Include autogenerated links when calculating the spam score

[rails.git] / lib / osm.rb

diff --git a/lib/osm.rb b/lib/osm.rb
index 554ebfaef7a6cb6e06d13710272fb9380b1a95b5..c4781ca417e7b828b382e42b057f409c6f247809 100644 (file)
--- a/lib/osm.rb
+++ b/lib/osm.rb
@@ -7,6 +7,14 @@ module OSM
   require 'xml/libxml'
   require 'digest/md5'
   require 'RMagick'
+  require 'nokogiri'
+
+  if defined?(SystemTimer)
+    Timer = SystemTimer
+  else
+    require 'timeout'
+    Timer = Timeout
+  end
 
   # The base class for API Errors.
   class APIError < RuntimeError
@@ -439,21 +447,27 @@ module OSM
       doc = XML::Document.new
       doc.encoding = XML::Encoding::UTF_8
       root = XML::Node.new 'osm'
-      root['version'] = API_VERSION
+      root['version'] = API_VERSION.to_s
       root['generator'] = GENERATOR
       doc.root = root
       return doc
     end
   end
 
-  def self.IPLocation(ip_address)
-    Timeout::timeout(4) do
-      Net::HTTP.start('api.hostip.info') do |http|
-        country = http.get("/country.php?ip=#{ip_address}").body
-        country = "GB" if country == "UK"
-        country = Country.find_by_code(country)
-        return { :minlon => country.min_lon, :minlat => country.min_lat, :maxlon => country.max_lon, :maxlat => country.max_lat }
+  def self.IPToCountry(ip_address)
+    Timer.timeout(4) do
+      ipinfo = Quova::IpInfo.new(ip_address)
+
+      if ipinfo.status == Quova::Success then
+        country = ipinfo.country_code
+      else
+        Net::HTTP.start('api.hostip.info') do |http|
+          country = http.get("/country.php?ip=#{ip_address}").body
+          country = "GB" if country == "UK"
+        end
       end
+      
+      return country.upcase
     end
 
     return nil
@@ -461,6 +475,16 @@ module OSM
     return nil
   end
 
+  def self.IPLocation(ip_address)
+    code = OSM.IPToCountry(ip_address)
+
+    if code and country = Country.find_by_code(code)
+      return { :minlon => country.min_lon, :minlat => country.min_lat, :maxlon => country.max_lon, :maxlat => country.max_lat }
+    end
+
+    return nil
+  end
+
   # Construct a random token of a given length
   def self.make_token(length = 30)
     chars = 'abcdefghijklmnopqrtuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789'
@@ -480,15 +504,38 @@ module OSM
   end
 
   # Return an SQL fragment to select a given area of the globe
-  def self.sql_for_area(minlat, minlon, maxlat, maxlon, prefix = nil)
-    tilesql = QuadTile.sql_for_area(minlat, minlon, maxlat, maxlon, prefix)
-    minlat = (minlat * 10000000).round
-    minlon = (minlon * 10000000).round
-    maxlat = (maxlat * 10000000).round
-    maxlon = (maxlon * 10000000).round
-
-    return "#{tilesql} AND #{prefix}latitude BETWEEN #{minlat} AND #{maxlat} AND #{prefix}longitude BETWEEN #{minlon} AND #{maxlon}"
+  def self.sql_for_area(bbox, prefix = nil)
+    tilesql = QuadTile.sql_for_area(bbox, prefix)
+    bbox = bbox.to_scaled
+
+    return "#{tilesql} AND #{prefix}latitude BETWEEN #{bbox.min_lat} AND #{bbox.max_lat} " +
+                      "AND #{prefix}longitude BETWEEN #{bbox.min_lon} AND #{bbox.max_lon}"
   end
 
+  # Return a spam score for a chunk of text
+  def self.spam_score(text)
+    link_count = 0
+    link_size = 0
+
+    doc = Nokogiri::HTML(Rinku.auto_link(text, :urls))
 
+    if doc.content.length > 0
+      doc.xpath("//a").each do |link|
+        link_count += 1
+        link_size += link.content.length
+      end
+
+      link_proportion = link_size.to_f / doc.content.length.to_f
+    else
+      link_proportion = 0
+    end
+
+    return [link_proportion - 0.2, 0.0].max * 200 + link_count * 20
+  end
+
+  def self.legal_text_for_country(country_code)
+    file_name = File.join(Rails.root, "config", "legales", country_code.to_s + ".yml")
+    file_name = File.join(Rails.root, "config", "legales", DEFAULT_LEGALE + ".yml") unless File.exist? file_name
+    YAML::load_file(file_name)
+  end
 end