scout-gear 7.3.0 → 8.0.0

data/test/scout/tsv/test_index.rb

@@ -48,19 +48,30 @@ row2    a    b    id3
 #: :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 +85,7 @@ 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 +118,27 @@ 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

@@ -172,6 +172,25 @@ 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 +211,5 @@ 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_transformer.rb

@@ -0,0 +1,108 @@
+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 TestTSVTransformer < Test::Unit::TestCase
+  def test_traverse
+    content =<<-EOF
+#: :sep=" "
+#ID    ValueA    ValueB
+row1    A1|A11    B1|B11
+row2    A2|A22    B2|B22
+    EOF
+
+    TmpFile.with_file(content) do |file|
+      parser = TSV::Parser.new file
+      dumper = TSV::Dumper.new :key_field => "Key", :fields => ["Values"], type: :flat
+      dumper.init
+
+      trans = TSV::Transformer.new parser, dumper
+      dumper = trans.traverse do |k,values|
+        [k, values.flatten]
+      end
+
+      tsv = trans.tsv
+      assert_equal %w(A1 A11 B1 B11), tsv['row1']
+    end
+  end
+
+  def test_each
+    content =<<-EOF
+#: :sep=" "
+#ID    ValueA    ValueB
+row1    A1|A11    B1|B11
+row2    A2|A22    B2|B22
+    EOF
+
+    TmpFile.with_file(content) do |file|
+      parser = TSV::Parser.new file
+      dumper = TSV::Dumper.new :key_field => "Key", :fields => ["Values"], type: :flat
+      dumper.init
+
+      trans = TSV::Transformer.new parser, dumper
+      dumper = trans.each do |k,values|
+        values.replace values.flatten
+      end
+
+      trans["row3"] = %w(A3 A33)
+
+      tsv = trans.tsv
+      assert_equal %w(A1 A11 B1 B11), tsv['row1']
+      assert_equal %w(A3 A33), tsv['row3']
+    end
+  end
+
+  def test_no_dumper_no_parser
+    content =<<-EOF
+#: :sep=" "
+#ID    ValueA    ValueB
+row1    A1|A11    B1|B11
+row2    A2|A22    B2|B22
+    EOF
+
+    TmpFile.with_file(content) do |file|
+      trans = TSV::Transformer.new file
+      trans.key_field = "Key"
+      trans.fields = ["Values"]
+      trans.type = :flat
+      trans.sep = "\t"
+
+
+      trans["row3"] = %w(A3 A33)
+
+      trans.each do |k,values|
+        values.replace values.flatten
+      end
+
+      tsv = trans.tsv
+      assert_equal %w(A1 A11 B1 B11), tsv['row1']
+      assert_equal %w(A3 A33), tsv['row3']
+    end
+  end
+
+  def test_to_list
+    content =<<-EOF
+#: :sep=" "
+#ID    ValueA    ValueB
+row1    A1|A11    B1|B11
+row2    A2|A22    B2|B22
+    EOF
+
+    tsv = TSV.open(content)
+    assert_equal "A1", tsv.to_list["row1"]["ValueA"]
+    assert_equal "B2", tsv.to_list["row2"]["ValueB"]
+  end
+
+  def test_to_single
+    content =<<-EOF
+#: :sep=" "
+#ID    ValueA    ValueB
+row1    A1|A11    B1|B11
+row2    A2|A22    B2|B22
+    EOF
+
+    tsv = TSV.open(content)
+    assert_equal "A1", tsv.to_single["row1"]
+    assert_equal "A2", tsv.to_single["row2"]
+  end
+end

data/test/scout/tsv/test_traverse.rb

@@ -13,7 +13,6 @@ row2    A    B    Id3
 row2    AA    BB    Id33
     EOF
 
-    sss 0
     tsv = TmpFile.with_file(content) do |filename|
       TSV.open(filename, :persist => true)
     end
@@ -31,18 +30,26 @@ row2    AA    BB    Id33
     assert_equal ["Id3", "B"], res["row2"]
 
     res = {}
-    tsv.traverse "OtherID", %w(Id ValueB), one2one: true do |k,v|
+    tsv.traverse "OtherID", %w(Id ValueB), one2one: :strict do |k,v|
       res[k] = v
     end
     assert_equal [[nil], %w(BB)], res["Id33"]
 
     res = {}
-    tsv.traverse "OtherID", %w(Id ValueB), one2one: true, type: :list do |k,v|
+    tsv.traverse "OtherID", %w(Id ValueB), one2one: :strict, type: :list do |k,v|
       res[k] = v
     end
     assert_equal ["row2", "B"], res["Id3"]
     assert_equal [nil, "BB"], res["Id33"]
 
