imw 0.1.1 → 0.2.0
Sign up to get free protection for your applications and to get access to all the features.
- data/.gitignore +4 -1
- data/Rakefile +10 -0
- data/TODO +18 -0
- data/VERSION +1 -1
- data/bin/imw +1 -1
- data/etc/imwrc.rb +0 -50
- data/examples/dataset.rb +12 -0
- data/lib/imw/boot.rb +55 -9
- data/lib/imw/dataset/paths.rb +15 -24
- data/lib/imw/dataset/workflow.rb +131 -72
- data/lib/imw/dataset.rb +94 -186
- data/lib/imw/parsers/html_parser.rb +1 -1
- data/lib/imw/parsers.rb +1 -1
- data/lib/imw/repository.rb +3 -27
- data/lib/imw/resource.rb +190 -0
- data/lib/imw/resources/archive.rb +97 -0
- data/lib/imw/resources/archives_and_compressed/bz2.rb +18 -0
- data/lib/imw/resources/archives_and_compressed/gz.rb +18 -0
- data/lib/imw/resources/archives_and_compressed/rar.rb +23 -0
- data/lib/imw/resources/archives_and_compressed/tar.rb +23 -0
- data/lib/imw/resources/archives_and_compressed/tarbz2.rb +78 -0
- data/lib/imw/resources/archives_and_compressed/targz.rb +78 -0
- data/lib/imw/resources/archives_and_compressed/zip.rb +57 -0
- data/lib/imw/resources/archives_and_compressed.rb +32 -0
- data/lib/imw/resources/compressed_file.rb +89 -0
- data/lib/imw/resources/compressible.rb +77 -0
- data/lib/imw/resources/formats/delimited.rb +92 -0
- data/lib/imw/resources/formats/excel.rb +125 -0
- data/lib/imw/resources/formats/json.rb +53 -0
- data/lib/imw/resources/formats/sgml.rb +72 -0
- data/lib/imw/resources/formats/yaml.rb +53 -0
- data/lib/imw/resources/formats.rb +32 -0
- data/lib/imw/resources/local.rb +198 -0
- data/lib/imw/resources/remote.rb +110 -0
- data/lib/imw/resources/schemes/hdfs.rb +242 -0
- data/lib/imw/resources/schemes/http.rb +161 -0
- data/lib/imw/resources/schemes/s3.rb +137 -0
- data/lib/imw/resources/schemes.rb +19 -0
- data/lib/imw/resources.rb +118 -0
- data/lib/imw/runner.rb +5 -4
- data/lib/imw/transforms/archiver.rb +215 -0
- data/lib/imw/transforms/transferer.rb +103 -0
- data/lib/imw/transforms.rb +8 -0
- data/lib/imw/utils/error.rb +26 -30
- data/lib/imw/utils/extensions/array.rb +5 -15
- data/lib/imw/utils/extensions/hash.rb +6 -16
- data/lib/imw/utils/extensions/hpricot.rb +0 -14
- data/lib/imw/utils/extensions/string.rb +5 -15
- data/lib/imw/utils/extensions/symbol.rb +0 -13
- data/lib/imw/utils/extensions.rb +65 -0
- data/lib/imw/utils/log.rb +14 -13
- data/lib/imw/utils/misc.rb +0 -6
- data/lib/imw/utils/paths.rb +101 -42
- data/lib/imw/utils/version.rb +8 -9
- data/lib/imw/utils.rb +2 -18
- data/lib/imw.rb +92 -17
- data/spec/data/sample.csv +1 -1
- data/spec/data/sample.json +1 -0
- data/spec/data/sample.tsv +1 -1
- data/spec/data/sample.txt +1 -1
- data/spec/data/sample.xml +1 -1
- data/spec/data/sample.yaml +1 -1
- data/spec/imw/dataset/paths_spec.rb +32 -0
- data/spec/imw/dataset/workflow_spec.rb +41 -0
- data/spec/imw/resource_spec.rb +79 -0
- data/spec/imw/resources/archive_spec.rb +69 -0
- data/spec/imw/resources/archives_and_compressed/bz2_spec.rb +15 -0
- data/spec/imw/resources/archives_and_compressed/gz_spec.rb +15 -0
- data/spec/imw/resources/archives_and_compressed/rar_spec.rb +16 -0
- data/spec/imw/resources/archives_and_compressed/tar_spec.rb +16 -0
- data/spec/imw/resources/archives_and_compressed/tarbz2_spec.rb +24 -0
- data/spec/imw/resources/archives_and_compressed/targz_spec.rb +21 -0
- data/spec/imw/resources/archives_and_compressed/zip_spec.rb +16 -0
- data/spec/imw/resources/compressed_file_spec.rb +48 -0
- data/spec/imw/resources/compressible_spec.rb +36 -0
- data/spec/imw/resources/formats/delimited_spec.rb +33 -0
- data/spec/imw/resources/formats/json_spec.rb +32 -0
- data/spec/imw/resources/formats/sgml_spec.rb +24 -0
- data/spec/imw/resources/formats/yaml_spec.rb +41 -0
- data/spec/imw/resources/local_spec.rb +98 -0
- data/spec/imw/resources/remote_spec.rb +35 -0
- data/spec/imw/resources/schemes/hdfs_spec.rb +61 -0
- data/spec/imw/resources/schemes/http_spec.rb +19 -0
- data/spec/imw/resources/schemes/s3_spec.rb +19 -0
- data/spec/imw/transforms/archiver_spec.rb +120 -0
- data/spec/imw/transforms/transferer_spec.rb +113 -0
- data/spec/imw/utils/paths_spec.rb +5 -33
- data/spec/imw/utils/shared_paths_spec.rb +29 -0
- data/spec/spec_helper.rb +5 -5
- data/spec/support/paths_matcher.rb +67 -0
- data/spec/support/random.rb +39 -36
- metadata +88 -75
- data/lib/imw/dataset/task.rb +0 -41
- data/lib/imw/files/archive.rb +0 -113
- data/lib/imw/files/basicfile.rb +0 -122
- data/lib/imw/files/binary.rb +0 -28
- data/lib/imw/files/compressed_file.rb +0 -93
