pmirror 0.0.1 → 0.0.2
Sign up to get free protection for your applications and to get access to all the features.
- data/.gitignore +1 -0
- data/README.md +11 -2
- data/features/pmirror.feature +19 -8
- data/lib/pmirror.rb +51 -30
- data/lib/pmirror/version.rb +1 -1
- metadata +2 -2
data/.gitignore
CHANGED
data/README.md
CHANGED
@@ -17,6 +17,9 @@ find it.
|
|
17
17
|
The tool is very new so it only has the bare minimum feature set:
|
18
18
|
|
19
19
|
- Specify multiple regex patterns to match against the remote directory
|
20
|
+
- Specify a list of URL's to look for patterns across, this means you
|
21
|
+
can look for the same patterns on multiple urls and aggregate those
|
22
|
+
files into one place
|
20
23
|
- Provides a progressbar download status indicator
|
21
24
|
- Specify a local directory to download files to
|
22
25
|
- Specify a command to execute on the local directory once all files are
|
@@ -53,6 +56,8 @@ Options:
|
|
53
56
|
-h, --help Show command line help
|
54
57
|
-p, --pattern PAT1,PAT2,PAT3 Regex to match files in remote dir,
|
55
58
|
may specify multiple patterns
|
59
|
+
-u --url One or more URL's to check for
|
60
|
+
files that match the defined patterns
|
56
61
|
-l, --localdir DIR Local directory to mirror files to
|
57
62
|
-e, --exec CMD Execute command after completion
|
58
63
|
-d, --debug Enable debugging
|
@@ -61,9 +66,13 @@ may specify multiple patterns
|
|
61
66
|
|
62
67
|
Usage should be pretty self explanatory but here are the details:
|
63
68
|
|
64
|
-
|
69
|
+
`--url` is the remote URL that you want to fetch files from. Right now
|
65
70
|
this is assumed to be an un-authenticated url. We do not recurse into
|
66
|
-
directories looking for files.
|
71
|
+
directories looking for files. You may specify more than one url and we
|
72
|
+
will look at each url and download any matching files from that url. If
|
73
|
+
the same filename is matched across multiple URL's only the first will
|
74
|
+
be downloaded, subsequent files will see that there is already a local
|
75
|
+
file with the same name and will not download.
|
67
76
|
|
68
77
|
`--pattern` allows you to specify a comma separated list of patterns to
|
69
78
|
match on the remote directly. We will iterate over each pattern and
|
data/features/pmirror.feature
CHANGED
@@ -14,25 +14,28 @@ Feature: My bootstrapped app kinda works
|
|
14
14
|
|--localdir|
|
15
15
|
|--exec|
|
16
16
|
|--version|
|
17
|
-
|
18
|
-
|
|
19
|
-
|
17
|
+
|--url|
|
18
|
+
|-p|
|
19
|
+
|-d|
|
20
|
+
|-l|
|
21
|
+
|-e|
|
22
|
+
|-u|
|
20
23
|
|
21
24
|
Scenario: Download a file
|
22
|
-
When I successfully run `pmirror -p meh -l ../foo http://localhost:55555`
|
25
|
+
When I successfully run `pmirror -p meh -l ../foo -u http://localhost:55555`
|
23
26
|
Then the exit status should be 0
|
24
27
|
And the following files should exist:
|
25
28
|
|../foo/meh.txt|
|
26
29
|
|
27
30
|
Scenario: Execute on local directory
|
28
|
-
When I successfully run `pmirror -p meh -l ../foo -e "touch test" http://localhost:55555`
|
31
|
+
When I successfully run `pmirror -p meh -l ../foo -e "touch test" -u http://localhost:55555`
|
29
32
|
Then the exit status should be 0
|
30
33
|
And the following files should exist:
|
31
34
|
|../foo/meh.txt|
|
32
35
|
|../foo/test |
|
33
36
|
|
34
37
|
Scenario: Match multiple files
|
35
|
-
When I successfully run `pmirror -p floo -l ../foo http://localhost:55555`
|
38
|
+
When I successfully run `pmirror -p floo -l ../foo -u http://localhost:55555`
|
36
39
|
Then the exit status should be 0
|
37
40
|
And the following files should exist:
|
38
41
|
| ../foo/floober.txt|
|
@@ -42,7 +45,7 @@ Feature: My bootstrapped app kinda works
|
|
42
45
|
| ../foo/meh.txt|
|
43
46
|
|
44
47
|
Scenario: Match multiple patterns
|
45
|
-
When I successfully run `pmirror -p '^floo.*','^mah.*' -l ../foo http://localhost:55555`
|
48
|
+
When I successfully run `pmirror -p '^floo.*','^mah.*' -l ../foo -u http://localhost:55555`
|
46
49
|
Then the exit status should be 0
|
47
50
|
And the following files should exist:
|
48
51
|
| ../foo/floober.txt|
|
@@ -51,5 +54,13 @@ Feature: My bootstrapped app kinda works
|
|
51
54
|
And the following files should not exist:
|
52
55
|
| ../foo/meh.txt|
|
53
56
|
|
54
|
-
|
57
|
+
Scenario: Match multiple urls
|
58
|
+
When I successfully run `pmirror -p '^floo.*','^mah.*' -l ../foo -u http://localhost:55555,http://localhost:55555`
|
59
|
+
Then the exit status should be 0
|
60
|
+
And the following files should exist:
|
61
|
+
| ../foo/floober.txt|
|
62
|
+
| ../foo/floobah.txt|
|
63
|
+
| ../foo/mah.txt|
|
64
|
+
And the following files should not exist:
|
65
|
+
| ../foo/meh.txt|
|
55
66
|
|
data/lib/pmirror.rb
CHANGED
@@ -9,15 +9,17 @@ module Pmirror
|
|
9
9
|
include Methadone::CLILogging
|
10
10
|
include Methadone::SH
|
11
11
|
|
12
|
-
main do
|
12
|
+
main do
|
13
|
+
d "Inside main"
|
13
14
|
|
14
|
-
|
15
|
+
download_list = get_download_list(options[:url], options[:pattern])
|
16
|
+
d "download_list: #{download_list.inspect}"
|
17
|
+
download_files(options[:localdir], download_list)
|
15
18
|
execute(options[:exec]) if options[:exec]
|
16
19
|
|
17
20
|
end
|
18
21
|
|
19
22
|
description "Mirror files on a remote http server based on pattern match"
|
20
|
-
arg("url", "Url or remote site", :one, :required)
|
21
23
|
on("-p", "--pattern PAT1,PAT2,PAT3", Array,
|
22
24
|
"Regex to match files in remote dir, may specify multiple patterns"
|
23
25
|
)
|
@@ -25,48 +27,66 @@ module Pmirror
|
|
25
27
|
on("-e", "--exec CMD", "Execute command after completion")
|
26
28
|
on("-d", "--debug", "Enable debugging")
|
27
29
|
on("-v", "--version", "Show version")
|
30
|
+
on("-u", "--url URL,URL", Array, "Url or remote site")
|
28
31
|
|
29
|
-
def self.
