rbitter 0.1.2-java
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gitignore +21 -0
- data/.rspec +2 -0
- data/.travis.yml +15 -0
- data/Gemfile +12 -0
- data/LICENSE.txt +22 -0
- data/README.md +72 -0
- data/Rakefile +8 -0
- data/XMLRPC.md +19 -0
- data/bin/rbitter +20 -0
- data/lib/rbitter.rb +86 -0
- data/lib/rbitter/arcserver.rb +130 -0
- data/lib/rbitter/console.rb +93 -0
- data/lib/rbitter/default/config_json.rb +37 -0
- data/lib/rbitter/dlthread.rb +63 -0
- data/lib/rbitter/env.rb +62 -0
- data/lib/rbitter/libtwitter_connection_override.rb +46 -0
- data/lib/rbitter/records.rb +121 -0
- data/lib/rbitter/records_migrate/.keep +0 -0
- data/lib/rbitter/records_migrate/20150327_add_index.rb +12 -0
- data/lib/rbitter/records_migrate/20150504_add_replyto_column.rb +12 -0
- data/lib/rbitter/streaming.rb +105 -0
- data/lib/rbitter/version.rb +20 -0
- data/lib/rbitter/xmlrpc.rb +4 -0
- data/lib/rbitter/xmlrpcd/base.rb +25 -0
- data/lib/rbitter/xmlrpcd/rpchandles.rb +12 -0
- data/lib/rbitter/xmlrpcd/xmlrpc_auth_server.rb +83 -0
- data/lib/rbitter/xmlrpcd/xmlrpcd.rb +69 -0
- data/rbitter.gemspec +46 -0
- data/spec/config/.keep +0 -0
- data/spec/config/default.json +33 -0
- data/spec/rbitter/arcserver_spec.rb +30 -0
- data/spec/rbitter/console_spec.rb +9 -0
- data/spec/rbitter/default/config_json_spec.rb +3 -0
- data/spec/rbitter/dlthread_spec.rb +8 -0
- data/spec/rbitter/env_spec.rb +62 -0
- data/spec/rbitter/libtwitter_connection_override_spec.rb +8 -0
- data/spec/rbitter/records_spec.rb +13 -0
- data/spec/rbitter/streaming_spec.rb +9 -0
- data/spec/rbitter/version_spec.rb +8 -0
- data/spec/rbitter/xmlrpc_spec.rb +8 -0
- data/spec/rbitter/xmlrpcd/base_spec.rb +29 -0
- data/spec/rbitter/xmlrpcd/rpchandles_spec.rb +10 -0
- data/spec/rbitter/xmlrpcd/xmlrpc_auth_server_spec.rb +8 -0
- data/spec/rbitter/xmlrpcd/xmlrpcd_spec.rb +9 -0
- data/spec/rbitter_spec.rb +42 -0
- data/spec/sample_data/.keep +0 -0
- data/spec/spec_helper.rb +39 -0
- metadata +265 -0
@@ -0,0 +1,37 @@
|
|
1
|
+
module Rbitter
|
2
|
+
DEFAULT_CONFIG_JSON = <<-ENDOFJSON
|
3
|
+
{
|
4
|
+
"twitter": {
|
5
|
+
"consumer_key": "",
|
6
|
+
"consumer_secret": "",
|
7
|
+
"access_token": "",
|
8
|
+
"access_token_secret": ""
|
9
|
+
},
|
10
|
+
"activerecord": "sqlite3",
|
11
|
+
"sqlite3": {
|
12
|
+
"dbfile": "rbitter.sqlite"
|
13
|
+
},
|
14
|
+
"mysql2": {
|
15
|
+
"host": "localhost",
|
16
|
+
"port": 3306,
|
17
|
+
"dbname": "archive",
|
18
|
+
"username": "",
|
19
|
+
"password": ""
|
20
|
+
},
|
21
|
+
"media_downloader": {
|
22
|
+
"large_image": true,
|
23
|
+
"download_dir": "imgs/"
|
24
|
+
},
|
25
|
+
"xmlrpc": {
|
26
|
+
"enable": true,
|
27
|
+
"bind_host": "0.0.0.0",
|
28
|
+
"bind_port": 1400,
|
29
|
+
"auth": {
|
30
|
+
"username": "username",
|
31
|
+
"password": "password"
|
32
|
+
},
|
33
|
+
"handles": ["/path/to/handles"]
|
34
|
+
}
|
35
|
+
}
|
36
|
+
ENDOFJSON
|
37
|
+
end
|
@@ -0,0 +1,63 @@
|
|
1
|
+
# encoding: utf-8
|
2
|
+
|
3
|
+
require "net/http"
|
4
|
+
require "openssl"
|
5
|
+
|
6
|
+
module Rbitter
|
7
|
+
class DLThread
|
8
|
+
def initialize(dlfolder, large_flag)
|
9
|
+
@dest = dlfolder
|
10
|
+
if not File.directory?(dlfolder)
|
11
|
+
warn "[dlthread] Given download location is not available for downloading."