- data/lib/imw/files/compressed_files_and_archives.rb +0 -334
- data/lib/imw/files/compressible.rb +0 -103
- data/lib/imw/files/csv.rb +0 -113
- data/lib/imw/files/directory.rb +0 -62
- data/lib/imw/files/excel.rb +0 -84
- data/lib/imw/files/json.rb +0 -41
- data/lib/imw/files/sgml.rb +0 -46
- data/lib/imw/files/text.rb +0 -68
- data/lib/imw/files/yaml.rb +0 -46
- data/lib/imw/files.rb +0 -125
- data/lib/imw/packagers/archiver.rb +0 -126
- data/lib/imw/packagers/s3_mover.rb +0 -36
- data/lib/imw/packagers.rb +0 -8
- data/lib/imw/utils/components.rb +0 -61
- data/lib/imw/utils/config.rb +0 -46
- data/lib/imw/utils/extensions/class/attribute_accessors.rb +0 -8
- data/lib/imw/utils/extensions/core.rb +0 -27
- data/lib/imw/utils/extensions/dir.rb +0 -24
- data/lib/imw/utils/extensions/file_core.rb +0 -64
- data/lib/imw/utils/extensions/typed_struct.rb +0 -22
- data/lib/imw/utils/extensions/uri.rb +0 -59
- data/lib/imw/utils/view/dump_csv.rb +0 -112
- data/lib/imw/utils/view/dump_csv_older.rb +0 -117
- data/lib/imw/utils/view.rb +0 -113
- data/spec/imw/dataset/datamapper/uri_spec.rb +0 -43
- data/spec/imw/dataset/datamapper_spec_helper.rb +0 -11
- data/spec/imw/files/archive_spec.rb +0 -118
- data/spec/imw/files/basicfile_spec.rb +0 -121
- data/spec/imw/files/bz2_spec.rb +0 -32
- data/spec/imw/files/compressed_file_spec.rb +0 -96
- data/spec/imw/files/compressible_spec.rb +0 -100
- data/spec/imw/files/file_spec.rb +0 -144
- data/spec/imw/files/gz_spec.rb +0 -32
- data/spec/imw/files/rar_spec.rb +0 -33
- data/spec/imw/files/tar_spec.rb +0 -31
- data/spec/imw/files/text_spec.rb +0 -23
- data/spec/imw/files/zip_spec.rb +0 -31
- data/spec/imw/files_spec.rb +0 -38
- data/spec/imw/packagers/archiver_spec.rb +0 -125
- data/spec/imw/packagers/s3_mover_spec.rb +0 -7
- data/spec/imw/utils/extensions/file_core_spec.rb +0 -72
- data/spec/imw/utils/extensions/find_spec.rb +0 -113
- data/spec/imw/workflow/rip/local_spec.rb +0 -89
- data/spec/imw/workflow/rip_spec.rb +0 -27
- data/spec/support/archive_contents_matcher.rb +0 -94
- data/spec/support/directory_contents_matcher.rb +0 -61
data/lib/imw/utils/log.rb
CHANGED
@@ -1,16 +1,16 @@
|
|
1
1
|
require 'logger'
|
2
2
|
|
3
3
|
module IMW
|
4
|
+
|
5
|
+
# Default log file.
|
4
6
|
LOG_FILE_DESTINATION = STDERR unless defined?(LOG_FILE_DESTINATION)
|
7
|
+
|
5
8
|
LOG_TIMEFORMAT = "%Y%m%d-%H:%M:%S " unless defined?(LOG_TIMEFORMAT)
|
6
9
|
|
7
10
|
class << self; attr_accessor :log end
|
8
|
-
|
9
|
-
# Create a Logger and point it at LOG_FILE_DESTINATION
|
10
|
-
#
|
11
|
-
# LOG_FILE_DESTINATION is STDOUT by default; redefine it in your
|
12
|
-
# ~/.imwrc, or set IMW.log yourself, if that's not cool.
|
13
|
-
#
|
11
|
+
|
12
|
+
# Create a Logger and point it at IMW::LOG_FILE_DESTINATION which is
|
13
|
+
# set in ~/.imwrc and defaults to STDERR.
|
14
14
|
def self.instantiate_logger!
|
15
15
|
IMW.log ||= Logger.new(LOG_FILE_DESTINATION)
|
16
16
|
IMW.log.datetime_format = "%Y%m%d-%H:%M:%S "
|
@@ -18,15 +18,19 @@ module IMW
|
|
18
18
|
end
|
19
19
|
|
20
20
|
def announce *events
|
21
|
-
options = events.extract_options!
|
21
|
+
options = events.flatten.extract_options!
|
22
22
|
options.reverse_merge! :level => Logger::INFO
|
23
|
-
# puts [options, events ].inspect, "*"*76
|
24
23
|
IMW.log.add options[:level], events.join("\n")
|
25
24
|
end
|
26
25
|
def banner *events
|
27
|
-
options = events.extract_options!
|
26
|
+
options = events.flatten.extract_options!
|
28
27
|
options.reverse_merge! :level => Logger::INFO
|
29
|
-
["*"*75, events, "*"*75]
|
28
|
+
announce(["*"*75, events, "*"*75], options)
|
29
|
+
end
|
30
|
+
def warn *events
|
31
|
+
options = events.flatten.extract_options!
|
32
|
+
options.reverse_merge! :level => Logger::WARN
|
33
|
+
announce events, options
|
30
34
|
end
|
31
35
|
|
32
36
|
PROGRESS_TRACKERS = {}
|
@@ -61,7 +65,4 @@ module IMW
|
|
61
65
|
end
|
62
66
|
end
|
63
67
|
|
64
|
-
#
|
65
|
-
# Make the default logger
|
66
|
-
#
|
67
68
|
IMW.instantiate_logger!
|
data/lib/imw/utils/misc.rb
CHANGED
@@ -1,10 +1,4 @@
|
|
1
1
|
module IMW
|
2
|
-
# Return a string representing the current UTC time in the IMW
|
3
|
-
# format.
|
4
|
-
def self.current_utc_time_string
|
5
|
-
Time.now.utc.strftime(IMW::STRFTIME_FORMAT)
|
6
|
-
end
|
7
|
-
|
8
2
|
|
9
3
|
# A simple counter. The +value+ and +add+ methods read and
|
10
4
|
# increment the counter's value.
|
data/lib/imw/utils/paths.rb
CHANGED
@@ -1,3 +1,5 @@
|
|
1
|
+
require 'pathname'
