imw 0.2.18 → 0.3.0
Sign up to get free protection for your applications and to get access to all the features.
- data/Gemfile +7 -26
- data/Gemfile.lock +13 -38
- data/{LICENSE → LICENSE.txt} +1 -1
- data/README.textile +35 -0
- data/Rakefile +45 -22
- data/VERSION +1 -1
- data/examples/foo.rb +19 -0
- data/examples/html_selector.rb +22 -0
- data/examples/nes_game_list.csv +625 -0
- data/examples/nes_gamespot.csv +1371 -0
- data/examples/nes_nintendo.csv +624 -0
- data/examples/nes_unlicensed.csv +89 -0
- data/examples/nes_wikipedia.csv +710 -0
- data/examples/nibbler_test.rb +24 -0
- data/examples/script.rb +19 -0
- data/lib/imw.rb +28 -140
- data/lib/imw/error.rb +9 -0
- data/lib/imw/recordizer.rb +8 -0
- data/lib/imw/recordizer/html_selector_recordizer.rb +86 -0
- data/lib/imw/recordizer/string_slice_recordizer.rb +39 -0
- data/lib/imw/resource.rb +3 -119
- data/lib/imw/serializer.rb +7 -0
- data/lib/imw/serializer/json_serializer.rb +17 -0
- data/lib/imw/uri.rb +41 -0
- data/spec/resource_spec.rb +78 -0
- data/spec/uri_spec.rb +55 -0
- metadata +81 -232
- data/README.rdoc +0 -371
- data/bin/imw +0 -5
- data/bin/tsv_to_json.rb +0 -29
- data/etc/imwrc.rb +0 -26
- data/examples/dataset.rb +0 -12
- data/examples/metadata.yml +0 -10
- data/lib/imw/archives.rb +0 -120
- data/lib/imw/archives/rar.rb +0 -19
- data/lib/imw/archives/tar.rb +0 -19
- data/lib/imw/archives/tarbz2.rb +0 -73
- data/lib/imw/archives/targz.rb +0 -73
- data/lib/imw/archives/zip.rb +0 -51
- data/lib/imw/boot.rb +0 -87
- data/lib/imw/compressed_files.rb +0 -94
- data/lib/imw/compressed_files/bz2.rb +0 -16
- data/lib/imw/compressed_files/compressible.rb +0 -75
- data/lib/imw/compressed_files/gz.rb +0 -16
- data/lib/imw/dataset.rb +0 -125
- data/lib/imw/dataset/paths.rb +0 -29
- data/lib/imw/dataset/workflow.rb +0 -195
- data/lib/imw/formats.rb +0 -33
- data/lib/imw/formats/delimited.rb +0 -170
- data/lib/imw/formats/excel.rb +0 -100
- data/lib/imw/formats/json.rb +0 -41
- data/lib/imw/formats/pdf.rb +0 -71
- data/lib/imw/formats/sgml.rb +0 -69
- data/lib/imw/formats/yaml.rb +0 -41
- data/lib/imw/metadata.rb +0 -83
- data/lib/imw/metadata/contains_metadata.rb +0 -54
- data/lib/imw/metadata/dsl.rb +0 -111
- data/lib/imw/metadata/field.rb +0 -37
- data/lib/imw/metadata/has_metadata.rb +0 -98
- data/lib/imw/metadata/has_summary.rb +0 -57
- data/lib/imw/metadata/schema.rb +0 -17
- data/lib/imw/parsers.rb +0 -8
- data/lib/imw/parsers/flat.rb +0 -44
- data/lib/imw/parsers/html_parser.rb +0 -387
- data/lib/imw/parsers/html_parser/matchers.rb +0 -289
- data/lib/imw/parsers/line_parser.rb +0 -87
- data/lib/imw/parsers/regexp_parser.rb +0 -72
- data/lib/imw/repository.rb +0 -12
- data/lib/imw/runner.rb +0 -118
- data/lib/imw/schemes.rb +0 -23
- data/lib/imw/schemes/ftp.rb +0 -142
- data/lib/imw/schemes/hdfs.rb +0 -251
- data/lib/imw/schemes/http.rb +0 -165
- data/lib/imw/schemes/local.rb +0 -409
- data/lib/imw/schemes/remote.rb +0 -119
- data/lib/imw/schemes/s3.rb +0 -143
- data/lib/imw/schemes/sql.rb +0 -129
- data/lib/imw/tools.rb +0 -12
