datashift 0.1.0 → 0.2.1

data/tasks/import/excel.rake

@@ -1,72 +1,72 @@
-# Copyright:: (c) Autotelik Media Ltd 2011
-# Author ::   Tom Statter
-# Date ::     Feb 2011
-# License::   TBD. Free, Open Source. MIT ?
-#
-# REQUIRES:   JRuby
-#
-# Usage::
-#
-#  In Rakefile:
-#
-#     require 'datashift'
-#
-#     DataShift::load_tasks
-#
-#  Cmd Line:
-#
-# => jruby -S rake datashift:import:excel model=<active record class> input=<file.xls>
-# => jruby -S rake datashift:import:excel model=<active record class> input=C:\MyProducts.xlsverbose=true
-#
-require 'datashift'
-require 'excel_loader'
-
-namespace :datashift do
-
-  namespace :import do
-
-    desc "Populate model's table with data from .xls (Excel) file"
-  
-    task :excel, [:model, :loader, :input, :config, :verbose] => [:environment] do |t, args|
-
-      # in familiar ruby style args seems to have been become empty using this new style for rake 0.9.2
-      #  whatever format i try, on both Win and OSX .. so had to revert back to ENV
-      model = ENV['model']
-      input = ENV['input']
-    
-      raise "USAGE: jruby -S rake datashift:import:excel input=excel_file.xls model=<Class>" unless(input)
-      raise "ERROR: Cannot process without AR Model - please supply model=<Class>" unless(model)
-      raise "ERROR: Could not find file #{args[:input]}" unless File.exists?(input)
-
-      begin
-        klass = Kernel.const_get(model)
-      rescue NameError
-        raise "ERROR: No such AR Model found - check valid model supplied via model=<Class>"
-      end
-
-      if(ENV['loader'])
-        begin
-          loader_klass = Kernel.const_get(ENV['loader'])
-
-          loader = loader_klass.new(klass)
-
-          puts "INFO: Using loader : #{loader.class}"
-        rescue
-          puts "INFO: No specific #{model}Loader found  - using generic ExcelLoader"
-          loader = DataShift::ExcelLoader.new(klass)
-        end
-      else
-        puts "INFO: No Loader specified - using generic ExcelLoader"
-        loader = DataShift::ExcelLoader.new(klass)
-      end
-
-      puts "ARGS #{args.inspect} #{args[:verbose]} [#{args.verbose}]"
-      loader.logger.verbose if(ENV['verbose'])
-      
-      loader.configure_from( ENV['config'] ) if(ENV['config'])
-       
-      loader.perform_load(input)
-    end
-
-  end
+# Copyright:: (c) Autotelik Media Ltd 2011
+# Author ::   Tom Statter
+# Date ::     Feb 2011
+# License::   TBD. Free, Open Source. MIT ?
+#
+# REQUIRES:   JRuby
+#
+# Usage::
+#
+#  In Rakefile:
+#
+#     require 'datashift'
+#
+#     DataShift::load_tasks
+#
+#  Cmd Line:
+#
+# => jruby -S rake datashift:import:excel model=<active record class> input=<file.xls>
+# => jruby -S rake datashift:import:excel model=<active record class> input=C:\MyProducts.xlsverbose=true
+#
+require 'datashift'
+require 'excel_loader'
+
+namespace :datashift do
+
+  namespace :import do
+
+    desc "Populate model's table with data from .xls (Excel) file"
+  
+    task :excel, [:model, :loader, :input, :config, :verbose] => [:environment] do |t, args|
+
+      # in familiar ruby style args seems to have been become empty using this new style for rake 0.9.2
+      #  whatever format i try, on both Win and OSX .. so had to revert back to ENV
+      model = ENV['model']
+      input = ENV['input']
+    
+      raise "USAGE: jruby -S rake datashift:import:excel input=excel_file.xls model=<Class>" unless(input)
+      raise "ERROR: Cannot process without AR Model - please supply model=<Class>" unless(model)
+      raise "ERROR: Could not find file #{args[:input]}" unless File.exists?(input)
+
+      begin
+        klass = Kernel.const_get(model)
+      rescue NameError
+        raise "ERROR: No such AR Model found - check valid model supplied via model=<Class>"
+      end
+
+      if(ENV['loader'])
+        begin
+          loader_klass = Kernel.const_get(ENV['loader'])
+
+          loader = loader_klass.new(klass)
+