|
2
|
+
|
1
3
|
module IMW
|
2
4
|
|
3
5
|
# Implements methods designed to work with an object's
|
@@ -9,19 +11,48 @@ module IMW
|
|
9
11
|
# <tt>@paths</tt>.
|
10
12
|
module Paths
|
11
13
|
|
12
|
-
# Expands a shorthand workflow path specification to an
|
13
|
-
#
|
14
|
+
# Expands a shorthand workflow path specification to an actual
|
15
|
+
# file path. Strings are interpreted literally but symbols are
|
16
|
+
# first resolved to the paths they represent.
|
17
|
+
#
|
18
|
+
# add_path :foo, '~/whoa'
|
19
|
+
# path_to :foo, 'my_thing'
|
20
|
+
# => '~/whoa/my_thing'
|
14
21
|
#
|
15
|
-
#
|
16
|
-
#
|
17
|
-
# => (...)/data/ripd/gd2.mlb.com/components/game/mlb/year_2008/month_06/day_08/miniscoreboard.xml
|
22
|
+
# @param [String, Symbol] pathsegs the path segments to join
|
23
|
+
# @return [String] the resulting expanded path
|
18
24
|
def path_to *pathsegs
|
19
|
-
|
20
|
-
|
21
|
-
|
22
|
-
|
23
|
-
|
24
|
-
|
25
|
+
path = Pathname.new path_to_helper(*pathsegs)
|
26
|
+
path.absolute? ? File.expand_path(path) : path.to_s
|
27
|
+
end
|
28
|
+
|
29
|
+
# Return the presently defined paths for this object.
|
30
|
+
#
|
31
|
+
# @return [Hash]
|
32
|
+
def paths
|
33
|
+
@paths ||= {}
|
34
|
+
end
|
35
|
+
|
36
|
+
# Adds a symbolic path for expansion by +path_to+.
|
37
|
+
#
|
38
|
+
# add_path :foo, '~/whoa'
|
39
|
+
# add_path :bar, :foo, 'baz'
|
40
|
+
# path_to :bar
|
41
|
+
# => '~/whoa/baz'
|
42
|
+
#
|
43
|
+
# @param [Symbol] sym the name of the path to store
|
44
|
+
# @param [Symbol, String] pathsegs the path segments to use to define the path to the name
|
45
|
+
# @return [String] the resulting path
|
46
|
+
def add_path sym, *pathsegs
|
47
|
+
paths[sym] = pathsegs.flatten
|
48
|
+
path_to(sym)
|
49
|
+
end
|
50
|
+
|
51
|
+
# Removes a symbolic path for expansion by +path_to+.
|
52
|
+
#
|
53
|
+
# @param [Symbol] sym the stored path symbol to remove
|
54
|
+
def remove_path sym
|
55
|
+
paths.delete sym if paths.include? sym
|
25
56
|
end
|
26
57
|
|
27
58
|
private
|
@@ -29,7 +60,7 @@ module IMW
|
|
29
60
|
# +path_to_helper+ handles the recursive calls for +path_to+.
|
30
61
|
expanded = pathsegs.flatten.compact.map do |pathseg|
|
31
62
|
case
|
32
|
-
when pathseg.is_a?(Symbol) &&
|
63
|
+
when pathseg.is_a?(Symbol) && paths.include?(pathseg) then path_to(paths[pathseg])
|
33
64
|
when pathseg.is_a?(Symbol) && IMW::PATHS.include?(pathseg) then path_to(IMW::PATHS[pathseg])
|
34
65
|
when pathseg.is_a?(Symbol) then raise IMW::PathError.new("No path expansion set for #{pathseg.inspect}")
|
35
66
|
else pathseg
|
@@ -37,29 +68,70 @@ module IMW
|
|
37
68
|
end
|
38
69
|
File.join(*expanded)
|
39
70
|
end
|
40
|
-
|
71
|
+
end
|
41
72
|
|
42
|
-
# Adds a symbolic path for expansion by +path_to+.
|
43
|
-
def add_path sym, *pathsegs
|
44
|
-
@paths[sym] = pathsegs.flatten
|
45
|
-
end
|
46
73
|
|
47
|
-
|
48
|
-
|
49
|
-
|
50
|
-
|
51
|
-
|
74
|
+
# Default paths for the IMW. Chosen to make sense on most *NIX
|
75
|
+
# distributions.
|
76
|
+
DEFAULT_PATHS = {
|
77
|
+
:home => ENV['HOME'],
|
78
|
+
:data_root => "/var/lib/imw",
|
79
|
+
:log_root => "/var/log/imw",
|
80
|
+
:scripts_root => "/usr/share/imw",
|
81
|
+
:tmp_root => "/tmp/imw",
|
82
|
+
|
83
|
+
# the imw library
|
84
|
+
:imw_root => File.expand_path(File.dirname(__FILE__) + "/../../.."),
|
85
|
+
:imw_bin => [:imw_root, 'bin'],
|
86
|
+
:imw_etc => [:imw_root, 'etc'],
|
87
|
+
:imw_lib => [:imw_root, 'lib'],
|
88
|
+
|
89
|
+
# workflow
|
90
|
+
:ripd_root => [:data_root, 'ripd'],
|
91
|
+
:rawd_root => [:data_root, 'rawd'],
|
92
|
+
:fixd_root => [:data_root, 'fixd'],
|
93
|
+
:pkgd_root => [:data_root, 'pkgd']
|
94
|
+
}
|
95
|
+
defined?(PATHS) ? PATHS.reverse_merge!(DEFAULT_PATHS) : PATHS = DEFAULT_PATHS
|
52
96
|
|
97
|
+
# Expands a shorthand workflow path specification to an actual
|
98
|
+
# file path. Strings are interpreted literally but symbols are
|
99
|
+
# first resolved to the paths they represent.
|
100
|
+
#
|
101
|
+
# IMW.add_path :foo, '~/whoa'
|
102
|
+
# IMW.path_to :foo, 'my_thing'
|
103
|
+
# => '~/whoa/my_thing'
|
104
|
+
#
|
105
|
+
# @param [String, Symbol] pathsegs the path segments to join
|
106
|
+
# @return [String] the resulting expanded path
|
53
107
|
def self.path_to *pathsegs
|
54
|
-
|
55
|
-
|
56
|
-
|
57
|
-
|
58
|
-
|
59
|
-
|
108
|
+
path = Pathname.new IMW.path_to_helper(*pathsegs)
