198
198
|
end
|
199
199
|
|
200
|
+
def bulk_import_upload_parts2(op)
|
201
|
+
puts "This command is moved to 'td import:upload' since 0.10.85."
|
202
|
+
end
|
203
|
+
|
200
204
|
# obsoleted
|
201
205
|
def bulk_import_delete_part(op)
|
202
206
|
name, part_name = op.cmd_parse
|
@@ -423,352 +427,8 @@ module Command
|
|
423
427
|
$stderr.puts normalized_message if has_bignum
|
424
428
|
end
|
425
429
|
|
426
|
-
def bulk_import_upload_parts2(op)
|
427
|
-
opts = upload_parts2_config(op)
|
428
|
-
|
429
|
-
# java command
|
430
|
-
javacmd = 'java'
|
431
|
-
|
432
|
-
# make jvm options
|
433
|
-
jvm_opts = []
|
434
|
-
jvm_opts << "-Xmx1024m" # TODO
|
435
|
-
|
436
|
-
# find java/*.jar and td.jar
|
437
|
-
base_path = File.expand_path('../../..', File.dirname(__FILE__)) # TODO
|
438
|
-
libjars = Dir.glob("#{base_path}/java/**/*.jar")
|
439
|
-
found = libjars.find { |path| File.basename(path) =~ /^td-bulk-import/ }
|
440
|
-
td_command_jar = libjars.delete(found)
|
441
|
-
|
442
|
-
# make application options
|
443
|
-
app_opts = []
|
444
|
-
app_opts << "-cp \"#{td_command_jar}\""
|
445
|
-
|
446
|
-
# make system properties
|
447
|
-
sysprops = []
|
448
|
-
sysprops.concat(upload_parts2_sysprops(opts))
|
449
|
-
|
450
|
-
# make application arguments
|
451
|
-
app_args = []
|
452
|
-
app_args << 'com.treasure_data.bulk_import.Main'
|
453
|
-
app_args << 'upload_parts'
|
454
|
-
app_args << opts[21]
|
455
|
-
|
456
|
-
# TODO consider parameters including spaces; don't use join(' ')
|
457
|
-
command = "#{javacmd} #{jvm_opts.join(' ')} #{app_opts.join(' ')} #{sysprops.join(' ')} #{app_args.join(' ')}"
|
458
|
-
|
459
|
-
exec command
|
460
|
-
end
|
461
|
-
|
462
430
|
def bulk_import_prepare_parts2(op)
|
463
|
-
|
464
|
-
|
465
|
-
# java command
|
466
|
-
javacmd = 'java'
|
467
|
-
|
468
|
-
# make jvm options
|
469
|
-
jvm_opts = []
|
470
|
-
jvm_opts << "-Xmx1024m" # TODO
|
471
|
-
|
472
|
-
# find java/*.jar and td.jar
|
473
|
-
base_path = File.expand_path('../../..', File.dirname(__FILE__)) # TODO
|
474
|
-
libjars = Dir.glob("#{base_path}/java/**/*.jar")
|
475
|
-
found = libjars.find { |path| File.basename(path) =~ /^td-bulk-import/ }
|
476
|
-
td_command_jar = libjars.delete(found)
|
477
|
-
|
478
|
-
# make application options
|
479
|
-
app_opts = []
|
480
|
-
app_opts << "-cp \"#{td_command_jar}\""
|
481
|
-
|
482
|
-
# make system properties
|
483
|
-
sysprops = []
|
484
|
-
sysprops.concat(prepare_parts2_sysprops(opts))
|
485
|
-
|
486
|
-
# make application arguments
|
487
|
-
app_args = []
|
488
|
-
app_args << 'com.treasure_data.bulk_import.Main'
|
489
|
-
app_args << 'prepare_parts'
|
490
|
-
app_args << opts[18]
|
491
|
-
|
492
|
-
# TODO consider parameters including spaces; don't use join(' ')
|
493
|
-
command = "#{javacmd} #{jvm_opts.join(' ')} #{app_opts.join(' ')} #{sysprops.join(' ')} #{app_args.join(' ')}"
|
494
|
-
|
495
|
-
exec command
|
496
|
-
end
|
497
|
-
|
498
|
-
private
|
499
|
-
def prepare_parts2_sysprops(opts)
|
500
|
-
sysprops = []
|
501
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.format=#{opts[0]}"
|
502
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.compression=#{opts[1]}"
|
503
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.encoding=#{opts[2]}"
|
504
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.time-column=#{opts[3]}"
|
505
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.time-format=#{opts[4]}"
|
506
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.time-value=#{opts[5].to_s}" if opts[5]
|
507
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.output-dir=#{opts[6]}"
|
508