- data/lib/imw/tools/aggregator.rb +0 -148
- data/lib/imw/tools/archiver.rb +0 -220
- data/lib/imw/tools/downloader.rb +0 -63
- data/lib/imw/tools/extension_analyzer.rb +0 -114
- data/lib/imw/tools/summarizer.rb +0 -83
- data/lib/imw/tools/transferer.rb +0 -167
- data/lib/imw/utils.rb +0 -74
- data/lib/imw/utils/dynamically_extendable.rb +0 -137
- data/lib/imw/utils/error.rb +0 -59
- data/lib/imw/utils/extensions/hpricot.rb +0 -34
- data/lib/imw/utils/has_uri.rb +0 -131
- data/lib/imw/utils/log.rb +0 -92
- data/lib/imw/utils/misc.rb +0 -57
- data/lib/imw/utils/paths.rb +0 -146
- data/lib/imw/utils/uri.rb +0 -59
- data/lib/imw/utils/uuid.rb +0 -33
- data/lib/imw/utils/validate.rb +0 -38
- data/lib/imw/utils/version.rb +0 -11
- data/spec/data/formats/delimited/sample.csv +0 -131
- data/spec/data/formats/delimited/sample.tsv +0 -131
- data/spec/data/formats/delimited/with_schema/ace-hardware-locations.tsv +0 -11
- data/spec/data/formats/delimited/with_schema/all-countries-ip-address-to-geolocation-data.tsv +0 -16
- data/spec/data/formats/delimited/with_schema/complete-list-of-starbucks-locations.tsv +0 -11
- data/spec/data/formats/delimited/with_schema/myspace-user-activity-stream-cumulative-word-count-from-from-dec.tsv +0 -22
- data/spec/data/formats/delimited/with_schema/myspace-user-activity-stream-myspace-application-adds-by-zip-cod.tsv +0 -22
- data/spec/data/formats/delimited/with_schema/myspace-user-activity-stream-myspace-application-counts.tsv +0 -12
- data/spec/data/formats/delimited/with_schema/myspace-user-activity-stream-user-count-by-latlong.tsv +0 -13
- data/spec/data/formats/delimited/with_schema/myspace-user-activity-stream-user-count-by-zip-code.tsv +0 -22
- data/spec/data/formats/delimited/with_schema/myspace-user-activity-stream-word-count-by-day-from-december-200.tsv +0 -22
- data/spec/data/formats/delimited/without_schema/ace-hardware-locations.tsv +0 -10
- data/spec/data/formats/delimited/without_schema/all-countries-ip-address-to-geolocation-data.tsv +0 -15
- data/spec/data/formats/delimited/without_schema/complete-list-of-starbucks-locations.tsv +0 -10
- data/spec/data/formats/delimited/without_schema/myspace-user-activity-stream-cumulative-word-count-from-from-dec.tsv +0 -21
- data/spec/data/formats/delimited/without_schema/myspace-user-activity-stream-myspace-application-adds-by-zip-cod.tsv +0 -21
- data/spec/data/formats/delimited/without_schema/myspace-user-activity-stream-myspace-application-counts.tsv +0 -11
- data/spec/data/formats/delimited/without_schema/myspace-user-activity-stream-user-count-by-latlong.tsv +0 -12
- data/spec/data/formats/delimited/without_schema/myspace-user-activity-stream-user-count-by-zip-code.tsv +0 -21
- data/spec/data/formats/delimited/without_schema/myspace-user-activity-stream-word-count-by-day-from-december-200.tsv +0 -21
- data/spec/data/formats/excel/sample.xls +0 -0
- data/spec/data/formats/json/sample.json +0 -1
- data/spec/data/formats/none/sample +0 -650
- data/spec/data/formats/sgml/sample.xml +0 -617
- data/spec/data/formats/text/sample.txt +0 -650