|
109
|
+
path.absolute? ? File.expand_path(path) : path.to_s
|
110
|
+
end
|
111
|
+
|
112
|
+
# Adds a symbolic path for expansion by +path_to+.
|
113
|
+
#
|
114
|
+
# IMW.add_path :foo, '~/whoa'
|
115
|
+
# IMW.add_path :bar, :foo, 'baz'
|
116
|
+
# IMW.path_to :bar
|
117
|
+
# => '~/whoa/baz'
|
118
|
+
#
|
119
|
+
# @param [Symbol] sym the name of the path to store
|
120
|
+
# @param [Symbol, String] pathsegs the path segments to use to define the path to the name
|
121
|
+
# @return [String] the resulting path
|
122
|
+
def self.add_path sym, *pathsegs
|
123
|
+
IMW::PATHS[sym] = pathsegs.flatten
|
124
|
+
path_to[sym]
|
60
125
|
end
|
61
126
|
|
62
|
-
|
127
|
+
# Removes a symbolic path for expansion by +path_to+.
|
128
|
+
#
|
129
|
+
# @param [Symbol] sym the stored path symbol to remove
|
130
|
+
def self.remove_path sym
|
131
|
+
IMW::PATHS.delete sym if IMW::PATHS.include? sym
|
132
|
+
end
|
133
|
+
|
134
|
+
protected
|
63
135
|
def self.path_to_helper *pathsegs # :nodoc:
|
64
136
|
# +path_to_helper+ handles the recursive calls for +path_to+.
|
65
137
|
expanded = pathsegs.flatten.compact.map do |pathseg|
|
@@ -71,17 +143,4 @@ module IMW
|
|
71
143
|
end
|
72
144
|
File.join(*expanded)
|
73
145
|
end
|
74
|
-
public
|
75
|
-
|
76
|
-
# Adds a symbolic path for expansion by +path_to+.
|
77
|
-
def self.add_path sym, *pathsegs
|
78
|
-
IMW::PATHS[sym] = pathsegs.flatten
|
79
|
-
end
|
80
|
-
|
81
|
-
# Removes a symbolic path for expansion by +path_to+.
|
82
|
-
def self.remove_path sym
|
83
|
-
IMW::PATHS.delete sym if IMW::PATHS.include? sym
|
84
|
-
end
|
85
146
|
end
|
86
|
-
|
87
|
-
# puts "#{File.basename(__FILE__)}: Your monkeywrench glows alternately dim then bright as you wander, suggesting to you which paths to take."
|
data/lib/imw/utils/version.rb
CHANGED
@@ -1,12 +1,11 @@
|
|
1
|
-
|
2
|
-
module IMWVersion #:nodoc:
|
1
|
+
module IMW
|
3
2
|
unless defined?(VERSION)
|
4
|
-
|
5
|
-
|
6
|
-
|
7
|
-
|
8
|
-
|
9
|
-
|
10
|
-
|
3
|
+
module VERSION #:nodoc:
|
4
|
+
MAJOR = 0
|
5
|
+
MINOR = 0
|
6
|
+
TINY = 0
|
7
|
+
|
8
|
+
STRING = [MAJOR, MINOR, TINY].join('.')
|
9
|
+
end
|
11
10
|
end
|
12
11
|
end
|
data/lib/imw/utils.rb
CHANGED
@@ -1,24 +1,8 @@
|
|
1
|
-
#
|
2
|
-
# h2. lib/imw/utils.rb -- utility functions
|
3
|
-
#
|
4
|
-
# == About
|
5
|
-
#
|
6
|
-
# Author:: (Philip flip Kromer, Dhruv Bansal) for Infinite Monkeywrench Project (mailto:coders@infochimps.org)
|
7
|
-
# Copyright:: Copyright (c) 2008 infochimps.org
|
8
|
-
# License:: GPL 3.0
|
9
|
-
# Website:: http://infinitemonkeywrench.org/
|
10
|
-
#
|
11
|
-
|
12
1
|
require 'rubygems'
|
2
|
+
require 'fileutils'
|
13
3
|
require 'imw/utils/error'
|
14
4
|
require 'imw/utils/log'
|
15
|
-
require 'imw/utils/config'
|
16
5
|
require 'imw/utils/paths'
|
17
6
|
require 'imw/utils/misc'
|
18
|
-
require 'imw/utils/
|
19
|
-
require 'imw/utils/extensions/core'
|
20
|
-
require 'fileutils'
|
21
|
-
require 'pathname'
|
22
|
-
|
7
|
+
require 'imw/utils/extensions'
|
23
8
|
|
24
|
-
# puts "#{File.basename(__FILE__)}: Early economists thought they would measure the utility of an action in units of `utils'. Really." # at bottom
|
data/lib/imw.rb
CHANGED
@@ -1,11 +1,6 @@
|
|
1
1
|
require 'rubygems'
|
2
2
|
require 'imw/boot'
|
3
3
|
require 'imw/utils'
|
4
|
-
require 'imw/dataset'
|
5
|
-
require 'imw/repository'
|
6
|
-
require 'imw/files'
|
7
|
-
require 'imw/parsers'
|
8
|
-
require 'imw/packagers'
|
9
4
|
|
10
5
|
# The Infinite Monkeywrench (IMW) is a Ruby library for ripping,
|
11
6
|
# extracting, parsing, munging, and packaging datasets. It allows you
|
@@ -13,19 +8,99 @@ require 'imw/packagers'