+          puts "INFO: Using loader : #{loader.class}"
+        rescue
+          puts "INFO: No specific #{model}Loader found  - using generic ExcelLoader"
+          loader = DataShift::ExcelLoader.new(klass)
+        end
+      else
+        puts "INFO: No Loader specified - using generic ExcelLoader"
+        loader = DataShift::ExcelLoader.new(klass)
+      end
+
+      puts "ARGS #{args.inspect} #{args[:verbose]} [#{args.verbose}]"
+      loader.logger.verbose if(ENV['verbose'])
+      
+      loader.configure_from( ENV['config'] ) if(ENV['config'])
+       
+      loader.perform_load(input)
+    end
+
+  end
 end

data/tasks/spree/image_load.rake

@@ -1,109 +1,109 @@
-# Copyright:: (c) Autotelik Media Ltd 2011
-# Author ::   Tom Statter
-# Date ::     Feb 2011
-# License::   MIT
-#
-# Usage::
-#
-# => rake datashift:spree:images input=vendor/extensions/site/fixtures/images
-# => rake datashift:spree:images input=C:\images\photos large dummy=true
-#
-# => rake datashift:spree:images input=C:\images\taxon_icons skip_if_no_assoc=true klass=Taxon
-#
-namespace :datashift do
-
-  namespace :spree do
-
-    desc "Populate the DB with images.\nDefault location db/image_seeds, or specify :input=<path> or dir under db/image_seeds with :folder"
-    # :dummy => dummy run without actual saving to DB
-    task :images, [:input, :folder, :dummy, :sku, :skip_if_no_assoc, :skip_if_loaded, :model] => :environment do |t, args|
-
-      require 'image_loader'
-
-      raise "USAGE: Please specify one of :input or :folder" if(args[:input] && args[:folder])
-      puts  "SKU not specified " if(args[:input] && args[:folder])
-
-      if args[:input]
-        @image_cache = args[:input]
-      else
-        @image_cache =  File.join(Rails.root, "db", "image_seeds")
-        @image_cache =  File.join(@image_cache, args[:folder]) if(args[:folder])
-      end
-
-      attachment_klazz = Product
-
-      begin
-        attachment_klazz = Kernel.const_get(args[:model]) if(args[:model])
-      rescue NameError
-        attachment_klazz = Product
-      end
-
-      image_loader = ImageLoader.new
-
-      if(File.exists? @image_cache )
-        puts "Loading images from #{@image_cache}"
-
-        missing_records = []
-        Dir.glob("#{@image_cache}/*.{jpg,png,gif}") do |image_name|
-
-          puts "Processing #{image_name} : #{File.exists?(image_name)}"
-          base_name = File.basename(image_name, '.*')
-
-          record = nil
-          if(attachment_klazz == Product && args[:sku])
-            sku = base_name.slice!(/\w+/)
-            sku.strip!
-            base_name.strip!
-
-            puts "Looking fo SKU #{sku}"
-            record = Variant.find_by_sku(sku)
-            if record
-              record = record.product   # SKU stored on Variant but we want it's master Product
-            else
-              puts "Looking for NAME [#{base_name}]"
-              record = attachment_klazz.find_by_name(base_name)
-            end
-          else
-            puts "Looking for #{attachment_klazz.name} with NAME [#{base_name}]"
-            record = attachment_klazz.find_by_name(base_name)
-          end
-      
-          if(record)
-            puts "Found record for attachment : #{record.inspect}"
-            exists = record.images.detect {|i| puts "COMPARE #{i.attachment_file_name} => #{image_name}"; i.attachment_file_name == image_name }
-            puts "Found existing attachments [#{exists}]" unless(exists.nil?)
-            if(args[:skip_if_loaded] && !exists.nil?)
-              puts "Skipping - Image #{image_name} already loaded for #{attachment_klazz}"
-              next
-            end
-          else
-            missing_records << image_name
-          end
-
-          # Now do actual upload to DB unless we are doing a dummy run,
-          # or the Image must have an associated record
-          unless(args[:dummy] == 'true' || (args[:skip_if_no_assoc] && record.nil?))
