hflr 1.1.0 → 1.2.0

data/hflr.gemspec

@@ -2,7 +2,7 @@
 
 Gem::Specification.new do |s|
   s.name = %q{hflr}
-  s.version = "1.1.0"
+  s.version = "1.2.0"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Colin Davis"]

data/lib/hflr/fl_record_file.rb

@@ -26,11 +26,18 @@ class FLRFile
   
   def set_fast
   @fast = !@record_type_labels.is_a?(Hash)                    
-  unless @fast
-    raise "Cannot set fast mode with more than one record type."
+  unless @fast    
+  raise "Cannot set fast mode with more than one record type."
   end
       if @fast
        @width = get_record_width_from_file
+       puts "record width #{@width.to_s}"
+       records_to_take =  50000001 / @width
+       puts "records_to_take: #{records_to_take}"
+       @buffer_size = @width * records_to_take       
+       puts "Buffer size: #{@buffer_size}"
+       @position=0
+       @current_buffer=nil
       end        
   end
   
@@ -41,6 +48,7 @@ class FLRFile
      end
      
       @offsets =offsets_to_read(ranges, @width)
+      puts "Chunks of data #{@offsets.inspect}"
       @ranges = ranges
   end
 
@@ -50,7 +58,11 @@ class FLRFile
   end
 
 def finished?
-  @file.eof? ||  (@fast && @offsets.empty?)
+  if @fast 
+  @offsets.empty? && @current_buffer.nil?
+  else
+    @file.eof?
+  end
 end
 
 def close
@@ -85,10 +97,31 @@ def get_next_known_line_type
 end
 
 def fast_get_next_known_line_type  
-  unless @offsets.nil? ||  @offsets.empty?
-    @file.pos = @offsets.shift
-    @file.read(@width)    
+  unless @current_buffer.nil? && (@offsets.nil? ||  @offsets.empty?)
+    if @current_buffer.nil?
+      chunk = @offsets.shift
+      puts "shifting first chunk #{chunk.inspect}"
+    @file.pos =  chunk.pos
+    @current_buffer=@file.read(chunk.width)        
+          record= @current_buffer.slice(@position,@width)
+          puts "Record was from #{@position} width of #{@width}"
+          puts "Getting record: #{record}"
+          @position += @width
+          puts "Position incremented to #{@position}"
+          puts "Current buffer size: #{@current_buffer.size}"
+          @current_buffer = nil if @position >= @current_buffer.size
+          puts "current_buffer changed to #{@current_buffer}"
+          return record
+    else
+      record= @current_buffer.slice(@position,@width)
+      puts "Getting subsequent record: #{record}"
+      @position += @width
+      @current_buffer = nil if @position >= @current_buffer.size
+      return record
+    end
+    
   else    
+  puts "At end of all chunks"
     nil
   end
 end
@@ -151,7 +184,22 @@ end
 private
 
 def offsets_to_read(ranges, width)
-  ranges.map{|r| r.map{|o| o * width}}.flatten.uniq
+  #ranges.map{|r| r.map{|o| o * width}}.flatten.uniq
+  chunk = Struct.new(:pos,:width)
+  chunks = []
+  ranges.each do |range|
+    offsets = range.map{|offset| offset}
+    taken = []
+    
+    while !offsets.empty?
+      taken << offsets.shift
+      if taken.size * width == @buffer_size  || offsets.empty?
+        chunks << chunk.new(taken.first, taken.size * width)
+        taken = []       
+      end # if
+    end # while              
+  end # each
+  chunks   
 end
 
 def get_record_width_from_file  

data/test/flrfile_test.rb

@@ -91,7 +91,7 @@ class FLRFileTest < Test::Unit::TestCase
            
   end
 
-def test_faast_next_record
+def test_fast_next_record
 sample_data_path = File.dirname(__FILE__) 
     
   layout = {:customer=>{

metadata

@@ -2,7 +2,7 @@
 name: hflr
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: 1.1.0
+  version: 1.2.0
 platform: ruby
 authors: 
   - Colin Davis