|
|
13
8
|
# transformations of data as a network of dependencies (a la Make or
|
14
9
|
# Rake).
|
15
10
|
#
|
16
|
-
#
|
17
|
-
#
|
18
|
-
# of File.open). These classes do a lot of work to ensure that all
|
19
|
-
# objects returned by IMW.open share methods (write, read, load, dump,
|
20
|
-
# parse, compress, extract, &c.) while continuing to use existing
|
21
|
-
# implementations of these concepts.
|
11
|
+
# IMW has a few central concepts: resources, datasets, workflows, and
|
12
|
+
# repositories.
|
22
13
|
#
|
23
|
-
#
|
24
|
-
#
|
25
|
-
#
|
26
|
-
#
|
14
|
+
# Resources represent individual data resources like local files,
|
15
|
+
# websites, databases, &c. Resources are typically instantiated via
|
16
|
+
# IMW.open, with IMW doing the work of figuring out what to return
|
17
|
+
# based on the URI passed in.
|
27
18
|
#
|
28
|
-
#
|
29
|
-
#
|
19
|
+
# Datasets represent collections of related data resources. An
|
20
|
+
# IMW::Dataset comes with a pre-defined (but customizable) workflow
|
21
|
+
# that takes data resources through several steps: rip, parse, munge,
|
22
|
+
# and package. The workflow leverages Rake and so the various tasks
|
23
|
+
# that are necessary to process the data till it is nice and pretty
|
24
|
+
# can all be linked with dependencies.
|
25
|
+
#
|
26
|
+
# Repositories are collections of datasets and it is on these
|
27
|
+
# collections that the +imw+ command line tool operates.
|
30
28
|
module IMW
|
29
|
+
autoload :Resource, 'imw/resource'
|
30
|
+
autoload :Resources, 'imw/resources'
|
31
|
+
autoload :Repository, 'imw/repository'
|
32
|
+
autoload :Dataset, 'imw/dataset'
|
33
|
+
autoload :Transforms, 'imw/transforms'
|
34
|
+
autoload :Parsers, 'imw/parsers'
|
35
|
+
|
36
|
+
# Open a resource at the given +uri+. The resource will
|
37
|
+
# automatically be extended by modules which make sense given the
|
38
|
+
# +uri+.
|
39
|
+
#
|
40
|
+
# See the documentation for IMW::Resource and the various modules
|
41
|
+
# within IMW::Resources for more information and options.
|
42
|
+
#
|
43
|
+
# Passing in an IMW::Resource will simply return it.
|
44
|
+
#
|
45
|
+
# @param [String, Addressable::URI, IMW::Resource] obj the URI to open
|
46
|
+
# @return [IMW::Resource] the resulting resource, property extended for the given URI
|
47
|
+
def self.open obj, options={}
|
48
|
+
return obj if obj.is_a?(IMW::Resource)
|
49
|
+
IMW::Resource.new(obj, options)
|
50
|
+
end
|
51
|
+
|
52
|
+
# Works the same way as IMW.open except opens the resource for
|
53
|
+
# writing.
|
54
|
+
#
|
55
|
+
# @param [String, Addressable::URI] uri the URI to open
|
56
|
+
# @return [IMW::Resource] the resultng resource, properly extended for the given URI and opened for writing.
|
57
|
+
def self.open! uri, options={}
|
58
|
+
IMW::Resource.new(uri, options.merge(:mode => 'w'))
|
59
|
+
end
|
60
|
+
|
61
|
+
# The default repository in which to place datasets. See the
|
62
|
+
# documentation for IMW::Repository for more information on how
|
63
|
+
# datasets and repositories fit together.
|
64
|
+
#
|
65
|
+
# @return [IMW::Repository] the default IMW repository
|
66
|
+
def self.repository
|
67
|
+
@@repository ||= IMW::Repository.new
|
68
|
+
end
|
69
|
+
|
70
|
+
# Create a dataset and put it in the default IMW repository. Also
|
71
|
+
# yields the dataset so you can define its workflow
|
72
|
+
#
|
73
|
+
# IMW.dataset :my_dataset do
|
74
|
+
#
|
75
|
+
# # Define some paths we're going to use
|
76
|
+
# add_path :raw_data, :ripd, 'raw_data.csv'
|
77
|
+
# add_path :fixd_data, :fixd, 'fixed_data.csv'
|
78
|
+
#
|
79
|
+
# # Copy a file from a website to this dataset's +ripd+ directory.
|
80
|
+
# rip do
|
81
|
+
# IMW.open('http://mysite.com/data_archives/2010/03/03.csv').cp(path_to(:raw_data))
|
82
|
+
# end
|
83
|
+
#
|
84
|
+
# # Filter the raw data to those values which match some criterion defined by <tt>accept?</tt>
|
85
|
+
# munge do
|
86
|
+
# IMW.open(path_to(:raw_data)).map do |row|
|
87
|
+
# row if accept?(row)
|
88
|
+
# end.compact.dump(path_to(:fixd_data))
|
89
|
+
# end
|
90
|
+
#
|
91
|
+
# # Compress this new data
|
92
|
+
# package do
|
93
|
+
# IMW.open(path_to(:fixd_data)).compress.mv(path_to(:pkgd))
|
94
|
+
# end
|
95
|
+
# end
|
96
|
+
#
|
97
|
+
# @param [Symbol, String] handle the handle to identify this dataset with
|
98
|
+
# @param [Hash] options a hash of options (see IMW::Dataset)
|
99
|
+
# @return [IMW::Dataset] the new dataset
|
100
|
+
def self.dataset handle, options={}, &block
|
101
|
+
d = IMW::Dataset.new(handle, options)
|
102
|
+
d.instance_eval(&block) if block_given?
|
103
|
+
d
|
104
|
+
end
|
105
|
+
|
31
106
|
end
|
data/spec/data/sample.csv
CHANGED
@@ -81,7 +81,7 @@ ID,Name,Genus,Species
|
|
81
81
|
080,Tonkean Macaque,Macaca,tonkeana
|
82
82
|
081,Heck's Macaque,Macaca,hecki
|
83
83
|
082,Gorontalo Macaque,Macaca,nigrescens
|
84
|
-
083,Celebes Crested Macaque or Black
|
84
|
+
083,Celebes Crested Macaque or Black Ape,Macaca,nigra
|
85
85
|
084,Crab-eating Macaque or Long-tailed Macaque or Kera,Macaca,fascicularis
|
86
86
|
085,Stump-tailed Macaque or Bear Macaque,Macaca,arctoides
|
87
87
|
086,Rhesus Macaque,Macaca,mulatta
|
@@ -0,0 +1 @@
