Home Explore Help
Register Sign In
J1X
/
huginn
Watch 2
Star 0
Fork 0

Browse Source
Merge pull request #210 from knu/ftpsite_agent

Implement FtpsiteAgent.

Andrew Cantino 10 years ago
parent
136cbf5c56 fc822e6c92
commit
7c98b1acce
2 changed files with 293 additions and 0 deletions
Show Diff Stats
  1. 214 0
      app/models/agents/ftpsite_agent.rb
  2. 79 0
      spec/models/agents/ftpsite_agent_spec.rb

+ 214 - 0
app/models/agents/ftpsite_agent.rb
View File 

@@ -0,0 +1,214 @@
1 
+require 'net/ftp'
2 
+require 'uri'
3 
+require 'time'
4 
+
5 
+module Agents
6 
+  class FtpsiteAgent < Agent
7 
+    cannot_receive_events!
8 
+
9 
+    default_schedule "every_12h"
10 
+
11 
+    description <<-MD
12 
+      The FtpsiteAgent checks a FTP site and creates Events based on newly uploaded files in a directory.
13 
+
14 
+      Specify a `url` that represents a directory of an FTP site to watch, and a list of `patterns` to match against file names.
15 
+
16 
+      Login credentials can be included in `url` if authentication is required.
17 
+
18 
+      Only files with a last modification time later than the `after` value, if specifed, are notified.
19 
+    MD
20 
+
21 
+    event_description <<-MD
22 
+      Events look like this:
23 
+
24 
+          {
25 
+            "url": "ftp://example.org/pub/releases/foo-1.2.tar.gz",
26 
+            "filename": "foo-1.2.tar.gz",
27 
+            "timestamp": "2014-04-10T22:50:00Z"
28 
+          }
29 
+    MD
30 
+
31 
+    def working?
32 
+      event_created_within?(options['expected_update_period_in_days']) && !recent_error_logs?
33 
+    end
34 
+
35 
+    def default_options
36 
+      {
37 
+          'expected_update_period_in_days' => "1",
38 
+          'url' => "ftp://example.org/pub/releases/",
39 
+          'patterns' => [
40 
+            'foo-*.tar.gz',
41 
+          ],
42 
+          'after' => Time.now.iso8601,
43 
+      }
44 
+    end
45 
+
46 
+    def validate_options
47 
+      # Check for required fields
48 
+      begin
49 
+        url = options['url']
50 
+        String === url or raise
51 
+        uri = URI(url)
52 
+        URI::FTP === uri or raise
53 
+        errors.add(:base, "url must end with a slash") unless uri.path.end_with?('/')
54 
+      rescue
55 
+        errors.add(:base, "url must be a valid FTP URL")
56 
+      end
57 
+
58 
+      patterns = options['patterns']
59 
+      case patterns
60 
+      when Array
61 
+        if patterns.empty?
62 
+          errors.add(:base, "patterns must not be empty")
63 
+        end
64 
+      when nil, ''
65 
+        errors.add(:base, "patterns must be specified")
66 
+      else
67 
+        errors.add(:base, "patterns must be an array")
68 
+      end
69 
+
70 
+      # Check for optional fields
71 
+      if (timestamp = options['timestamp']).present?
72 
+        begin
73 
+          Time.parse(timestamp)
74 
+        rescue
75 
+          errors.add(:base, "timestamp cannot be parsed as time")
76 
+        end
77 
+      end
78 
+
79 
+      if options['expected_update_period_in_days'].present?
80 
+        errors.add(:base, "Invalid expected_update_period_in_days format") unless is_positive_integer?(options['expected_update_period_in_days'])
81 
+      end
82 
+    end
83 
+
84 
+    def check
85 
+      saving_entries do |found|
86 
+        each_entry { |filename, mtime|
87 
+          found[filename, mtime]
88 
+        }
89 
+      end
90 
+    end
91 
+
92 
+    def each_entry
93 
+      patterns = options['patterns']
94 
+
95 
+      after =
96 
+        if str = options['after']