+    res = {}
+    tsv.traverse "OtherID", %w(Id ValueB), one2one: true, type: :list do |k,v|
+      res[k] = v
+    end
+    assert_equal ["row2", "B"], res["Id3"]
+    assert_equal ["row2", "BB"], res["Id33"]
+
+
     tsv.traverse "OtherID", %w(Id ValueB), one2one: false, type: :list do |k,v|
       res[k] = v
     end
@@ -57,6 +64,36 @@ row2    AA    BB    Id33
     assert_equal %w(Id ValueA ValueB), field_names
   end
 
+  def test_tsv_traverse_all
+    content =<<-'EOF'
+#: :sep=/\s+/#:type=:double
+#Id    ValueA    ValueB    OtherID
+row1    a|aa|aaa    b    Id1|Id2
+row2    A    B    Id3
+row2    AA    BB    Id33
+    EOF
+
+    tsv = TmpFile.with_file(content) do |filename|
+      TSV.open(filename, :persist => true)
+    end
+
+    all_values = []
+    tsv.traverse "ValueA", :all do |k,v|
+      all_values.concat(v)
+    end
+    assert_include all_values.flatten, "row1"
+    assert_include all_values.flatten, "a"
+    assert_include all_values.flatten, "aaa"
+
+    all_values = []
+    tsv.traverse "Id", :all do |k,v|
+      all_values.concat(v)
+    end
+    assert_include all_values.flatten, "row1"
+    assert_include all_values.flatten, "a"
+    assert_include all_values.flatten, "aaa"
+  end
+
   def test_tsv_traverse_list
     content =<<-'EOF'
 #: :sep=/\s+/#:type=:list
@@ -112,5 +149,53 @@ row2    A
     end
     assert_equal [["row1"]], res["a"]
   end
+
+  def test_traverse_reorder_one2one
+    content =<<-'EOF'
+#: :sep=" "
+#ID    ValueA    ValueB
+row1    A1|A11    B1|B11
+row2    A2|A22    B2|B22
+    EOF
+
+    tsv = TmpFile.with_file(content) do |filename|
+      TSV.open(filename, :persist => true)
+    end
+
+    res = {}
+    tsv.traverse "ValueA", one2one: true do |k,v|
+      res[k] = v
+    end
+
+    assert_include res["A2"][0], "row2"
+
+  end
+
+  def test_traverse_select
+    content =<<-'EOF'
+#: :sep=" "
+#ID    ValueA    ValueB
+row1    A1|A11    B1|B11
+row2    A2|A22    B2|B22
+    EOF
+
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.open(filename, :persist => true, select: "B1")
+
+      assert_equal %w(row1), tsv.keys
+
+      tsv = TSV.open(filename, :persist => true, select: "B2")
+
+      assert_equal %w(row2), tsv.keys
+
+      tsv = TSV.open(filename, :persist => true, select: "B1")
+
+      assert_equal %w(row1), tsv.keys
+
+      tsv = TSV.open(filename, :persist => true, select: "B2")
+
+      assert_equal %w(row2), tsv.keys
+    end
+  end
 end
 

data/test/scout/tsv/test_util.rb

@@ -17,6 +17,7 @@ row2    a    a    id3
     end
     assert_equal %w(row1 row2), tsv.collect{|k,v| k }
     assert NamedArray === tsv.collect{|k,v| v }.first
+    assert "row1", tsv["row1"].key
   end
 
