addcrawlobjects

app/models/crawlobject.rb

@@ -0,0 +1,58 @@
+class Crawlobject < ActiveRecord::Base
+  attr_accessible :children_count, :crawltime, :crawlurl, :depth, :lft, :name, :parent_id, :published_at, :raw, :referenced, :rgt, :schematype, :text, :type, :url
+ acts_as_nested_set
+
+  belongs_to :something, :polymorphic=>true
+  def parse_children
+    if self.objtype == 1 # ET Forum Article loaded
+      self.json["comments"].each do |com|
+        if self.children.where(:objhash=>Digest::SHA512.hexdigest(com.to_json)).empty?
+          cco = self.children.new(:raw=>com.to_json,:crawlurl=>self.crawlurl)
+          cco.objtype=2
+          cco.parse_object
+          cco.calc_hash
+          
+          if self.children.where(:objhash2=>cco.objhash2).empty?
+            cco.save
+          else
+            cco=self.children.where(:objhash2=>cco.objhash2).first
+            cco.raw=com.to_json
+            cco.parse_object
+            cco.calc_hash
+            cco.save
+          end
+        end
+      end
+    end
+  end
+  def parse_object
+      
+    if self.objtype == 1 # ET Forum Article loaded
+      # ET - Forum 
+      self.schematype = self.json["type"].first
+      self.name = self.json["properties"]["name"].first
+      self.url = self.json["properties"]["url"].first
+      self.published_at = self.json["properties"]["dateCreated"].first
+      
+    end
+    if self.objtype == 2 # ET Forum Comment loaded is part of Article 
+      self.schematype = self.json["type"].first
+      self.url = self.json["properties"]["replyToUrl"].first
+      self.name = self.json["properties"]["name"].try(:first)
+      self.published_at = self.json["properties"]["commentTime"].try(:first)
+      self.text = self.json["properties"]["commentText"].try(:first)
+      
+      
+
+    end
+  end
+  def calc_hash
+    self.objhash=Digest::SHA512.hexdigest(self.raw)
+    self.objhash2=Digest::SHA512.hexdigest(self.url.to_s+self.try(:name).to_s+self.schematype.to_s+self.published_at.utc.to_s)
+    
+  end
+  def json
+    JSON.parse(self.raw)
+  end
+
+end

app/models/lva.rb

@@ -46,7 +46,7 @@ class Lva < ActiveRecord::Base
   has_many :beispiele , :class_name => "Beispiel"
   has_and_belongs_to_many :lecturers
   has_many :nlinks, as: :link
-  
+  has_many :crawlobjects, :as=>:something
 #  scope :search, ->(query) {where("name like ? or lvas.desc like ?", "%#{query}%", "%#{query}%")} 
   
   validates :lvanr,:format=>{ :with => /^[0-9][0-9][0-9]\.[0-9A][0-9][0-9]$/}, :presence=>true, :uniqueness=>true # , :uniqueness=>true # LVA-Nummer muss das Format 000.000 besitzen (uniqueness?) oder 000 für nicht 
@@ -234,5 +234,33 @@ class Lva < ActiveRecord::Base
     newlvas
     
   end
+
+
+  def read_et_forum
+    lva=self
+    url=lva.forumlink
+    ans = JSON.parse(`python ../microdata/downloadlogin.py #{url}`)
+    ans.each do |a|
+      if Crawlobject.where(:objhash=>Digest::SHA512.hexdigest(a.to_json), :objtype=>1).count ==0
+        aa = Crawlobject.new(:raw=>a.to_json)
+        aa.objtype=1
+        aa.parse_object
+        aa.calc_hash
+        aa.something=lva
+        if Crawlobject.where(:objhash2=>aa.objhash2, :objtype=>1).count==0
+            aa.save
+        else
+          aa=Crawlobject.where(:objhash2=>aa.objhash2).first
+          aa.raw=a.to_json
+          aa.parse_object
+          aa.calc_hash
+          aa.save
+        end
+      end
+    end
+  end
+
+
  
 end
+