97 
+          Time.parse(str)
98 
+        else
99 
+          Time.at(0)
100 
+        end
101 
+
102 
+      open_ftp(base_uri) do |ftp|
103 
+        log "Listing the directory"
104 
+        # Do not use a block style call because we need to call other
105 
+        # commands during iteration.
106 
+        list = ftp.list('-a')
107 
+
108 
+        month2year = {}
109 
+
110 
+        list.each do |line|
111 
+          mon, day, smtn, rest = line.split(' ', 9)[5..-1]
112 
+
113 
+          # Remove symlink target part if any
114 
+          filename = rest[/\A(.+?)(?:\s+->\s|\z)/, 1]
115 
+
116 
+          patterns.any? { |pattern|
117 
+            File.fnmatch?(pattern, filename)
118 
+          } or next
119 
+
120 
+          case smtn
121 
+          when /:/
122 
+            if year = month2year[mon]
123 
+              mtime = Time.parse("#{mon} #{day} #{year} #{smtn} GMT")
124 
+            else
125 
+              log "Getting mtime of #{filename}"
126 
+              mtime = ftp.mtime(filename)
127 
+              month2year[mon] = mtime.year
128 
+            end
129 
+          else
130 
+            # Do not bother calling MDTM for old files.  Losing the
131 
+            # time part only makes a timestamp go backwards, meaning
132 
+            # that it will trigger no new event.
133 
+            mtime = Time.parse("#{mon} #{day} #{smtn} GMT")
134 
+          end
135 
+
136 
+          after < mtime or next
137 
+
138 
+          yield filename, mtime
139 
+        end
140 
+      end
141 
+    end
142 
+
143 
+    def open_ftp(uri)
144 
+      ftp = Net::FTP.new
145 
+
146 
+      log "Connecting to #{uri.host}#{':%d' % uri.port if uri.port != uri.default_port}"
147 
+      ftp.connect(uri.host, uri.port)
148 
+
149 
+      user =
150 
+        if str = uri.user
151 
+          URI.decode(str)
152 
+        else
153 
+          'anonymous'
154 
+        end
155 
+      password =
156 
+        if str = uri.password
157 
+          URI.decode(str)
158 
+        else
159 
+          'anonymous@'
160 
+        end
161 
+      log "Logging in as #{user}"
162 
+      ftp.login(user, password)
163 
+
164 
+      ftp.passive = true
165 
+
166 
+      path = uri.path.chomp('/')
167 
+      log "Changing directory to #{path}"
168 
+      ftp.chdir(path)
169 
+
170 
+      yield ftp
171 
+    ensure
172 
+      log "Closing the connection"
173 
+      ftp.close
174 
+    end
175 
+
176 
+    def base_uri
177 
+      @base_uri ||= URI(options['url'])
178 
+    end
179 
+
180 
+    def saving_entries
181 
+      known_entries = memory['known_entries'] || {}
182 
+      found_entries = {}
183 
+      new_files = []
184 
+
185 
+      yield proc { |filename, mtime|
186 
+        found_entries[filename] = misotime = mtime.utc.iso8601
187 
+        unless (prev = known_entries[filename]) && misotime <= prev
188 
+          new_files << filename
189 
+        end
190 
+      }
191 
+
192 
+      new_files.sort_by { |filename|
193 
+        found_entries[filename]
194 
+      }.each { |filename|
195 
+        create_event :payload => {
196 
+          'url' => (base_uri + filename).to_s,
197 
+          'filename' => filename,
198 
+          'timestamp' => found_entries[filename],
199 
+        }
200 
+      }
201 
+
202 
+      memory['known_entries'] = found_entries
203 
+      save!
204 
+    end
205 
+
206 
+    private
207 
+
208 
+    def is_positive_integer?(value)
209 
+      Integer(value) >= 0
210 
+    rescue
211 
+      false
212 
+    end
213 
+  end
214 
+end