|
|
1
|
+
{"monkeys":[{"monkey":{"name":"Gray-bellied Night Monkey","id":1,"genus":"Aotus","species":"lemurinus"}},{"monkey":{"name":"Panamanian Night Monkey","id":2,"genus":"Aotus","species":"zonalis"}},{"monkey":{"name":"Hern\u00e1ndez-Camacho's Night Monkey","id":3,"genus":"Aotus","species":"jorgehernandezi"}},{"monkey":{"name":"Gray-handed Night Monkey","id":4,"genus":"Aotus","species":"griseimembra"}},{"monkey":{"name":"Hershkovitz's Night Monkey","id":5,"genus":"Aotus","species":"hershkovitzi"}},{"monkey":{"name":"Brumback's Night Monkey","id":6,"genus":"Aotus","species":"brumbacki"}},{"monkey":{"name":"Three-striped Night Monkey","id":7,"genus":"Aotus","species":"trivirgatus"}},{"monkey":{"name":"Spix's Night Monkey","id":"008","genus":"Aotus","species":"vociferans"}},{"monkey":{"name":"Malaysian Lar Gibbon","id":"009","genus":"Hylobates","species":"lar lar"}},{"monkey":{"name":"Carpenter's Lar Gibbon","id":8,"genus":"Hylobates","species":"lar carpenteri"}},{"monkey":{"name":"Central Lar Gibbon","id":9,"genus":"Hylobates","species":"lar entelloides"}},{"monkey":{"name":"Sumatran Lar Gibbon","id":10,"genus":"Hylobates","species":"lar vestitus"}},{"monkey":{"name":"Yunnan Lar Gibbon","id":11,"genus":"Hylobates","species":"lar yunnanensis"}},{"monkey":{"name":"Mountain Agile Gibbon","id":12,"genus":"Hylobates","species":"agilis agilis"}},{"monkey":{"name":"Bornean White-bearded Gibbon","id":13,"genus":"Hylobates","species":"agilis albibarbis"}},{"monkey":{"name":"Lowland Agile Gibbon","id":14,"genus":"Hylobates","species":"agilis unko"}},{"monkey":{"name":"M\u00fcller's Gray Gibbon","id":15,"genus":"Hylobates","species":"muelleri muelleri"}},{"monkey":{"name":"Abbott's Gray Gibbon","id":"018","genus":"Hylobates","species":"muelleri abbotti"}},{"monkey":{"name":"Northern Gray Gibbon","id":"019","genus":"Hylobates","species":"muelleri funereus"}},{"monkey":{"name":"Black Tamarin","id":16,"genus":"Saguinas","species":"niger"}},{"monkey":{"name":"Black-mantled Tamarin","id":17,"genus":"Saguinas","species":"nigricollis"}},{"monkey":{"name":"Brown-mantled Tamarin","id":18,"genus":"Saguinas","species":"fuscicollis"}},{"monkey":{"name":"Cottontop Tamarin or Pinch\u00e9 Tamarin","id":19,"genus":"Saguinas","species":"oedipus"}},{"monkey":{"name":"Emperor Tamarin","id":20,"genus":"Saguinas","species":"imperator"}},{"monkey":{"name":"Geoffroy's Tamarin","id":21,"genus":"Saguinas","species":"geoffroyi"}},{"monkey":{"name":"Golden-mantled Tamarin","id":22,"genus":"Saguinas","species":"tripartitus"}},{"monkey":{"name":"Graells's Tamarin","id":23,"genus":"Saguinas","species":"graellsi"}},{"monkey":{"name":"Martins's Tamarin","id":"028","genus":"Saguinas","species":"martinsi"}},{"monkey":{"name":"Mottle-faced Tamarin","id":"029","genus":"Saguinas","species":"inustus"}},{"monkey":{"name":"Moustached Tamarin","id":24,"genus":"Saguinas","species":"mystax"}},{"monkey":{"name":"Pied Tamarin","id":25,"genus":"Saguinas","species":"bicolor"}},{"monkey":{"name":"Red-capped Tamarin","id":26,"genus":"Saguinas","species":"pileatus"}},{"monkey":{"name":"Red-handed Tamarin","id":27,"genus":"Saguinas","species":"midas"}},{"monkey":{"name":"White-footed Tamarin","id":28,"genus":"Saguinas","species":"leucopus"}},{"monkey":{"name":"White-lipped Tamarin","id":29,"genus":"Saguinas","species":"labiatus"}},{"monkey":{"name":"White-mantled Tamarin","id":30,"genus":"Saguinas","species":"melanoleucus"}},{"monkey":{"name":"Allen's Swamp Monkey","id":31,"genus":"Allenopithecus","species":"nigroviridis"}},{"monkey":{"name":"Angolan Talapoin","id":"038","genus":"Miopithecus","species":"talapoin"}},{"monkey":{"name":"Gabon Talapoin","id":"039","genus":"Miopithecus","species":"ogouensis"}},{"monkey":{"name":"Patas Monkey","id":32,"genus":"Erythrocebus","species":"patas"}},{"monkey":{"name":"Green Monkey","id":33,"genus":"Chlorocebus","species":"sabaeus"}},{"monkey":{"name":"Grivet","id":34,"genus":"Chlorocebus","species":"aethiops"}},{"monkey":{"name":"Bale Mountains Vervet","id":35,"genus":"Chlorocebus","species":"djamdjamensis"}},{"monkey":{"name":"Tantalus Monkey","id":36,"genus":"Chlorocebus","species":"tantalus"}},{"monkey":{"name":"Vervet Monkey","id":37,"genus":"Chlorocebus","species":"pygerythrus"}},{"monkey":{"name":"Malbrouck","id":38,"genus":"Chlorocebus","species":"cynosuros"}},{"monkey":{"name":"Dryas Monkey or Salongo Monkey","id":39,"genus":"Cercopithecus","species":"dryas"}},{"monkey":{"name":"Diana Monkey","id":"048","genus":"Cercopithecus","species":"diana"}},{"monkey":{"name":"Roloway Monkey","id":"049","genus":"Cercopithecus","species":"roloway"}},{"monkey":{"name":"Greater Spot-nosed Monkey","id":40,"genus":"Cercopithecus","species":"nictitans"}},{"monkey":{"name":"Blue Monkey","id":41,"genus":"Cercopithecus","species":"mitis"}},{"monkey":{"name":"Silver Monkey","id":42,"genus":"Cercopithecus","species":"doggetti"}},{"monkey":{"name":"Golden Monkey","id":43,"genus":"Cercopithecus","species":"kandti"}},{"monkey":{"name":"Sykes's Monkey","id":44,"genus":"Cercopithecus","species":"albogularis"}},{"monkey":{"name":"Mona Monkey","id":45,"genus":"Cercopithecus","species":"mona"}},{"monkey":{"name":"Campbell's Mona