Squashed 'projects/bad-nlp/name-database/' content from commit 46d2e01a

git-subtree-dir: projects/bad-nlp/name-database
git-subtree-split: 46d2e01a4953b8f54e8625c9111cf5b372d80b49

tasks/importers/census1990.rake

@@ -0,0 +1,59 @@
+
+desc "imports the census 2000 data files and merged the data with the existing data files"
+task "import:census1990" => "import:census1990:run"
+
+namespace "import:census1990" do
+    
+  task :run => [:db, "import:census1990:male", "import:census1990:female", "import:census1990:family"] do
+    @db.write
+  end
+  
+  task :male => :db do
+    
+    open("sources/census-1990/dist.male.first", 'r') do |file|        
+      file.each do |line|
+        fields = line.split(/\s+/)
+        name = fields.first
+        entry = @db.given_names.get name
+        
+        
+        entry.meta[:gender] = case entry.meta[:gender]
+                              when "female", "unisex" then
+                                "unisex"
+                              else
+                                "male"
+                              end
+      end
+    end
+    
+  end
+    
+  task :female => :db do
+    open("sources/census-1990/dist.female.first", 'r') do |file|        
+      file.each do |line|
+        fields = line.split(/\s+/)
+        name = fields.first
+        entry = @db.given_names.get name
+        
+        entry.meta[:gender] = case entry.meta[:gender]
+                              when "male", "unisex" then
+                                "unisex"
+                              else
+                                "female"
+                              end
+      end
+    end
+  end
+      
+  task :family => :db do
+    open("sources/census-1990/dist.all.last", 'r') do |file|        
+      file.each do |line|
+        fields = line.split(/\s+/)
+        name = fields.first
+        @db.family_names.get name
+      end
+    end
+  end
+    
+    
+end

tasks/importers/census2000.rake

@@ -0,0 +1,20 @@
+desc "imports the census 2000 data files and merged the data with the existing data files"
+task "import:census2000" => "import:census2000:run"
+
+namespace "import:census2000" do
+  task :run => :db do
+    
+    
+    open("sources/census-2000/app_c.csv", 'r') do |file|
+      
+      file.gets #skip the header
+      file.each do |line|
+        fields = line.split(",")
+        name = fields.first
+        @db.family_names.get name
+      end
+    end
+    
+    @db.write
+  end
+end