+ 79 - 0
spec/models/agents/ftpsite_agent_spec.rb
View File 

@@ -0,0 +1,79 @@
1 
+require 'spec_helper'
2 
+require 'time'
3 
+
4 
+describe Agents::FtpsiteAgent do
5 
+  describe "checking anonymous FTP" do
6 
+    before do
7 
+      @site = {
8 
+        'expected_update_period_in_days' => 1,
9 
+        'url' => "ftp://ftp.example.org/pub/releases/",
10 
+        'patterns' => ["example-*.tar.gz"],
11 
+      }
12 
+      @checker = Agents::FtpsiteAgent.new(:name => "Example", :options => @site, :keep_events_for => 2)
13 
+      @checker.user = users(:bob)
14 
+      @checker.save!
15 
+      stub(@checker).each_entry.returns { |block|
16 
+        block.call("example-latest.tar.gz", Time.parse("2014-04-01T10:00:01Z"))
17 
+        block.call("example-1.0.tar.gz",    Time.parse("2013-10-01T10:00:00Z"))
18 
+        block.call("example-1.1.tar.gz",    Time.parse("2014-04-01T10:00:00Z"))
19 
+      }
20 
+    end
21 
+
22 
+    describe "#check" do
23 
+      it "should validate the integer fields" do
24 
+        @checker.options['expected_update_period_in_days'] = "nonsense"
25 
+        lambda { @checker.save! }.should raise_error;
26 
+        @checker.options = @site
27 
+      end
28 
+
29 
+      it "should check for changes and save known entries in memory" do
30 
+        lambda { @checker.check }.should change { Event.count }.by(3)
31 
+        @checker.memory['known_entries'].tap { |known_entries|
32 
+          known_entries.size.should == 3
33 
+          known_entries.sort_by(&:last).should == [
34 
+            ["example-1.0.tar.gz",    "2013-10-01T10:00:00Z"],
35 
+            ["example-1.1.tar.gz",    "2014-04-01T10:00:00Z"],
36 
+            ["example-latest.tar.gz", "2014-04-01T10:00:01Z"],
37 
+          ]
38 
+        }
39 
+
40 
+        Event.last(2).first.payload.should == {
41 
+          'url' => 'ftp://ftp.example.org/pub/releases/example-1.1.tar.gz',
42 
+          'filename' => 'example-1.1.tar.gz',
43 
+          'timestamp' => '2014-04-01T10:00:00Z',
44 
+        }
45 
+
46 
+        lambda { @checker.check }.should_not change { Event.count }
47 
+
48 
+        stub(@checker).each_entry.returns { |block|
49 
+          block.call("example-latest.tar.gz", Time.parse("2014-04-02T10:00:01Z"))
50 
+
51 
+          # In the long list format the timestamp may look going
52 
+          # backwards after six months: Oct 01 10:00 -> Oct 01 2013
53 
+          block.call("example-1.0.tar.gz",    Time.parse("2013-10-01T00:00:00Z"))
54 
+
55 
+          block.call("example-1.1.tar.gz",    Time.parse("2014-04-01T10:00:00Z"))
56 
+          block.call("example-1.2.tar.gz",    Time.parse("2014-04-02T10:00:00Z"))
57 
+        }
58 
+        lambda { @checker.check }.should change { Event.count }.by(2)
59 
+        @checker.memory['known_entries'].tap { |known_entries|
60 
+          known_entries.size.should == 4
61 
+          known_entries.sort_by(&:last).should == [
62 
+            ["example-1.0.tar.gz",    "2013-10-01T00:00:00Z"],
63 
+            ["example-1.1.tar.gz",    "2014-04-01T10:00:00Z"],
64 
+            ["example-1.2.tar.gz",    "2014-04-02T10:00:00Z"],
65 
+            ["example-latest.tar.gz", "2014-04-02T10:00:01Z"],
66 
+          ]
67 
+        }
68 
+
69 
+        Event.last(2).first.payload.should == {
70 
+          'url' => 'ftp://ftp.example.org/pub/releases/example-1.2.tar.gz',
71 
+          'filename' => 'example-1.2.tar.gz',
72 
+          'timestamp' => '2014-04-02T10:00:00Z',
73 
+        }
74 
+
75 
+        lambda { @checker.check }.should_not change { Event.count }
76 
+      end
77 
+    end
78 
+  end
79 
+end