remote_table 0.2.3 → 0.2.4

data/VERSION

@@ -1 +1 @@
-0.2.3
+0.2.4

data/lib/remote_table.rb

@@ -1,5 +1,6 @@
 require 'rubygems'
 require 'digest/md5'
+require 'iconv'
 require 'active_support'
 require 'tempfile'
 require 'fastercsv'

data/lib/remote_table/file.rb

@@ -1,6 +1,7 @@
 class RemoteTable
   class File
     attr_accessor :filename, :format, :delimiter, :skip, :cut, :crop, :sheet, :headers, :schema, :schema_name, :trap
+    attr_accessor :encoding
     attr_accessor :path
     attr_accessor :keep_blank_rows
     
@@ -17,9 +18,23 @@ class RemoteTable
       @schema = bus[:schema]
       @schema_name = bus[:schema_name]
       @trap = bus[:trap]
+      @encoding = bus[:encoding] || 'UTF-8'
       extend "RemoteTable::#{format.to_s.camelcase}".constantize
     end
     
+    class << self
+      # http://santanatechnotes.blogspot.com/2005/12/matching-iso-8859-1-strings-with-ruby.html
+      def convert_to_utf8(str, encoding)
+        if encoding == 'UTF-8'
+          str.toutf8 # just in case
+        else
+          @_iconv ||= Hash.new
+          @_iconv[encoding] ||= Iconv.new 'UTF-8', encoding
+          @_iconv[encoding].iconv(str).toutf8
+        end
+      end
+    end
+    
     def tabulate(path)
       define_fixed_width_schema! if format == :fixed_width and schema.is_a?(Array) # TODO move to generic subclass callback
       self.path = path
@@ -47,6 +62,38 @@ class RemoteTable
       end
     end
     
+    def backup_file!
+      FileUtils.cp path, "#{path}.backup"
+    end
+    
+    def skip_rows!
+      return unless skip
+      `cat #{path} | tail -n +#{skip + 1} > #{path}.tmp`
+      FileUtils.mv "#{path}.tmp", path
+    end
+    
+    def convert_file_to_utf8!
+      return if encoding == 'UTF8' or encoding == 'UTF-8'
+      `iconv -c -f #{encoding} -t UTF8 #{path} > #{path}.tmp`
+      FileUtils.mv "#{path}.tmp", path
+    end
+    
+    def restore_file!
+      FileUtils.mv "#{path}.backup", path if ::File.readable? "#{path}.backup"
+    end
+    
+    def cut_columns!
+      return unless cut
+      `cat #{path} | cut -c #{cut} > #{path}.tmp`
+      FileUtils.mv "#{path}.tmp", path
+    end
+    
+    def crop_rows!
+      return unless crop
+      `cat #{path} | tail -n +#{crop.first} | head -n #{crop.last - crop.first + 1} > #{path}.tmp`
+      FileUtils.mv "#{path}.tmp", path
+    end
+    
     def format_from_filename
       extname = ::File.extname(filename).gsub('.', '')
       return :csv if extname.blank?

data/lib/remote_table/file/csv.rb

@@ -1,6 +1,8 @@
 class RemoteTable
   module Csv
     def each_row(&block)
+      backup_file!
+      convert_file_to_utf8!
       skip_rows!
       FasterCSV.foreach(path, fastercsv_options) do |row|
         ordered_hash = ActiveSupport::OrderedHash.new
@@ -27,13 +29,13 @@ class RemoteTable
         yield ordered_hash if keep_blank_rows or filled_values.nonzero?
       end
     ensure
-      restore_rows!
+      restore_file!
     end
     
     private
     
     def fastercsv_options
-      fastercsv_options = { :skip_blanks => !keep_blank_rows, :header_converters => lambda { |k| k.to_s.toutf8 } }
+      fastercsv_options = { :skip_blanks => !keep_blank_rows, :header_converters => lambda { |k| RemoteTable::File.convert_to_utf8 k.to_s, encoding } }
       if headers == false
         fastercsv_options.merge!(:headers => nil)
       else
@@ -42,17 +44,5 @@ class RemoteTable
       fastercsv_options.merge!(:col_sep => delimiter) if delimiter
       fastercsv_options
     end
-    
-    def skip_rows!
-      return unless skip
-      original = "#{path}.original"
-      FileUtils.cp(path, original)
-      `cat #{original} | tail -n +#{skip + 1} > #{path}`
-    end
-    
-    def restore_rows!
-      return unless skip
-      FileUtils.mv "#{path}.original", path
-    end
   end
 end

data/lib/remote_table/file/fixed_width.rb

@@ -1,6 +1,8 @@
 class RemoteTable
   module FixedWidth
     def each_row(&block)
+      backup_file!
+      convert_file_to_utf8!
       crop_rows!
       skip_rows!
       cut_columns!
@@ -10,47 +12,7 @@ class RemoteTable
         yield hash if keep_blank_rows or hash.any? { |k, v| v.present? }
       end
     ensure
-      uncut_columns!
-      unskip_rows!
-      uncrop_rows!
-    end
-    
-    private
-    
-    def cut_columns!
-      return unless cut
-      original = "#{path}.uncut"
-      FileUtils.cp(path, original)
-      `cat #{original} | cut -c #{cut} > #{path}`
-    end
-    
-    def uncut_columns!
-      return unless cut
-      FileUtils.mv "#{path}.uncut", path
-    end
-    
-    def skip_rows!
-      return unless skip
-      original = "#{path}.unskipped"
-      FileUtils.cp(path, original)
-      `cat #{original} | tail -n +#{skip + 1} > #{path}`
-    end
-    
-    def unskip_rows!
-      return unless skip
-      FileUtils.mv "#{path}.unskipped", path
-    end
-    
-    def crop_rows!
-      return unless crop
-      original = "#{path}.uncropped"
-      FileUtils.cp(path, original)
-      `cat #{original} | tail -n +#{crop.first} | head -n #{crop.last - crop.first + 1} > #{path}`
-    end
-    
-    def uncrop_rows!
-      return unless crop
-      FileUtils.mv "#{path}.uncropped", path
+      restore_file!
     end
   end
 end

data/remote_table.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = %q{remote_table}
-  s.version = "0.2.3"
+  s.version = "0.2.4"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Seamus Abshere", "Andy Rossmeissl"]
-  s.date = %q{2010-03-16}
+  s.date = %q{2010-03-24}
   s.description = %q{Remotely open and parse Excel XLS, ODS, CSV and fixed-width tables.}
   s.email = %q{seamus@abshere.net}
   s.extra_rdoc_files = [

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: remote_table
 version: !ruby/object:Gem::Version 
-  version: 0.2.3
+  version: 0.2.4
 platform: ruby
 authors: 
 - Seamus Abshere
@@ -10,7 +10,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-03-16 00:00:00 -04:00
+date: 2010-03-24 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency