scout-gear 7.3.0 → 8.1.0

data/test/scout/tsv/test_change_id.rb

@@ -0,0 +1,98 @@
+require File.expand_path(__FILE__).sub(%r(/test/.*), '/test/test_helper.rb')
+require File.expand_path(__FILE__).sub(%r(.*/test/), '').sub(/test_(.*)\.rb/,'\1')
+
+require 'scout/tsv'
+class TestChangeID < Test::Unit::TestCase
+  def test_simple_reorder
+    content1 =<<-EOF
+#: :sep=" "
+#ID    ValueA    ValueB
+row1    A1|A11    B1|B11
+row2    A2|A22    B2|B22
+    EOF
+
+    tsv = TSV.open StringIO.new(content1)
+
+    res = tsv.change_key "ValueA", keep: true
+    assert_equal ["row1"], res["A1"]["ID"]
+    assert_equal ["row1"], res["A11"]["ID"]
+    assert_equal ["row2"], res["A2"]["ID"]
+
+    res = tsv.change_key "ValueA", keep: false, one2one: true
+    assert_equal ["B1"], res["A1"]["ValueB"]
+    assert_equal ["B11"], res["A11"]["ValueB"]
+  end
+
+  def test_simple_reorder_file
+    content1 =<<-EOF
+#: :sep=" "
+#ID    ValueA    ValueB
+row1    A1|A11    B1|B11
+row2    A2|A22    B2|B22
+    EOF
+
+    TmpFile.with_file(content1) do |file1|
+      res = TSV.change_key file1, "ValueA", keep: true
+      assert_equal ["row1"], res["A1"]["ID"]
+      assert_equal ["row1"], res["A11"]["ID"]
+      assert_equal ["row2"], res["A2"]["ID"]
+      assert_equal ["B1","B11"], res["A1"]["ValueB"]
+
+      res = TSV.change_key file1, "ValueA", one2one: true, keep: true
+      assert_equal ["row1"], res["A1"]["ID"]
+      assert_equal ["row1"], res["A11"]["ID"]
+      assert_equal ["row2"], res["A2"]["ID"]
+      assert_equal ["B1"], res["A1"]["ValueB"]
+    end
+  end
+
+  def test_change_key_identifiers
+    content1 =<<-EOF
+#: :sep=" "
+#ID    ValueA    ValueB
+row1    A1|A11    B1|B11
+row2    A2|A22    B2|B22
+    EOF
+
+    identifiers_content =<<-EOF
+#: :sep=" "
+#ID    ValueC    ValueD
+row1    C1|C11    D1|D11
+row2    C2|C22    D2|D22
+    EOF
+
+
+    tsv = TSV.open StringIO.new(content1)
+    identifiers = TSV.open StringIO.new(identifiers_content)
+
+    res = tsv.change_key "ValueC", identifiers: identifiers, keep: true
+    assert_equal ["row1"], res["C1"]["ID"]
+    assert_equal ["row1"], res["C11"]["ID"]
+    assert_equal ["row2"], res["C2"]["ID"]
+  end
+
+  def test_change_id_identifiers
+    content1 =<<-EOF
+#: :sep=" "
+#ID    ValueA    ValueB
+row1    A1|A11    B1|B11
+row2    A2|A22    B2|B22
+    EOF
+
+    identifiers_content =<<-EOF
+#: :sep=" "
+#ID    ValueC    ValueD
+row1    C1|C11    D1|D11
+row2    C2|C22    D2|D22
+    EOF
+
+
+    tsv = TSV.open StringIO.new(content1)
+    identifiers = TSV.open StringIO.new(identifiers_content)
+
+    res = tsv.change_id "ValueA", "ValueC", identifiers: identifiers
+    assert_equal ["C1","C11"], res["row1"]["ValueC"]
+    assert_equal ["C2","C22"], res["row2"]["ValueC"]
+  end
+end
+

data/test/scout/tsv/test_dumper.rb

@@ -37,7 +37,7 @@ a\t1|11\t2|22
     end
 
     assert_raise ScoutException do
-      dumper.stream.read
+      TSV.open(dumper.stream, bar: true)
     end
   end
 end

