Data commit

Task/Inverted-index/Ruby/inverted-index-1.rb

@@ -0,0 +1,29 @@
+if File.exist? "index.dat"
+  @data = Marshal.load open("index.dat")
+else
+  @data = {}
+end
+
+# Let's give the string class the ability to tokenize itsself into lowercase
+# words with no punctuation.
+class String
+  def index_sanitize
+    self.split.collect do |token|
+      token.downcase.gsub(/\W/, '')
+    end
+  end
+end
+
+# Just implementing a simple inverted index here.
+ARGV.each do |filename|
+  open filename do |file|
+    file.read.index_sanitize.each do |word|
+      @data[word] ||= []
+      @data[word] << filename unless @data[word].include? filename
+    end
+  end
+end
+
+open("index.dat", "w") do |index|
+  index.write Marshal.dump(@data)
+end

Task/Inverted-index/Ruby/inverted-index-2.rb

@@ -0,0 +1,22 @@
+if File.exist? "index.dat"
+  @data = Marshal.load open("index.dat")
+else
+  raise "The index data file could not be located."
+end
+
+class String
+  def index_sanitize
+    self.split.collect do |token|
+      token.downcase.gsub(/\W/, '')
+    end
+  end
+end
+
+# Take anything passed in on the command line in any form and break it
+# down the same way we did when making the index.
+ARGV.join(' ').index_sanitize.each do |word|
+  @result ||= @data[word]
+  @result &= @data[word]
+end
+
+p @result