 end

data/test/scout/tsv/util/test_reorder.rb

@@ -0,0 +1,94 @@
+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 TestReorder < Test::Unit::TestCase
+  def test_reorder
+    content =<<-'EOF'
+#: :sep=/\s+/#:type=:double
+#Id    ValueA    ValueB    OtherID
+row1    a1|a2    b1|b2    Id1|Id2
+row2    A1|A3    B1|B3    Id1|Id3
+    EOF
+
+    tsv = TmpFile.with_file(content) do |filename|
+      TSV.open(filename)
+    end
+
+    r = tsv.reorder "OtherID", %w(ValueB Id)
+
+    assert_equal %w(row1 row2), r["Id1"]["Id"]
+    assert_equal %w(row2), r["Id3"]["Id"]
+  end
+
+  def test_reorder_list
+    content =<<-'EOF'
+#: :sep=/\s+/#:type=:list
+#Id    ValueA    ValueB    OtherID
+row1    a1    b1    Id1
+row2    A1    B1    Id1
+    EOF
+
+    tsv = TmpFile.with_file(content) do |filename|
+      TSV.open(filename)
+    end
+
+    r = tsv.reorder "ValueB"
+
+    assert_equal "row1", r["b1"]["Id"]
+    assert_equal "row2", r["B1"]["Id"]
+  end
+
+  def test_reorder_single
+    content =<<-'EOF'
+#: :sep=/\s+/#:type=:single
+#Id    ValueA
+row1    a1
+row2    A1
+    EOF
+
+    tsv = TmpFile.with_file(content) do |filename|
+      TSV.open(filename)
+    end
+
+    r = tsv.reorder "ValueA"
+
+    assert_equal "row1", r["a1"]
+    assert_equal "row2", r["A1"]
+  end
+
+  def test_transpose
+     content =<<-EOF
+#: :type=:list
+#Row   vA   vB   vID
+row1    a    b    Id1
+row2    A    B    Id3
+row3    a    C    Id4
+    EOF
+ 
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.open(filename, :sep => /\s+/)
+
+      assert_equal %w(vA vB vID),  tsv.transpose("Values").keys
+      assert_equal %w(Id1 Id3 Id4),  tsv.transpose("Values")["vID"]
+    end
+  end
+
+  def test_column
+    content =<<-EOF
+#Id    ValueA    ValueB ValueC
+rowA    A|AA    B|BB  C|CC
+rowa    a|aa    b|BB  C|CC
+    EOF
+
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.open(File.open(filename), :sep => /\s+/, :type => :double)
+      tsv = tsv.column("ValueA", cast: :downcase)
+      assert_equal %w(a aa), tsv["rowA"]
+      assert_equal %w(a aa), tsv["rowa"]
+    end
+  end
+
+end
+

data/test/scout/tsv/util/test_select.rb

@@ -23,22 +23,36 @@ row2    AA    BB    Id33
     assert_equal ['row2'], s.keys
   end
 
-  def test_reorder
+  def test_select_values
     content =<<-'EOF'
-#: :sep=/\s+/#:type=:double
-#Id    ValueA    ValueB    OtherID
-row1    a1|a2    b1|b2    Id1|Id2
-row2    A1|A3    B1|B3    Id1|Id3
+#: :sep=" "
+#ID    ValueA    ValueB
+row1    A1|A11    B1|B11
+row2    A2|A22    B2|B22
     EOF
 
-    tsv = TmpFile.with_file(content) do |filename|
-      TSV.open(filename)
-    end
+    tsv = TSV.open(content)
+
+    assert TSV.select("row1", tsv["row1"], "A1")
+    refute TSV.select("row2", tsv["row2"], "A1")
+    assert TSV.select("row2", tsv["row2"], "A2")
 
-    r = tsv.reorder "OtherID", %w(ValueB Id)
+    assert TSV.select("row1", tsv["row1"], "B1")
+    refute TSV.select("row2", tsv["row2"], "B1")
+
+    refute TSV.select("row1", tsv["row1"], "B1", invert: true)
+    assert TSV.select("row2", tsv["row2"], "B1", invert: true)
+
+    assert TSV.select("row1", tsv["row1"], {"ValueB" => "B1"})
+    refute TSV.select("row2", tsv["row2"], {"ValueB" => "B1"})
+
+    tsv.with_unnamed do
+      assert TSV.select("row1", tsv["row1"], {"ValueB" => "B1"}, fields: tsv.fields)
+      refute TSV.select("row2", tsv["row2"], {"ValueB" => "B1"}, fields: tsv.fields)
+    end
 
-    assert_equal %w(row1 row2), r["Id1"]["Id"]
-    assert_equal %w(row2), r["Id3"]["Id"]
+    assert TSV.select("row1", tsv["row1"], "B1", field: "ValueB")
+    refute TSV.select("row2", tsv["row2"], "B1", field: "ValueB")
   end
 end