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.split-size=#{opts[7]}"
|
509
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.error-record-output=#{opts[8]}" if opts[8]
|
510
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.dry-run=#{opts[9]}"
|
511
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.delimiter=#{opts[10]}"
|
512
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.newline=#{opts[11]}"
|
513
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.column-header=#{opts[12]}" if opts[3]
|
514
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.columns=#{opts[13]}" if opts[1]
|
515
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.column-types=#{opts[14]}" if opts[2]
|
516
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.type-conversion-error=#{opts[15]}" if opts[15]
|
517
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.exclude-columns=#{opts[16]}"
|
518
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.only-columns=#{opts[17]}"
|
519
|
-
sysprops
|
520
|
-
end
|
521
|
-
|
522
|
-
private
|
523
|
-
def upload_parts2_sysprops(opts)
|
524
|
-
sysprops = []
|
525
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.format=#{opts[0]}"
|
526
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.compression=#{opts[1]}"
|
527
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.encoding=#{opts[2]}"
|
528
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.time-column=#{opts[3]}"
|
529
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.time-format=#{opts[4]}"
|
530
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.time-value=#{opts[5].to_s}" if opts[5]
|
531
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.output-dir=#{opts[6]}"
|
532
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.split-size=#{opts[7]}"
|
533
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.error-record-output=#{opts[8]}" if opts[8]
|
534
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.dry-run=#{opts[9]}"
|
535
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.delimiter=#{opts[10]}"
|
536
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.newline=#{opts[11]}"
|
537
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.column-header=#{opts[12]}" if opts[3]
|
538
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.columns=#{opts[13]}" if opts[1]
|
539
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.column-types=#{opts[14]}" if opts[2]
|
540
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.type-conversion-error=#{opts[15]}" if opts[15]
|
541
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.exclude-columns=#{opts[16]}"
|
542
|
-
sysprops << "-Dtd.bulk_import.prepare_parts.only-columns=#{opts[17]}"
|
543
|
-
|
544
|
-
sysprops << "-Dtd.bulk_import.upload_parts.auto-perform=#{opts[18]}"
|
545
|
-
sysprops << "-Dtd.bulk_import.upload_parts.auto-commit=#{opts[19]}"
|
546
|
-
sysprops << "-Dtd.bulk_import.upload_parts.parallel=#{opts[20]}"
|
547
|
-
sysprops << "-Dtd.bulk_import.upload_parts.retrycount=10"
|
548
|
-
sysprops << "-Dtd.bulk_import.upload_parts.waitsec=1"
|
549
|
-
sysprops << "-Dtd.api.key=#{get_client.apikey}"
|
550
|
-
sysprops
|
551
|
-
end
|
552
|
-
|
553
|
-
private
|
554
|
-
def prepare_parts2_config(op)
|
555
|
-
format = 'csv'
|
556
|
-
compress = 'none'
|
557
|
-
encoding = 'utf-8'
|
558
|
-
time_column = 'time'
|
559
|
-
time_format = nil
|
560
|
-
time_value = nil
|
561
|
-
outdir = nil
|
562
|
-
split_size_kb = PART_SPLIT_SIZE / 1024 # kb
|
563
|
-
error_record_output = nil
|
564
|
-
dry_run = false
|
565
|
-
|
566
|
-
delimiter = ','
|
567
|
-
newline = 'CRLF'
|
568
|
-
