dansguardian 0.2.3 → 0.4.0

data/Changelog

@@ -1,3 +1,11 @@
+0.4.0 (2011-02-20)
+
+* filtergroup: :weightedphrasemode (may override the main config) 
+
+0.3.0 (2011-02-17)
+
+* lists: file encoding support
+
 0.2.3 (2011-02-16)
 
 * slight doc fixes

data/README.rdoc

@@ -21,14 +21,18 @@ the same author.
 
   dgconf.filtergroup(1, :cached => true) 
 
-	listfile = dgconf.filtergroup(1)[:weightedphraselist]
-	listobject = DansGuardian::List.new(:file => listfile)
-	pp listobject
-
-	listfile2 = "/etc/dansguardian/lists/phraselists/goodphrases/weighted_news"
-	listobject2 = DansGuardian::List.new(listfile2)
-	listobject2.read!
-	pp listobject2
+  listfile = dgconf.filtergroup(1)[:weightedphraselist]
+  listobject = DansGuardian::List.new(:file => listfile)
+  pp listobject
+
+  listfile2 = "/etc/dansguardian/lists/phraselists/goodphrases/weighted_news"
+  listobject2 = DansGuardian::List.new(
+      :file           => listfile2,
+      :file_encoding  => Encoding.find('ISO-8859-5')
+  )
+
+  listobject2.read!
+  pp listobject2
 
 == License
 

data/example.rb

@@ -26,8 +26,11 @@ listfile = dgconf.filtergroup(1)[:weightedphraselist]
 listobject = DansGuardian::List.new(:file => listfile)
 pp listobject
 
-listfile2 = "/etc/dansguardian/lists/phraselists/goodphrases/weighted_news"
-listobject2 = DansGuardian::List.new(listfile2)
+listfile2 = "/etc/dansguardian/lists/phraselists/pornography/weighted_russian"
+listobject2 = DansGuardian::List.new(
+  :file           => listfile2,
+  :file_encoding  => Encoding.find('ISO-8859-5')
+)
 listobject2.read!
 pp listobject2
 

data/lib/dansguardian.rb

@@ -5,7 +5,7 @@ autoload :Set,    'set'
 require 'configfiles'
 
 module DansGuardian
-  VERSION = '0.2.3'
+  VERSION = '0.4.0'
 
   autoload :Config,     'dansguardian/config'
   autoload :Parser,     'dansguardian/parser'

data/lib/dansguardian/config/filtergroup.rb

@@ -58,6 +58,12 @@ module DansGuardian
         parameter name 
       end
 
+      # may override the main config file
+      parameter :weightedphrasemode,
+        '0'   => false,
+        '1'   => :normal,
+        '2'   => :singular
+
       parameter :naughtynesslimit, :to_i
 
       virtual   :naughtyness do |confdata|

data/lib/dansguardian/list.rb

@@ -7,25 +7,26 @@ module DansGuardian
     #   DansGuardian::List.new(:file = '/path/to/list')
     def initialize(h={})
       if h.is_a? String
-        @init         = {:file => h}
+        @init           = {:file => h}
+        @file_encoding  = Encoding::BINARY
       else
-        @init         = h      
+        @init           = h      
       end
-      @items        = []
-      @includes     = []
-      @listcategory = nil
-
+      @items          = []
+      @includes       = []
+      @listcategory   = nil
       #read! if @init[:file] 
     end
 
     def filename; @init[:file]; end
+    def file_encoding; @init[:file_encoding]; end
 
     # Reads the file and fill @items ad @includes . 
     # This method might be overridden for non-trivial list types (?)
     def read!
       File.foreach(@init[:file]) do |line|
+        line.force_encoding file_encoding
         line.strip!
-        line.force_encoding Encoding::BINARY
         # Special comment used to "categorize" DG message pages
         if line =~ /^#listcategory:\s*"(.*)"/ 
           @listcategory = $1
@@ -41,5 +42,27 @@ module DansGuardian
       end
     end
 
+    def each_line(opts={:exclude => []}) 
+      Enumerator.new do |yielder|
+        File.foreach(@init[:file]) do |line|
+          line.strip!
+          line.force_encoding file_encoding
+          case line
+          when /^#listcategory:\s*"(.*)"/
+            next if opts[:exclude].include? :listcategory
+          when /^[\s#]*\.Include/
+            next if opts[:exclude].include? :includes
+          when /^[\s#]*</
+            next if opts[:exclude].include? :items
+          when /^\s*#/
+            next if opts[:exclude].include? :comment_lines
+          when /^\s*$/
+            next if opts[:exclude].include? :blank_lines
+          end
+          yielder.yield line 
+        end
+      end
+    end
+
   end
 end

metadata

@@ -4,9 +4,9 @@ version: !ruby/object:Gem::Version
   prerelease: false
   segments: 
   - 0
-  - 2
-  - 3
-  version: 0.2.3
+  - 4
+  - 0
+  version: 0.4.0
 platform: ruby
 authors: 
 - Guido De Rosa
@@ -14,7 +14,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-02-16 00:00:00 +00:00
+date: 2011-02-20 00:00:00 +01:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency