← Back to branch summary

~ubuntu-branches/ubuntu/saucy/tdiary/saucy-proposed

« back to all changes in this revision

Viewing changes to contrib2/plugin/antirefspam/filter/antirefspam.rb

  • Committer: Bazaar Package Importer
  • Author(s): Daigo Moriwaki
  • Date: 2011-04-11 21:53:16 UTC
  • mfrom: (1.2.3 upstream)
  • Revision ID: james.westby@ubuntu.com-20110411215316-ih4gt4q8p29d2wf8
Tags: 3.0.1-1
http://bugs.debian.org/542801
http://bugs.debian.org/594947
* New upstream release (Closes: #542801, #594947)
* debian/control:
 - Bumped up Standards-Version to 3.9.1.
 - Updated version dependency.
* debian/tdiary-setup.rb: Followed the upstream changes, incorporating js and
  index.fcgi

Show diffs side-by-side

added added

removed removed

Lines of Context:
contrib2/plugin/antirefspam/filter/antirefspam.rb
1
 
#
2
 
# antirefspamfilter.rb
3
 
#
4
 
# Copyright (c) 2004-2005 T.Shimomura <redbug@netlife.gr.jp>
5
 
# You can redistribute it and/or modify it under GPL2.
6
 
# Please use version 1.0.0 (not 1.0.0G) if GPL doesn't want to be forced on me.
7
 
#
8
 
 
9
 
require 'net/http'
10
 
require 'uri'
11
 
 
12
 
module TDiary
13
 
  module Filter
14
 
 
15
 
    class AntirefspamFilter < Filter
16
 
      # ͭ���ˤ���Ȼ��ꤷ���ե�����˥ǥХå�����ʸ������ɵ�����
17
 
      def debug_out(filename, str)
18
 
        if $debug
19
 
          filename = File.join(@conf.data_path,"AntiRefSpamFilter",filename)
20
 
          File::open(filename, "a+") {|f|
21
 
            f.puts str
22
 
          }
23
 
        end
24
 
      end
25
 
 
26
 
      # str �˻��ꤵ�줿ʸ����Ŭ�ڤʥ�����ޤ�Ǥ��뤫������å�
27
 
      def isIncludeMyUrl(str)
28
 
        # str ��������URL���ޤޤ�Ƥ��뤫�ɤ���
29
 
        base_url = @conf.base_url
30
 
        unless base_url.empty?
31
 
          if str.include? base_url
32
 
            return true
33
 
          end
34
 
        end
35
 
 
36
 
        # str �˥ȥåץڡ���URL���ޤޤ�Ƥ��뤫�ɤ���
37
 
        unless @conf.index_page.empty?
38
 
          if /\Ahttps?:\/\// =~ @conf.index_page
39
 
            if str.include? @conf.index_page
40
 
              return true
41
 
            end
42
 
          end
43
 
        end
44
 
 
45
 
        # str �˵��Ƥ������褬�ޤޤ�Ƥ��뤫�ɤ���
46
 
        if (myurl = @conf['antirefspam.myurl']) && !myurl.empty?
47
 
          if str.include? myurl
48
 
            return true
49
 
          end
50
 
          
51
 
          url = myurl.gsub("/", "\\/").gsub(":", "\\:")
52
 
          exp = Regexp.new(url)
53
 
          if exp =~ str
54
 
            return true
55
 
          end
56
 
        end
57
 
 
58
 
        return false
59
 
      end
60
 
 
61
 
      def referer_filter(referer)
62
 
        conf_disable = @conf['antirefspam.disable'] != nil ? @conf['antirefspam.disable'].to_s : ''
63
 
        conf_checkreftable = @conf['antirefspam.checkreftable'] != nil ? @conf['antirefspam.checkreftable'].to_s : ''
64
 
        conf_trustedurl = @conf['antirefspam.trustedurl'] != nil ? @conf['antirefspam.trustedurl'].to_s : ''
65
 
        conf_proxy_server = @conf['antirefspam.proxy_server'] != nil && @conf['antirefspam.proxy_server'].size > 0 ? @conf['antirefspam.proxy_server'].to_s : nil
66
 
        conf_proxy_port = @conf['antirefspam.proxy_port'] != nil && @conf['antirefspam.proxy_port'].size > 0 ? @conf['antirefspam.proxy_port'].to_s : nil
67
 