data/test/scout/tsv/test_index.rb

@@ -43,24 +43,49 @@ row2    a    b    id3
     end
   end
 
+  def test_index_from_flat
+    content =<<-'EOF'
+#: :sep=" "#:type=:flat
+#Id    ValueA
+row1    a aa aaa
+row2    b bb bbb
+    EOF
+
+    TmpFile.with_file(content) do |filename|
+      index = TSV.index(filename, :target => "Id")
+      assert_equal "row1", index["aa"]
+    end
+  end
+
   def test_from_tsv
     content =<<-'EOF'
 #: :sep=/\s+/#:type=:double#:merge=:concat
 #Id    ValueA    ValueB    OtherID
 row1    a|aa|aaa    b    Id1|Id2
-row2    A    B    Id3|a
+row2    A|b    B    Id3|a
 row2    a    b    id3
     EOF
 
     TmpFile.with_file(content) do |filename|
       tsv = TSV.open(filename)
+      index = TSV.index(tsv, :target => "ValueB")
+      assert_equal 'b', index["a"]
+      assert_equal 'B', index["B"]
+      assert_equal 'b', index["b"]
+
+      index = tsv.index(:target => "ValueB")
+      assert_equal 'b', index["a"]
+      assert_equal 'B', index["B"]
+      assert_equal 'b', index["b"]
+
+
       index = TSV.index(tsv, :target => "ValueB", :fields => "OtherID")
       assert_equal 'B', index["a"]
       assert_nil index["B"]
 
       index = tsv.index(:target => "ValueB", :fields => "OtherID")
       assert_equal 'B', index["a"]
-      assert_nil index["B"]
+      assert_nil  index["B"]
     end
   end
 
@@ -74,7 +99,7 @@ row2    A    B    Id3|a
 row2    a    b    id3
     EOF
     tsv = TmpFile.with_file(content) do |filename|
-      index = TSV.index(filename, :target => "ValueB", :persist => true)
+      index = TSV.index(filename, :target => "ValueB", :persist => true, bar: true)
       assert_equal 'b', index["row1"]
       assert_equal 'b', index["a"]
       assert_equal 'b', index["aaa"]
@@ -107,6 +132,27 @@ g:         ____
     end
   end
 
+  def test_pos_index
+    data =<<-EOF
+# 012345678901234567890
+#ID:Range
+a:   ______
+b: ______
+c:    _______
+d:  ____
+e:    ______
+f:             ___
+g:         ____
+    EOF
+    TmpFile.with_file(data) do |datafile|
+      tsv = load_segment_data(datafile)
+      f   = tsv.pos_index("Start", :persist => true)
+
+      assert_equal %w(), f[0].sort
+      assert_equal %w(a c d e), f[(2..4)].sort
+    end
+  end
+
   def test_range_index_persistent
     data =<<-EOF
 # 012345678901234567890

data/test/scout/tsv/test_open.rb

@@ -1,9 +1,167 @@
 require File.expand_path(__FILE__).sub(%r(/test/.*), '/test/test_helper.rb')
 require File.expand_path(__FILE__).sub(%r(.*/test/), '').sub(/test_(.*)\.rb/,'\1')
 