-            image_loader.reset()
-            puts "Process Image"
-            image_loader.process( image_name, record )
-          end
-
-        end
-
-        unless missing_records.empty?
-          FileUtils.mkdir_p('MissingRecords') unless File.directory?('MissingRecords')
-        
-          puts '\nMISSING Records Report>>'
-          missing_records.each do |i|
-            puts "Copy #{i} to MissingRecords folder"
-            FileUtils.cp( i, 'MissingRecords')  unless(args[:dummy] == 'true')
-          end
-        end
-
-      else
-        puts "ERROR: Supplied Path #{@image_cache} not accesible"
-        exit(-1)
-      end
-    end
-  end
+# Copyright:: (c) Autotelik Media Ltd 2011
+# Author ::   Tom Statter
+# Date ::     Feb 2011
+# License::   MIT
+#
+# Usage::
+#
+# => rake datashift:spree:images input=vendor/extensions/site/fixtures/images
+# => rake datashift:spree:images input=C:\images\photos large dummy=true
+#
+# => rake datashift:spree:images input=C:\images\taxon_icons skip_if_no_assoc=true klass=Taxon
+#
+namespace :datashift do
+
+  namespace :spree do
+
+    desc "Populate the DB with images.\nDefault location db/image_seeds, or specify :input=<path> or dir under db/image_seeds with :folder"
+    # :dummy => dummy run without actual saving to DB
+    task :images, [:input, :folder, :dummy, :sku, :skip_if_no_assoc, :skip_if_loaded, :model] => :environment do |t, args|
+
+      require 'image_loader'
+
+      raise "USAGE: Please specify one of :input or :folder" if(args[:input] && args[:folder])
+      puts  "SKU not specified " if(args[:input] && args[:folder])
+
+      if args[:input]
+        @image_cache = args[:input]
+      else
+        @image_cache =  File.join(Rails.root, "db", "image_seeds")
+        @image_cache =  File.join(@image_cache, args[:folder]) if(args[:folder])
+      end
+
+      attachment_klazz = Product
+
+      begin
+        attachment_klazz = Kernel.const_get(args[:model]) if(args[:model])
+      rescue NameError
+        attachment_klazz = Product
+      end
+
+      image_loader = ImageLoader.new
+
+      if(File.exists? @image_cache )
+        puts "Loading images from #{@image_cache}"
+
+        missing_records = []
+        Dir.glob("#{@image_cache}/*.{jpg,png,gif}") do |image_name|
+
+          puts "Processing #{image_name} : #{File.exists?(image_name)}"
+          base_name = File.basename(image_name, '.*')
+
+          record = nil
+          if(attachment_klazz == Product && args[:sku])
+            sku = base_name.slice!(/\w+/)
+            sku.strip!
+            base_name.strip!
+
+            puts "Looking fo SKU #{sku}"
+            record = Variant.find_by_sku(sku)
+            if record
+              record = record.product   # SKU stored on Variant but we want it's master Product
+            else
+              puts "Looking for NAME [#{base_name}]"
+              record = attachment_klazz.find_by_name(base_name)
+            end
+          else
+            puts "Looking for #{attachment_klazz.name} with NAME [#{base_name}]"
+            record = attachment_klazz.find_by_name(base_name)
+          end
+      
+          if(record)
+            puts "Found record for attachment : #{record.inspect}"
+            exists = record.images.detect {|i| puts "COMPARE #{i.attachment_file_name} => #{image_name}"; i.attachment_file_name == image_name }
+            puts "Found existing attachments [#{exists}]" unless(exists.nil?)
+            if(args[:skip_if_loaded] && !exists.nil?)
+              puts "Skipping - Image #{image_name} already loaded for #{attachment_klazz}"
+              next
+            end
+          else
+            missing_records << image_name
+          end
+
+          # Now do actual upload to DB unless we are doing a dummy run,
+          # or the Image must have an associated record
+          unless(args[:dummy] == 'true' || (args[:skip_if_no_assoc] && record.nil?))
+            image_loader.reset()
+            puts "Process Image"
+            image_loader.process( image_name, record )
+          end
+
+        end
+
+        unless missing_records.empty?