column_header = nil
|
569
|
-
columns = nil
|
570
|
-
column_types = nil
|
571
|
-
type_conversion_error = 'skip'
|
572
|
-
exclude_columns = nil
|
573
|
-
only_columns = nil
|
574
|
-
|
575
|
-
# prepare_parts
|
576
|
-
op.on('-f', '--format NAME', 'source file format [csv, tsv]; default=csv') {|s|
|
577
|
-
format = s
|
578
|
-
}
|
579
|
-
op.on('-C', '--compress TYPE', 'compressed type [gzip, none]; default=auto detect') {|s|
|
580
|
-
compress = s
|
581
|
-
}
|
582
|
-
op.on('-e', '--encoding TYPE', 'encoding type [utf-8]') {|s|
|
583
|
-
encoding = s
|
584
|
-
}
|
585
|
-
op.on('-t', '--time-column NAME', 'name of the time column') {|s|
|
586
|
-
time_column = s
|
587
|
-
}
|
588
|
-
op.on('-T', '--time-format', 'STRF_FORMAT; default=auto detect') {|s|
|
589
|
-
time_format = s
|
590
|
-
}
|
591
|
-
op.on('--time-value TIME', 'long value of the time column') {|s|
|
592
|
-
if s.to_i.to_s == s
|
593
|
-
time_value = s.to_i
|
594
|
-
else
|
595
|
-
require 'time'
|
596
|
-
time_value = Time.parse(s).to_i
|
597
|
-
end
|
598
|
-
}
|
599
|
-
op.on('-o', '--output DIR', 'output directory') {|s|
|
600
|
-
outdir = s
|
601
|
-
}
|
602
|
-
op.on('-s', '--split-size SIZE_IN_KB', "size of each parts (default: #{split_size_kb})", Integer) {|i|
|
603
|
-
split_size_kb = i
|
604
|
-
}
|
605
|
-
op.on('--error-record-output FILE', 'error records output file; default=NULL output stream') {|s|
|
606
|
-
error_record_outdir = s
|
607
|
-
}
|
608
|
-
op.on('--dry-run', 'show samples as JSON and exit', FalseClass) {|b|
|
609
|
-
dry_run = b
|
610
|
-
}
|
611
|
-
op.on('--delimiter CHAR', 'delimiter CHAR; default="," at csv, "\t" at tsv') {|s|
|
612
|
-
delimiter = s
|
613
|
-
}
|
614
|
-
op.on('--newline', 'newline [CRLR, LR, CR]; default=CRLF') {|s|
|
615
|
-
newline = s
|
616
|
-
}
|
617
|
-
op.on('-H', '--column-header', 'first line includes column names', TrueClass) {|b|
|
618
|
-
column_header = b
|
619
|
-
}
|
620
|
-
op.on('-h', '--columns NAME,NAME,...', 'column names (use --column-header instead if the first line has column names)') {|s|
|
621
|
-
columns = s
|
622
|
-
}
|
623
|
-
op.on('--column-types TYPE,TYPE,...', 'column types [string, int, long]') {|s|
|
624
|
-
column_types = s
|
625
|
-
}
|
626
|
-
op.on('--type-conversion-error TYPE', 'type conversion error [skip,null]; default=skip') {|s|
|
627
|
-
type_conversion_error = s
|
628
|
-
}
|
629
|
-
op.on('--exclude-columns NAME,NAME,...', 'exclude columns') {|s|
|
630
|
-
exclude_columns = s
|
631
|
-
}
|
632
|
-
op.on('--only-columns NAME,NAME,...', 'only columns') {|s|
|
633
|
-
only_columns = s
|
634
|
-
}
|
635
|
-
|
636
|
-
files = op.cmd_parse
|
637
|
-
files = [files] unless files.is_a?(Array) # TODO ruby 1.9
|
638
|
-
|
639
|
-
# options validation
|
640
|
-
#unless column_types
|
641
|
-
# $stderr.puts "--column-types TYPE,TYPE,... option is required."
|
642
|
-
# exit 1
|
643
|
-
#end
|
644
|
-
unless outdir
|
645
|
-
$stderr.puts "-o, --output DIR option is required."
|
646
|
-
exit 1
|
647
|
-
end
|
648
|
-
|
649
|
-
return [ format, compress, encoding,
|
650
|
-
time_column, time_format, time_value,
|
651
|
-
outdir, split_size_kb, error_record_output, dry_run,
|
652
|
-
delimiter, newline, column_header, columns, column_types,
|
653
|
-
type_conversion_error, exclude_columns, only_columns, files ]
|
654
|
-
end
|
655
|
-
|
656
|
-
private
|
657
|
-
def upload_parts2_config(op)
|
658
|
-
format = 'csv'
|
659
|
-
compress = 'none'
|
660
|
-
encoding = 'utf-8'
|
661
|
-
time_column = 'time'
|
662
|
-
time_format = nil
|
663
|
-
time_value = nil
|
664
|
-
outdir = nil
|
665
|
-
split_size_kb = PART_SPLIT_SIZE / 1024 # kb
