rbbt-util 1.1.0 → 1.2.1

data/lib/rbbt/util/workflow.rb

@@ -1,4 +1,3 @@
-require 'tsv'
-
 module WorkFlow
+
 end

data/share/lib/R/util.R

@@ -0,0 +1,89 @@
+rbbt.glob <- function(d, pattern){
+    d=sub("/$", '', d);
+    sapply(dir(d, pattern), function(file){paste(d,file,sep="/")});
+}
+
+rbbt.png_plot <- function(filename, width, height, p){
+    png(filename="temp.png", width=width, height=height);
+    eval(p);
+    dev.off();
+}
+
+rbbt.load.data <- function(filename, sep = "\t",  ...){
+  data=read.table(file=filename, sep=sep, fill=TRUE,  as.is=TRUE, ...);
+  return(data);
+}
+
+rbbt.tsv <- function(filename, sep = "\t", comment.char ="#",  ...){
+  data=read.table(file=filename, sep=sep, fill=TRUE,  as.is=TRUE, row.names=1, comment.char = comment.char, ...);
+  f = file(filename, 'r');
+  headers = readLines(f, 1);
+  close(f);
+  if (grep(paste("^", comment.char, sep=""), headers)){
+      fields = strsplit(headers, sep)[[1]];
+      fields = fields[2:length(fields)];
+      names(data) <- fields;
+  }
+  return(data);
+}
+
+rbbt.tsv.write <- function(filename, data, key.field = NULL){
+  if (is.null(key.field)){ key.field = "ID";}
+
+  f = file(filename, 'w');
+
+  header = paste("#", key.field, sep="");
+  for (name in colnames(data)){ header = paste(header, name, sep="\t");}
+  header = paste(header, "\n", sep="");
+  cat(header, file=f);
+  cat(header, file=stderr());
+  
+  close(f);
+
+  write.table(data, file=filename, quote=FALSE, append=TRUE, col.names=FALSE, row.names=TRUE, sep="\t");
+
+  return(NULL);
+}
+
+rbbt.print.data <- function(data, file="", ...){
+    write.table(data, quote=FALSE, row.name=FALSE,col.name=FALSE,file=file, ...);
+}
+
+rbbt.percent <- function(values){
+    values=values/sum(values);
+}
+
+rbbt.sort_by_field <- function(data, field, is.numeric=TRUE){
+    if (is.numeric){
+        field.data=as.numeric(data[,field]);
+    }else{
+        field.data=data[,field];
+    }
+    index = sort(field.data, index.return = TRUE)$ix;
+    return(data[index,]);
+}
+
+rbbt.add <- function(data, new){
+    if (is.null(data)){
+        return(new);
+    }else{
+        return(c(data, new));
+    }
+}
+
+rbbt.acc <- function(data, new){
+    if (is.null(data)){
+        return(new);
+    }else{
+        return(unique(c(data, new)));
+    }
+}
+
+rbbt.init <- function(data, new){
+    if (is.null(data)){
+        return(new);
+    }else{
+        return(data);
+    }
+}
+

data/test/rbbt/util/test_R.rb

@@ -0,0 +1,9 @@
+require File.join(File.expand_path(File.dirname(__FILE__)), '../..', 'test_helper.rb')
+require 'rbbt/util/R'
+
+class TestR < Test::Unit::TestCase
+  def test_sum
+    assert_equal "6", R.run('cat(3+3)').read
+  end
+end
+

data/test/rbbt/util/test_bed.rb

@@ -0,0 +1,136 @@
+require File.join(File.expand_path(File.dirname(__FILE__)), '../..', 'test_helper.rb')
+require 'rbbt/util/bed'
+require 'rbbt/util/tmpfile'
+
+class TestBed < Test::Unit::TestCase
+  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
+ 
+    tsv
+  end
+
+  def test_point
+    data =<<-EOF
+#ID Pos
+a 1
+b 10
+c 20
+d 12
+e 26
+f 11
+g 25
+    EOF
+
+    TmpFile.with_file(data) do |f|
+      bed = Bed.new TSV.new(f, :list, :sep=>" "), :key => "Pos" , :value => "ID"
+
+      assert_equal %w(), bed[0].sort
+      assert_equal %w(b), bed[10].sort
+      assert_equal %w(a b c d f), bed[(0..20)].sort
+    end
+  end
+
+  def test_range
+    data =<<-EOF
+#ID:Range
+#:012345678901234567890
+a:   ______
+b: ______
+c:    _______
+d:  ____
+e:    ______
+f:             ___
+g:         ____
+    EOF
+
+    TmpFile.with_file(data) do |f|
+      bed = Bed.new load_data(f), :range => ["Start" , "End"], :value => "ID"
+
+      assert_equal %w(), bed[0].sort
+      assert_equal %w(b), bed[1].sort
+      assert_equal %w(), bed[20].sort
+      assert_equal %w(), bed[(20..100)].sort
+      assert_equal %w(a b d), bed[3].sort
+      assert_equal %w(a b c d e), bed[(3..4)].sort
+    end
+  end
+
+  def test_format
+    entry = Bed::Entry.new "Hola", 0, 10, nil
+    format = Bed::FixWidthTable.format(entry,1, 100)
+    assert_equal entry, Bed::FixWidthTable.unformat(format, 100)
+  end
+
+  def test_table
+    TmpFile.with_file do |f|
+      table = Bed::FixWidthTable.new(f, 100, false)
+      table.add Bed::Entry.new("Entry1", 0, 10, nil), 0
+      table.add Bed::Entry.new("Entry2", 10, 20, nil), 10
+      table.add Bed::Entry.new("Entry3", 10, 20, nil), 10
+
+      table.read 
+
+      assert_equal "Entry1", table[0].value
+      assert_equal "Entry2", table[1].value
+      assert_equal "Entry3", table[2].value
+    end
+
+  end
+
+  def test_range_point
+     data =<<-EOF
+#ID Pos
+a 1
+b 10
+c 20
+d 12
+e 26
+f 11
+g 25
+    EOF
+
+    TmpFile.with_file(data) do |f|
+      bed = Bed.new TSV.new(f,:list, :sep=>" "), :key => "Pos" , :value => "ID", :persistence => true
+
+      bed = Bed.new TSV.new(f,:list, :sep=>" "), :key => "Pos" , :value => "ID", :persistence => true
+
+      assert_equal %w(), bed[0].sort
+      assert_equal %w(a b c d f), bed[(0..20)].sort
+      assert_equal %w(b), bed[10].sort
+    end
+   end
+
+  def test_range_persistence
+    data =<<-EOF
+#ID:Range
+#:012345678901234567890
+a:   ______
+b: ______
+c:    _______
+d:  ____
+e:    ______
+f:             ___
+g:         ____
+    EOF
+
+    TmpFile.with_file(data) do |f|
+      bed = Bed.new load_data(f), :range => ["Start" , "End"], :value => "ID", :persistence => true
+
+      bed = Bed.new load_data(f), :range => ["Start" , "End"], :value => "ID", :persistence => true
+      
+      assert_equal %w(), bed[0].sort
+      assert_equal %w(b), bed[1].sort
+      assert_equal %w(), bed[20].sort
+      assert_equal %w(), bed[(20..100)].sort
+      assert_equal %w(a b d), bed[3].sort
+      assert_equal %w(a b c d e), bed[(3..4)].sort
+    end
+  end
+end

