From 95644c74e8ec3d99d8273454b252b1cb79309e7e Mon Sep 17 00:00:00 2001
From: Luke Cowell <lcowell@gmail.com>
Date: Tue, 9 Oct 2012 15:05:43 -0700
Subject: [PATCH] add capability to process a file and return a count of
 imported rows

This is useful when importing large data files as #import builds an array with
all results and returns that.
---
 lib/csv-mapper.rb       | 47 +++++++++++++++++++++++++++++------------
 spec/csv-mapper_spec.rb |  7 ++++++
 2 files changed, 41 insertions(+), 13 deletions(-)

diff --git a/lib/csv-mapper.rb b/lib/csv-mapper.rb
index eaf4bb0..85af926 100644
--- a/lib/csv-mapper.rb
+++ b/lib/csv-mapper.rb
@@ -95,28 +95,49 @@ def map_csv(&map_block)
   # <tt>:map</tt>:: Specify an instance of a RowMap to take presidence over a given block defintion.
   #
   def import(data, options={}, &map_block)
-    csv_data = options[:type] == :io ? data : File.new(data, 'r')
-
-    config = { :type => :file_path,
-               :map => map_csv_with_data(csv_data, &map_block) }.merge!(options)
-
-    map = config[:map]
-    
     results = []
-    FasterCSV.new(csv_data, map.parser_options ).each_with_index do |row, i|
-      results << map.parse(row) if i >= map.start_at_row && i <= map.stop_at_row
+    build_results(data, map_block, options) do |result|
+      results << result
     end
-    
     results
-  end  
+  end
+
+  # Process works the same was as import except that it returns a count of
+  # imported rows instead of the results. This is useful for processing large
+  # files.
+  def process(data, options={}, &map_block)
+    count = 0
+    build_results(data, map_block, options) do |result|
+      count = count.succ
+    end
+
+    count
+  end
 
   protected
   # Create a new RowMap instance from the definition in the given block and pass the csv_data.
   def map_csv_with_data(csv_data, &map_block) # :nodoc:
     CsvMapper::RowMap.new(self, csv_data, &map_block)
   end
-  
+
+  private
+  # Process each row and yield result to caller
+  def build_results(data, map_block, options={})
+    csv_data = options[:type] == :io ? data : File.new(data, 'r')
+
+    config = { :type => :file_path,
+               :map => map_csv_with_data(csv_data, &map_block) }.merge!(options)
+
+    map = config[:map]
+
+    FasterCSV.new(csv_data, map.parser_options ).each_with_index do |row, i|
+      if i >= map.start_at_row && i <= map.stop_at_row
+        yield map.parse(row)
+      end
+    end
+  end
+
   extend self
 end
 
-require 'csv-mapper/row_map'
\ No newline at end of file
+require 'csv-mapper/row_map'
diff --git a/spec/csv-mapper_spec.rb b/spec/csv-mapper_spec.rb
index e12628c..97689ca 100644
--- a/spec/csv-mapper_spec.rb
+++ b/spec/csv-mapper_spec.rb
@@ -89,6 +89,13 @@ def upcase_name(row)
       
       results[0].first_name.should == 'JOHN'
     end
+
+    it "should allow us to not capture the output and return a row count instead" do
+      io = "foo,bar,00,01\na,b,c,d"
+      results = @mapped.process(io, :type => :io, :capture_output => false)
+
+      results.should == 2
+    end
   end
   
   describe "extended" do