|
12
|
+
warn "[dlthread] Fallback to current directory."
|
13
|
+
@dest = "./"
|
14
|
+
end
|
15
|
+
|
16
|
+
if large_flag.nil?
|
17
|
+
@large_image = false
|
18
|
+
else
|
19
|
+
@large_image = large_flag
|
20
|
+
end
|
21
|
+
|
22
|
+
@pool = Array.new
|
23
|
+
end
|
24
|
+
|
25
|
+
def <<(url_array)
|
26
|
+
download_task = Thread.new {
|
27
|
+
url_array.each { |url|
|
28
|
+
uri = URI.parse(@large_image ? url + ":large" : url)
|
29
|
+
ssl = uri.scheme.downcase == 'https'
|
30
|
+
|
31
|
+
Net::HTTP.start(uri.host, uri.port, :use_ssl => ssl) { |h|
|
32
|
+
req = Net::HTTP::Get.new uri.request_uri
|
33
|
+
h.request(req) { |res|
|
34
|
+
case res
|
35
|
+
when Net::HTTPOK
|
36
|
+
fname = File.basename(url)
|
37
|
+
|
38
|
+
puts "[fetch] remote: #{uri.path} => local: #{fname}"
|
39
|
+
open(File.join(@dest, fname), "wb") { |file|
|
40
|
+
res.read_body { |chunk| file.write(chunk) }
|
41
|
+
}
|
42
|
+
end
|
43
|
+
}
|
44
|
+
}
|
45
|
+
}
|
46
|
+
}
|
47
|
+
|
48
|
+
@pool.push download_task
|
49
|
+
end
|
50
|
+
|
51
|
+
def job_cleanup
|
52
|
+
until @pool.empty?
|
53
|
+
dlthrd = @pool.shift
|
54
|
+
|
55
|
+
if dlthrd.alive?
|
56
|
+
puts "[dlthread] Thread forceful cleaning up [remains: #{@pool.length}]"
|
57
|
+
dlthrd.terminate
|
58
|
+
dlthrd.join
|
59
|
+
end
|
60
|
+
end
|
61
|
+
end
|
62
|
+
end
|
63
|
+
end
|
data/lib/rbitter/env.rb
ADDED
@@ -0,0 +1,62 @@
|
|
1
|
+
# encoding: utf-8
|
2
|
+
|
3
|
+
require "json"
|
4
|
+
|
5
|
+
module Rbitter
|
6
|
+
@@env = Hash.new
|
7
|
+
|
8
|
+
class ConfigFileError < StandardError; end
|
9
|
+
|
10
|
+
def self.[](k)
|
11
|
+
@@env[k]
|
12
|
+
end
|
13
|
+
|
14
|
+
module_function
|
15
|
+
def env
|
16
|
+
@@env
|
17
|
+
end
|
18
|
+
|
19
|
+
def env_reset
|
20
|
+
@@env.clear
|
21
|
+
end
|
22
|
+
|
23
|
+
def env_validate?
|
24
|
+
# TODO: Add validator
|
25
|
+
true
|
26
|
+
end
|
27
|
+
|
28
|
+
def config_initialize json_path=nil
|
29
|
+
env_reset
|
30
|
+
|
31
|
+
unless json_path.nil?
|
32
|
+
begin
|
33
|
+
open(json_path, 'r') { |file|
|
34
|
+
@@env = JSON.parse(file.read)
|
35
|
+
}
|
36
|
+
|
37
|
+
return @@env if env_validate?
|
38
|
+
fail StandardError, "Invalid configuration"
|
39
|
+
rescue => e
|
40
|
+
fail ConfigFileError, "Load Failure (#{json_path}): #{e.to_s}"
|
41
|
+
end
|
42
|
+
end
|
43
|
+
|
44
|
+
# Configuration default location
|
45
|
+
# 1. (current_dir)/config.json
|
46
|
+
# 2. (current_dir)/.rbitter/config.json
|
47
|
+
locations = ["config.json", ".rbitter/config.json"]
|
48
|
+
locations.collect! { |base| File.join(Dir.pwd, base) }
|
49
|
+
|
50
|
+
for location in locations
|
51
|
+
next unless File.file?(location)
|
52
|
+
open(location, 'r') { |file|
|
53
|
+
@@env = JSON.parse(file.read)
|
54
|
+
}
|
55
|
+
break if env_validate?