- data/spec/data/formats/yaml/sample.yaml +0 -410
- data/spec/data/schema-tabular.yaml +0 -11
- data/spec/imw/archives/rar_spec.rb +0 -16
- data/spec/imw/archives/tar_spec.rb +0 -16
- data/spec/imw/archives/tarbz2_spec.rb +0 -24
- data/spec/imw/archives/targz_spec.rb +0 -21
- data/spec/imw/archives/zip_spec.rb +0 -16
- data/spec/imw/archives_spec.rb +0 -77
- data/spec/imw/compressed_files/bz2_spec.rb +0 -15
- data/spec/imw/compressed_files/compressible_spec.rb +0 -36
- data/spec/imw/compressed_files/gz_spec.rb +0 -15
- data/spec/imw/compressed_files_spec.rb +0 -47
- data/spec/imw/dataset/paths_spec.rb +0 -32
- data/spec/imw/dataset/workflow_spec.rb +0 -41
- data/spec/imw/formats/delimited_spec.rb +0 -44
- data/spec/imw/formats/excel_spec.rb +0 -55
- data/spec/imw/formats/json_spec.rb +0 -18
- data/spec/imw/formats/sgml_spec.rb +0 -24
- data/spec/imw/formats/yaml_spec.rb +0 -19
- data/spec/imw/metadata/contains_metadata_spec.rb +0 -56
- data/spec/imw/metadata/field_spec.rb +0 -25
- data/spec/imw/metadata/has_metadata_spec.rb +0 -58
- data/spec/imw/metadata/has_summary_spec.rb +0 -32
- data/spec/imw/metadata/schema_spec.rb +0 -24
- data/spec/imw/metadata_spec.rb +0 -86
- data/spec/imw/parsers/line_parser_spec.rb +0 -96
- data/spec/imw/parsers/regexp_parser_spec.rb +0 -42
- data/spec/imw/resource_spec.rb +0 -32
- data/spec/imw/schemes/hdfs_spec.rb +0 -67
- data/spec/imw/schemes/http_spec.rb +0 -19
- data/spec/imw/schemes/local_spec.rb +0 -165
- data/spec/imw/schemes/remote_spec.rb +0 -38
- data/spec/imw/schemes/s3_spec.rb +0 -31
- data/spec/imw/schemes/sql_spec.rb +0 -3
- data/spec/imw/tools/aggregator_spec.rb +0 -71
- data/spec/imw/tools/archiver_spec.rb +0 -120
- data/spec/imw/tools/extension_analyzer_spec.rb +0 -153
- data/spec/imw/tools/summarizer_spec.rb +0 -8
- data/spec/imw/tools/transferer_spec.rb +0 -195
- data/spec/imw/utils/dynamically_extendable_spec.rb +0 -69
- data/spec/imw/utils/has_uri_spec.rb +0 -61
- data/spec/imw/utils/paths_spec.rb +0 -10
- data/spec/imw/utils/shared_paths_spec.rb +0 -29
- data/spec/imw_spec.rb +0 -14
- data/spec/rcov.opts +0 -1
- data/spec/spec_helper.rb +0 -31
- data/spec/support/custom_matchers.rb +0 -28
- data/spec/support/file_contents_matcher.rb +0 -30
- data/spec/support/paths_matcher.rb +0 -66
- data/spec/support/random.rb +0 -213
- data/spec/support/without_regard_to_order_matcher.rb +0 -41
data/lib/imw/schemes/remote.rb
DELETED
@@ -1,119 +0,0 @@
|
|
1
|
-
module IMW
|
2
|
-
module Schemes
|
3
|
-
|
4
|
-
# Contains modules which define methods appropriate for remote
|
5
|
-
# resources, no matter the protocol.
|
6
|
-
module Remote
|
7
|
-
|
8
|
-
# Defines methods appropriate for accessing a remote resource,
|
9
|
-
# no matter the protocol.
|
10
|
-
module Base
|
11
|
-
|
12
|
-
#
|
13
|
-
# TODO -- self.extended should extend by RemoteDirectory when appropriate
|
14
|
-
#
|
15
|
-
|
16
|
-
def self.extended obj
|
17
|
-
obj.extend(RemoteFile)
|
18
|
-
end
|
19
|
-
|
20
|
-
# Is this resource on a remote host?
|
21
|
-
#
|
22
|
-
# @return [true,false]
|
23
|
-
def is_remote?
