s3-sync 1.2.6
Sign up to get free protection for your applications and to get access to all the features.
- data/History.txt +184 -0
- data/README.rdoc +318 -0
- data/bin/s3cmd +338 -0
- data/bin/s3sync +731 -0
- data/lib/s3sync.rb +39 -0
- data/lib/s3sync/HTTPStreaming.rb +107 -0
- data/lib/s3sync/S3.rb +714 -0
- data/lib/s3sync/S3_s3sync_mod.rb +143 -0
- data/lib/s3sync/S3encoder.rb +50 -0
- data/lib/s3sync/s3config.rb +27 -0
- data/lib/s3sync/s3try.rb +178 -0
- data/lib/s3sync/thread_generator.rb +383 -0
- data/test/test_helper.rb +3 -0
- data/test/test_s3sync.rb +11 -0
- metadata +67 -0
@@ -0,0 +1,143 @@
|
|
1
|
+
# This software code is made available "AS IS" without warranties of any
|
2
|
+
# kind. You may copy, display, modify and redistribute the software
|
3
|
+
# code either by itself or as incorporated into your code; provided that
|
4
|
+
# you do not remove any proprietary notices. Your use of this software
|
5
|
+
# code is at your own risk and you waive any claim against Amazon
|
6
|
+
# Digital Services, Inc. or its affiliates with respect to your use of
|
7
|
+
# this software code. (c) 2006 Amazon Digital Services, Inc. or its
|
8
|
+
# affiliates.
|
9
|
+
#
|
10
|
+
# This software code is made available "AS IS" without warranties of any
|
11
|
+
# kind. You may copy, display, modify and redistribute the software
|
12
|
+
# code either by itself or as incorporated into your code; provided that
|
13
|
+
# you do not remove any proprietary notices. Your use of this software
|
14
|
+
# code is at your own risk and you waive any claim against the author
|
15
|
+
# with respect to your use of this software code.
|
16
|
+
# (c) 2007 s3sync.net
|
17
|
+
#
|
18
|
+
require 's3sync/S3'
|
19
|
+
require 's3sync/HTTPStreaming'
|
20
|
+
|
21
|
+
# The purpose of this file is to overlay the S3 library from AWS
|
22
|
+
# to add some functionality
|
23
|
+
# (without changing the file itself or requiring a specific version)
|
24
|
+
# It still isn't perfectly robust, i.e. if radical changes are made
|
25
|
+
# to the underlying lib this stuff will need updating.
|
26
|
+
|
27
|
+
module S3
|
28
|
+
class AWSAuthConnection
|
29
|
+
|
30
|
+
def make_http(bucket='', host='', proxy_host=nil, proxy_port=nil, proxy_user=nil, proxy_pass=nil)
|
31
|
+
|
32
|
+
# build the domain based on the calling format
|
33
|
+
server = ''
|
34
|
+
if host != ''
|
35
|
+
server = host
|
36
|
+
elsif bucket.empty?
|
37
|
+
# for a bucketless request (i.e. list all buckets)
|
38
|
+
# revert to regular domain case since this operation
|
39
|
+
# does not make sense for vanity domains
|
40
|
+
server = @server
|
41
|
+
elsif @calling_format == CallingFormat::SUBDOMAIN
|
42
|
+
server = "#{bucket}.#{@server}"
|
43
|
+
elsif @calling_format == CallingFormat::VANITY
|
44
|
+
server = bucket
|
45
|
+
else
|
46
|
+
server = @server
|
47
|
+
end
|
48
|
+
# automatically does the right thing when no proxy
|
49
|
+
http = Net::HTTP::Proxy(proxy_host, proxy_port, proxy_user, proxy_pass).new(server, @port)
|
50
|
+
#http = Net::HTTP.new(server, @port)
|
51
|
+
http.use_ssl = @is_secure
|
52
|
+
http.verify_mode=@verify_mode
|
53
|
+
http.ca_file=@ca_file
|
54
|
+
http.ca_path=@ca_path
|
55
|
+
http.start
|
56
|
+
return http
|
57
|
+
end
|
58
|
+
|
59
|
+
# add support for streaming the response body to an IO stream
|
60
|
+
alias __make_request__ make_request
|
61
|
+
def make_request(method, bucket='', key='', path_args={}, headers={}, data='', metadata={}, streamOut=nil)
|
62
|
+
# build the path based on the calling format
|
63
|
+
path = ''
|
64
|
+
if (not bucket.empty?) and (@calling_format == CallingFormat::REGULAR)
|
65
|
+
path << "/#{bucket}"
|
66
|
+
end
|
67
|
+
# add the slash after the bucket regardless
|
68
|
+
# the key will be appended if it is non-empty
|
69
|
+
path << "/#{key}"
|
70
|
+
|
71
|
+
# build the path_argument string
|
72
|
+
# add the ? in all cases since
|
73
|
+
# signature and credentials follow path args
|
74
|
+
path << '?'