+          FileUtils.mkdir_p('MissingRecords') unless File.directory?('MissingRecords')
+        
+          puts '\nMISSING Records Report>>'
+          missing_records.each do |i|
+            puts "Copy #{i} to MissingRecords folder"
+            FileUtils.cp( i, 'MissingRecords')  unless(args[:dummy] == 'true')
+          end
+        end
+
+      else
+        puts "ERROR: Supplied Path #{@image_cache} not accesible"
+        exit(-1)
+      end
+    end
+  end
 end

data/tasks/spree/product_loader.rake

@@ -1,44 +1,44 @@
-# Copyright:: (c) Autotelik Media Ltd 2011
-# Author ::   Tom Statter
-# Date ::     Feb 2011
-# License::   MIT. Free, Open Source.
-#
-# REQUIRES:   JRuby access to Java
-#
-# Usage::
-#
-# e.g.  => jruby -S rake datashift:spree:products input=vendor/extensions/autotelik/fixtures/SiteSpreadsheetInfo.xls
-#       => jruby -S rake datashift:spree:products input=C:\MyProducts.xls verbose=true
-#
-require 'datashift'
-
-namespace :datashift do
-
-  namespace :spree do
-
-    desc "Populate Spree Product/Variant data from .xls (Excel) or CSV file"
-    task :products, [:input, :verbose, :sku_prefix] => :environment do |t, args|
-
-      input = ENV['input']
-
-      raise "USAGE: jruby -S rake  datashift:spree:products input=excel_file.xls" unless input
-      raise "ERROR: Could not find file #{args[:input]}" unless File.exists?(input)
-
-      require 'product_loader'
-
-      # COLUMNS WITH DEFAULTS - TODO create YAML configuration file to drive defaults etc
-
-      loader = DataShift::ProductLoader.new
-
-      loader.set_default_value('available_on', Time.now.to_s(:db) )
-      loader.set_default_value('cost_price', 0.0 )
-
-      loader.set_prefix('sku', args[:sku_prefix] ) if(args[:sku_prefix])
-      
-      puts "Loading from file: #{input}"
-
-      loader.perform_load(input, :mandatory => ['sku', 'name', 'price'] )
-    end
-  end 
-
+# Copyright:: (c) Autotelik Media Ltd 2011
+# Author ::   Tom Statter
+# Date ::     Feb 2011
+# License::   MIT. Free, Open Source.
+#
+# REQUIRES:   JRuby access to Java
+#
+# Usage::
+#
+# e.g.  => jruby -S rake datashift:spree:products input=vendor/extensions/autotelik/fixtures/SiteSpreadsheetInfo.xls
+#       => jruby -S rake datashift:spree:products input=C:\MyProducts.xls verbose=true
+#
+require 'datashift'
+
+namespace :datashift do
+
+  namespace :spree do
+
+    desc "Populate Spree Product/Variant data from .xls (Excel) or CSV file"
+    task :products, [:input, :verbose, :sku_prefix] => :environment do |t, args|
+
+      input = ENV['input']
+
+      raise "USAGE: jruby -S rake  datashift:spree:products input=excel_file.xls" unless input
+      raise "ERROR: Could not find file #{args[:input]}" unless File.exists?(input)
+
+      require 'product_loader'
+
+      # COLUMNS WITH DEFAULTS - TODO create YAML configuration file to drive defaults etc
+
+      loader = DataShift::ProductLoader.new
+
+      loader.set_default_value('available_on', Time.now.to_s(:db) )
+      loader.set_default_value('cost_price', 0.0 )
+
+      loader.set_prefix('sku', args[:sku_prefix] ) if(args[:sku_prefix])
+      
+      puts "Loading from file: #{input}"
+
+      loader.perform_load(input, :mandatory => ['sku', 'name', 'price'] )
+    end
+  end 
+
 end

data/tasks/word_to_seedfu.rake

@@ -1,167 +1,167 @@
-# Copyright:: (c) Autotelik Media Ltd 2011
-# Author ::   Tom Statter
-# Date ::     Aug 2010
-#
-# License::   Free, OpenSource... MIT ?
-#
-# About::     Rake tasks to read Word documents, containing product descriptions,
-#             convert to HTML, tidy the HTML and then create seed_fu ready fixtures,
-#             from a template, with product description supplied by the HTML
-#
-#             Note cleanest HTML is produced by this combination : saving with WdFormatHTML
-#             not WdFormatFilteredHTML and using the '--word-2000', 'y' option to tidy
-#             (don't use the '--bare' option)
-#
-#             Not currently available for JRuby due to Win32Ole requirement
-#
-#             Requires local exes available in PATH for :
-#             Microsoft Word
-#             HTML Tidy - http://tidy.sourceforge.net  (Free)
-#
-require 'erb'
-
-namespace :datashift do
-
-  desc "Convert MS Word to HTML and seed_fu fixtures. help=true for detailed usage."
-
-  task :word2html, [:help] => [:environment] do |t, args|
-    x =<<-EOS
-
-  USAGE::
-      Convert MS Word docs to HTML and seed_fu fixtures, by default searches for docs
-      in RAILS_ROOT/doc/copy
-
-      You can change the directory where Word document files are located
-      with the COPY_PATH  environment variable.
-
-      Examples:
-        # default, to convert all Word files for the current environment
-        rake datashift:word2seedfu
-
-        # to load seed files matching orders or customers
-        rake db:seed SEED=orders,customers
-
-        # to load files from RAILS_ROOT/features/fixtures
-        rake db:seed FIXTURE_PATH=features/fixtures
-    EOS
-
-    if(args[:help])
-      puts x
-      exit(0)
-    end
-
-    site_extension_lib = File.join(SiteExtension.root, 'lib')
-
-    require File.join(site_extension_lib, 'word')
-
-    copy_path = ENV["COPY_PATH"] ? ENV["COPY_PATH"] : File.join(RAILS_ROOT, "doc", "copy")
-    fixtures_path = ENV["FIXTURES_PATH"] ? ENV["FIXTURES_PATH"] : File.join(RAILS_ROOT, "db", "fixtures")
-
-    copy_files = Dir[File.join(copy_path, '*.doc')]
-
-    copy_files.each do |file|
-
-      name = File.basename(file, '.doc')
-
-      puts "\n== Generate raw HTML from #{name}.doc =="
-
-      @word = Word.new(true)
-
-      @word.open( file )
-
-      html_file = File.join(copy_path, "#{name}.ms.html")
-
-      @word.save_as_html( html_file )
-
-      tidy_file = File.join(copy_path, "#{name}.html")
-
-      tidy_config = File.join(site_extension_lib, 'tasks', 'tidy_config.txt')
-
-      puts "tidy cmd line:", "tidy -config #{tidy_config} -clean --show-body-only y --word-2000 y --indent-spaces 2 -output #{tidy_file} #{html_file}"
-
-      result = system("tidy", '-config', "#{tidy_config}", '-clean', '--show-body-only', 'y', '--word-2000', 'y', '--indent-spaces', '2', '-output', "#{tidy_file}", "#{html_file}")
-
-      # TODO maybe report on result, $?
-
-      File.open( tidy_file ) do |f|
-        puts f.read
-      end
-
-      @word.quit
-    end
-  end
-
-  desc "Convert MS Word to HTML and seed_fu fixtures. help=true for detailed usage."