+require 'scout/tsv'
 class TestOpenTraverse < Test::Unit::TestCase
-  def test_true
-    assert true
+  def test_array
+    num_lines = 100
+    lines = num_lines.times.collect{|i| "line-#{i}" }
+
+    r = TSV.traverse lines, :into => [] do |l|
+      l + "-" + Process.pid.to_s
+    end
+
+    assert_equal num_lines, r.length
+  end
+
+  def test_array_cpus
+    num_lines = 1000
+    lines = num_lines.times.collect{|i| "line-#{i}" }
+
+    r = TSV.traverse lines, :into => [], :cpus => 2 do |l|
+      l + "-" + Process.pid.to_s
+    end
+
+    assert_equal num_lines, r.length
+    assert_equal 2, r.collect{|l| l.split("-").last}.uniq.length
+  end
+
+  def test_tsv_cpus
+    num_lines = 10000
+    lines = num_lines.times.collect{|i| "line-#{i}" }
+
+    tsv  = TSV.setup({}, key_field: "Line", :fields => %w(Prefix Number), :type => :list)
+    lines.each do |line|
+      tsv[line] = ["LINE", line.split("-").last]
+    end
+
+    r = TSV.traverse tsv, :into => [], :cpus => 2, :bar => {desc: "Process", severity: 0} do |l,v|
+      pre, num = v
+      pre + "-" + num.to_s + "-" + Process.pid.to_s
+    end
+
+    assert_equal num_lines, r.length
+    assert_equal 2, r.collect{|l| l.split("-").last}.uniq.length
+    assert_equal "LINE", r.collect{|l| l.split("-").first}.first
+  end
+
+  def test_into_stream
+    num_lines = 100
+    lines = num_lines.times.collect{|i| "line-#{i}" }
+
+    r = TSV.traverse lines, :into => :stream do |l|
+      l + "-" + Process.pid.to_s
+    end
+
+    assert_equal num_lines, r.read.split("\n").length
+  end
+
+  def test_into_stream_error
+    num_lines = 100
+    lines = num_lines.times.collect{|i| "line-#{i}" }
+
+    assert_raise ScoutException do
+      Log.with_severity 7 do
+        i = 0
+        r = TSV.traverse lines, :into => :stream, cpus: 3 do |l|
+          raise ScoutException if i > 10
+          i += 1
+          l + "-" + Process.pid.to_s
+        end
+
+        r.read
+      end
+    end
+  end
+
+  def test_into_dumper_error
+    num_lines = 100
+    lines = num_lines.times.collect{|i| "line-#{i}" }
+
+    assert_raise ScoutException do 
+      i = 0
+      Log.with_severity 7 do
+        dumper = TSV::Dumper.new :key_field => "Key", :fields => ["Value"], :type => :single
+        dumper.init
+        dumper = TSV.traverse lines, :into => dumper, :cpus => 3 do |l|
+          raise ScoutException if i > 10
+          i += 1
+          value = l + "-" + Process.pid.to_s
+
+          [i.to_s, value]
+        end
+        ppp dumper.stream.read
+      end
+    end
+  end
+
+  def test_traverse_line
+    text=<<-EOF
+#: :sep=" "
+#Row LabelA LabelB LabelC
+row1 A B C
+row1 a b c
+row2 AA BB CC
+row2 aa bb cc
+    EOF
+
+    TmpFile.with_file(text) do |file|
+      lines = Open.traverse file, :type => :line, :into => [] do |line|
+        line
+      end
+      assert_include lines, "row2 AA BB CC"
+    end
+  end
+
+  def test_collapse_stream
+    text=<<-EOF
+#: :sep=" "
+#Row LabelA LabelB LabelC
+row1 A B C
+row1 a b c
+row2 AA BB CC
+row2 aa bb cc
+    EOF
+
+    s = StringIO.new text
+    collapsed = TSV.collapse_stream(s)
+    tsv = TSV.open collapsed 
+    assert_equal ["A", "a"], tsv["row1"][0]
+    assert_equal ["BB", "bb"], tsv["row2"][1]
+  end
+
+  def test_cpus_error_dumper
+    num_lines = 100
+    lines = num_lines.times.collect{|i| "line-#{i}" }
+
+    dumper =  TSV::Dumper.new :key_field => "Key", :fields => ["Field"], type: :single
+    dumper.init
+    assert_raise ScoutException do
+      Log.with_severity 0 do
+        i = 0
+        TSV.traverse lines, :into => dumper, cpus: 3 do |l|
+          raise ScoutException if i > 10
+          i += 1
+          [Process.pid.to_s, l + "-" + Process.pid.to_s]
+        end
+
+      end
+      ppp dumper.stream.read
+    end
+  end
+
+  def test_step_travese_cpus
+
+    size = 1000
+    step = Step.new tmpdir.step[__method__] do
+      lines = size.times.collect{|i| "line-#{i}" }
+      Open.traverse lines, :type => :array, :into => :stream, :cpus => 3 do |line|
+        line.reverse
+      end
+    end
+    step.type = :array
+
+    assert_equal size, step.run.length
   end
 end
 

data/test/scout/tsv/test_parser.rb

@@ -119,6 +119,20 @@ k a|A b|B
     assert_equal [%w(b B), %w(k)], tsv['k']
   end
 