|
75
|
+
path << S3.path_args_hash_to_string(path_args)
|
76
|
+
|
77
|
+
req = method_to_request_class(method).new("#{path}")
|
78
|
+
|
79
|
+
set_headers(req, headers)
|
80
|
+
set_headers(req, metadata, METADATA_PREFIX)
|
81
|
+
set_headers(req, {'Connection' => 'keep-alive', 'Keep-Alive' => '300'})
|
82
|
+
|
83
|
+
set_aws_auth_header(req, @aws_access_key_id, @aws_secret_access_key, bucket, key, path_args)
|
84
|
+
|
85
|
+
http = $S3syncHttp
|
86
|
+
|
87
|
+
if req.request_body_permitted?
|
88
|
+
return http.request(req, data, streamOut)
|
89
|
+
else
|
90
|
+
return http.request(req, nil, streamOut)
|
91
|
+
end
|
92
|
+
end
|
93
|
+
|
94
|
+
# a "get" operation that sends the body to an IO stream
|
95
|
+
def get_stream(bucket, key, headers={}, streamOut=nil)
|
96
|
+
return GetResponse.new(make_request('GET', bucket, CGI::escape(key), {}, headers, '', {}, streamOut))
|
97
|
+
end
|
98
|
+
|
99
|
+
# a "get" operation that sends the body to an IO stream
|
100
|
+
def get_query_stream(bucket, key, path_args={}, headers={}, streamOut=nil)
|
101
|
+
return GetResponse.new(make_request('GET', bucket, CGI::escape(key), path_args, headers, '', {}, streamOut))
|
102
|
+
end
|
103
|
+
|
104
|
+
def head(bucket, key=nil, headers={})
|
105
|
+
return GetResponse.new(make_request('HEAD', bucket, CGI::escape(key), {}, headers, '', {}))
|
106
|
+
end
|
107
|
+
undef create_bucket
|
108
|
+
def create_bucket(bucket, object)
|
109
|
+
object = S3Object.new(object) if not object.instance_of? S3Object
|
110
|
+
return Response.new(
|
111
|
+
make_request('PUT', bucket, '', {}, {}, object.data, object.metadata)
|
112
|
+
)
|
113
|
+
end
|
114
|
+
# no, because internally the library does not support the header,wait,body paradigm, so this is useless
|
115
|
+
#alias __put__ put
|
116
|
+
#def put(bucket, key, object, headers={})
|
117
|
+
# headers['Expect'] = "100-continue"
|
118
|
+
# __put__(bucket, key, object, headers)
|
119
|
+
#end
|
120
|
+
|
121
|
+
|
122
|
+
# allow ssl validation
|
123
|
+
attr_accessor :verify_mode
|
124
|
+
attr_accessor :ca_path
|
125
|
+
attr_accessor :ca_file
|
126
|
+
|
127
|
+
end
|
128
|
+
module CallingFormat
|
129
|
+
def CallingFormat.string_to_format(s)
|
130
|
+
case s
|
131
|
+
when 'REGULAR'
|
132
|
+
return CallingFormat::REGULAR
|
133
|
+
when 'SUBDOMAIN'
|
134
|
+
return CallingFormat::SUBDOMAIN
|
135
|
+
when 'VANITY'
|
136
|
+
return CallingFormat::VANITY
|
137
|
+
else
|
138
|
+
raise "Unsupported calling format #{s}"