|
24
|
-
true
|
25
|
-
end
|
26
|
-
|
27
|
-
# The host of this resource.
|
28
|
-
#
|
29
|
-
# @return [String]
|
30
|
-
def host
|
31
|
-
@host ||= uri.host
|
32
|
-
end
|
33
|
-
|
34
|
-
# Return the query string part of this resource's URI. Will
|
35
|
-
# likely be +nil+ for local resources.
|
36
|
-
#
|
37
|
-
# @return [String]
|
38
|
-
def query_string
|
39
|
-
@query_string ||= uri.query
|
40
|
-
end
|
41
|
-
|
42
|
-
# Return the path part of this resource's URI. Will _not_
|
43
|
-
# include the +query_string+ or +fragment+.
|
44
|
-
#
|
45
|
-
# @return [String]
|
46
|
-
def path
|
47
|
-
@path ||= uri.path
|
48
|
-
end
|
49
|
-
|
50
|
-
end
|
51
|
-
|
52
|
-
module RemoteFile
|
53
|
-
|
54
|
-
# Return the IO object for this remote file.
|
55
|
-
#
|
56
|
-
# The mode of this resource is ignored.
|
57
|
-
#
|
58
|
-
# @return [StringIO]
|
59
|
-
def io
|
60
|
-
require 'open-uri'
|
61
|
-
@io ||= open(uri.to_s) # ignore mode
|
62
|
-
end
|
63
|
-
|
64
|
-
# Read the contents of this remote file.
|
65
|
-
#
|
66
|
-
# @return [String]
|
67
|
-
def read
|
68
|
-
io.read
|
69
|
-
end
|
70
|
-
|
71
|
-
# Return the lines of this remote file.
|
72
|
-
#
|
73
|
-
# If passed a block then yield each line to the block.
|
74
|
-
#
|
75
|
-
# @return [Array] the lines of this remote file
|
76
|
-
# @yield [String] each line of this remote file
|
77
|
-
def load &block
|
78
|
-
if block_given?
|
79
|
-
io.each do |line|
|
80
|
-
yield line
|
81
|
-
end
|
82
|
-
else
|
83
|
-
read.split("\n")
|
84
|
-
end
|
85
|
-
end
|
86
|
-
|
87
|
-
# Map over the lines in this remote file.
|
88
|
-
#
|
89
|
-
# @yield [String] each line of the file
|
90
|
-
def map &block
|
91
|
-
io.map(&block)
|
92
|
-
end
|
93
|
-
end
|
94
|
-
|
95
|
-
|
96
|
-
module RemoteDirectory
|
97
|
-
|
98
|
-
# Return the resource at the base path of this resource joined
|
99
|
-
# to +path+.
|
100
|
-
#
|
101
|
-
# IMW.open('http://example.com/path/to/dir').join('subdir')
|
102
|
-
# #=> IMW::Resource at 'http://example.com/path/to/dir/subdir'
|
103
|
-
#
|
104
|
-
# @param [Array<String>] paths
|
105
|
-
# @return [IMW::Resource]
|
106
|
-
def join *paths
|
107
|
-
IMW.open(File.join(stripped_uri.to_s, *paths))
|
108
|
-
end
|
109
|
-
|
110
|
-
#
|
111
|
-
# TODO -- bloody everything. what's the best way to tell if
|
112
|
-
# the remote URL is a directory?
|
113
|
-
#
|
114
|
-
|
115
|
-
|
116
|
-
end
|
117
|
-
end
|
118
|
-
end
|
119
|
-
end
|
data/lib/imw/schemes/s3.rb
DELETED
@@ -1,143 +0,0 @@
|
|
1
|
-
module IMW
|
2
|
-
module Schemes
|
3
|
-
|
4
|
-
# Defines methods for reading and writing data to {Amazon
|
5
|
-
# S3}[http://aws.amazon.com/s3] buckets.
|
6
|
-
#
|
7
|
-
# IMW.open('s3://my_bucket/path/to/some/file.csv')
|
8
|
-
#
|
9
|
-
# Learn more about {Amazon Web Services}[http://aws.amazon.com].
|
10
|
-
module S3
|
11
|
-
|
12
|
-
# For an S3 resource, the bucket is just the hostname.
|
13
|
-
#
|
14
|
-
# @return [String]
|
15
|
-
def bucket
|
16
|
-
host
|
17
|
-
end
|
18
|
-
|
19
|
-
# Is this resource an S3 resource?
|
20
|
-
#
|
21
|
-
# @return [true, false]
|
22
|
-
def on_s3?
|
23
|
-
true
|
24
|
-
end
|
25
|
-
alias_method :is_s3?, :on_s3?
|
26
|
-
|
27
|
-
# Copy this resource to the +new_uri+.
|
28
|
-
#
|
29
|
-
# @param [String, IMW::Resource] new_uri
|
30
|
-
# @return [IMW::Resource] the new resource
|
31
|
-
def cp new_uri
|
32
|
-
#IMW::Tools::Transferer.new(:cp, self, new_uri).transfer!
|
33
|
-
IMW::Schemes::S3.get(self, new_uri)
|
34
|
-
end
|
35
|
-
|
36
|
-
# Does this resource exist on S3?
|
37
|
-
#
|
38
|
-
# @return [true, false]
|
39
|
-
def exist?
|
40
|
-
AWS::S3::S3Object.exists?(raw_path, bucket)
|
41
|
-
end
|
42
|
-
alias_method :exists?, :exist?