+  def test_parse_flat
+    content =<<-EOF
+#: :sep=" "#:type=:flat
+#Key ValueA
+row1 a aa aaa
+row2 b bb bbb
+    EOF
+
+    tsv = TSV.open(content)
+    assert_equal %w(a aa aaa), tsv["row1"]
+    tsv = TSV.open(content, :fields => ["ValueA"])
+    assert_equal %w(a aa aaa), tsv["row1"]
+  end
+
   def test_parse_key
     content =<<-EOF
 #: :sep=" "#:type=:double
@@ -172,6 +186,25 @@ k a|A b|B
     assert_equal [["k", [%w(b B)]]], values
   end
 
+  def test_parser_traverse_all
+    content =<<-EOF
+Key ValueA ValueB
+k a|A b|B
+    EOF
+    content = StringIO.new content
+
+    parser = TSV::Parser.new content, sep: " ", header_hash: ''
+
+    assert_equal "Key", parser.key_field
+
+    values = []
+    parser.traverse key_field: "ValueA", fields: :all, type: :double do |k,v|
+      values << v
+    end
+
+    assert_include values.flatten, 'a'
+  end
+
 
   def test_parse_persist_serializer
     content =<<-EOF
@@ -192,7 +225,5 @@ k 1 2
       TSV.parse content, sep: " ", header_hash: '', data: data, cast: :to_i, type: :list, serializer: :float_array
       assert_equal [1.0, 2.0], data["k"]
     end
-
   end
-
 end

data/test/scout/tsv/test_persist.rb

@@ -29,6 +29,8 @@ row2    a    a    id3
       end
     end
 
+    assert_include tsv.keys, 'row1'
+    assert_include tsv.keys, 'row2'
 
     assert_nothing_raised do
       tsv = Persist.persist("TEST Persist TSV", :tsv) do 