|
139
|
+
end
|
140
|
+
end
|
141
|
+
end
|
142
|
+
|
143
|
+
end
|
@@ -0,0 +1,50 @@
|
|
1
|
+
# This software code is made available "AS IS" without warranties of any
|
2
|
+
# kind. You may copy, display, modify and redistribute the software
|
3
|
+
# code either by itself or as incorporated into your code; provided that
|
4
|
+
# you do not remove any proprietary notices. Your use of this software
|
5
|
+
# code is at your own risk and you waive any claim against the author
|
6
|
+
# with respect to your use of this software code.
|
7
|
+
# (c) 2007 s3sync.net
|
8
|
+
#
|
9
|
+
|
10
|
+
# The purpose of this file is to overlay the cgi class
|
11
|
+
# to add some functionality
|
12
|
+
# (without changing the file itself or requiring a specific version)
|
13
|
+
# It still isn't perfectly robust, i.e. if radical changes are made
|
14
|
+
# to the underlying lib this stuff will need updating.
|
15
|
+
|
16
|
+
require 'cgi'
|
17
|
+
require 'iconv' # for UTF-8 conversion
|
18
|
+
|
19
|
+
# thanks to http://www.redhillconsulting.com.au/blogs/simon/archives/000326.html
|
20
|
+
module S3ExtendCGI
|
21
|
+
def self.included(base)
|
22
|
+
base.extend(ClassMethods)
|
23
|
+
base.class_eval do
|
24
|
+
class << self
|
25
|
+
alias_method :S3Extend_escape_orig, :escape unless method_defined?(:S3Extend_escape_orig)
|
26
|
+
alias_method :escape, :S3Extend_escape
|
27
|
+
end
|
28
|
+
end
|
29
|
+
end
|
30
|
+
module ClassMethods
|
31
|
+
@@exemptSlashesInEscape = false
|
32
|
+
attr_writer :exemptSlashesInEscape
|
33
|
+
@@usePercent20InEscape = false
|
34
|
+
attr_writer :usePercent20InEscape
|
35
|
+
@@nativeCharacterEncoding = "ISO-8859-1"
|
36
|
+
attr_writer :nativeCharacterEncoding
|
37
|
+
@@useUTF8InEscape = false
|
38
|
+
attr_writer :useUTF8InEscape
|
39
|
+
|
40
|
+
def S3Extend_escape(string)
|
41
|
+
result = string
|
42
|
+
result = Iconv.iconv("UTF-8", @nativeCharacterEncoding, string).join if @useUTF8InEscape
|
43
|
+
result = S3Extend_escape_orig(result)
|
44
|
+
result.gsub!(/%2f/i, "/") if @exemptSlashesInEscape
|
45
|
+
result.gsub!("+", "%20") if @usePercent20InEscape
|
46
|
+
result
|
47
|
+
end
|
48
|
+
end
|
49
|
+
end
|
50
|
+
CGI.send(:include, S3ExtendCGI)
|
@@ -0,0 +1,27 @@
|
|
1
|
+
#!/usr/bin/ruby
|
2
|
+
# This software code is made available "AS IS" without warranties of any
|
3
|
+
# kind. You may copy, display, modify and redistribute the software
|
4
|
+
# code either by itself or as incorporated into your code; provided that
|
5
|
+
# you do not remove any proprietary notices. Your use of this software
|
6
|
+
# code is at your own risk and you waive any claim against the author
|
7
|
+
# with respect to your use of this software code.
|
8
|
+
# (c) 2007 alastair brunton
|
9
|
+
#
|
10
|
+
# modified to search out the yaml in several places, thanks wkharold.
|
11
|
+
require 'yaml'
|
12
|
+
|
13
|
+
module S3Config
|
14
|
+
|
15
|
+
confpath = ["#{ENV['S3CONF']}", "#{ENV['HOME']}/.s3conf", "/etc/s3conf"]
|
16
|
+
|
17
|
+
confpath.each do |path|
|
18
|
+
if File.exists?(path) and File.directory?(path) and File.exists?("#{path}/s3config.yml")
|
19
|
+
config = YAML.load_file("#{path}/s3config.yml")
|
20
|
+
config.each_pair do |key, value|
|
21
|
+
eval("$#{key.upcase} = '#{value}'")
|
22
|
+
end
|
23
|
+
break
|
24
|
+
end
|
25
|
+
end
|
26
|
+
|
27
|
+
end
|
data/lib/s3sync/s3try.rb
ADDED
@@ -0,0 +1,178 @@
|
|
1
|
+
#!/usr/bin/env ruby
|
2
|
+
# This software code is made available "AS IS" without warranties of any
|
3
|
+
# kind. You may copy, display, modify and redistribute the software
|
4
|
+
# code either by itself or as incorporated into your code; provided that
|
