rbbt-util 3.2.1 → 4.0.0

data/test/rbbt/tsv/test_accessor.rb

@@ -0,0 +1,163 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../test_helper')
+require 'rbbt/tsv'
+require 'rbbt/util/tmpfile'
+require 'test/unit'
+
+class TestTSV < Test::Unit::TestCase
+
+  def _test_tsv
+    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.open(filename, :sep => /\s+/)
+      assert_equal 2, tsv.keys.length
+      assert_equal 2, tsv.values.length
+      assert_equal 2, tsv.collect.length
+    end
+  end
+
+  def _test_named_values
+    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.open(filename, :sep => /\s+/)
+      assert_equal ["A"], tsv["row2"]["ValueA"]
+    end
+  end
+
+  def _test_to_s
+    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.open(filename, :sep => /\s+/)
+      assert tsv.to_s =~ /row1\ta|aa|aaa/
+      assert tsv.to_s =~ /:type=:double/
+    end
+  end
+  
+  def _test_entries
+    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.open(filename, :sep => /\s+/)
+      assert_equal filename, tsv.options[:filename]
+    end
+ 
+  end
+
+  def _test_marshal
+    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.open(filename, :sep => /\s+/, :persist_serializer => :marshal, :persist => true)
+      assert_equal filename, tsv.options[:filename]
+    end
+  end
+
+  def _test_zip_fields
+    a = [%w(1 2), %w(a b)]
+    assert_equal a, TSV.zip_fields(TSV.zip_fields(a))
+  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.open(File.open(filename), :double, :sep => /\s/)
+      assert_equal :key, tsv.identify_field("ID")
+    end
+  end
+
+  def _test_sort
+    content =<<-EOF
+#ID ValueA ValueB Comment
+row1 a B c
+row2 A b C
+    EOF
+
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.open(File.open(filename), :double, :sep => /\s/)
+      assert_equal %w(row2 row1), tsv.sort{|a,b|
+        a[1]["ValueA"] <=> b[1]["ValueA"]
+      }.collect{|k,v| k}
+      assert_equal %w(row1 row2), tsv.sort{|a,b|
+        a[1]["ValueB"] <=> b[1]["ValueB"]
+      }.collect{|k,v| k}
+    end
+  end
+
+  def _test_sort_by
+    content =<<-EOF
+#ID ValueA ValueB Comment
+row1 a B c
+row2 A b C
+    EOF
+
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.open(File.open(filename), :list, :sep => /\s/)
+      assert_equal %w(row2 row1), tsv.sort_by("ValueA").collect{|k,v| k}
+      assert_equal %w(row1 row2), tsv.sort_by("ValueB").collect{|k,v| k}
+    end
+  end
+
+
+  def test_page
+    content =<<-EOF
+#ID ValueA ValueB Comment
+row1 a B f
+row2 A b e
+row3 A b d
+row4 A b c
+row5 A b b
+row6 A b a
+    EOF
+
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.open(File.open(filename), :list, :sep => /\s/)
+      assert_equal 3, tsv.page(1,3).size
+      assert_equal %w(row1 row2 row3).sort, tsv.page(1,3).keys.sort
+      assert_equal %w(row6 row5 row4).sort, tsv.page(1,3, "Comment").keys.sort
+      assert_equal %w(row4 row3).sort, tsv.page(2,2, "Comment").keys.sort
+    end
+  end
+
+
+  def _test_sort_by_with_proc
+    content =<<-EOF
+#Id    ValueA    ValueB    OtherID    Pos
+row1    a|aa|aaa    b    Id1|Id2    2
+row2    aA    B    Id3    1
+row3    A|AA|AAA|AAA    B    Id3    3
+    EOF
+
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.open(File.open(filename), :sep => /\s+/)
+      assert_equal %w(row3 row1 row2), (tsv.sort_by("ValueA") do |key, value| value.length end).collect{|k,v| k}.reverse
+    end
+  end
+
+end