|
56
|
+
end
|
57
|
+
|
58
|
+
if @@env.empty?
|
59
|
+
fail ConfigFileError, "Can not load any configuration in [#{locations.join(', ')}]"
|
60
|
+
end
|
61
|
+
end
|
62
|
+
end
|
@@ -0,0 +1,46 @@
|
|
1
|
+
# encoding: utf-8
|
2
|
+
|
3
|
+
require 'http/parser'
|
4
|
+
require 'openssl'
|
5
|
+
require 'resolv'
|
6
|
+
|
7
|
+
module Twitter
|
8
|
+
module Streaming
|
9
|
+
class Connection
|
10
|
+
MODIFIED = true
|
11
|
+
attr_reader :tcp_socket_class, :ssl_socket_class
|
12
|
+
|
13
|
+
def initialize(options = {})
|
14
|
+
@tcp_socket_class = options.fetch(:tcp_socket_class) { TCPSocket }
|
15
|
+
@ssl_socket_class = options.fetch(:ssl_socket_class) { OpenSSL::SSL::SSLSocket }
|
16
|
+
end
|
17
|
+
|
18
|
+
def stream(request, response)
|
19
|
+
client_context = OpenSSL::SSL::SSLContext.new
|
20
|
+
client = @tcp_socket_class.new(Resolv.getaddress(request.uri.host), request.uri.port)
|
21
|
+
ssl_client = @ssl_socket_class.new(client, client_context)
|
22
|
+
ssl_client.connect
|
23
|
+
request.stream(ssl_client)
|
24
|
+
|
25
|
+
loop {
|
26
|
+
begin
|
27
|
+
body = ssl_client.read_nonblock(1024) # rubocop:disable AssignmentInCondition, WhileUntilModifier
|
28
|
+
response << body
|
29
|
+
rescue IO::WaitReadable
|
30
|
+
# The reason for setting 90 seconds as a timeout is documented on:
|
31
|
+
# https://dev.twitter.com/streaming/overview/connecting
|
32
|
+
r, w, e = IO.select([ssl_client], [], [], 90)
|
33
|
+
if r.nil?
|
34
|
+
# If timeout occurs
|
35
|
+
ssl_client.close
|
36
|
+
raise Twitter::Error::ServerError.new("Connection stalled")
|
37
|
+
else
|
38
|
+
# If socket is readable
|
39
|
+
retry
|
40
|
+
end
|
41
|
+
end
|
42
|
+
}
|
43
|
+
end
|
44
|
+
end
|
45
|
+
end
|
46
|
+
end
|
@@ -0,0 +1,121 @@
|
|
1
|
+
# encoding: utf-8
|
2
|
+
|
3
|
+
require "active_record"
|
4
|
+
require "date"
|
5
|
+
|
6
|
+
module Rbitter
|
7
|
+
class Record < ActiveRecord::Base
|
8
|
+
end
|
9
|
+
end
|
10
|
+
|
11
|
+
module ARSupport
|
12
|
+
SCHEME_VERSION = 20150504
|
13
|
+
SCHEME = {
|
14
|
+
:marker => :integer, # 0 normal, 1 begin 2 halt
|
15
|
+
:marker_msg => :string,
|
16
|
+
:userid => :integer,
|
17
|
+
:username => :string,
|
18
|
+
:tweetid => :integer,
|
19
|
+
:replyto => :integer,
|
20
|
+
:tweet => :text, # with url unpacked
|
21
|
+
:date => :datetime,
|
22
|
+
:rt_count => :integer,
|
23
|
+
:fav_count => :integer
|
24
|
+
}
|
25
|
+
|
26
|
+
module_function
|
27
|
+
def prepared?
|
28
|
+
ActiveRecord::Base.connection.table_exists?(:records)
|
29
|
+
end
|
30
|
+
|
31
|
+
def connect_database
|
32
|
+
if Rbitter['activerecord'] == 'sqlite3'
|
33
|
+
warn "Warning: If you enable XMLRPC access, using sqlite is not recommended."
|
34
|
+
warn "Warning: Random crash can happen because of concurrency."