5
|
+
# you do not remove any proprietary notices. Your use of this software
|
6
|
+
# code is at your own risk and you waive any claim against the author
|
7
|
+
# with respect to your use of this software code.
|
8
|
+
# (c) 2007 s3sync.net
|
9
|
+
#
|
10
|
+
module S3sync
|
11
|
+
|
12
|
+
$AWS_ACCESS_KEY_ID = ENV["AWS_ACCESS_KEY_ID"]
|
13
|
+
$AWS_SECRET_ACCESS_KEY = ENV["AWS_SECRET_ACCESS_KEY"]
|
14
|
+
$AWS_S3_HOST = (ENV["AWS_S3_HOST"] or "s3.amazonaws.com")
|
15
|
+
$HTTP_PROXY_HOST = ENV["HTTP_PROXY_HOST"]
|
16
|
+
$HTTP_PROXY_PORT = ENV["HTTP_PROXY_PORT"]
|
17
|
+
$HTTP_PROXY_USER = ENV["HTTP_PROXY_USER"]
|
18
|
+
$HTTP_PROXY_PASSWORD = ENV["HTTP_PROXY_PASSWORD"]
|
19
|
+
$SSL_CERT_DIR = ENV["SSL_CERT_DIR"]
|
20
|
+
$SSL_CERT_FILE = ENV["SSL_CERT_FILE"]
|
21
|
+
$S3SYNC_RETRIES = (ENV["S3SYNC_RETRIES"] or 100).to_i # number of errors to tolerate
|
22
|
+
$S3SYNC_WAITONERROR = (ENV["S3SYNC_WAITONERROR"] or 5).to_i # seconds
|
23
|
+
$S3SYNC_NATIVE_CHARSET = (ENV["S3SYNC_NATIVE_CHARSET"] or "ISO-8859-1")
|
24
|
+
$AWS_CALLING_FORMAT = (ENV["AWS_CALLING_FORMAT"] or "SUBDOMAIN")
|
25
|
+
|
26
|
+
require 's3sync/S3'
|
27
|
+
require 's3sync/HTTPStreaming'
|
28
|
+
require 's3sync/S3encoder'
|
29
|
+
|
30
|
+
CGI::exemptSlashesInEscape = true
|
31
|
+
CGI::usePercent20InEscape = true
|
32
|
+
CGI::useUTF8InEscape = true
|
33
|
+
CGI::nativeCharacterEncoding = $S3SYNC_NATIVE_CHARSET
|
34
|
+
require 's3sync/S3_s3sync_mod'
|
35
|
+
|
36
|
+
|
37
|
+
$S3syncRetriesLeft = $S3SYNC_RETRIES.to_i
|
38
|
+
|
39
|
+
|
40
|
+
def S3sync.s3trySetup
|
41
|
+
|
42
|
+
# ---------- CONNECT ---------- #
|
43
|
+
|
44
|
+
$S3syncConnection = S3::AWSAuthConnection.new($AWS_ACCESS_KEY_ID, $AWS_SECRET_ACCESS_KEY, $S3syncOptions['--ssl'], $AWS_S3_HOST)
|
45
|
+
$S3syncConnection.calling_format = S3::CallingFormat::string_to_format($AWS_CALLING_FORMAT)
|
46
|
+
if $S3syncOptions['--ssl']
|
47
|
+
if $SSL_CERT_DIR
|
48
|
+
$S3syncConnection.verify_mode = OpenSSL::SSL::VERIFY_PEER
|
49
|
+
$S3syncConnection.ca_path = $SSL_CERT_DIR
|
50
|
+
elsif $SSL_CERT_FILE
|
51
|
+
$S3syncConnection.verify_mode = OpenSSL::SSL::VERIFY_PEER
|
52
|
+
$S3syncConnection.ca_file = $SSL_CERT_FILE
|
53
|
+
end
|
54
|
+
end
|
55
|
+
end
|
56
|
+
|
57
|
+
def S3sync.s3urlSetup
|
58
|
+
$S3syncGenerator = S3::QueryStringAuthGenerator.new($AWS_ACCESS_KEY_ID, $AWS_SECRET_ACCESS_KEY, $S3syncOptions['--ssl'], $AWS_S3_HOST)
|
59
|
+
$S3syncGenerator.calling_format = S3::CallingFormat::string_to_format($AWS_CALLING_FORMAT)
|
60
|
+
$S3syncGenerator.expires_in = $S3syncOptions['--expires-in']
|
61
|
+
end
|
62
|
+
|
63
|
+
def S3sync.S3tryConnect(bucket, host='')
|
64
|
+
$S3syncHttp = $S3syncConnection.make_http(bucket, host, $HTTP_PROXY_HOST, $HTTP_PROXY_PORT, $HTTP_PROXY_USER, $HTTP_PROXY_PASSWORD)
|
65
|
+
end
|
66
|
+
|
67
|
+
def S3sync.S3try(command, bucket, *args)
|
68
|
+
if(not $S3syncHttp or (bucket != $S3syncLastBucket))
|
69
|
+
$stderr.puts "Creating new connection" if $S3syncOptions['--debug']
|
70
|
+
$S3syncLastBucket = bucket
|
71
|
+
while $S3syncRetriesLeft > 0 do
|
72
|
+
begin
|
73
|
+
S3sync.S3tryConnect(bucket)
|
74
|
+
break
|
75
|
+
rescue Errno::ECONNRESET => e
|
76
|
+
$stderr.puts "Connection reset: #{e}"
|
77
|
+
rescue Errno::ECONNABORTED => e
|
78
|
+
$stderr.puts "Connection aborted: #{e}"
|
79
|
+
rescue Errno::ETIMEDOUT => e
|
80
|
+
$stderr.puts "Connection timed out: #{e}"
|
81
|
+
rescue Timeout::Error => e
|
82
|
+
$stderr.puts "Connection timed out: #{e}"
|
83
|
+
end
|
84
|
+
$S3syncRetriesLeft -= 1
|
85
|
+
$stderr.puts "#{$S3syncRetriesLeft} retries left, sleeping for #{$S3SYNC_WAITONERROR} seconds"