data/test/rbbt/{util/tsv → tsv}/test_attach.rb

@@ -1,6 +1,6 @@
-require File.join(File.expand_path(File.dirname(__FILE__)), '../../..', 'test_helper.rb')
-require 'rbbt/util/tsv'
-require 'rbbt/util/tsv/attach'
+require File.join(File.expand_path(File.dirname(__FILE__)), '../..', 'test_helper.rb')
+require 'rbbt/tsv'
+require 'rbbt/tsv/attach'
 require 'rbbt'
 
 class TestAttach < Test::Unit::TestCase
@@ -19,11 +19,11 @@ row3    B    Id3
 
     tsv1 = tsv2 = nil
     TmpFile.with_file(content1) do |filename|
-      tsv1 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+      tsv1 = TSV.open(File.open(filename), :double, :sep => /\s+/)
     end
 
     TmpFile.with_file(content2) do |filename|
-      tsv2 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+      tsv2 = TSV.open(File.open(filename), :double, :sep => /\s+/)
     end
 
     tsv1.attach_same_key tsv2, "OtherID"
@@ -32,7 +32,7 @@ row3    B    Id3
     assert_equal %w(Id1 Id2), tsv1["row1"]["OtherID"]
 
     TmpFile.with_file(content1) do |filename|
-      tsv1 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+      tsv1 = TSV.open(File.open(filename), :double, :sep => /\s+/)
     end
 
     tsv1.attach_same_key tsv2
@@ -41,11 +41,11 @@ row3    B    Id3
 
     tsv1 = tsv2 = nil
     TmpFile.with_file(content1) do |filename|
-      tsv1 = TSV.new(File.open(filename), :list, :sep => /\s+/)
+      tsv1 = TSV.open(File.open(filename), :list, :sep => /\s+/)
     end
 
     TmpFile.with_file(content2) do |filename|
-      tsv2 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+      tsv2 = TSV.open(File.open(filename), :double, :sep => /\s+/)
     end
 
     tsv1.attach_same_key tsv2, "OtherID"
@@ -69,11 +69,11 @@ B    Id3
 
     tsv1 = tsv2 = nil
     TmpFile.with_file(content1) do |filename|
-      tsv1 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+      tsv1 = TSV.open(File.open(filename), :double, :sep => /\s+/)
     end
 
     TmpFile.with_file(content2) do |filename|
-      tsv2 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+      tsv2 = TSV.open(File.open(filename), :double, :sep => /\s+/)
     end
 
     tsv1.attach_source_key tsv2, "ValueB"
@@ -82,7 +82,7 @@ B    Id3
     assert_equal %w(Id1 Id2), tsv1["row1"]["OtherID"]
 
     TmpFile.with_file(content1) do |filename|
-      tsv1 = TSV.new(File.open(filename), :list, :sep => /\s+/)
+      tsv1 = TSV.open(File.open(filename), :list, :sep => /\s+/)
     end
 
     tsv1.attach_source_key tsv2, "ValueB"
@@ -113,15 +113,15 @@ row2    E
 
     tsv1 = tsv2 = index = nil
     TmpFile.with_file(content1) do |filename|
-      tsv1 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+      tsv1 = TSV.open(File.open(filename), :double, :sep => /\s+/)
     end
 
     TmpFile.with_file(content2) do |filename|
-      tsv2 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+      tsv2 = TSV.open(File.open(filename), :double, :sep => /\s+/)
     end
 
     TmpFile.with_file(content_index) do |filename|
-      index = TSV.new(File.open(filename), :flat, :sep => /\s+/)
+      index = TSV.open(File.open(filename), :flat, :sep => /\s+/)
     end
 
     tsv1.attach_index tsv2, index
@@ -130,7 +130,7 @@ row2    E
     assert_equal %w(Id1 Id2), tsv1["row1"]["OtherID"]
 
     TmpFile.with_file(content1) do |filename|
-      tsv1 = TSV.new(File.open(filename), :list, :sep => /\s+/)
+      tsv1 = TSV.open(File.open(filename), :list, :sep => /\s+/)
     end
 
     tsv1.attach_index tsv2, index
@@ -159,15 +159,15 @@ B    Id3
     EOF
    tsv1 = tsv2 = tsv3 = nil
     TmpFile.with_file(content1) do |filename|
-      tsv1 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+      tsv1 = TSV.open(File.open(filename), :double, :sep => /\s+/)
     end
 