|
666
|
-
error_record_output = nil
|
667
|
-
dry_run = false
|
668
|
-
|
669
|
-
delimiter = ','
|
670
|
-
newline = 'CRLF'
|
671
|
-
column_header = nil
|
672
|
-
columns = nil
|
673
|
-
column_types = nil
|
674
|
-
type_conversion_error = 'skip'
|
675
|
-
exclude_columns = nil
|
676
|
-
only_columns = nil
|
677
|
-
|
678
|
-
auto_perform = false
|
679
|
-
auto_commit = false
|
680
|
-
parallel = 2
|
681
|
-
|
682
|
-
# prepare_parts
|
683
|
-
op.on('-f', '--format NAME', 'source file format [csv, tsv]; default=csv') {|s|
|
684
|
-
format = s
|
685
|
-
}
|
686
|
-
op.on('-C', '--compress TYPE', 'compressed type [gzip, none]; default=auto detect') {|s|
|
687
|
-
compress = s
|
688
|
-
}
|
689
|
-
op.on('-e', '--encoding TYPE', 'encoding type [utf-8]') {|s|
|
690
|
-
encoding = s
|
691
|
-
}
|
692
|
-
op.on('-t', '--time-column NAME', 'name of the time column') {|s|
|
693
|
-
time_column = s
|
694
|
-
}
|
695
|
-
op.on('-T', '--time-format', 'STRF_FORMAT; default=auto detect') {|s|
|
696
|
-
time_format = s
|
697
|
-
}
|
698
|
-
op.on('--time-value TIME', 'long value of the time column') {|s|
|
699
|
-
if s.to_i.to_s == s
|
700
|
-
time_value = s.to_i
|
701
|
-
else
|
702
|
-
require 'time'
|
703
|
-
time_value = Time.parse(s).to_i
|
704
|
-
end
|
705
|
-
}
|
706
|
-
op.on('-o', '--output DIR', 'output directory') {|s|
|
707
|
-
outdir = s
|
708
|
-
}
|
709
|
-
op.on('-s', '--split-size SIZE_IN_KB', "size of each parts (default: #{split_size_kb})", Integer) {|i|
|
710
|
-
split_size_kb = i
|
711
|
-
}
|
712
|
-
op.on('--error-record-output FILE', 'error records output file; default=NULL output stream') {|s|
|
713
|
-
error_record_outdir = s
|
714
|
-
}
|
715
|
-
op.on('--dry-run', 'show samples as JSON and exit', FalseClass) {|b|
|
716
|
-
dry_run = b
|
717
|
-
}
|
718
|
-
op.on('--delimiter CHAR', 'delimiter CHAR; default="," at csv, "\t" at tsv') {|s|
|
719
|
-
delimiter = s
|
720
|
-
}
|
721
|
-
op.on('--newline', 'newline [CRLR, LR, CR]; default=CRLF') {|s|
|
722
|
-
newline = s
|
723
|
-
}
|
724
|
-
op.on('-H', '--column-header', 'first line includes column names', TrueClass) {|b|
|
725
|
-
column_header = b
|
726
|
-
}
|
727
|
-
op.on('-h', '--columns NAME,NAME,...', 'column names (use --column-header instead if the first line has column names)') {|s|
|
728
|
-
columns = s
|
729
|
-
}
|
730
|
-
op.on('--column-types TYPE,TYPE,...', 'column types [string, int, long]') {|s|
|
731
|
-
column_types = s
|
732
|
-
}
|
733
|
-
op.on('--type-conversion-error TYPE', 'type conversion error [skip,null]; default=skip') {|s|
|
734
|
-
type_conversion_error = s
|
735
|
-
}
|
736
|
-
op.on('--exclude-columns NAME,NAME,...', 'exclude columns') {|s|
|
737
|
-
exclude_columns = s
|
738
|
-
}
|
739
|
-
op.on('--only-columns NAME,NAME,...', 'only columns') {|s|
|
740
|
-
only_columns = s
|
741
|
-
}
|
742
|
-
# upload_parts
|
743
|
-
op.on('--auto-perform', 'perform bulk import job automatically', TrueClass) {|b|
|
744
|
-
auto_perform = b
|
745
|
-
}
|
746
|
-
op.on('--auto-commit', 'commit bulk import job automatically', FalseClass) {|b|
|
747
|
-
auto_commit = b
|
748
|
-
}
|
749
|
-
op.on('--parallel NUM', 'upload in parallel (default: 2; max 8)', Integer) {|i|
|
750
|
-
parallel = i
|
751
|
-
}
|
752
|
-
|
753
|
-
files = op.cmd_parse
|
754
|
-
files = [files] unless files.is_a?(Array) # TODO ruby 1.9
|
755
|
-
|
756
|
-
# options validation
|
757
|
-
#unless column_types
|
758
|
-
# $stderr.puts "--column-types TYPE,TYPE,... option is required."
|
759
|
-
# exit 1
|
760
|
-
#end
|
761
|
-
unless outdir
|
762
|
-
$stderr.puts "-o, --output DIR option is required."