|
43
|
-
|
44
|
-
# Remove this resource from S3.
|
45
|
-
#
|
46
|
-
# @return [IMW::Resource] the deleted object
|
47
|
-
def rm
|
48
|
-
AWS::S3::S3Object.delete(raw_path, bucket)
|
49
|
-
end
|
50
|
-
alias_method :rm!, :rm
|
51
|
-
|
52
|
-
# Return the S3N URL for this S3 object
|
53
|
-
#
|
54
|
-
# resource = IMW.open('s3://my_bucket/path/to/some/obj')
|
55
|
-
# resource.s3n_url
|
56
|
-
# => 's3n://my_bucket/path/to/some/obj'
|
57
|
-
#
|
58
|
-
# @return [String]
|
59
|
-
def s3n_url
|
60
|
-
uri.to_s.gsub(/^s3:/, 's3n:')
|
61
|
-
end
|
62
|
-
|
63
|
-
# Return the contents of this S3 object.
|
64
|
-
#
|
65
|
-
# @return [String]
|
66
|
-
def read
|
67
|
-
AWS::S3::S3Object.value(raw_path, bucket)
|
68
|
-
end
|
69
|
-
|
70
|
-
# Store +source+ into +destination+.
|
71
|
-
#
|
72
|
-
# @param [String, IMW::Resource, #io] source
|
73
|
-
# @param [String, IMW::Resource, #path, #bucket] destination
|
74
|
-
# @return [IMW::Resource] the new S3 object
|
75
|
-
def self.put source, destination
|
76
|
-
source = IMW.open(source)
|
77
|
-
destintation = IMW.open(destination)
|
78
|
-
raise IMW::ArgumentError.new("destination must be on S3 -- #{destination} given") unless destination.on_s3?
|
79
|
-
make_connection!
|
80
|
-
AWS::S3::S3Object.store(destination.raw_path, source.io, destination.bucket)
|
81
|
-
destination
|
82
|
-
end
|
83
|
-
|
84
|
-
# Download +source+ from S3 into +destination+.
|
85
|
-
#
|
86
|
-
# @param [String, IMW::Resource, #path, #bucket] source
|
87
|
-
# @param [String, IMW::Resource, #write] destination
|
88
|
-
# @return [IMW::Resource] the new resource
|
89
|
-
def self.get source, destination
|
90
|
-
source = IMW.open(source)
|
91
|
-
destination = IMW.open!(destination)
|
92
|
-
raise IMW::ArgumentError.new("source must be on S3 -- #{source} given") unless source.on_s3?
|
93
|
-
make_connection!
|
94
|
-
AWS::S3::S3Object.stream(source.raw_path, source.bucket) do |chunk|
|
95
|
-
destination.write(chunk)
|
96
|
-
end
|
97
|
-
destination.close
|
98
|
-
destination.reopen
|
99
|
-
end
|
100
|
-
|
101
|
-
# Copy S3 resource +source+ to +destination+.
|
102
|
-
#
|
103
|
-
# @param [String, IMW::Resource, #path, #bucket] source
|
104
|
-
# @param [String, IMW::Resource, #path, #bucket] destination
|
105
|
-
# @return [IMW::Resource] the new resource
|
106
|
-
def self.copy source, destination
|
107
|
-
source = IMW.open(source)
|
108
|
-
destination = IMW.open(destination)
|
109
|
-
raise IMW::PathError.new("Bucket names must be non-blank and match to 'copy'") unless source.bucket.present? && destination.bucket.present? && source.bucket == destination.bucket
|
110
|
-
make_connection!
|
111
|
-
AWS::S3::Object.copy(source.raw_path, destination.raw_path, destination.bucket)
|
112
|
-
destination
|
113
|
-
end
|
114
|
-
|
115
|
-
# Return the resource at the base path of this resource joined
|
116
|
-
# to +path+.
|
117
|
-
#
|
118
|
-
# IMW.open('s3:://bucket/path/to/dir').join('subdir')
|
119
|
-
# #=> IMW::Resource at 's3://bucket/path/to/dir/subdir'
|
120
|
-
#
|
121
|
-
# @param [Array<String>] paths
|
122
|
-
# @return [IMW::Resource]
|
123
|
-
def join *paths
|
124
|
-
IMW.open(File.join(stripped_uri.to_s, *paths))
|
125
|
-
end
|
126
|
-
|
127
|
-
protected
|
128
|
-
# Make an S3 connection.
|
129
|
-
#
|
130
|
-
# Uses settings defined in IMW::AWS_CREDENTIALS.
|
131
|
-
#
|
132
|
-
# @return [AWS
|
133
|
-
def self.make_connection!