-  task :word2seedfu => :environment do
-    site_extension_lib = File.join(SiteExtension.root, 'lib')
-
-    require File.join(site_extension_lib, 'word')
-
-    sku_id     = ENV["INITIAL_SKU_ID"] ? ENV["INITIAL_SKU_ID"] : 0
-    sku_prefix = ENV["SKU_PREFIX"] ? ENV["SKU_PREFIX"] : File.basename( RAILS_ROOT )
-
-    seedfu_template = File.join(site_extension_lib, 'tasks', 'seed_fu_product_template.erb')
-
-    begin
-      File.open( seedfu_template ) do |f|
-        @template = ERB.new(f.read)
-      end
-    rescue => e
-      puts "ERROR: #{e.inspect}"
-      puts "Cannot open or read template #{seedfu_template}"
-      raise e
-    end
-
-    copy_path = ENV["COPY_PATH"] ? ENV["COPY_PATH"] : File.join(RAILS_ROOT, "doc", "copy")
-    fixtures_path = ENV["FIXTURES_PATH"] ? ENV["FIXTURES_PATH"] : File.join(RAILS_ROOT, "db", "fixtures")
-
-    copy_files = Dir[File.join(copy_path, '*.doc')]
-
-    copy_files.each do |file|
-
-      name = File.basename(file, '.doc')
-
-      puts "\n== Generate raw HTML from #{name}.doc =="
-
-      @word = Word.new(true)
-
-      @word.open( file )
-
-      html_file = File.join(copy_path, "#{name}.ms.html")
-
-      @word.save_as_html( html_file )
-
-      tidy_file = File.join(copy_path, "#{name}.html")
-
-      tidy_config = File.join(site_extension_lib, 'tasks', 'tidy_config.txt')
-
-      puts "tidy cmd line:", "tidy -config #{tidy_config} -clean --show-body-only y --word-2000 y --indent-spaces 2 -output #{tidy_file} #{html_file}"
-
-      result = system("tidy", '-config', "#{tidy_config}", '-clean', '--show-body-only', 'y', '--word-2000', 'y', '--indent-spaces', '2', '-output', "#{tidy_file}", "#{html_file}")
-
-      # TODO maybe report on result, $?
-
-      File.open( tidy_file ) do |f|
-        @description = f.read
-      end
-
-      sku_id_str  = "%03d" % sku_id
-
-      seed_file = "#{sku_id_str}_#{name.gsub(' ', '_')}.rb"
-      puts "\n== Generate seed fu file #{seed_file} =="
-
-      @sku  = "#{sku_prefix}_#{sku_id_str}"
-      @name = 'TODO'
-  
-      File.open( File.join(fixtures_path, seed_file), 'w' ) do |f|
-        f.write @template.result(binding)
-        puts "\nFile created: #{File.join(fixtures_path, seed_file)}"
-      end
-
-      sku_id += 1
-
-      @word.quit
-    end
-
-  end
+# Copyright:: (c) Autotelik Media Ltd 2011
+# Author ::   Tom Statter
+# Date ::     Aug 2010
+#
+# License::   Free, OpenSource... MIT ?
+#
+# About::     Rake tasks to read Word documents, containing product descriptions,
+#             convert to HTML, tidy the HTML and then create seed_fu ready fixtures,
+#             from a template, with product description supplied by the HTML
+#
+#             Note cleanest HTML is produced by this combination : saving with WdFormatHTML
+#             not WdFormatFilteredHTML and using the '--word-2000', 'y' option to tidy
+#             (don't use the '--bare' option)
+#
+#             Not currently available for JRuby due to Win32Ole requirement
+#
+#             Requires local exes available in PATH for :
+#             Microsoft Word
+#             HTML Tidy - http://tidy.sourceforge.net  (Free)
+#
+require 'erb'
+
+namespace :datashift do
+
+  desc "Convert MS Word to HTML and seed_fu fixtures. help=true for detailed usage."
+
+  task :word2html, [:help] => [:environment] do |t, args|
+    x =<<-EOS
+
+  USAGE::
+      Convert MS Word docs to HTML and seed_fu fixtures, by default searches for docs
+      in RAILS_ROOT/doc/copy
+
+      You can change the directory where Word document files are located
+      with the COPY_PATH  environment variable.