|
86
|
+
Kernel.sleep $S3SYNC_WAITONERROR.to_i
|
87
|
+
end
|
88
|
+
end
|
89
|
+
|
90
|
+
result = nil
|
91
|
+
delim = $,
|
92
|
+
$,=' '
|
93
|
+
while $S3syncRetriesLeft > 0 do
|
94
|
+
$stderr.puts "Trying command #{command} #{bucket} #{args} with #{$S3syncRetriesLeft} retries left" if $S3syncOptions['--debug']
|
95
|
+
forceRetry = false
|
96
|
+
now = false
|
97
|
+
hush = false
|
98
|
+
begin
|
99
|
+
result = $S3syncConnection.send(command, bucket, *args)
|
100
|
+
rescue Errno::EPIPE => e
|
101
|
+
forceRetry = true
|
102
|
+
$stderr.puts "Broken pipe: #{e}"
|
103
|
+
rescue Errno::ECONNRESET => e
|
104
|
+
forceRetry = true
|
105
|
+
$stderr.puts "Connection reset: #{e}"
|
106
|
+
rescue Errno::ECONNABORTED => e
|
107
|
+
forceRetry = true
|
108
|
+
$stderr.puts "Connection aborted: #{e}"
|
109
|
+
rescue Errno::ETIMEDOUT => e
|
110
|
+
forceRetry = true
|
111
|
+
$stderr.puts "Connection timed out: #{e}"
|
112
|
+
rescue Timeout::Error => e
|
113
|
+
forceRetry = true
|
114
|
+
$stderr.puts "Connection timed out: #{e}"
|
115
|
+
rescue EOFError => e
|
116
|
+
# i THINK this is happening like a connection reset
|
117
|
+
forceRetry = true
|
118
|
+
$stderr.puts "EOF error: #{e}"
|
119
|
+
rescue OpenSSL::SSL::SSLError => e
|
120
|
+
forceRetry = true
|
121
|
+
$stderr.puts "SSL Error: #{e}"
|
122
|
+
rescue NoMethodError => e
|
123
|
+
# we get this when using --progress, and the local item is something unreadable
|
124
|
+
$stderr.puts "Null stream error: #{e}"
|
125
|
+
break
|
126
|
+
end
|
127
|
+
if forceRetry
|
128
|
+
# kill and reset connection when we receive a non-50x yet retry-able error
|
129
|
+
S3sync.S3tryConnect(bucket)
|
130
|
+
end
|
131
|
+
begin
|
132
|
+
debug("Response code: #{result.http_response.code}")
|
133
|
+
break if ((200...300).include? result.http_response.code.to_i) and (not forceRetry)
|
134
|
+
if result.http_response.code.to_i == 301
|
135
|
+
$stderr.puts "Permanent redirect received. Try setting AWS_CALLING_FORMAT to SUBDOMAIN"
|
136
|
+
elsif result.http_response.code.to_i == 307
|
137
|
+
# move to the other host
|
138
|
+
host = %r{https?://([^/]+)}.match(result.http_response['Location'])[1]
|
139
|
+
$stderr.puts("Temporary Redirect to: " + host)
|
140
|
+
debug("Host: " + host)
|
141
|
+
S3sync.S3tryConnect(bucket, host)
|
142
|
+
$S3syncRetriesLeft = $S3syncRetriesLeft+1 # don't use one up below
|
143
|
+
forceRetry = true
|
144
|
+
now = true
|
145
|
+
hush = true
|
146
|
+
else
|
147
|
+
$stderr.puts "S3 command failed:\n#{command} #{args}"
|
148
|
+
$stderr.puts "With result #{result.http_response.code} #{result.http_response.message}\n"
|
149
|
+
debug(result.http_response.body)
|
150
|
+
end
|
151
|
+
# only retry 500's, per amazon
|
152
|
+
break unless ((500...600).include? result.http_response.code.to_i) or forceRetry
|
153
|
+
rescue NoMethodError
|
154
|
+
debug("No result available")
|
155
|
+
end
|
156
|
+
$S3syncRetriesLeft -= 1
|
157
|
+
$stderr.puts "#{$S3syncRetriesLeft} retries left, sleeping for #{$S3SYNC_WAITONERROR} seconds" unless hush
|
158
|
+
Kernel.sleep $S3SYNC_WAITONERROR.to_i unless now
|
159
|
+
end
|
160
|
+
if $S3syncRetriesLeft <= 0
|
161
|
+
$stderr.puts "Ran out of retries; operations did not complete!"
|
162
|
+
end
|
163
|
+
$, = delim
|
164
|
+
result
|
165
|
+
end
|
166
|
+
|
167
|
+
def S3sync.S3url(command, bucket, *args)
|
168
|
+
S3sync.s3urlSetup() unless $S3syncGenerator
|
169
|
+
result = nil
|
170
|
+
delim = $,
|
171
|
+
$,=' '
|
172
|
+
$stderr.puts "Calling command #{command} #{bucket} #{args}" if $S3syncOptions['--debug']
|
173
|
+
result = $S3syncGenerator.send(command, bucket, *args)
