rbbt-util 1.2.1 → 2.0.1

data/test/rbbt/util/tsv/test_accessor.rb

@@ -0,0 +1,109 @@
+require File.join(File.expand_path(File.dirname(__FILE__)), '../../..', 'test_helper.rb')
+require 'rbbt/util/tsv/accessor'
+require 'rbbt/util/tsv'
+
+class TestTSVAcessor < Test::Unit::TestCase
+
+  def test_zip_fields
+    a = [%w(1 2), %w(a b)]
+    assert_equal a, TSV.zip_fields(TSV.zip_fields(a))
+  end
+
+  def test_values_at
+    content =<<-EOF
+#Id    ValueA    ValueB    OtherID
+row1    a|aa|aaa    b    Id1|Id2
+row2    A    B    Id3
+    EOF
+
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.new(File.open(filename), :list, :sep => /\s+/, :key => "OtherID", :persistence => true)
+      assert_equal "row2", tsv.values_at("Id1", "Id3").last.first 
+    end
+  end
+
+  def test_to_s
+    content =<<-EOF
+#Id	ValueA	ValueB	OtherID
+row1	a|aa|aaa	b	Id1|Id2
+row2	A	B	Id3
+    EOF
+
+    content2 =<<-EOF
+#Id	ValueA	ValueB	OtherID
+row1	a|aa|aaa	b	Id1|Id2
+row2	A	B	Id3
+    EOF
+ 
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.new(File.open(filename), :sep => /\s+/)
+      assert_equal content, tsv.to_s.sub(/^#: [^\n]*\n/s,'')
+    end
+  end
+
+  def test_to_s_ordered
+    content =<<-EOF
+#Id	ValueA	ValueB	OtherID
+row1	a|aa|aaa	b	Id1|Id2
+row2	A	B	Id3
+    EOF
+
+    content2 =<<-EOF
+#Id	ValueA	ValueB	OtherID
+row2	A	B	Id3
+row1	a|aa|aaa	b	Id1|Id2
+    EOF
+
+
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.new(File.open(filename), :sep => /\s+/)
+      assert_equal content, tsv.to_s(%w(row1 row2)).sub(/^#: [^\n]*\n/s,'')
+      assert_not_equal content, tsv.to_s(%w(row2 row1)).sub(/^#: [^\n]*\n/s,'')
+      assert_equal content2, tsv.to_s(%w(row2 row1)).sub(/^#: [^\n]*\n/s,'')
+    end
+  end
+
+  def test_field_compare
+     content =<<-EOF
+#Id    Letter:LetterValue    Other:LetterValue    OtherID
+row1    a|aa|aaa    b    Id1|Id2
+row2    A    B    Id3
+row3    a    C    Id4
+    EOF
+    
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.new(filename + '#:sep=/\s+/')
+
+      assert tsv.fields.include?("LetterValue")
+    end
+  end
+
+  def test_indentify_fields
+    content =<<-EOF
+#ID ValueA ValueB Comment
+row1 a b c
+row2 A B C
+    EOF
+
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.new(File.open(filename), :double, :sep => /\s/)
+      assert_equal :key, tsv.identify_field("ID")
+    end
+  end
+
+  def test_named_fields
+    content =<<-EOF
+#ID ValueA ValueB Comment
+row1 a b c
+row2 A B C
+    EOF
+
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.new(File.open(filename), :double, :sep => /\s/)
+
+      assert_equal "ValueA", tsv.fields["ValueA"]
+    end
+  end
+
+end
+

data/test/rbbt/util/tsv/test_attach.rb

@@ -0,0 +1,271 @@
+require File.join(File.expand_path(File.dirname(__FILE__)), '../../..', 'test_helper.rb')
+require 'rbbt/util/tsv'
+require 'rbbt/util/tsv/attach'
+require 'rbbt'
+
+class TestAttach < Test::Unit::TestCase
+  def test_attach_same_key
+    content1 =<<-EOF
+#Id    ValueA    ValueB
+row1    a|aa|aaa    b
+row2    A    B
+    EOF
+
+    content2 =<<-EOF
+#ID    ValueB    OtherID
+row1    b    Id1|Id2
+row3    B    Id3
+    EOF
+
+    tsv1 = tsv2 = nil
+    TmpFile.with_file(content1) do |filename|
+      tsv1 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    TmpFile.with_file(content2) do |filename|