     TmpFile.with_file(content2) do |filename|
-      tsv2 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+      tsv2 = TSV.open(File.open(filename), :double, :sep => /\s+/)
     end
 
     TmpFile.with_file(content3) do |filename|
-      tsv3 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+      tsv3 = TSV.open(File.open(filename), :double, :sep => /\s+/)
     end
 
     tsv1.attach tsv2, "OtherID"
@@ -176,7 +176,7 @@ B    Id3
     assert_equal %w(Id1 Id2), tsv1["row1"]["OtherID"]
 
     TmpFile.with_file(content1) do |filename|
-      tsv1 = TSV.new(File.open(filename), :double, :sep => /\s+/)
+      tsv1 = TSV.open(File.open(filename), :double, :sep => /\s+/)
     end
 
     tsv1.attach tsv3
@@ -188,12 +188,14 @@ B    Id3
 
   def test_attach_using_index
     content1 =<<-EOF
+#: :sep=/\\s+/
 #Id    ValueA    ValueB
 row1    a|aa|aaa    b
 row2    A    B
     EOF
 
     content2 =<<-EOF
+#: :sep=/\\s+/
 #ValueE    OtherID
 e    Id1|Id2
 E    Id3
@@ -206,19 +208,19 @@ row1    e
 row2    E
     EOF
 
-    Rbbt.tmp.test.test1.data.define_as_string content1
-    Rbbt.tmp.test.test2.data.define_as_string content2
-    Rbbt.tmp.test.test2.identifiers.define_as_string content_index
+    Rbbt.claim Rbbt.tmp.test.test1.data, :string, content1
+    Rbbt.claim Rbbt.tmp.test.test2.data, :string, content2
+    Rbbt.claim Rbbt.tmp.test.test2.identifiers, :string, content_index
 
     tsv1 = tsv2 = nil
 
     tsv1 = Rbbt.tmp.test.test1.data.tsv :double,  :sep => /\s+/
     tsv2 = Rbbt.tmp.test.test2.data.tsv :double,  :sep => /\s+/
 
-    tsv2.identifiers = Rbbt.tmp.test.test2.identifiers.produce
-
-    tsv1.attach tsv2, "OtherID", :in_namespace => false
+    tsv2.identifiers = Rbbt.tmp.test.test2.identifiers.produce.find.to_s
 
+    tsv1.attach tsv2, :fields => ["OtherID"], :persist_input => true
+    
     assert_equal tsv1.fields,%w(ValueA ValueB OtherID)
     assert_equal %w(Id1 Id2), tsv1["row1"]["OtherID"]
   end
@@ -247,24 +249,25 @@ row2    E
 
     tsv1 = tsv2 = identifiers = nil
     TmpFile.with_file(content1) do |filename|
-      tsv1 = TSV.new(File.open(filename), :key => "Id")
+      tsv1 = TSV.open(Path.setup(filename), :key => "Id")
     end
 
     TmpFile.with_file(content2) do |filename|
-      tsv2 = TSV.new(File.open(filename), :double)
+      tsv2 = TSV.open(Path.setup(filename), :double)
     end
 
     TmpFile.with_file(content_identifiers) do |filename|
-      identifiers = TSV.new(File.open(filename), :flat, :sep => /\s+/)
+      identifiers = TSV.open(Path.setup(filename), :flat, :sep => /\s+/)
     end
 
     tsv1.identifiers = identifiers
+
     tsv1.attach tsv2
  
     assert_equal %w(ValueA ValueB ValueE), tsv1.fields
   end
 
-  def test_paste
+  def test_merge_different_rows
     file1 =<<-EOF
 row6 dd dd ee
 row1 a b c
@@ -287,12 +290,12 @@ row6 dd dd ee
     EOF
 
     TmpFile.with_file do |f|
-      TSV.paste(StringIO.new(file1), StringIO.new(file2), f, " ")
+      TSV.merge_different_fields(StringIO.new(file1), StringIO.new(file2), f, " ")
       assert_equal result, Open.read(f)
     end
   end
 