|
174
|
+
$, = delim
|
175
|
+
result
|
176
|
+
end
|
177
|
+
end #module
|
178
|
+
|
@@ -0,0 +1,383 @@
|
|
1
|
+
#!/usr/bin/env ruby
|
2
|
+
#--
|
3
|
+
# $Idaemons: /home/cvs/rb/generator.rb,v 1.8 2001/10/03 08:54:32 knu Exp $
|
4
|
+
# $RoughId: generator.rb,v 1.10 2003/10/14 19:36:58 knu Exp $
|
5
|
+
# $Id: generator.rb,v 1.12 2005/12/31 02:56:46 ocean Exp $
|
6
|
+
#++
|
7
|
+
#
|
8
|
+
# = generator.rb: convert an internal iterator to an external one
|
9
|
+
#
|
10
|
+
# Copyright (c) 2001,2003 Akinori MUSHA <knu@iDaemons.org>
|
11
|
+
#
|
12
|
+
# All rights reserved. You can redistribute and/or modify it under
|
13
|
+
# the same terms as Ruby.
|
14
|
+
#
|
15
|
+
# == Overview
|
16
|
+
#
|
17
|
+
# This library provides the Generator class, which converts an
|
18
|
+
# internal iterator (i.e. an Enumerable object) to an external
|
19
|
+
# iterator. In that form, you can roll many iterators independently.
|
20
|
+
#
|
21
|
+
# The SyncEnumerator class, which is implemented using Generator,
|
22
|
+
# makes it easy to roll many Enumerable objects synchronously.
|
23
|
+
#
|
24
|
+
# See the respective classes for examples of usage.
|
25
|
+
|
26
|
+
|
27
|
+
#
|
28
|
+
# Generator converts an internal iterator (i.e. an Enumerable object)
|
29
|
+
# to an external iterator.
|
30
|
+
#
|
31
|
+
# == Example
|
32
|
+
#
|
33
|
+
# require 'generator'
|
34
|
+
#
|
35
|
+
# # Generator from an Enumerable object
|
36
|
+
# g = Generator.new(['A', 'B', 'C', 'Z'])
|
37
|
+
#
|
38
|
+
# while g.next?
|
39
|
+
# puts g.next
|
40
|
+
# end
|
41
|
+
#
|
42
|
+
# # Generator from a block
|
43
|
+
# g = Generator.new { |g|
|
44
|
+
# for i in 'A'..'C'
|
45
|
+
# g.yield i
|
46
|
+
# end
|
47
|
+
#
|
48
|
+
# g.yield 'Z'
|
49
|
+
# }
|
50
|
+
#
|
51
|
+
# # The same result as above
|
52
|
+
# while g.next?
|
53
|
+
# puts g.next
|
54
|
+
# end
|
55
|
+
#
|
56
|
+
class Generator
|
57
|
+
include Enumerable
|
58
|
+
|
59
|
+
# Creates a new generator either from an Enumerable object or from a
|
60
|
+
# block.
|
61
|
+
#
|
62
|
+
# In the former, block is ignored even if given.
|
63
|
+
#
|
64
|
+
# In the latter, the given block is called with the generator
|
65
|
+
# itself, and expected to call the +yield+ method for each element.
|
66
|
+
def initialize(enum = nil, &block)
|
67
|
+
if enum
|
68
|
+
@block = proc{|g| enum.each{|value| g.yield value}}
|
69
|
+
else
|
70
|
+
@block = block
|
71
|
+
end
|
72
|
+
@index = 0
|
73
|
+
@queue = []
|
74
|
+
@main_thread = nil
|
75
|
+
@loop_thread.kill if defined?(@loop_thread)
|
76
|
+
@loop_thread = Thread.new do
|
77
|
+
Thread.stop
|
78
|
+
begin
|
79
|
+
@block.call(self)
|
80
|
+
rescue
|
81
|
+
@main_thread.raise $!
|
82
|
+
ensure
|
83
|
+
@main_thread.wakeup
|
84
|
+
end
|
85
|
+
end
|
86
|
+
Thread.pass until @loop_thread.stop?
|
87
|
+
self
|
88
|
+
end
|
89
|
+
|
90
|
+
# Yields an element to the generator.
|
91
|
+
def yield(value)
|
92
|
+
if Thread.current != @loop_thread
|
93
|
+
raise "should be called in Generator.new{|g| ... }"
|
94
|
+
end
|
95
|
+
Thread.critical = true
|
96
|
+
begin
|
97
|
+
@queue << value
|
98
|
+
@main_thread.wakeup
|
99
|
+
Thread.stop
|
100
|
+
ensure
|
101
|
+
Thread.critical = false
|
102
|
+
end
|
103
|
+
self
|
104
|
+
end
|
105
|
+
|
106
|
+
# Returns true if the generator has reached the end.
|
107
|
+
def end?
|
108
|
+
if @queue.empty?
|
109
|
+
if @main_thread
|
110
|
+
raise "should not be called in Generator.new{|g| ... }"
|
111
|
+
end
|
112
|
+
Thread.critical = true
|
113
|
+
begin
|
114
|
+
@main_thread = Thread.current
|
115
|
+
@loop_thread.wakeup
|
116
|
+
Thread.stop
|
117
|
+
rescue ThreadError
|
118
|
+
# ignore
|
119
|
+
ensure
|
120
|
+
@main_thread = nil
|
121
|
+
Thread.critical = false
|
122
|
+
end
|
123
|
+
end
|
124
|
+
@queue.empty?