+      tsv2 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    tsv1.attach_same_key tsv2, "OtherID"
+
+    assert_equal %w(ValueA ValueB OtherID), tsv1.fields
+    assert_equal %w(Id1 Id2), tsv1["row1"]["OtherID"]
+
+    TmpFile.with_file(content1) do |filename|
+      tsv1 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    tsv1.attach_same_key tsv2
+
+    assert_equal %w(ValueA ValueB OtherID), tsv1.fields
+
+    tsv1 = tsv2 = nil
+    TmpFile.with_file(content1) do |filename|
+      tsv1 = TSV.new(File.open(filename), :list, :sep => /\s+/)
+    end
+
+    TmpFile.with_file(content2) do |filename|
+      tsv2 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    tsv1.attach_same_key tsv2, "OtherID"
+
+    assert_equal %w(ValueA ValueB OtherID), tsv1.fields
+    assert_equal "Id1", tsv1["row1"]["OtherID"]
+  end
+
+  def test_attach_source_field
+    content1 =<<-EOF
+#Id    ValueA    ValueB
+row1    a|aa|aaa    b
+row2    A    B
+    EOF
+
+    content2 =<<-EOF
+#ValueB    OtherID
+b    Id1|Id2
+B    Id3
+    EOF
+
+    tsv1 = tsv2 = nil
+    TmpFile.with_file(content1) do |filename|
+      tsv1 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    TmpFile.with_file(content2) do |filename|
+      tsv2 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    tsv1.attach_source_key tsv2, "ValueB"
+
+    assert_equal %w(ValueA ValueB OtherID), tsv1.fields
+    assert_equal %w(Id1 Id2), tsv1["row1"]["OtherID"]
+
+    TmpFile.with_file(content1) do |filename|
+      tsv1 = TSV.new(File.open(filename), :list, :sep => /\s+/)
+    end
+
+    tsv1.attach_source_key tsv2, "ValueB"
+
+
+    assert_equal %w(ValueA ValueB OtherID), tsv1.fields
+    assert_equal "Id1", tsv1["row1"]["OtherID"]
+  end
+
+  def test_attach_index
+    content1 =<<-EOF
+#Id    ValueA    ValueB
+row1    a|aa|aaa    b
+row2    A    B
+    EOF
+
+    content2 =<<-EOF
+#ValueE    OtherID
+e    Id1|Id2
+E    Id3
+    EOF
+
+    content_index =<<-EOF
+#Id    ValueE
+row1    e
+row2    E
+    EOF
+
+    tsv1 = tsv2 = index = nil
+    TmpFile.with_file(content1) do |filename|
+      tsv1 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    TmpFile.with_file(content2) do |filename|
+      tsv2 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    TmpFile.with_file(content_index) do |filename|
+      index = TSV.new(File.open(filename), :flat, :sep => /\s+/)
+    end
+
+    tsv1.attach_index tsv2, index
+
+    assert_equal %w(ValueA ValueB OtherID), tsv1.fields
+    assert_equal %w(Id1 Id2), tsv1["row1"]["OtherID"]
+
+    TmpFile.with_file(content1) do |filename|
+      tsv1 = TSV.new(File.open(filename), :list, :sep => /\s+/)
+    end
+
+    tsv1.attach_index tsv2, index
+
+    assert_equal %w(ValueA ValueB OtherID), tsv1.fields
+    assert_equal "Id1", tsv1["row1"]["OtherID"]
+  end
+
+  def test_attach
+    content1 =<<-EOF
+#Id    ValueA    ValueB
+row1    a|aa|aaa    b
+row2    A    B
+    EOF
+
+    content2 =<<-EOF
+#Id    ValueB    OtherID
+row1    b    Id1|Id2
+row3    B    Id3
+    EOF
+
+    content3 =<<-EOF
+#ValueB    OtherID
+b    Id1|Id2
+B    Id3
+    EOF
+   tsv1 = tsv2 = tsv3 = nil
+    TmpFile.with_file(content1) do |filename|
+      tsv1 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    TmpFile.with_file(content2) do |filename|