|
763
|
-
exit 1
|
764
|
-
end
|
765
|
-
|
766
|
-
return [ format, compress, encoding,
|
767
|
-
time_column, time_format, time_value,
|
768
|
-
outdir, split_size_kb, error_record_output, dry_run,
|
769
|
-
delimiter, newline, column_header, columns, column_types,
|
770
|
-
type_conversion_error, exclude_columns, only_columns,
|
771
|
-
auto_perform, auto_commit, parallel, files ]
|
431
|
+
puts "This command is moved to 'td import:prepare' since 0.10.85."
|
772
432
|
end
|
773
433
|
|
774
434
|
private
|
data/lib/td/command/import.rb
CHANGED
@@ -35,6 +35,10 @@ module Command
|
|
35
35
|
import_generic(APP_OPTION_UPLOAD)
|
36
36
|
end
|
37
37
|
|
38
|
+
def import_auto(op)
|
39
|
+
import_generic(APP_OPTION_UPLOAD, true)
|
40
|
+
end
|
41
|
+
|
38
42
|
def import_perform(op)
|
39
43
|
require 'td/command/bulk_import'
|
40
44
|
bulk_import_perform(op)
|
@@ -66,7 +70,7 @@ module Command
|
|
66
70
|
end
|
67
71
|
|
68
72
|
private
|
69
|
-
def import_generic(subcmd)
|
73
|
+
def import_generic(subcmd, auto=false)
|
70
74
|
# has java runtime
|
71
75
|
check_java
|
72
76
|
|
@@ -91,6 +95,10 @@ module Command
|
|
91
95
|
else
|
92
96
|
java_args << ARGV
|
93
97
|
end
|
98
|
+
if auto
|
99
|
+
java_args << "--auto-perform"
|
100
|
+
java_args << "--auto-commit"
|
101
|
+
end
|
94
102
|
|
95
103
|
# TODO consider parameters including spaces; don't use join(' ')
|
96
104
|
cmd = "#{JAVA_COMMAND} #{jvm_opts.join(' ')} #{java_opts.join(' ')} #{sysprops.join(' ')} #{java_args.join(' ')}"
|
data/lib/td/command/list.rb
CHANGED
@@ -234,10 +234,8 @@ module List
|
|
234
234
|
add_list 'bulk_import:show', %w[name], 'Show list of uploaded parts', 'bulk_import:show'
|
235
235
|
add_list 'bulk_import:create', %w[name db table], 'Create a new bulk import session to the the table', 'bulk_import:create logs_201201 example_db event_logs'
|
236
236
|
add_list 'bulk_import:prepare_parts', %w[files_], 'Convert files into part file format', 'bulk_import:prepare_parts logs/*.csv --format csv --columns time,uid,price,count --time-column "time" -o parts/'
|
237
|
-
add_list 'bulk_import:prepare_parts2', %w[files_], 'Convert files into part file format', 'bulk_import:prepare_parts2 logs/*.csv --format csv --columns time,uid,price,count --column-types long,string,long,int --time-column "time" -o parts/'
|
238
237
|
add_list 'bulk_import:upload_part', %w[name id path.msgpack.gz], 'Upload or re-upload a file into a bulk import session', 'bulk_import:upload_part logs_201201 01h data-201201-01.msgpack.gz'
|
239
238
|
add_list 'bulk_import:upload_parts', %w[name files_], 'Upload or re-upload files into a bulk import session', 'bulk_import:upload_parts parts/* --parallel 4'
|
240
|
-
add_list 'bulk_import:upload_parts2', %w[name files_], 'Upload or re-upload files into a bulk import session', 'bulk_import:upload_parts parts/* --parallel 4'
|
241
239
|
add_list 'bulk_import:delete_part', %w[name id], 'Delete a uploaded file from a bulk import session', 'bulk_import:delete_part logs_201201 01h'
|
242
240
|
add_list 'bulk_import:delete_parts', %w[name ids_], 'Delete uploaded files from a bulk import session', 'bulk_import:delete_parts logs_201201 01h 02h 03h'
|
243
241
|
add_list 'bulk_import:perform', %w[name], 'Start to validate and convert uploaded files', 'bulk_import:perform logs_201201'