|
125
|
+
end
|
126
|
+
|
127
|
+
# Returns true if the generator has not reached the end yet.
|
128
|
+
def next?
|
129
|
+
!end?
|
130
|
+
end
|
131
|
+
|
132
|
+
# Returns the current index (position) counting from zero.
|
133
|
+
def index
|
134
|
+
@index
|
135
|
+
end
|
136
|
+
|
137
|
+
# Returns the current index (position) counting from zero.
|
138
|
+
def pos
|
139
|
+
@index
|
140
|
+
end
|
141
|
+
|
142
|
+
# Returns the element at the current position and moves forward.
|
143
|
+
def next
|
144
|
+
raise EOFError.new("no more elements available") if end?
|
145
|
+
@index += 1
|
146
|
+
@queue.shift
|
147
|
+
end
|
148
|
+
|
149
|
+
# Returns the element at the current position.
|
150
|
+
def current
|
151
|
+
raise EOFError.new("no more elements available") if end?
|
152
|
+
@queue.first
|
153
|
+
end
|
154
|
+
|
155
|
+
# Rewinds the generator.
|
156
|
+
def rewind
|
157
|
+
initialize(nil, &@block) if @index.nonzero?
|
158
|
+
self
|
159
|
+
end
|
160
|
+
|
161
|
+
# Rewinds the generator and enumerates the elements.
|
162
|
+
def each
|
163
|
+
rewind
|
164
|
+
until end?
|
165
|
+
yield self.next
|
166
|
+
end
|
167
|
+
self
|
168
|
+
end
|
169
|
+
end
|
170
|
+
|
171
|
+
#
|
172
|
+
# SyncEnumerator creates an Enumerable object from multiple Enumerable
|
173
|
+
# objects and enumerates them synchronously.
|
174
|
+
#
|
175
|
+
# == Example
|
176
|
+
#
|
177
|
+
# require 'generator'
|
178
|
+
#
|
179
|
+
# s = SyncEnumerator.new([1,2,3], ['a', 'b', 'c'])
|
180
|
+
#
|
181
|
+
# # Yields [1, 'a'], [2, 'b'], and [3,'c']
|
182
|
+
# s.each { |row| puts row.join(', ') }
|
183
|
+
#
|
184
|
+
class SyncEnumerator
|
185
|
+
include Enumerable
|
186
|
+
|
187
|
+
# Creates a new SyncEnumerator which enumerates rows of given
|
188
|
+
# Enumerable objects.
|
189
|
+
def initialize(*enums)
|
190
|
+
@gens = enums.map { |e| Generator.new(e) }
|
191
|
+
end
|
192
|
+
|
193
|
+
# Returns the number of enumerated Enumerable objects, i.e. the size
|
194
|
+
# of each row.
|
195
|
+
def size
|
196
|
+
@gens.size
|
197
|
+
end
|
198
|
+
|
199
|
+
# Returns the number of enumerated Enumerable objects, i.e. the size
|
200
|
+
# of each row.
|
201
|
+
def length
|
202
|
+
@gens.length
|
203
|
+
end
|
204
|
+
|
205
|
+
# Returns true if the given nth Enumerable object has reached the
|
206
|
+
# end. If no argument is given, returns true if any of the
|
207
|
+
# Enumerable objects has reached the end.
|
208
|
+
def end?(i = nil)
|
209
|
+
if i.nil?
|
210
|
+
@gens.detect { |g| g.end? } ? true : false
|
211
|
+
else
|
212
|
+
@gens[i].end?
|
213
|
+
end
|
214
|
+
end
|
215
|
+
|
216
|
+
# Enumerates rows of the Enumerable objects.
|
217
|
+
def each
|
218
|
+
@gens.each { |g| g.rewind }
|
219
|
+
|
220
|
+
loop do
|
221
|
+
count = 0
|
222
|
+
|
223
|
+
ret = @gens.map { |g|
|
224
|
+
if g.end?
|
225
|
+
count += 1
|
226
|
+
nil
|
227
|
+
else
|
228
|
+
g.next
|
229
|
+
end
|
230
|
+
}
|
231
|
+
|
232
|
+
if count == @gens.size
|
233
|
+
break
|
234
|
+
end
|
235
|
+
|
236
|
+
yield ret
|
237
|
+
end
|
238
|
+
|
239
|
+
self
|
240
|
+
end
|
241
|
+
end
|
242
|
+
|
243
|
+
if $0 == __FILE__
|
244
|
+
eval DATA.read, nil, $0, __LINE__+4
|
245
|
+
end
|
246
|
+
|
247
|
+
__END__
|
248
|
+
|
249
|
+
require 'test/unit'
|
250
|
+
|
251
|
+
class TC_Generator < Test::Unit::TestCase
|
252
|
+
def test_block1
|
253
|
+
g = Generator.new { |g|
|
254
|
+
# no yield's
|
255
|
+
}
|
256
|
+
|
257
|
+
assert_equal(0, g.pos)
|
258
|
+
assert_raises(EOFError) { g.current }
|
259
|
+
end
|
260
|
+
|
261
|
+
def test_block2
|
262
|
+
g = Generator.new { |g|
|
263
|
+
for i in 'A'..'C'
|
264
|
+
g.yield i
|
265
|
+
end
|
266
|
+
|
267
|
+
g.yield 'Z'
|
268
|
+
}
|
269
|
+
|
270
|
+
assert_equal(0, g.pos)
|
271
|
+
assert_equal('A', g.current)
|
272
|
+
|
273
|
+
assert_equal(true, g.next?)