-  def test_paste
+  def test_merge_different_rows_tsv
     file1 =<<-EOF
 row6 dd dd ee
 row1 a b c
@@ -315,14 +318,14 @@ row6 dd dd ee
     EOF
 
     TmpFile.with_file do |f|
-      tsv1 = TSV.new StringIO.new(file1), :sep => " "
-      tsv2 = TSV.new StringIO.new(file2), :sep => " "
-      tsv_r = tsv1.paste tsv2
+      tsv1 = TSV.open StringIO.new(file1), :sep => " "
+      tsv2 = TSV.open StringIO.new(file2), :sep => " "
+      tsv_r = tsv1.merge_different_fields tsv2
       assert_equal result, tsv_r.to_s(tsv_r.keys.sort, true).gsub(/\t/,' ')
     end
   end
 
-  def test_paste_merge
+  def test_merge_different_rows_split_lines
     file1 =<<-EOF
 row6,dd,dd,ee
 row1,a,b,c
@@ -336,23 +339,34 @@ row1,d,e
 row2,D,E
 row4,x,y
     EOF
-    result =<<-EOF
+
+    # Might be slightly different ...
+    result1 =<<-EOF
 row1,aa|a,bb|b,cc|c,d,e
 row2,A,B,C,D,E
 row20,,,,rr,rr
 row3,1,2,3,,
 row4,,,,x,y
+row6,dd,dd,ee,,
+    EOF
+    result2 =<<-EOF
+row1,a|aa,b|bb,c|cc,d,e
+row2,A,B,C,D,E
+row20,,,,rr,rr
+row3,1,2,3,,
+row4,,,,x,y
 row6,dd,dd,ee,,
     EOF
 
     TmpFile.with_file do |f|
-      TSV.paste_merge StringIO.new(file1), StringIO.new(file2), f, ','
+      TSV.merge_different_fields StringIO.new(file1), StringIO.new(file2), f, ','
 
-      assert_equal result, Open.read(f)
+      # ... so check for either
+      assert(Open.read(f) == result1 || Open.read(f) == result2)
     end
   end
 
-  def test_paste
+  def test_merge_different_rows_split_lines_tsv
     file1 =<<-EOF
 row6,dd,dd,ee
 row1,a,b,c
@@ -376,9 +390,9 @@ row6,dd,dd,ee,,
     EOF
 
     TmpFile.with_file do |f|
-      data1 = TSV.new StringIO.new(file1), :sep => ',', :merge => true
-      data2 = TSV.new StringIO.new(file2), :sep => ',', :merge => true
-      data3 = data1.paste(data2)
+      data1 = TSV.open StringIO.new(file1), :sep => ',', :merge => true
+      data2 = TSV.open StringIO.new(file2), :sep => ',', :merge => true
+      data3 = data1.merge_different_fields(data2)
       data3.each do |key, list|
         list.each do |l| l.replace l.sort_by{|v| v.length}.reverse end
       end
@@ -396,12 +410,41 @@ row3,1,2,3
     EOF
     TmpFile.with_file(file1) do |input|
       TmpFile.with_file() do |output|
-        TSV.merge_rows Open.open(input), output
+        TSV.merge_row_fields Open.open(input), output
         assert Open.read(output) =~ /a|aa/
       end
     end
 
 
   end
+
+  def test_one2one
+    content1 =<<-EOF
+#Id    ValueA    ValueB
+row1    a|aa|aaa    b
+row2    A    B
+row3    A    b|bb
+    EOF
+
+    content2 =<<-EOF
+#ValueB    OtherID
+b    Id1|Id2
+B    Id3
+bb   Id4
+    EOF
+   tsv1 = tsv2 = nil
+    TmpFile.with_file(content1) do |filename|
+      tsv1 = TSV.open(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    TmpFile.with_file(content2) do |filename|
+      tsv2 = TSV.open(File.open(filename), :double, :sep => /\s+/)
+    end
+
+    tsv1.attach tsv2, :one2one => true
+
+    assert_equal %w(ValueA ValueB OtherID), tsv1.fields
+    assert_equal %w(Id1 Id4), tsv1["row3"]["OtherID"]
+  end
 end