|
@@ -252,6 +250,7 @@ module List
|
|
252
250
|
add_list 'import:create', %w[name db table], 'Create a new bulk import session to the the table', 'import:create logs_201201 example_db event_logs'
|
253
251
|
add_list 'import:prepare', %w[files_], 'Convert files into part file format', 'import:prepare logs/*.csv --format csv --columns time,uid,price,count --time-column "time" -o parts/'
|
254
252
|
add_list 'import:upload', %w[name files_], 'Upload or re-upload files into a bulk import session', 'import:upload parts/* --parallel 4'
|
253
|
+
add_list 'import:auto', %w[name files_], 'Upload files and automatically perform and commit the data', 'import:auto parts/* --parallel 4'
|
255
254
|
add_list 'import:perform', %w[name], 'Start to validate and convert uploaded files', 'import:perform logs_201201'
|
256
255
|
add_list 'import:error_records', %w[name], 'Show records which did not pass validations', 'import:error_records logs_201201'
|
257
256
|
add_list 'import:commit', %w[name], 'Start to commit a performed bulk import session', 'import:commit logs_201201'
|
@@ -374,6 +373,8 @@ module List
|
|
374
373
|
add_alias 'bulk_import', 'bulk_import:show'
|
375
374
|
add_alias 'bulk_imports', 'bulk_import:list'
|
376
375
|
|
376
|
+
add_alias 'import', 'import:show'
|
377
|
+
|
377
378
|
add_alias 'job', 'job:show'
|
378
379
|
add_alias 'jobs', 'job:list'
|
379
380
|
add_alias 'kill', 'job:kill'
|
@@ -426,7 +427,6 @@ module List
|
|
426
427
|
add_guess 'show-job', 'job:show'
|
427
428
|
add_guess 'show-jobs', 'job:list'
|
428
429
|
add_guess 'server-status', 'server:status'
|
429
|
-
add_alias 'import', 'import:upload'
|
430
430
|
|
431
431
|
finishup
|
432
432
|
end
|
data/lib/td/command/runner.rb
CHANGED
@@ -42,7 +42,8 @@ Basic commands:
|
|
42
42
|
table # create/delete/list/import/export/tail tables
|
43
43
|
query # issue a query
|
44
44
|
job # show/kill/list jobs
|
45
|
-
|
45
|
+
import # manage bulk import sessions (Java based fast processing)
|
46
|
+
bulk_import # manage bulk import sessions (Old Ruby-based implementation)
|
46
47
|
result # create/delete/list result URLs
|
47
48
|
|
48
49
|
Additional commands:
|
data/lib/td/version.rb
CHANGED
metadata
CHANGED
@@ -1,34 +1,32 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: td
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.10.
|
5
|
-
prerelease:
|
4
|
+
version: 0.10.86
|
6
5
|
platform: ruby
|
7
6
|
authors:
|
8
7
|
- Treasure Data, Inc.
|
9
8
|
autorequire:
|
10
9
|
bindir: bin
|
11
10
|
cert_chain: []
|
12
|
-
date: 2013-08-
|
11
|
+
date: 2013-08-26 00:00:00.000000000 Z
|
13
12
|
dependencies:
|
14
13
|
- !ruby/object:Gem::Dependency
|
15
14
|
name: msgpack
|
16
15
|
requirement: !ruby/object:Gem::Requirement
|
17
|
-
none: false
|
18
16
|
requirements:
|
19
|
-
- -
|
17
|
+
- - '>='
|
20
18
|
- !ruby/object:Gem::Version
|
21
19
|
version: 0.4.4
|
22
|
-
- -
|
20
|
+
- - '!='
|
23
21
|
- !ruby/object:Gem::Version
|
24
22
|
version: 0.5.0
|
25
|
-
- -
|
23
|
+
- - '!='
|
26
24
|
- !ruby/object:Gem::Version
|
27
25
|
version: 0.5.1
|
28
|
-
- -
|
26
|
+
- - '!='
|
29
27
|
- !ruby/object:Gem::Version
|
30
28
|
version: 0.5.2
|
31
|
-
- -
|
29
|
+
- - '!='
|
32
30
|
- !ruby/object:Gem::Version
|
33
31
|
version: 0.5.3
|
34
32
|
- - <
|
@@ -37,21 +35,20 @@ dependencies:
|