|
274
|
+
assert_equal(0, g.pos)
|
275
|
+
assert_equal('A', g.current)
|
276
|
+
assert_equal(0, g.pos)
|
277
|
+
assert_equal('A', g.next)
|
278
|
+
|
279
|
+
assert_equal(1, g.pos)
|
280
|
+
assert_equal(true, g.next?)
|
281
|
+
assert_equal(1, g.pos)
|
282
|
+
assert_equal('B', g.current)
|
283
|
+
assert_equal(1, g.pos)
|
284
|
+
assert_equal('B', g.next)
|
285
|
+
|
286
|
+
assert_equal(g, g.rewind)
|
287
|
+
|
288
|
+
assert_equal(0, g.pos)
|
289
|
+
assert_equal('A', g.current)
|
290
|
+
|
291
|
+
assert_equal(true, g.next?)
|
292
|
+
assert_equal(0, g.pos)
|
293
|
+
assert_equal('A', g.current)
|
294
|
+
assert_equal(0, g.pos)
|
295
|
+
assert_equal('A', g.next)
|
296
|
+
|
297
|
+
assert_equal(1, g.pos)
|
298
|
+
assert_equal(true, g.next?)
|
299
|
+
assert_equal(1, g.pos)
|
300
|
+
assert_equal('B', g.current)
|
301
|
+
assert_equal(1, g.pos)
|
302
|
+
assert_equal('B', g.next)
|
303
|
+
|
304
|
+
assert_equal(2, g.pos)
|
305
|
+
assert_equal(true, g.next?)
|
306
|
+
assert_equal(2, g.pos)
|
307
|
+
assert_equal('C', g.current)
|
308
|
+
assert_equal(2, g.pos)
|
309
|
+
assert_equal('C', g.next)
|
310
|
+
|
311
|
+
assert_equal(3, g.pos)
|
312
|
+
assert_equal(true, g.next?)
|
313
|
+
assert_equal(3, g.pos)
|
314
|
+
assert_equal('Z', g.current)
|
315
|
+
assert_equal(3, g.pos)
|
316
|
+
assert_equal('Z', g.next)
|
317
|
+
|
318
|
+
assert_equal(4, g.pos)
|
319
|
+
assert_equal(false, g.next?)
|
320
|
+
assert_raises(EOFError) { g.next }
|
321
|
+
end
|
322
|
+
|
323
|
+
def test_each
|
324
|
+
a = [5, 6, 7, 8, 9]
|
325
|
+
|
326
|
+
g = Generator.new(a)
|
327
|
+
|
328
|
+
i = 0
|
329
|
+
|
330
|
+
g.each { |x|
|
331
|
+
assert_equal(a[i], x)
|
332
|
+
|
333
|
+
i += 1
|
334
|
+
|
335
|
+
break if i == 3
|
336
|
+
}
|
337
|
+
|
338
|
+
assert_equal(3, i)
|
339
|
+
|
340
|
+
i = 0
|
341
|
+
|
342
|
+
g.each { |x|
|
343
|
+
assert_equal(a[i], x)
|
344
|
+
|
345
|
+
i += 1
|
346
|
+
}
|
347
|
+
|
348
|
+
assert_equal(5, i)
|
349
|
+
end
|
350
|
+
end
|
351
|
+
|
352
|
+
class TC_SyncEnumerator < Test::Unit::TestCase
|
353
|
+
def test_each
|
354
|
+
r = ['a'..'f', 1..10, 10..20]
|
355
|
+
ra = r.map { |x| x.to_a }
|
356
|
+
|
357
|
+
a = (0...(ra.map {|x| x.size}.max)).map { |i| ra.map { |x| x[i] } }
|
358
|
+
|
359
|
+
s = SyncEnumerator.new(*r)
|
360
|
+
|
361
|
+
i = 0
|
362
|
+
|
363
|
+
s.each { |x|
|
364
|
+
assert_equal(a[i], x)
|
365
|
+
|
366
|
+
i += 1
|
367
|
+
|
368
|
+
break if i == 3
|
369
|
+
}
|
370
|
+
|
371
|
+
assert_equal(3, i)
|
372
|
+
|
373
|
+
i = 0
|
374
|
+
|
375
|
+
s.each { |x|
|
376
|
+
assert_equal(a[i], x)
|
377
|
+
|
378
|
+
i += 1
|
379
|
+
}
|
380
|
+
|
381
|
+
assert_equal(a.size, i)
|
382
|
+
end
|
383
|
+
end
|