+
+      Examples:
+        # default, to convert all Word files for the current environment
+        rake datashift:word2seedfu
+
+        # to load seed files matching orders or customers
+        rake db:seed SEED=orders,customers
+
+        # to load files from RAILS_ROOT/features/fixtures
+        rake db:seed FIXTURE_PATH=features/fixtures
+    EOS
+
+    if(args[:help])
+      puts x
+      exit(0)
+    end
+
+    site_extension_lib = File.join(SiteExtension.root, 'lib')
+
+    require File.join(site_extension_lib, 'word')
+
+    copy_path = ENV["COPY_PATH"] ? ENV["COPY_PATH"] : File.join(RAILS_ROOT, "doc", "copy")
+    fixtures_path = ENV["FIXTURES_PATH"] ? ENV["FIXTURES_PATH"] : File.join(RAILS_ROOT, "db", "fixtures")
+
+    copy_files = Dir[File.join(copy_path, '*.doc')]
+
+    copy_files.each do |file|
+
+      name = File.basename(file, '.doc')
+
+      puts "\n== Generate raw HTML from #{name}.doc =="
+
+      @word = Word.new(true)
+
+      @word.open( file )
+
+      html_file = File.join(copy_path, "#{name}.ms.html")
+
+      @word.save_as_html( html_file )
+
+      tidy_file = File.join(copy_path, "#{name}.html")
+
+      tidy_config = File.join(site_extension_lib, 'tasks', 'tidy_config.txt')
+
+      puts "tidy cmd line:", "tidy -config #{tidy_config} -clean --show-body-only y --word-2000 y --indent-spaces 2 -output #{tidy_file} #{html_file}"
+
+      result = system("tidy", '-config', "#{tidy_config}", '-clean', '--show-body-only', 'y', '--word-2000', 'y', '--indent-spaces', '2', '-output', "#{tidy_file}", "#{html_file}")
+
+      # TODO maybe report on result, $?
+
+      File.open( tidy_file ) do |f|
+        puts f.read
+      end
+
+      @word.quit
+    end
+  end
+
+  desc "Convert MS Word to HTML and seed_fu fixtures. help=true for detailed usage."
+  task :word2seedfu => :environment do
+    site_extension_lib = File.join(SiteExtension.root, 'lib')
+
+    require File.join(site_extension_lib, 'word')
+
+    sku_id     = ENV["INITIAL_SKU_ID"] ? ENV["INITIAL_SKU_ID"] : 0
+    sku_prefix = ENV["SKU_PREFIX"] ? ENV["SKU_PREFIX"] : File.basename( RAILS_ROOT )
+
+    seedfu_template = File.join(site_extension_lib, 'tasks', 'seed_fu_product_template.erb')
+
+    begin
+      File.open( seedfu_template ) do |f|
+        @template = ERB.new(f.read)
+      end
+    rescue => e
+      puts "ERROR: #{e.inspect}"
+      puts "Cannot open or read template #{seedfu_template}"
+      raise e
+    end
+
+    copy_path = ENV["COPY_PATH"] ? ENV["COPY_PATH"] : File.join(RAILS_ROOT, "doc", "copy")
+    fixtures_path = ENV["FIXTURES_PATH"] ? ENV["FIXTURES_PATH"] : File.join(RAILS_ROOT, "db", "fixtures")
+
+    copy_files = Dir[File.join(copy_path, '*.doc')]
+
+    copy_files.each do |file|
+
+      name = File.basename(file, '.doc')
+
+      puts "\n== Generate raw HTML from #{name}.doc =="
+
+      @word = Word.new(true)
+
+      @word.open( file )
+
+      html_file = File.join(copy_path, "#{name}.ms.html")
+
+      @word.save_as_html( html_file )
+
+      tidy_file = File.join(copy_path, "#{name}.html")
+
+      tidy_config = File.join(site_extension_lib, 'tasks', 'tidy_config.txt')
+
+      puts "tidy cmd line:", "tidy -config #{tidy_config} -clean --show-body-only y --word-2000 y --indent-spaces 2 -output #{tidy_file} #{html_file}"
+
+      result = system("tidy", '-config', "#{tidy_config}", '-clean', '--show-body-only', 'y', '--word-2000', 'y', '--indent-spaces', '2', '-output', "#{tidy_file}", "#{html_file}")
+
+      # TODO maybe report on result, $?
+
+      File.open( tidy_file ) do |f|
+        @description = f.read
+      end
+
+      sku_id_str  = "%03d" % sku_id
+
+      seed_file = "#{sku_id_str}_#{name.gsub(' ', '_')}.rb"
+      puts "\n== Generate seed fu file #{seed_file} =="
+
+      @sku  = "#{sku_prefix}_#{sku_id_str}"
+      @name = 'TODO'
+  
+      File.open( File.join(fixtures_path, seed_file), 'w' ) do |f|
+        f.write @template.result(binding)
+        puts "\nFile created: #{File.join(fixtures_path, seed_file)}"
+      end
+
+      sku_id += 1
+
+      @word.quit
+    end
+
+  end
 end