file.rb in remote_table-0.2.12

- old
+ new

@@ -2,10 +2,12 @@
   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
+    attr_accessor :row_xpath
+    attr_accessor :column_xpath
     
     def initialize(bus)
       @filename = bus[:filename]
       @format = bus[:format] || format_from_filename
       @delimiter = bus[:delimiter]
@@ -17,26 +19,15 @@
       @headers = bus[:headers]
       @schema = bus[:schema]
       @schema_name = bus[:schema_name]
       @trap = bus[:trap]
       @encoding = bus[:encoding] || 'UTF-8'
+      @row_xpath = bus[:row_xpath]
+      @column_xpath = bus[:column_xpath]
       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' or encoding == 'UTF8'
-          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
       self
     end
@@ -96,9 +87,10 @@
     
     def format_from_filename
       extname = ::File.extname(filename).gsub('.', '')
       return :csv if extname.blank?
       format = [ :xls, :ods ].detect { |i| i == extname.to_sym }
+      format = :html if extname =~ /\Ahtm/
       format = :csv if format.blank?
       format
     end
   end
 end