README.rdoc

Ruby4Knime¶ ↑

Ruby4Knime is a realization of special node types for knime.org that allow to use Ruby scripts for interactive data analytic.

The main aims of this project are:

• provide possibility to use Ruby language for interactive data analysis;

• make the analysis more effective using clear and laconic Ruby style.

knime.rb is the simple mediator that allows to use KNIME`s classes written in Java in the Ruby programming style and also use all Java classes without any changes.

Simple examples¶ ↑

These samples illustrate how to generate output data from input data of the node.

Copy existing rows¶ ↑

$in_data_0.each do |row|
  $out_data_0 << row
end

Copy existing rows from all inputs¶ ↑

This example is applicated for Ruby Script 2x2. Note: all inputs and outputs are accessible by names like $in_data_X, $out_data_X or by number in appropriate array $output_datatable_arr, $input_datatable_arr.

(0..1).each do |i|
  out = $output_datatable_arr[i]
  $input_datatable_arr[i].each do |row|
    out << row
  end
end

Add new two columns with String and Int types¶ ↑

This example also illustrates the progress state updating.

table = $input_datatable_arr[0]
count = table.length
table.each_with_index do |row, i|
  $out_data_0 << (row << Cells.new.string('Hi!').
                                   int(row[0].to_s.length))

  setProgress "#{i*100/count}%" if i%100 != 0
end

Create new rows¶ ↑

In this example new rows creates with new uniq rowkeys. Therefore it is possible to create output table with any number of rows.

$in_data_0.each do |row|
  $out_data_0 << Cells.new.string( row.getCell(0).to_s.length.to_s )
end

In this example cell 0 copies without changes and adds difference between cells with indexes 1 and 2. Index in Ruby style.

$in_data_0.each do |row|
  $out_data_0 << Cells.new.int(row[0].to_i).
                           double(row[1].to_f - row[2].to_f)
end

Ruby Snippet examples¶ ↑

Ruby snippet realizes as the body of a lambda-function in the following template:

func = ->(row) do
  <<SNIPPET CODE>>
end
snippet_runner &func

Simple snippet examples¶ ↑

Copy all rows without changes. Code may contains only row-variable.

row

Add difference between columns:

row << Cells.new.
  double(row[1].to_f - row[2].to_f)

Access to cells of the input row¶ ↑

Now is realized an access to cells of the row by indexes and by names. E.g. we have a table with following columns ‘x’, ‘Y1’, ‘Y2’, ‘y1(1)’, ‘y2(1)’. We can access by indexes from 0 to 4 or from -1 (last column) to -5 (first column). For these names a special methods are generated. All names are translated in low case. All symbols except alhabetic and numbers are changed to underline symbol. Multiple undeline symbols are changed by a single. Last one is deleted. That is we have a following set of methods: #x, #y1, #y2, #y1_1, #y2_1 available for instances of BlobSupportDataRow class. Following code fragments are available:

row << Cells.new.double(row[0].to_f)
                 double(row.y1.to_f - row.y2.to_f)

and

$in_data_0.each do |row|
  $out_data_0 << Cells.new.string(row.x.to_s)
end

Other experiments¶ ↑

Samples that add new cells in functional style.

count = $in_data_0.length
$in_data_0.each_with_index do |row,i|
  $out_data_0 << row.string('Hi!').int(row.getCell(0).to_s.length).append
  setProgress "#{i*100/count}%" if i%100 != 0
end

-

$in_data_0.each do |row|
  $out_data_0 << $out_data_0.createRowKey.
    stringCell(row.getCell(0).to_s.length.to_s).new_row
end

Other notes¶ ↑

Now it is possible to use any types of KNIME in the DataOutput table. Simply input a full qualified Java class name in the configuration dialog of the Ruby node. E.g. org.knime.core.data.def.ComplexNumberCell or org.knime.ext.textprocessing.data.DocumentCell.