+      tsv2 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    TmpFile.with_file(content3) do |filename|
+      tsv3 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    tsv1.attach tsv2, "OtherID"
+
+    assert_equal %w(ValueA ValueB OtherID), tsv1.fields
+    assert_equal %w(Id1 Id2), tsv1["row1"]["OtherID"]
+
+    TmpFile.with_file(content1) do |filename|
+      tsv1 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    tsv1.attach tsv3
+
+    assert_equal %w(ValueA ValueB OtherID), tsv1.fields
+    assert_equal %w(Id1 Id2), tsv1["row1"]["OtherID"]
+
+  end
+
+  def test_attach_using_index
+    content1 =<<-EOF
+#Id    ValueA    ValueB
+row1    a|aa|aaa    b
+row2    A    B
+    EOF
+
+    content2 =<<-EOF
+#ValueE    OtherID
+e    Id1|Id2
+E    Id3
+    EOF
+
+    content_index =<<-EOF
+#: :sep=/\\s+/
+#Id    ValueE
+row1    e
+row2    E
+    EOF
+
+
+    require 'rbbt/sources/organism'
+
+    Rbbt.claim "data", StringIO.new(content1), "Test1"
+    Rbbt.claim "data", StringIO.new(content2), "Test2"
+    Rbbt.claim "identifiers", StringIO.new(content_index), "Test2"
+
+    tsv1 = tsv2 = nil
+
+    tsv1 = Rbbt.files.Test1.data.tsv :double,  :sep => /\s+/
+    tsv2 = Rbbt.files.Test2.data.tsv :double,  :sep => /\s+/
+
+    tsv2.identifiers = Rbbt.files.Test2.identifiers
+
+    tsv1.attach tsv2, "OtherID", :in_namespace => false
+
+    assert_equal tsv1.fields,%w(ValueA ValueB OtherID)
+    assert_equal %w(Id1 Id2), tsv1["row1"]["OtherID"]
+
+  end
+
+  def test_find_path
+     content1 =<<-EOF
+#: :sep=/\\s+/#:case_insensitive=false
+#Id    ValueA    ValueB
+row1    a|aa|aaa    b
+row2    A    B
+    EOF
+
+    content2 =<<-EOF
+#: :sep=/\\s+/#:case_insensitive=false
+#OtherID ValueE
+Id1|Id2    e    
+Id3        E    
+    EOF
+
+    content_identifiers =<<-EOF
+#: :sep=/\\s+/#:case_insensitive=false
+#Id    ValueE
+row1    e
+row2    E
+    EOF
+
+    tsv1 = tsv2 = identifiers = nil
+    TmpFile.with_file(content1) do |filename|
+      tsv1 = TSV.new(File.open(filename), :key => "Id")
+    end
+
+    TmpFile.with_file(content2) do |filename|
+      tsv2 = TSV.new(File.open(filename), :double)
+    end
+
+    TmpFile.with_file(content_identifiers) do |filename|
+      identifiers = TSV.new(File.open(filename), :flat, :sep => /\s+/)
+    end
+
+    tsv1.identifiers = identifiers
+    tsv1.attach tsv2
+ 
+    assert_equal %w(ValueA ValueB ValueE), tsv1.fields
+  end
+end
+

data/test/rbbt/util/tsv/test_index.rb

@@ -0,0 +1,158 @@
+require File.join(File.expand_path(File.dirname(__FILE__)), '../../..', 'test_helper.rb')
+require 'rbbt/util/tsv'
+require 'rbbt/util/tsv/index'
+
+class TestTSVManipulate < Test::Unit::TestCase
+
+  def test_index
+    content =<<-EOF
+#Id    ValueA    ValueB    OtherID
+row1    a|aa|aaa    b    Id1|Id2
+row2    A    B    Id3
+    EOF
+
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.new(File.open(filename), :sep => /\s+/, :key => "OtherID", :persistence => false)
+      index = tsv.index(:case_insensitive => true, :persistence => true)
+      assert index["row1"].include? "Id1"
+      assert_equal "OtherID", index.fields.first
+    end
+
+#    TmpFile.with_file(content) do |filename|
+#      tsv = TSV.new(File.open(filename), :sep => /\s+/, :key => "OtherID")
+#      index = tsv.index(:case_insensitive => true)
+#      assert index["row1"].include? "Id1"