data/test/scout/tsv/test_stream.rb

@@ -0,0 +1,200 @@
+require File.expand_path(__FILE__).sub(%r(/test/.*), '/test/test_helper.rb')
+require File.expand_path(__FILE__).sub(%r(.*/test/), '').sub(/test_(.*)\.rb/,'\1')
+
+require 'scout/tsv'
+class TestTSVStream < Test::Unit::TestCase
+  def test_paste_stream
+    text1=<<-EOF
+#: :sep=" "
+#Row LabelA LabelB LabelC
+row1 A B C
+row2 AA BB CC
+row3 AAA BBB CCC
+    EOF
+
+    text2=<<-EOF
+#: :sep=" "
+#Row Labela Labelb 
+row1 a b
+row2 aa bb
+row3 aaa bbb
+    EOF
+
+    text3=<<-EOF
+#: :sep=" "
+#Row LabelC
+row1 c
+row2 cc
+row3 ccc
+    EOF
+
+    s1 = StringIO.new text1
+    s2 = StringIO.new text2
+    s3 = StringIO.new text3
+    tsv = TSV.open TSV.paste_streams([s1,s2,s3], :sep => " ", :type => :list)
+    assert_equal ["A", "B", "C", "a", "b", "c"], tsv["row1"]
+    assert_equal ["AA", "BB", "CC", "aa", "bb", "cc"], tsv["row2"]
+    assert_equal ["AAA", "BBB", "CCC", "aaa", "bbb", "ccc"], tsv["row3"]
+  end
+
+  def test_paste_stream_sort
+    text1=<<-EOF
+#: :sep=" "
+#Row LabelA LabelB LabelC
+row2 AA BB CC
+row1 A B C
+row3 AAA BBB CCC
+    EOF
+
+    text2=<<-EOF
+#: :sep=" "
+#Row Labela Labelb
+row1 a b
+row3 aaa bbb
+row2 aa bb
+    EOF
+
+    text3=<<-EOF
+#: :sep=" "
+#Row Labelc
+row3 ccc
+row1 c
+row2 cc
+    EOF
+
+    s1 = StringIO.new text1
+    s2 = StringIO.new text2
+    s3 = StringIO.new text3
+    tsv = TSV.open TSV.paste_streams([s1,s2,s3], :sep => " ", :type => :list, :sort => true)
+    assert_equal "Row", tsv.key_field
+    assert_equal %w(LabelA LabelB LabelC Labela Labelb Labelc), tsv.fields
+    assert_equal ["A", "B", "C", "a", "b", "c"], tsv["row1"]
+    assert_equal ["AA", "BB", "CC", "aa", "bb", "cc"], tsv["row2"]
+    assert_equal ["AAA", "BBB", "CCC", "aaa", "bbb", "ccc"], tsv["row3"]
+  end
+
+  def test_paste_stream_missing_2
+    text1=<<-EOF
+#: :sep=" "
+#Row LabelA LabelB LabelC
+row2 AA BB CC
+row1 A B C
+    EOF
+
+    text2=<<-EOF
+#: :sep=" "
+#Row Labela Labelb
+row2 aa bb
+    EOF
+
+    text3=<<-EOF
+#: :sep=" "
+#Row Labelc
+row3 ccc
+row2 cc
+    EOF
+
+    s1 = StringIO.new text1
+    s2 = StringIO.new text2
+    s3 = StringIO.new text3
+    tsv = TSV.open TSV.paste_streams([s1,s2,s3], :sep => " ", :type => :list, :sort => true)
+    assert_equal "Row", tsv.key_field
+    assert_equal %w(LabelA LabelB LabelC Labela Labelb Labelc), tsv.fields
+    assert_equal ["A", "B", "C", "", "", ""], tsv["row1"]
+    assert_equal ["AA", "BB", "CC", "aa", "bb", "cc"], tsv["row2"]
+    assert_equal ["", "", "", "", "", "ccc"], tsv["row3"]
+  end
+
+  def test_paste_stream_missing
+    text1=<<-EOF
+#: :sep=" "
+#Row LabelA LabelB LabelC
+row2 AA BB CC
+row1 A B C
+    EOF
+
+    text2=<<-EOF
+#: :sep=" "
+#Row Labela Labelb
+row2 aa bb
+    EOF
+
+    text3=<<-EOF
+#: :sep=" "
+#Row Labelc
+row3 ccc
+row2 cc
+    EOF
+
+    s1 = StringIO.new text1
+    s2 = StringIO.new text2
+    s3 = StringIO.new text3
+    tsv = TSV.open TSV.paste_streams([s1,s2,s3], :sep => " ", :type => :list, :sort => true)
+    assert_equal "Row", tsv.key_field
+    assert_equal %w(LabelA LabelB LabelC Labela Labelb Labelc), tsv.fields
+    assert_equal ["A", "B", "C", "", "", ""], tsv["row1"]
+    assert_equal ["AA", "BB", "CC", "aa", "bb", "cc"], tsv["row2"]
+    assert_equal ["", "", "", "", "", "ccc"], tsv["row3"]
+  end
+
+  def test_paste_stream_missing_3
+    text1=<<-EOF
+#: :sep=" "
+#Row LabelA LabelB LabelC
+row2 AA BB CC
+row1 A B C
+    EOF
+
+    text2=<<-EOF
+#: :sep=" "
+#Row Labelc
+    EOF
+
+    s1 = StringIO.new text1
+    s2 = StringIO.new text2
+    tsv = TSV.open TSV.paste_streams([s1,s2], :sep => " ", :type => :list, :sort => true)
+    assert_equal "Row", tsv.key_field
+    assert_equal %w(LabelA LabelB LabelC Labelc), tsv.fields
+    assert_equal ["A", "B", "C", ""], tsv["row1"]
+    assert_equal ["AA", "BB", "CC", ""], tsv["row2"]
+  end
+
+  def test_paste_stream_same_field
+    text1=<<-EOF
+#: :sep=" "
+#Row LabelA
+row1 A
+row2 AA
+    EOF
+
+    text2=<<-EOF
+#: :sep=" "
+#Row LabelA
+row2 AAA
+    EOF
+
+    s1 = StringIO.new text1
+    s2 = StringIO.new text2
+    tsv = TSV.open TSV.paste_streams([s1,s2], :sep => " ", :type => :double, :sort => false, :same_fields => true)
+    assert_equal "Row", tsv.key_field
+    assert_equal ["AA", "AAA"], tsv["row2"][0]
+  end
+
+  def test_paste_stream_nohead
+    text1=<<-EOF
+row1\tA
+row2\tAA
+    EOF
+
+    text2=<<-EOF
+row2\tAAA
+    EOF
+
+    s1 = StringIO.new text1
+    s2 = StringIO.new text2
+    tsv = TSV.open TSV.paste_streams([s1,s2], :type => :double, :sort => false, :same_fields => true)
+    assert_equal ["AA", "AAA"], tsv["row2"][0]
+  end
+
+end
+