|
35
|
+
|
36
|
+
if RUBY_PLATFORM == 'java'
|
37
|
+
require "jdbc/sqlite3"
|
38
|
+
Jdbc::SQLite3.load_driver
|
39
|
+
ActiveRecord::Base.establish_connection(
|
40
|
+
adapter: 'jdbcsqlite3',
|
41
|
+
database: Rbitter['sqlite3']['dbfile'],
|
42
|
+
timeout: 10000) # Long timeout for slow computer
|
43
|
+
else
|
44
|
+
ActiveRecord::Base.establish_connection(
|
45
|
+
adapter: 'sqlite3',
|
46
|
+
database: Rbitter['sqlite3']['dbfile'],
|
47
|
+
timeout: 10000) # Long timeout for slow computer
|
48
|
+
end
|
49
|
+
elsif Rbitter['activerecord'] == 'mysql2'
|
50
|
+
Jdbc::MySQL.load_driver if RUBY_PLATFORM == 'java'
|
51
|
+
|
52
|
+
ActiveRecord::Base.establish_connection(
|
53
|
+
adapter: (RUBY_PLATFORM == 'java' ? 'jdbcmysql' : 'mysql2'),
|
54
|
+
host: Rbitter['mysql2']['host'],
|
55
|
+
port: Rbitter['mysql2']['port'],
|
56
|
+
database: Rbitter['mysql2']['dbname'],
|
57
|
+
username: Rbitter['mysql2']['username'],
|
58
|
+
password: Rbitter['mysql2']['password'],
|
59
|
+
encoding: "utf8mb4",
|
60
|
+
collation: "utf8mb4_unicode_ci")
|
61
|
+
else
|
62
|
+
raise RuntimeException.new("Unknown configuration value. 'activerecord' value should be sqlite3 or mysql2.")
|
63
|
+
end
|
64
|
+
end
|
65
|
+
|
66
|
+
def update_database_scheme
|
67
|
+
current_version = ActiveRecord::Migrator.current_version
|
68
|
+
if current_version < SCHEME_VERSION
|
69
|
+
warn "[records] Your ActiveRecord scheme is outdated."
|
70
|
+
warn "[records] Migrate... #{current_version} => #{SCHEME_VERSION}"
|
71
|
+
ActiveRecord::Migrator.migrate(File.expand_path("../records_migrate", __FILE__), SCHEME_VERSION)
|
72
|
+
end
|
73
|
+
end
|
74
|
+
|
75
|
+
def prepare option_string=""
|
76
|
+
ActiveRecord::Schema.define(version: SCHEME_VERSION) {
|
77
|
+
# MySQL specific option_string:
|
78
|
+
# utf8mb4 -> supporting UTF-8 4-byte characters (i.e. Emoji)
|
79
|
+
create_table(:records, { :options => option_string }) do |t|
|
80
|
+
SCHEME.each_key { |column|
|
81
|
+
case SCHEME[column]
|
82
|
+
when :string
|
83
|
+
t.string column
|
84
|
+
when :integer
|
85
|
+
t.integer column, :limit => 8
|
86
|
+
when :datetime
|
87
|
+
t.datetime column
|
88
|
+
when :text
|
89
|
+
t.text column
|
90
|
+
else
|
91
|
+
puts "Unexpected column type '#{SCHEME[column]}' of #{column}"
|
92
|
+
end
|
93
|
+
}
|
94
|
+
end
|
95
|
+
|
96
|
+
add_index :records, :tweetid
|
97
|
+
}
|
98
|
+
end
|
99
|
+
|
100
|
+
def any_to_datestring(obj)
|
101
|
+
if obj.is_a?(String)
|
102
|
+
# try to parse it
|
103
|
+
DateTime.parse(obj).strftime("%Y-%m-%d %H:%M:%S")
|
104
|
+
elsif obj.is_a?(DateTime) or obj.is_a?(Time)
|
105
|
+
obj.strftime("%Y-%m-%d %H:%M:%S")
|
106
|
+
else
|
107
|
+
raise ArgumentError.new("Can\'t automatically extract DateTime info")
|
108
|
+
end
|
109
|
+
end
|
110
|
+
|
111
|
+
def export_to_csv(csvfile)
|
112
|
+
open(csvfile, 'w') { |f|
|
113
|
+
f.write("marker,marker_msg,userid,username,tweetid,replyto,tweet,date,rt_count,fav_count")
|
114
|
+
f.write("\n")
|
115
|
+
Rbitter::Record.find_each { |t|
|
116
|
+
f.write("#{t.marker},#{t.marker_msg},#{t.userid},#{t.username},#{t.tweetid},")