|
30
|
-
|
31
|
-
|
32
|
-
|
32
|
+
def self.d(msg)
|
33
|
+
if options[:debug]
|
34
|
+
puts "[DEBUG]: #{msg}"
|
35
|
+
end
|
36
|
+
end
|
37
|
+
|
38
|
+
def self.get_download_list(url_list, pattern)
|
39
|
+
d "inside get_download_list"
|
40
|
+
downloads = {}
|
41
|
+
url_list.each do |single_url|
|
42
|
+
downloads[single_url] = []
|
43
|
+
d "Getting download list for url: #{single_url}"
|
44
|
+
page = Nokogiri::HTML(open(single_url))
|
33
45
|
|
34
|
-
|
35
|
-
|
36
|
-
|
37
|
-
|
38
|
-
|
46
|
+
page.css("a").each do |link|
|
47
|
+
file_name = link.attributes['href'].value
|
48
|
+
pattern.each do |matcher|
|
49
|
+
if /#{matcher}/.match(file_name)
|
50
|
+
d "Found match: #{file_name}"
|
51
|
+
downloads[single_url] << file_name
|
52
|
+
end
|
39
53
|
end
|
40
54
|
end
|
55
|
+
d "Returning downloads: #{downloads.inspect}"
|
41
56
|
end
|
42
57
|
downloads
|
43
58
|
end
|
44
59
|
|
45
|
-
def self.download_files(
|
46
|
-
|
47
|
-
|
60
|
+
def self.download_files(local_dir, url_hash={})
|
61
|
+
d "Inside download_files"
|
62
|
+
url_hash.each_key do |single_url|
|
63
|
+
d "Working on #{single_url}"
|
64
|
+
url_hash[single_url].each do |file|
|
65
|
+
local_fn = "#{local_dir}/#{file}"
|
48
66
|
|
49
|
-
|
50
|
-
|
51
|
-
|
52
|
-
|
53
|
-
|
54
|
-
|
67
|
+
unless Dir.exist? options[:localdir]
|
68
|
+
d "PWD: #{Dir.pwd}"
|
69
|
+
puts Dir.open(Dir.pwd).read
|
70
|
+
puts "Destination directory '#{options[:localdir]}' does not exist!"
|
71
|
+
exit 1
|
72
|
+
end
|
55
73
|
|
56
|
-
|
57
|
-
|
58
|
-
|
59
|
-
|
60
|
-
|
61
|
-
|
62
|
-
|
63
|
-
|
64
|
-
|
74
|
+
remote_fn = "#{single_url}/#{file}"
|
75
|
+
unless File.exist?(local_fn)
|
76
|
+
puts "Downloading File: #{file}"
|
77
|
+
puts "#{remote_fn} ==> #{local_fn}"
|
78
|
+
http_to_file(local_fn, remote_fn)
|
79
|
+
# File.write(local_fn, open(remote_fn).read)
|
80
|
+
puts "Download Complete for #{file}"
|
81
|
+
else
|
82
|
+
puts "Skipping #{file}, already exists"
|
83
|
+
end
|
65
84
|
end
|
66
85
|
end
|
67
86
|
end
|
68
87
|
|
69
88
|
def self.http_to_file(filename,url)
|
89
|
+
d "Inside http_to_file"
|
70
90
|
pbar = nil
|
71
91
|
File.open(filename, 'wb') do |save_file|
|
72
92
|
open(url, 'rb',
|
@@ -84,6 +104,7 @@ module Pmirror
|
|
84
104
|
end
|
85
105
|
|
86
106
|
def self.execute(cmd)
|
107
|
+
d "Inside execute"
|
87
108
|
puts "Executing: #{cmd}"
|
88
109
|
sh("cd #{options[:localdir]} && #{cmd}")
|
89
110
|
end
|
data/lib/pmirror/version.rb
CHANGED
metadata
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: pmirror
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.0.
|
4
|
+
version: 0.0.2
|
5
5
|
prerelease:
|
6
6
|
platform: ruby
|
7
7
|
authors:
|
@@ -9,7 +9,7 @@ authors:
|
|
9
9
|
autorequire:
|
10
10
|
bindir: bin
|
11
11
|
cert_chain: []
|
12
|
-
date: 2014-01-
|
12
|
+
date: 2014-01-23 00:00:00.000000000 Z
|
13
13
|
dependencies:
|
14
14
|
- !ruby/object:Gem::Dependency
|
15
15
|
name: bundler
|