+#      assert_equal "OtherID", index.fields.first
+#    end
+  end
+
+  def test_index_headerless
+    content =<<-EOF
+row1    a|aa|aaa    b    Id1|Id2
+row2    A    B    Id3
+    EOF
+
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.new(File.open(filename), :sep => /\s+/)
+      index = tsv.index(:case_insensitive => true, :target => 2)
+      assert index["row1"].include? "Id1"
+    end
+  end
+
+
+  def test_best_index
+    content =<<-EOF
+#Id    ValueA    ValueB    OtherID
+row1    a|aa|aaa    b|A    Id1
+row2    A    a|B    Id3
+row3    A    a|B    Id4
+    EOF
+
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.new(File.open(filename), :sep => /\s+/, :key => "OtherID", :persistence => true)
+      index = tsv.index(:case_insensitive => false, :order => true)
+      ddd index
+      assert_equal "Id1", index['a'].first
+      assert_equal "Id3", index['A'].first
+      assert_equal "OtherID", index.fields.first
+    end
+
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.new(File.open(filename), :sep => /\s+/, :key => "OtherID")
+      index = tsv.index(:case_insensitive => true)
+      assert index["row1"].include? "Id1"
+      assert_equal "OtherID", index.fields.first
+    end
+  end
+
+  #{{{ Test Attach
+
+  def ___test_smart_merge_single
+    content1 =<<-EOF
+#Id    ValueA    ValueB
+row1    a|aa|aaa    b
+row2    A    B
+    EOF
+
+    content2 =<<-EOF
+#ValueC    ValueB    OtherID
+c|cc|ccc    b    Id1|Id2
+C    B    Id3
+    EOF
+
+    tsv1 = tsv2 = nil
+    TmpFile.with_file(content1) do |filename|
+      tsv1 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    TmpFile.with_file(content2) do |filename|
+      tsv2 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    tsv1 = tsv1.smart_merge tsv2, "ValueB"
+
+    assert_equal "C", tsv1["row2"]["ValueC"]
+    assert %w(c cc ccc).include? tsv1["row1"]["ValueC"]
+    assert_equal "Id1", tsv1["row1"]["OtherID"]
+  end
+
+  def test_index_to_key
+     content =<<-EOF
+#: :sep=/\\s+/
+#Id    ValueA    ValueB
+row1    a|aa|aaa    b
+row2    A    B
+    EOF
+
+    tsv1 = tsv2 = nil
+    TmpFile.with_file(content) do |filename|
+      tsv1 = TSV.new(File.open(filename), :double, :sep => /\s+/, :key => "ValueA", :case_insensitive =>  true)
+    end
+  end
+
+  # {{{ Test sorted index
+
+  def load_data(data)
+    Log.debug("Data:\n#{Open.read(data)}")
+    tsv = TSV.new(data, :list, :sep=>":", :cast => proc{|e| e =~ /(\s*)(_*)/; ($1.length..($1.length + $2.length - 1))})
+    tsv.add_field "Start" do |key, values|
+      values["Range"].first
+    end
+    tsv.add_field "End" do |key, values|
+      values["Range"].last
+    end
+
+    ddd tsv.fields
+    tsv = tsv.slice ["Start", "End"]
+    ddd tsv.fields
+ 
+    tsv
+  end
+
+  def test_sorted_index
+    data =<<-EOF
+#ID:Range
+#:012345678901234567890
+a:   ______
+b: ______
+c:    _______
+d:  ____
+e:    ______
+f:             ___
+g:         ____
+    EOF
+    TmpFile.with_file(data) do |datafile|
+      tsv = load_data(datafile)
+      f   = tsv.sorted_index
+
+      assert_equal %w(), f[0].sort
+      assert_equal %w(b), f[1].sort
+      assert_equal %w(), f[20].sort
+      assert_equal %w(), f[(20..100)].sort
+      assert_equal %w(a b d), f[3].sort
+      assert_equal %w(a b c d e), f[(3..4)].sort
+    end
+  end
+
+end
+