Monkey","id":46,"genus":"Cercopithecus","species":"campbelli"}},{"monkey":{"name":"Lowe's Mona Monkey","id":47,"genus":"Cercopithecus","species":"lowei"}},{"monkey":{"name":"Crested Mona Monkey","id":"058","genus":"Cercopithecus","species":"pogonias"}},{"monkey":{"name":"Wolf's Mona Monkey","id":"059","genus":"Cercopithecus","species":"wolfi"}},{"monkey":{"name":"Dent's Mona Monkey","id":48,"genus":"Cercopithecus","species":"denti"}},{"monkey":{"name":"Lesser Spot-nosed Monkey","id":49,"genus":"Cercopithecus","species":"petaurista"}},{"monkey":{"name":"White-throated Guenon","id":50,"genus":"Cercopithecus","species":"erythrogaster"}},{"monkey":{"name":"Sclater's Guenon","id":51,"genus":"Cercopithecus","species":"sclateri"}},{"monkey":{"name":"Red-eared Guenon","id":52,"genus":"Cercopithecus","species":"erythrotis"}},{"monkey":{"name":"Moustached Guenon","id":53,"genus":"Cercopithecus","species":"cephus"}},{"monkey":{"name":"Red-tailed Monkey","id":54,"genus":"Cercopithecus","species":"ascanius"}},{"monkey":{"name":"L'Hoest's Monkey","id":55,"genus":"Cercopithecus","species":"lhoesti"}},{"monkey":{"name":"Preuss's Monkey","id":"068","genus":"Cercopithecus","species":"preussi"}},{"monkey":{"name":"Sun-tailed Monkey","id":"069","genus":"Cercopithecus","species":"solatus"}},{"monkey":{"name":"Hamlyn's Monkey","id":56,"genus":"Cercopithecus","species":"hamlyni"}},{"monkey":{"name":"De Brazza's Monkey","id":57,"genus":"Cercopithecus","species":"neglectus"}},{"monkey":{"name":"Barbary Macaque","id":58,"genus":"Macaca","species":"sylvanus"}},{"monkey":{"name":"Lion-tailed Macaque","id":59,"genus":"Macaca","species":"silenus"}},{"monkey":{"name":"Southern Pig-tailed Macaque or Beruk","id":60,"genus":"Macaca","species":"nemestrina"}},{"monkey":{"name":"Northern Pig-tailed Macaque","id":61,"genus":"Macaca","species":"leonina"}},{"monkey":{"name":"Pagai Island Macaque or Bokkoi","id":62,"genus":"Macaca","species":"pagensis"}},{"monkey":{"name":"Siberut Macaque","id":63,"genus":"Macaca","species":"siberu"}},{"monkey":{"name":"Moor Macaque","id":"078","genus":"Macaca","species":"maura"}},{"monkey":{"name":"Booted Macaque","id":"079","genus":"Macaca","species":"ochreata"}},{"monkey":{"name":"Tonkean Macaque","id":"080","genus":"Macaca","species":"tonkeana"}},{"monkey":{"name":"Heck's Macaque","id":"081","genus":"Macaca","species":"hecki"}},{"monkey":{"name":"Gorontalo Macaque","id":"082","genus":"Macaca","species":"nigrescens"}},{"monkey":{"name":"Celebes Crested Macaque or Black Ape","id":"083","genus":"Macaca","species":"nigra"}},{"monkey":{"name":"Crab-eating Macaque or Long-tailed Macaque or Kera","id":"084","genus":"Macaca","species":"fascicularis"}},{"monkey":{"name":"Stump-tailed Macaque or Bear Macaque","id":"085","genus":"Macaca","species":"arctoides"}},{"monkey":{"name":"Rhesus Macaque","id":"086","genus":"Macaca","species":"mulatta"}},{"monkey":{"name":"Formosan Rock Macaque","id":"087","genus":"Macaca","species":"cyclopis"}},{"monkey":{"name":"Japanese Macaque","id":"088","genus":"Macaca","species":"fuscata"}},{"monkey":{"name":"Toque Macaque","id":"089","genus":"Macaca","species":"sinica"}},{"monkey":{"name":"Bonnet Macaque","id":"090","genus":"Macaca","species":"radiata"}},{"monkey":{"name":"Assam Macaque","id":"091","genus":"Macaca","species":"assamensis"}},{"monkey":{"name":"Tibetan Macaque or Milne-Edwards' Macaque","id":"092","genus":"Macaca","species":"thibetana"}},{"monkey":{"name":"Arunachal Macaque or Munzala","id":"093","genus":"Macaca","species":"munzala"}},{"monkey":{"name":"Grey-cheeked Mangabey","id":"094","genus":"Lophocebus","species":"albigena"}},{"monkey":{"name":"Black Crested Mangabey","id":"095","genus":"Lophocebus","species":"aterrimus"}},{"monkey":{"name":"Opdenbosch's Mangabey","id":"096","genus":"Lophocebus","species":"opdenboschi"}},{"monkey":{"name":"Uganda Mangabey","id":"097","genus":"Lophocebus","species":"ugandae"}},{"monkey":{"name":"Johnston's Mangabey","id":"098","genus":"Lophocebus","species":"johnstoni"}},{"monkey":{"name":"Osman Hill's Mangabey","id":"099","genus":"Lophocebus","species":"osmani"}},{"monkey":{"name":"Kipunji","id":100,"genus":"Rungwecebus","species":"kipunji"}},{"monkey":{"name":"Hamadryas Baboon","id":101,"genus":"Papio","species":"hamadryas"}},{"monkey":{"name":"Guinea Baboon","id":102,"genus":"Papio","species":"papio"}},{"monkey":{"name":"Olive Baboon","id":103,"genus":"Papio","species":"anubis"}},{"monkey":{"name":"Yellow Baboon","id":104,"genus":"Papio","species":"cynocephalus"}},{"monkey":{"name":"Chacma Baboon","id":105,"genus":"Papio","species":"ursinus"}},{"monkey":{"name":"Gelada","id":106,"genus":"Theropithecus","species":"gelada"}},{"monkey":{"name":"Sooty Mangabey","id":107,"genus":"Cercocebus","species":"atys"}},{"monkey":{"name":"Collared Mangabey","id":108,"genus":"Cercocebus","species":"torquatus"}},{"monkey":{"name":"Agile Mangabey","id":109,"genus":"Cercocebus","species":"agilis"}},{"monkey":{"name":"Golden-bellied Mangabey","id":110,"genus":"Cercocebus","species":"chrysogaster"}},{"monkey":{"name":"Tana River Mangabey","id":111,"genus":"Cercocebus","species":"galeritus"}},{"monkey":{"name":"Sanje