|
37
35
|
type: :runtime
|
38
36
|
prerelease: false
|
39
37
|
version_requirements: !ruby/object:Gem::Requirement
|
40
|
-
none: false
|
41
38
|
requirements:
|
42
|
-
- -
|
39
|
+
- - '>='
|
43
40
|
- !ruby/object:Gem::Version
|
44
41
|
version: 0.4.4
|
45
|
-
- -
|
42
|
+
- - '!='
|
46
43
|
- !ruby/object:Gem::Version
|
47
44
|
version: 0.5.0
|
48
|
-
- -
|
45
|
+
- - '!='
|
49
46
|
- !ruby/object:Gem::Version
|
50
47
|
version: 0.5.1
|
51
|
-
- -
|
48
|
+
- - '!='
|
52
49
|
- !ruby/object:Gem::Version
|
53
50
|
version: 0.5.2
|
54
|
-
- -
|
51
|
+
- - '!='
|
55
52
|
- !ruby/object:Gem::Version
|
56
53
|
version: 0.5.3
|
57
54
|
- - <
|
@@ -60,7 +57,6 @@ dependencies:
|
|
60
57
|
- !ruby/object:Gem::Dependency
|
61
58
|
name: yajl-ruby
|
62
59
|
requirement: !ruby/object:Gem::Requirement
|
63
|
-
none: false
|
64
60
|
requirements:
|
65
61
|
- - ~>
|
66
62
|
- !ruby/object:Gem::Version
|
@@ -68,7 +64,6 @@ dependencies:
|
|
68
64
|
type: :runtime
|
69
65
|
prerelease: false
|
70
66
|
version_requirements: !ruby/object:Gem::Requirement
|
71
|
-
none: false
|
72
67
|
requirements:
|
73
68
|
- - ~>
|
74
69
|
- !ruby/object:Gem::Version
|
@@ -76,23 +71,20 @@ dependencies:
|
|
76
71
|
- !ruby/object:Gem::Dependency
|
77
72
|
name: hirb
|
78
73
|
requirement: !ruby/object:Gem::Requirement
|
79
|
-
none: false
|
80
74
|
requirements:
|
81
|
-
- -
|
75
|
+
- - '>='
|
82
76
|
- !ruby/object:Gem::Version
|
83
77
|
version: 0.4.5
|
84
78
|
type: :runtime
|
85
79
|
prerelease: false
|
86
80
|
version_requirements: !ruby/object:Gem::Requirement
|
87
|
-
none: false
|
88
81
|
requirements:
|
89
|
-
- -
|
82
|
+
- - '>='
|
90
83
|
- !ruby/object:Gem::Version
|
91
84
|
version: 0.4.5
|
92
85
|
- !ruby/object:Gem::Dependency
|
93
86
|
name: parallel
|
94
87
|
requirement: !ruby/object:Gem::Requirement
|
95
|
-
none: false
|
96
88
|
requirements:
|
97
89
|
- - ~>
|
98
90
|
- !ruby/object:Gem::Version
|
@@ -100,7 +92,6 @@ dependencies:
|
|
100
92
|
type: :runtime
|
101
93
|
prerelease: false
|
102
94
|
version_requirements: !ruby/object:Gem::Requirement
|
103
|
-
none: false
|
104
95
|
requirements:
|
105
96
|
- - ~>
|
106
97
|
- !ruby/object:Gem::Version
|
@@ -108,7 +99,6 @@ dependencies:
|
|
108
99
|
- !ruby/object:Gem::Dependency
|
109
100
|
name: td-client
|
110
101
|
requirement: !ruby/object:Gem::Requirement
|
111
|
-
none: false
|
112
102
|
requirements:
|
113
103
|
- - ~>
|
114
104
|
- !ruby/object:Gem::Version
|
@@ -116,7 +106,6 @@ dependencies:
|
|
116
106
|
type: :runtime
|
117
107
|
prerelease: false
|
118
108
|
version_requirements: !ruby/object:Gem::Requirement
|
119
|
-
none: false
|
120
109
|
requirements:
|
121
110
|
- - ~>
|
122
111
|
- !ruby/object:Gem::Version
|
@@ -124,7 +113,6 @@ dependencies:
|
|
124
113
|
- !ruby/object:Gem::Dependency
|
125
114
|
name: td-logger
|
126
115
|
requirement: !ruby/object:Gem::Requirement
|
127
|
-
none: false
|
128
116
|
requirements:
|
129
117
|
- - ~>
|
130
118
|
- !ruby/object:Gem::Version
|
@@ -132,7 +120,6 @@ dependencies:
|
|
132
120
|
type: :runtime
|
133
121
|
prerelease: false
|
134
122
|
version_requirements: !ruby/object:Gem::Requirement
|
135
|
-
none: false
|
136
123
|
requirements:
|
137
124
|
- - ~>
|
138
125
|
- !ruby/object:Gem::Version
|
@@ -140,7 +127,6 @@ dependencies:
|
|
140
127
|