|
134
|
-
return @connection if @connection
|
135
|
-
raise IMW::Error.new("Must define a constant IMW::AWS_CREDENTIALS with an :access_key_id and a :secret_access_key before using S3 resources") unless defined?(IMW::AWS_CREDENTIALS)
|
136
|
-
require 'aws/s3'
|
137
|
-
@connection = AWS::S3::Base.establish_connection!(IMW::AWS_CREDENTIALS)
|
138
|
-
end
|
139
|
-
|
140
|
-
end
|
141
|
-
end
|
142
|
-
end
|
143
|
-
|
data/lib/imw/schemes/sql.rb
DELETED
@@ -1,129 +0,0 @@
|
|
1
|
-
require 'dbi'
|
2
|
-
|
3
|
-
module IMW
|
4
|
-
module Schemes
|
5
|
-
|
6
|
-
# Encapsulates a connection to a relational database.
|
7
|
-
#
|
8
|
-
# Calling
|
9
|
-
#
|
10
|
-
# IMW.open('sql://host:port/database_name')
|
11
|
-
#
|
12
|
-
# shold create a connection to a database at the given +port+ on
|
13
|
-
# the given +host+ using the given +database_name+.
|
14
|
-
module SQL
|
15
|
-
|
16
|
-
# A base implementation of a connection to a relational
|
17
|
-
# database.
|
18
|
-
#
|
19
|
-
# The Base#extended method will examine the +scheme+ of an
|
20
|
-
# object extended with this module and choose a more specific
|
21
|
-
# database adaptor module to extend with as well.
|
22
|
-
module Base
|
23
|
-
|
24
|
-
# When an IMW::Resource is extended use URI's scheme to choose
|
25
|
-
# which other module inside IMW::Schemes::SQL to extend with.
|
26
|
-
def self.extended obj
|
27
|
-
case obj.scheme
|
28
|
-
when 'mysql' then obj.extend(IMW::Schemes::SQL::MySQL)
|
29
|
-
when 'postgresql' then obj.extend(IMW::Schemes::SQL::PostgreSQL)
|
30
|
-
else raise IMW::ArgumentError.new("Unknown database type: #{obj.scheme}")
|
31
|
-
end
|
32
|
-
end
|
33
|
-
|
34
|
-
# For an SQL connection the database will be the same as the
|
35
|
-
# path.
|
36
|
-
#
|
37
|
-
# @return [String]
|
38
|
-
def database
|
39
|
-
@database ||= path.tr('/','')
|
40
|
-
end
|
41
|
-
|
42
|
-
# Redefineeach method inappropriate for databases.
|
43
|
-
[:dirname, :basename, :extname, :extension, :name].each do |method|
|
44
|
-
define_method(method) do
|
45
|
-
nil
|
46
|
-
end
|
47
|
-
end
|
48
|
-
|
49
|
-
# Return a summary of this database.
|
50
|
-
#
|
51
|
-
# Purposefully does not call +super+.
|
52
|
-
#
|
53
|
-
# @return [Hash]
|
54
|
-
def external_summary
|
55
|
-
{
|
56
|
-
:uri => uri.to_s,
|
57
|
-
:database => database
|
58
|
-
}
|
59
|
-
end
|
60
|
-
|
61
|
-
# The (cached) database connection for this resource.
|
62
|
-
#
|
63
|
-
# @return [DBI::DatabaseHandle]
|
64
|
-
def connection
|
65
|
-
@connection ||= DBI.connect("#{dbi_module}:#{database}:#{host}", user, password)
|
66
|
-
end
|
67
|
-
|
68
|
-
# Return the password associated with user's account on the
|
69
|
-
# given database.
|
70
|
-
#
|
71
|
-
# @return [String]
|
72
|
-
def password
|
73
|
-
@password ||= resource_options[:password]