Mangabey","id":112,"genus":"Cercocebus","species":"sanjei"}},{"monkey":{"name":"Mandrill","id":113,"genus":"Mandrillus","species":"sphinx"}},{"monkey":{"name":"Drill","id":114,"genus":"Mandrillus","species":"leucophaeus"}},{"monkey":{"name":"Black Colobus","id":115,"genus":"Colobus","species":"satanas"}},{"monkey":{"name":"Angola Colobus","id":116,"genus":"Colobus","species":"angolensis"}},{"monkey":{"name":"King Colobus","id":117,"genus":"Colobus","species":"polykomos"}},{"monkey":{"name":"Ursine Colobus","id":118,"genus":"Colobus","species":"vellerosus"}},{"monkey":{"name":"Mantled Guereza","id":119,"genus":"Colobus","species":"guereza"}},{"monkey":{"name":"Western Red Colobus","id":120,"genus":"Piliocolobus","species":"badius"}},{"monkey":{"name":"Pennant's Colobus","id":121,"genus":"Piliocolobus","species":"pennantii"}},{"monkey":{"name":"Preuss's Red Colobus","id":122,"genus":"Piliocolobus","species":"preussi"}},{"monkey":{"name":"Thollon's Red Colobus","id":123,"genus":"Piliocolobus","species":"tholloni"}},{"monkey":{"name":"Central African Red Colobus","id":124,"genus":"Piliocolobus","species":"foai"}},{"monkey":{"name":"Ugandan Red Colobus","id":125,"genus":"Piliocolobus","species":"tephrosceles"}},{"monkey":{"name":"Uzungwa Red Colobus","id":126,"genus":"Piliocolobus","species":"gordonorum"}},{"monkey":{"name":"Zanzibar Red Colobus","id":127,"genus":"Piliocolobus","species":"kirkii"}},{"monkey":{"name":"Tana River Red Colobus","id":128,"genus":"Piliocolobus","species":"rufomitratus"}},{"monkey":{"name":"Olive Colobus","id":129,"genus":"Procolobus","species":"verus"}},{"monkey":{"name":"Maroon Leaf Monkey","id":130,"genus":"Presbytis","species":"rubicunda"}}]}
|
data/spec/data/sample.tsv
CHANGED
@@ -81,7 +81,7 @@ ID Name Genus Species
|
|
81
81
|
080 Tonkean Macaque Macaca tonkeana
|
82
82
|
081 Heck's Macaque Macaca hecki
|
83
83
|
082 Gorontalo Macaque Macaca nigrescens
|
84
|
-
083 Celebes Crested Macaque or Black
|
84
|
+
083 Celebes Crested Macaque or Black Ape Macaca nigra
|
85
85
|
084 Crab-eating Macaque or Long-tailed Macaque or Kera Macaca fascicularis
|
86
86
|
085 Stump-tailed Macaque or Bear Macaque Macaca arctoides
|
87
87
|
086 Rhesus Macaque Macaca mulatta
|
data/spec/data/sample.txt
CHANGED
@@ -81,7 +81,7 @@ ID,Name,Genus,Species
|
|
81
81
|
080,Tonkean Macaque,Macaca,tonkeana
|
82
82
|
081,Heck's Macaque,Macaca,hecki
|
83
83
|
082,Gorontalo Macaque,Macaca,nigrescens
|
84
|
-
083,Celebes Crested Macaque or Black
|
84
|
+
083,Celebes Crested Macaque or Black Ape,Macaca,nigra
|
85
85
|
084,Crab-eating Macaque or Long-tailed Macaque or Kera,Macaca,fascicularis
|
86
86
|
085,Stump-tailed Macaque or Bear Macaque,Macaca,arctoides
|
87
87
|
086,Rhesus Macaque,Macaca,mulatta
|
data/spec/data/sample.xml
CHANGED
data/spec/data/sample.yaml
CHANGED
@@ -0,0 +1,32 @@
|
|
1
|
+
require File.dirname(__FILE__) + "/../../spec_helper"
|
2
|
+
require File.dirname(__FILE__) + "/../utils/shared_paths_spec"
|
3
|
+
|
4
|
+
describe IMW::Dataset do
|
5
|
+
|
6
|
+
describe 'setting default paths' do
|
7
|
+
|
8
|
+
before do
|
9
|
+
@dataset = IMW::Dataset.new(:testing, :root => IMWTest::TMP_DIR)
|
10
|
+
end
|
11
|
+
|
12
|
+
it "should set its root path to the value given" do
|
13
|
+
@dataset.path_to(:root).should == IMWTest::TMP_DIR
|
14
|
+
end
|
15
|
+
|
16
|
+
it "should set paths for each workflow dir" do
|
17
|
+
@dataset.workflow_dirs.each do |dir|
|
18
|
+
@dataset.path_to(dir).should == File.join(IMWTest::TMP_DIR, dir.to_s)
|
19
|
+
end
|
20
|
+
end
|
21
|
+
|
22
|
+
before do
|
23
|
+
@path_manager = @dataset
|
24
|
+
end
|
25
|
+
it_should_behave_like "an object that manages paths"
|
26
|
+
|
27
|
+
end
|
28
|
+
end
|
29
|
+
|
30
|
+
|
31
|
+
|
32
|
+
|
@@ -0,0 +1,41 @@
|
|
1
|
+
require File.dirname(__FILE__) + "/../../spec_helper"
|
2
|
+
require 'imw/dataset/workflow'
|
3
|
+
describe IMW::Workflow do
|
4
|
+
|
5
|
+
before do
|
6
|
+
@dataset = IMW::Dataset.new :testing
|
7
|
+
end
|
8
|
+
|
9
|
+
it "should dynamically define methods for each workflow step" do
|
10
|
+
@dataset.workflow_steps.each do |step|
|
11
|
+
@dataset.respond_to?(step).should be_true
|
12
|
+
end
|
13
|
+
end
|
14
|
+
|
15
|
+
describe "initializing workflow" do
|
16
|
+
it "should not make any directories if no tasks are invoked" do
|
17
|
+
@dataset.path_to(:root).should_not contain(*@dataset.workflow_dirs.map(&:to_s))
|
18
|
+
end
|
19
|
+
|
20
|
+
it "should only make directories once a task is invoked" do
|
21
|
+
@dataset[:initialize].invoke
|
22
|
+
@dataset.path_to(:root).should contain(*@dataset.workflow_dirs.map(&:to_s))
|
23
|
+
end
|
24
|
+
end
|
25
|
+
|
26
|
+
describe "cleaning workflow directories" do
|
27
|
+
it "should clean without error even if there's nothing to clean" do
|
28
|
+
@dataset[:clean].invoke
|
29
|
+
@dataset.path_to(:root).should_not contain(*@dataset.workflow_dirs.map(&:to_s))
|
30
|
+
end
|
31
|
+
|
32
|
+
it "should remove workflow directories when invoked" do
|
33
|
+
@dataset[:initialize].invoke
|
34
|
+
IMWTest::Random.file(@dataset.path_to(:ripd, 'foobar.txt')) # put a file in
|
35
|
+
@dataset[:clean].invoke
|
36
|
+
@dataset.path_to(:root).should_not contain(*@dataset.workflow_dirs.map(&:to_s))
|
37
|
+
end
|
38
|
+
end
|
39
|
+
|
40
|
+
end
|
41
|
+
|