- !ruby/object:Gem::Dependency
|
141
128
|
name: rubyzip
|
142
129
|
requirement: !ruby/object:Gem::Requirement
|
143
|
-
none: false
|
144
130
|
requirements:
|
145
131
|
- - ~>
|
146
132
|
- !ruby/object:Gem::Version
|
@@ -148,7 +134,6 @@ dependencies:
|
|
148
134
|
type: :runtime
|
149
135
|
prerelease: false
|
150
136
|
version_requirements: !ruby/object:Gem::Requirement
|
151
|
-
none: false
|
152
137
|
requirements:
|
153
138
|
- - ~>
|
154
139
|
- !ruby/object:Gem::Version
|
@@ -156,7 +141,6 @@ dependencies:
|
|
156
141
|
- !ruby/object:Gem::Dependency
|
157
142
|
name: rake
|
158
143
|
requirement: !ruby/object:Gem::Requirement
|
159
|
-
none: false
|
160
144
|
requirements:
|
161
145
|
- - ~>
|
162
146
|
- !ruby/object:Gem::Version
|
@@ -164,7 +148,6 @@ dependencies:
|
|
164
148
|
type: :development
|
165
149
|
prerelease: false
|
166
150
|
version_requirements: !ruby/object:Gem::Requirement
|
167
|
-
none: false
|
168
151
|
requirements:
|
169
152
|
- - ~>
|
170
153
|
- !ruby/object:Gem::Version
|
@@ -172,7 +155,6 @@ dependencies:
|
|
172
155
|
- !ruby/object:Gem::Dependency
|
173
156
|
name: rspec
|
174
157
|
requirement: !ruby/object:Gem::Requirement
|
175
|
-
none: false
|
176
158
|
requirements:
|
177
159
|
- - ~>
|
178
160
|
- !ruby/object:Gem::Version
|
@@ -180,7 +162,6 @@ dependencies:
|
|
180
162
|
type: :development
|
181
163
|
prerelease: false
|
182
164
|
version_requirements: !ruby/object:Gem::Requirement
|
183
|
-
none: false
|
184
165
|
requirements:
|
185
166
|
- - ~>
|
186
167
|
- !ruby/object:Gem::Version
|
@@ -188,7 +169,6 @@ dependencies:
|
|
188
169
|
- !ruby/object:Gem::Dependency
|
189
170
|
name: simplecov
|
190
171
|
requirement: !ruby/object:Gem::Requirement
|
191
|
-
none: false
|
192
172
|
requirements:
|
193
173
|
- - ~>
|
194
174
|
- !ruby/object:Gem::Version
|
@@ -196,7 +176,6 @@ dependencies:
|
|
196
176
|
type: :development
|
197
177
|
prerelease: false
|
198
178
|
version_requirements: !ruby/object:Gem::Requirement
|
199
|
-
none: false
|
200
179
|
requirements:
|
201
180
|
- - ~>
|
202
181
|
- !ruby/object:Gem::Version
|
@@ -281,33 +260,26 @@ files:
|
|
281
260
|
- td.gemspec
|
282
261
|
homepage: http://treasure-data.com/
|
283
262
|
licenses: []
|
263
|
+
metadata: {}
|
284
264
|
post_install_message:
|
285
265
|
rdoc_options: []
|
286
266
|
require_paths:
|
287
267
|
- lib
|
288
268
|
required_ruby_version: !ruby/object:Gem::Requirement
|
289
|
-
none: false
|
290
269
|
requirements:
|
291
|
-
- -
|
270
|
+
- - '>='
|
292
271
|
- !ruby/object:Gem::Version
|
293
272
|
version: '0'
|
294
|
-
segments:
|
295
|
-
- 0
|
296
|
-
hash: -2194512597141251482
|
297
273
|
required_rubygems_version: !ruby/object:Gem::Requirement
|
298
|
-
none: false
|
299
274
|
requirements:
|
300
|
-
- -
|
275
|
+
- - '>='
|
301
276
|
- !ruby/object:Gem::Version
|
302
277
|
version: '0'
|
303
|
-
segments:
|
304
|
-
- 0
|
305
|
-
hash: -2194512597141251482
|
306
278
|
requirements: []
|
307
279
|
rubyforge_project:
|
308
|
-
rubygems_version:
|
280
|
+
rubygems_version: 2.0.2
|
309
281
|
signing_key:
|
310
|
-
specification_version:
|
282
|
+
specification_version: 4
|
311
283
|
summary: CLI to manage data on Treasure Data, the Hadoop-based cloud data warehousing
|
312
284
|
test_files:
|
313
285
|
- spec/file_reader/filter_spec.rb
|