|
74
|
-
end
|
75
|
-
|
76
|
-
# Return an array of the table names in the current database.
|
77
|
-
#
|
78
|
-
# @return [Array<String>]
|
79
|
-
def tables
|
80
|
-
[].tap do |table_names|
|
81
|
-
execute("SHOW TABLES") do |row|
|
82
|
-
table_names << row.first
|
83
|
-
end
|
84
|
-
end
|
85
|
-
end
|
86
|
-
|
87
|
-
# Execute the (joined) +query_string_parts+ using this
|
88
|
-
# resource's cached connection.
|
89
|
-
#
|
90
|
-
# If passed a block, yield each row of the result set to the
|
91
|
-
# block.
|
92
|
-
#
|
93
|
-
# @param [Array<String>] query_string_parts
|
94
|
-
# @yield [DBI::Row]
|
95
|
-
# @return [DBI::StatementHandle]
|
96
|
-
def execute *query_string_parts, &block
|
97
|
-
query = query_string_parts.join(' ')
|
98
|
-
IMW.announce_if_verbose "Querying #{self}: #{query}"
|
99
|
-
statement = connection.execute(query)
|
100
|
-
block_given? ? statement.fetch(&block) : statement
|
101
|
-
end
|
102
|
-
end
|
103
|
-
|
104
|
-
# Module for MySQL databases.
|
105
|
-
module MySQL
|
106
|
-
|
107
|
-
# Return the name of the DBI module used to connect to MySQL.
|
108
|
-
#
|
109
|
-
# @return [String]
|
110
|
-
def dbi_module
|
111
|
-
"DBI:Mysql"
|
112
|
-
end
|
113
|
-
end
|
114
|
-
|
115
|
-
# Module for PostgreSQL databases.
|
116
|
-
module PostgreSQL
|
117
|
-
|
118
|
-
# Return the name of the DBI module used to connect to PostgreSQL.
|
119
|
-
#
|
120
|
-
# @return [String]
|
121
|
-
def dbi_module
|
122
|
-
"DBI:Pg"
|
123
|
-
end
|
124
|
-
end
|
125
|
-
|
126
|
-
end
|
127
|
-
end
|
128
|
-
end
|
129
|
-
|
data/lib/imw/tools.rb
DELETED
@@ -1,12 +0,0 @@
|
|
1
|
-
module IMW
|
2
|
-
module Tools
|
3
|
-
autoload :Aggregator, 'imw/tools/aggregator'
|
4
|
-
autoload :Archiver, 'imw/tools/archiver'
|
5
|
-
autoload :Transferer, 'imw/tools/transferer'
|
6
|
-
autoload :Summarizer, 'imw/tools/summarizer'
|
7
|
-
autoload :ExtensionAnalyzer, 'imw/tools/extension_analyzer'
|
8
|
-
autoload :Downloader, 'imw/tools/downloader'
|
9
|
-
end
|
10
|
-
end
|
11
|
-
|
12
|
-
|
data/lib/imw/tools/aggregator.rb
DELETED
@@ -1,148 +0,0 @@
|
|
1
|
-
require 'imw/resource'
|
2
|
-
|
3
|
-
module IMW
|
4
|
-
module Tools
|
5
|
-
|
6
|
-
# Aggregates resources into a single local directory.
|
7
|
-
#
|
8
|
-
# The directory should already exist.
|
9
|
-
#
|
10
|
-
# Any local resources will be copied into the directory.
|
11
|
-
#
|
12
|
-
# Any remote resources will be downloaded into the directory.
|
13
|
-
#
|
14
|
-
# If any of the resources are archives, they will first be
|
15
|
-
# extracted, with only their contents winding up in the final
|
16
|
-
# directory (the file hierarchy of the archive will be preserved).
|
17
|
-
#
|
18
|
-
# If any of the resources are compressed, they will first be
|
19
|
-
# uncompressed before being added to the directory.
|
20
|
-
#
|
21
|
-
# As an example:
|
22
|
-
#
|
23
|
-
# aggregator = IMW::Tools::Aggregator.new '/path/to/agg_dir'
|
24
|
-
# aggregator.aggregate '/path/to/my/regular_file.tsv', '/path/to/an/archive.tar.bz2', '/path/to/my_compressed_file.gz', 'http://mywebsite.com/index.html'