 
68
 
        if conf_disable == 'true'  or    # ��󥯸������å���ͭ���ǤϤʤ����ϥ��롼����
69
 
           referer == nil          or    # ��󥯸���̵��
70
 
           referer.size <= 1       or    # �����Υ���ƥʤǹ������郎���ʤ��ʤ�������б����뤿�ᡢ��󥯸�����ʸ������ξ��ϵ���
71
 
           isIncludeMyUrl(referer)       # ��ʬ�������⤫��Υ�󥯤Ͽ��ꤹ��
72
 
        then
73
 
          return true
74
 
        end
75
 
 
76
 
        # "����Ǥ���URL" �򣱤Ĥ��ļ��Ф���referer�ȹ��פ��뤫�����å�����
77
 
        conf_trustedurl.each_line do |trusted|
78
 
          trusted.sub!(/\r?\n/,'')
79
 
          next if trusted =~ /\A(\#|\s*)\z/  # #�ޤ��϶���ǻϤޤ�Ԥ��ɤ����Ф�
80
 
          
81
 
          # �ޤ��� "����Ǥ��� URL" �� referer �˴ޤޤ�뤫�ɤ���
82
 
          if referer.include? trusted
83
 
            debug_out("trusted", trusted+" (include?) "+referer)
84
 
            return true
85
 
          end
86
 
          
87
 
          # �ޤޤ�ʤ��ä����� "����Ǥ��� URL" ������ɽ���Ȥߤʤ��ƺƥ����å�
88
 
          begin
89
 
            if referer =~ Regexp.new( trusted.gsub("/", "\\/").gsub(":", "\\:") )
90
 
              debug_out("trusted", trusted+" (=~) "+referer)
91
 
              return true
92
 
            end
93
 
          rescue
94
 
            debug_out("error_config", "trustedurl: "+trusted)
95
 
          end
96
 
        end
97
 
 
98
 
        # URL�ִ��ꥹ�Ȥ򸫤�
99
 
        if conf_checkreftable == 'true'
100
 
          # "URL�ִ��ꥹ��" �򣱤Ĥ��ļ��Ф���referer�ȹ��פ��뤫�����å�����
101
 
          @conf.referer_table.each do |url, name|
102
 
            begin
103
 
              if /#{url}/i =~ referer && url != '^(.{50}).*$'
104
 
                debug_out("trusted", url+" (=~referer_table)  "+referer)
105
 
                return true
106
 
              end
107
 
            rescue
108
 
              debug_out("error_config", "referer_table: "+url)
109
 
            end
110
 
          end
111
 
        end
112
 
 
113
 
        @work_path = File.join(@conf.data_path,"AntiRefSpamFilter")
114
 
        @spamurl_list = File.join(@work_path,"spamurls")  # referer spam �Υ�󥯸�����
115
 
        @spamip_list  = File.join(@work_path,"spamips")   # referer spam ��IP����
116
 
        @safeurl_list = File.join(@work_path,"safeurls")  # �����餯������Τʤ���󥯸�����
117
 
 
118
 
        # �ǥ��쥯�ȥ�/�ե����뤬¸�ߤ��ʤ���к��
119
 
        unless File.exist? @work_path
120
 
          Dir::mkdir(@work_path)
121
 
        end
122
 
        unless File.exist? @spamurl_list
123
 
          File::open(@spamurl_list, "a").close
124
 
        end
125
 
        unless File.exist? @safeurl_list
126
 
          File::open(@safeurl_list, "a").close
127
 
        end
128
 
 
129
 
        uri = URI.parse(referer)
130
 
        temp_filename = File.join(@work_path,uri.host)
131
 
        # �����å����ˤ��оݤΥɥᥤ��̾����ä�����ե��������
132
 
        begin
133
 
          File::open(temp_filename, File::RDONLY | File::CREAT | File::EXCL).close
134
 
 
135
 
          # ���� SPAM URL �Ȥߤʤ�����ʸ�ϰʸ�ϵ���
136
 
          spamurls = IO::readlines(@spamurl_list).map {|url| url.chomp }
137
 
          if spamurls.include? referer
138
 
            return false
139
 
          end
140
 
 
141
 
          # ���� SPAM URL �Ǥʤ���Ƚ�Ǥ�����ʸ�ϵ���
142
 
          safeurls = IO::readlines(@safeurl_list).map {|url| url.chomp }
143
 
          if safeurls.include? referer
144
 
            return true
145
 
          end
146
 
 
147
 
          # ��󥯸� URL �� HTML �����ĥ�äƤ���
148
 
          Net::HTTP.version_1_2   # ���ޤ��ʤ��餷��
149
 
          body = ""
150
 
          begin
151
 
            Net::HTTP::Proxy(conf_proxy_server, conf_proxy_port).start(uri.host, uri.port) do |http|
152
 
              if uri.path == ""
153
 
                response, = http.get("/")
154
 
              else
155
 
                response, = http.get(uri.request_uri)
156
 
              end
157
 
              body = response.body
158
 
            end
159
 
 
160
 
            # body �������� URL ���ޤޤ�Ƥ��ʤ���� SPAM �Ȥߤʤ�
161
 
            unless isIncludeMyUrl(body)
162
 
              File::open(@spamurl_list, "a+") {|f|
163
 
                f.puts referer
164
 
              }
165
 
              File::open(@spamip_list, "a+") {|f|
166
 
                f.puts [@cgi.remote_addr, Time.now.utc.strftime("%Y/%m/%d %H:%M:%S UTC")].join("\t")
167
 