|
117
|
+
f.write("#{t.replyto},#{t.tweet},#{t.date},#{t.rt_count},#{t.fav_count}\n")
|
118
|
+
}
|
119
|
+
}
|
120
|
+
end
|
121
|
+
end
|
File without changes
|
@@ -0,0 +1,105 @@
|
|
1
|
+
# encoding: utf-8
|
2
|
+
|
3
|
+
require 'twitter'
|
4
|
+
|
5
|
+
module Rbitter
|
6
|
+
class DummyStreamClient
|
7
|
+
def initialize(tokens); end
|
8
|
+
|
9
|
+
def run(&operation_block)
|
10
|
+
internal(&operation_block)
|
11
|
+
end
|
12
|
+
|
13
|
+
private
|
14
|
+
def internal(&operation_block)
|
15
|
+
tweets = [{
|
16
|
+
"tweetid" => 1,
|
17
|
+
"userid" => 1,
|
18
|
+
"replyto" => nil,
|
19
|
+
"tweet" => "test",
|
20
|
+
"rt_count" => 0,
|
21
|
+
"fav_count" => 0,
|
22
|
+
"screen_name" => "twitter",
|
23
|
+
"date" => "2015-01-01 12:11:10",
|
24
|
+
"media_urls" => ["https://pbs.twimg.com/media/CEPWFtgUgAEmbcV.png"],
|
25
|
+
"web_urls" => ["https://www.google.com/"]
|
26
|
+
}]
|
27
|
+
|
28
|
+
tweets.each { |tweet|
|
29
|
+
yield tweet
|
30
|
+
}
|
31
|
+
end
|
32
|
+
end
|
33
|
+
|
34
|
+
class StreamClient
|
35
|
+
def initialize(tokens)
|
36
|
+
@t = Twitter::Streaming::Client.new do |object|
|
37
|
+
object.consumer_key = tokens['consumer_key']
|
38
|
+
object.consumer_secret = tokens['consumer_secret']
|
39
|
+
object.access_token = tokens['access_token']
|
40
|
+
object.access_token_secret = tokens['access_token_secret']
|
41
|
+
end
|
42
|
+
end
|
43
|
+
|
44
|
+
def run(&operation_block)
|
45
|
+
begin
|
46
|
+
internal(&operation_block)
|
47
|
+
rescue EOFError => e
|
48
|
+
puts "Network unreachable. Retry in 3 seconds..."
|
49
|
+
sleep 3
|
50
|
+
retry
|
51
|
+
end
|
52
|
+
end
|
53
|
+
|
54
|
+
private
|
55
|
+
def internal(&operation_block)
|
56
|
+
@t.user do |tweet|
|
57
|
+
if tweet.is_a?(Twitter::Tweet)
|
58
|
+
if tweet.retweet?
|
59
|
+
tweet = tweet.retweeted_tweet
|
60
|
+
end
|
61
|
+
|
62
|
+
text = tweet.full_text.gsub(/(\r\n|\n)/, '')
|
63
|
+
|
64
|
+
# unpack uris and media links
|
65
|
+
media_urls = Array.new
|
66
|
+
web_urls = Array.new
|
67
|
+
|
68
|
+
if tweet.entities?
|
69
|
+
if tweet.media?
|
70
|
+
tweet.media.each { |media|
|
71
|
+
media_urls.push("#{media.media_uri_https}")
|
72
|
+
text.gsub!("#{media.url}", "#{media.display_url}")
|
73
|
+
}
|
74
|
+
end
|
75
|
+
|
76
|
+
text += " "
|
77
|
+
text += media_urls.join(" ")
|
78
|
+
|
79
|
+
if tweet.uris?
|
80
|
+
tweet.uris.each { |uri|
|
81
|
+
web_urls.push("#{uri.expanded_url}")
|
82
|
+
text.gsub!("#{uri.url}", "#{uri.expanded_url}")
|
83
|
+
}
|
84
|
+
end
|
85
|
+
end
|
86
|
+
|
87
|
+
res = {
|
88
|
+
"tweetid" => tweet.id,
|
89
|
+
"userid" => tweet.user.id,
|
90
|
+
"replyto" => tweet.in_reply_to_status_id? ? tweet.in_reply_to_status_id : nil,
|
91
|
+
"tweet" => text,
|
92
|
+
"rt_count" => tweet.retweet_count,
|
93
|
+
"fav_count" => tweet.favorite_count,
|
94
|
+
"screen_name" => tweet.user.screen_name,
|
95
|
+
"date" => tweet.created_at,
|
96
|
+
"media_urls" => media_urls,
|
97
|
+
"web_urls" => web_urls
|
98
|
+
}
|
99
|
+
|
100
|
+
yield res
|
101
|
+
end
|
102
|
+
end
|
103
|
+
end
|
104
|
+
end
|
105
|
+
end
|