|
25
|
-
#
|
26
|
-
# This will create a directory at <tt>/path/to/agg_dir</tt> which
|
27
|
-
# looks like
|
28
|
-
#
|
29
|
-
# path_to_agg_dir
|
30
|
-
# |-- regular_file.tsv
|
31
|
-
# |-- archive
|
32
|
-
# | |-- internal_archive_file_1
|
33
|
-
# | |-- internal_archive_file_2
|
34
|
-
# | ...
|
35
|
-
# | `-- internal_archive_file_N
|
36
|
-
# |-- my_compressed_file
|
37
|
-
# `-- index.html
|
38
|
-
#
|
39
|
-
# Notice that
|
40
|
-
#
|
41
|
-
# - the local file was copied over
|
42
|
-
#
|
43
|
-
# - the remote file was downloaded and copied over
|
44
|
-
#
|
45
|
-
# - the tar archive was first exctracted
|
46
|
-
#
|
47
|
-
# - the compressed file was aggregated
|
48
|
-
#
|
49
|
-
# This process can take a while when the constituent files are
|
50
|
-
# large.
|
51
|
-
class Aggregator
|
52
|
-
|
53
|
-
attr_reader :dir
|
54
|
-
|
55
|
-
def initialize dir
|
56
|
-
self.dir = IMW.open(dir)
|
57
|
-
end
|
58
|
-
|
59
|
-
# Set the directory for this Aggregator.
|
60
|
-
#
|
61
|
-
# Will raise unless +new_dir+ is an existing, local directory.
|
62
|
-
#
|
63
|
-
# @param [String, IMW::Resource] new_dir
|
64
|
-
# @return [IMW::Resource]
|
65
|
-
def dir= new_dir
|
66
|
-
@dir = IMW.open(new_dir)
|
67
|
-
raise IMW::SchemError.new("Aggregator requires a local directory, not #{@dir}") unless @dir.is_local?
|
68
|
-
@dir.should_exist! "Aggregator requires the aggregation directory to already exist"
|
69
|
-
raise IMW::PathError.new("Aggregator requires a directory, not #{@dir}") unless @dir.is_directory?
|
70
|
-
@dir
|
71
|
-
end
|
72
|
-
|
73
|
-
# Return a list of error messages for this Aggregator.
|
74
|
-
#
|
75
|
-
# @return [Array] the error messages
|
76
|
-
def errors
|
77
|
-
@errors ||= []
|
78
|
-
end
|
79
|
-
|
80
|
-
# Was this archiver successful (did it not have any errors)?
|
81
|
-
#
|
82
|
-
# @return [true, false]
|
83
|
-
def success?
|
84
|
-
errors.empty?
|
85
|
-
end
|
86
|
-
|
87
|
-
# Aggregate the given inputs into this Aggregator's +dir+.
|
88
|
-
#
|
89
|
-
# @param [Array<IMW::Resource,String>] inputs
|
90
|
-
# @return [IMW::Tools::Aggregator]
|
91
|
-
def aggregate *paths_or_inputs
|
92
|
-
@errors = []
|
93
|
-
paths_or_inputs.flatten.compact.each do |path_or_input|
|
94
|
-
input = IMW.open(path_or_input)
|
95
|
-
if input.is_local?
|
96
|
-
aggregate_local_input(input)
|
97
|
-
else
|
98
|
-
download = download_remote_input(input)
|
99
|
-
if download.is_compressed? || download.is_archive?
|
100
|
-
aggregate_local_input(download)
|
101
|
-
download.rm!
|
102
|
-
end
|
103
|
-
end
|
104
|
-
end
|
105
|
-
end
|
106
|
-
|
107
|
-
protected
|
108
|
-
|
109
|
-
# Aggregate a local input.
|
110
|
-
#
|
111
|
-
# Will extract archives, decompress compressed files, and copy
|
112
|
-
# regular files and directories (but will not recurse into
|
113
|
-
# directories to find archives or compressed files).
|
114
|
-
#
|
115
|
-
# @param [IMW::Resource] input
|
116
|
-
def aggregate_local_input input
|
117
|
-
new_path = File.join(dir.path, input.basename)
|
118
|
-
case
|
119
|
-
when input.is_archive?
|
120
|
-
IMW.announce_if_verbose("Aggregating and extracting #{input} to #{dir}...")
|
121
|
-
FileUtils.cd(dir.path) do
|
122
|
-
input.extract
|
123
|
-
end
|
124
|
-
when input.is_compressed?
|
125
|
-
IMW.announce_if_verbose("Decompressing #{input}...")
|
126
|
-
input.cp(new_path).decompress!
|
127
|
-
else
|
128
|
-
IMW.announce_if_verbose("Copying #{input}...")
|
129
|
-
input.cp(new_path)
|
130
|
-
end
|
131
|
-
end
|
132
|
-
|
133
|
-
# Download a remote input to this Aggregator's +dir+.
|
134
|
-
#
|
135
|
-
# @param [IMW::Resource] input
|
136
|
-
def download_remote_input input
|
137
|
-
IMW.announce_if_verbose("Downloading #{input}...")
|
138
|
-
input.cp(File.join(dir.path, input.effective_basename))
|
139
|
-
end
|
140
|
-
|
141
|
-
def add_processing_error error # :nodoc:
|
142
|
-
IMW.logger.warn error
|
143
|
-
errors << error
|
144
|
-
end
|
145
|
-
|
146
|
-
end
|
147
|
-
end
|
148
|
-
end
|