              }
168
 
              return false
169
 
            else
170
 
              File::open(@safeurl_list, "a+") {|f|
171
 
                f.puts referer
172
 
              }
173
 
            end
174
 
          rescue
175
 
            # ���顼���Ф����� @spamurl_list ������ʤ�����󥯸��ˤ�����ʤ�
176
 
            return false
177
 
          end
178
 
 
179
 
        rescue StandardError, TimeoutError
180
 
          # ���ߥ����å���ʤ顢����ϥ�󥯸��˴��ꤷ�ʤ�
181
 
          return false
182
 
        ensure
183
 
          begin
184
 
            File::delete(temp_filename)
185
 
          rescue
186
 
          end
187
 
        end
188
 
 
189
 
        return true
190
 
      end
191
 
 
192
 
 
193
 
 
194
 
      def log_spamcomment( diary, comment )
195
 
        @work_path = File.join(@conf.data_path,"AntiRefSpamFilter")
196
 
        @spamcomment_list = File.join(@work_path,"spamcomments")  # comment spam �ΰ���
197
 
 
198
 
        # �ǥ��쥯�ȥ�/�ե����뤬¸�ߤ��ʤ���к��
199
 
        unless File.exist? @work_path
200
 
          Dir::mkdir(@work_path)
201
 
        end
202
 
        unless File.exist? @spamcomment_list
203
 
          File::open(@spamcomment_list, "a").close
204
 
        end
205
 
 
206
 
        File::open(@spamcomment_list, "a+") {|f|
207
 
          f.puts "From: "+comment.name+" <"+comment.mail+">"
208
 
          f.puts "To: "+diary.date.to_s
209
 
          f.puts "Date: "+comment.date.to_s
210
 
          f.puts comment.body
211
 
          f.puts ".\n\n"
212
 
        }
213
 
      end
214
 
 
215
 
      def comment_filter( diary, comment )
216
 
        # �ĥå��ߤ����ܸ�(�Ҥ餬��/��������)���ޤޤ�Ƥ��ʤ�����Ե���
217
 
        if @conf['antirefspam.comment_kanaonly'] != nil
218
 
          if @conf['antirefspam.comment_kanaonly'].to_s == 'true'
219
 
            unless comment.body =~ /[��-��-����]/
220
 
              log_spamcomment( diary, comment )
221
 
              return false
222
 
            end
223
 
          end
224
 
        end
225
 
 
226
 
        # �ĥå��ߤ�ʸ���������ꤷ����°���Ǥʤ��ʤ��Ե���
227
 
        maxsize = @conf['antirefspam.comment_maxsize'].to_i
228
 
        if maxsize > 0
229
 
          unless comment.body.size <= maxsize
230
 
            log_spamcomment( comment )
231
 
            return false
232
 
          end
233
 
        end
234
 
 
235
 
        # NG��ɤ����ĤǤ�ޤޤ�Ƥ������Ե���
236
 
        if @conf['antirefspam.comment_ngwords'] != nil
237
 
          ngwords = @conf['antirefspam.comment_ngwords']
238
 
          ngwords.to_s.each_line do |ngword|
239
 
            ngword.sub!(/\r?\n/,'')
240
 
            if comment.body.downcase.include? ngword.downcase
241
 
              log_spamcomment( comment )
242
 
              return false
243
 
            end
244
 
 
245
 
            # �ޤޤ�ʤ��ä����� "NG���" ������ɽ���Ȥߤʤ��ƺƥ����å�
246
 
            begin
247
 
              if comment.body =~ Regexp.new( ngword, Regexp::MULTILINE )
248
 
                log_spamcomment( comment )
249
 
                return false
250
 
              end
251
 
            rescue
252
 
              debug_out("error_config", "comment_ngwords: "+ngword)
253
 
            end
254
 
          end
255
 
        end
256
 
 
257
 
        return true
258
 
      end
259
 
    end
260
 
  end
261
 
end