data/test/rbbt/util/test_data_module.rb

@@ -4,6 +4,12 @@ require 'rbbt/util/data_module'
 require 'test/unit'
 require 'fileutils'
 
+SHAREDIR = File.join(PKGData.sharedir_for_file(__FILE__), 'install/DataTest')
+FileUtils.mkdir_p SHAREDIR
+File.open(File.join(SHAREDIR, 'Rakefile'), 'w') do |f|
+  f.puts "file :file1 do |t| File.open(t.name, 'w') do |f| f.write 'File 1' end end"
+end
+
 module DataTest  
   extend DataModule
 
@@ -16,28 +22,22 @@ end
 
 class TestDataModule < Test::Unit::TestCase
 
-  SHAREDIR = File.join(PKGData.sharedir_for_file(__FILE__), 'install', 'DataTest')
-
   def setup
+
     FileUtils.mkdir_p SHAREDIR
     File.open(File.join(SHAREDIR, 'Rakefile'), 'w') do |f|
-      f.puts "task :file1 do |t| File.open(t.name, 'w') do |f| f.write 'File 1' end end"
+      f.puts "file :file1 do |t| File.open(t.name, 'w') do |f| f.write 'File 1' end end"
     end
   end
 
-  def test_module
-    assert_equal "File 1", Open.read(DataTest.file1).chomp
-    assert_equal "Hello world", DataTest.salute("world")
+  def test_rakefile
+    assert_equal "File 1", Rbbt.files.DataTest.file1.read
     assert_equal "Hello world", DataTest.salute("world")
     assert_equal "Hello world", DataTest::with_key("world").salute
     assert_equal "Hello world", DataTest::World.salute
     FileUtils.rm_rf File.join(Rbbt.datadir, 'DataTest')
   end
 
-  def test_method_missing
-    assert_raise NoMethodError do DataTest.missing end
-  end
-
   def teardown
     FileUtils.rm_rf SHAREDIR
   end

data/test/rbbt/util/test_misc.rb

@@ -24,6 +24,7 @@ class TestMisc < Test::Unit::TestCase
     assert(Misc.string2hash("a=b")["a"] == 'b')
     assert(Misc.string2hash("a=b#c=d#:h=j")["c"] == 'd')
     assert(Misc.string2hash("a=b#c=d#:h=j")[:h] == 'j')
+    assert(Misc.string2hash("a=b#c=d#:h=:j")[:h] == :j)
   end
   
   def test_named_array

data/test/rbbt/util/test_persistence.rb

@@ -0,0 +1,60 @@
+require File.join(File.expand_path(File.dirname(__FILE__)), '../..', 'test_helper.rb')
+require 'rbbt/util/persistence'
+
+class TestPersistence < Test::Unit::TestCase
+
+  def test_string
+    string = "test string"
+    TmpFile.with_file do |f|
+      Persistence.persist("token_file", :Test, :string, :persistence_file => f) do string end
+      assert File.exists? f
+      assert_equal string, Open.read(f)
+
+      rm f
+    end
+  end
+
+  def test_yaml
+    object = [1,2,2]
+    TmpFile.with_file do |f|
+      Persistence.persist("token_file", :Test, :yaml, :persistence_file => f) do object end
+      assert File.exists? f
+      assert_equal object, YAML.load(File.open(f))
+      assert_equal YAML.dump(object), Open.read(f)
+      
+      rm f
+    end
+  end
+
+  def test_marshal
+    object = [1,2,2]
+    TmpFile.with_file do |f|
+      Persistence.persist("token_file", :Test, :marshal, :persistence_file => f) do object end
+      assert File.exists? f
+      assert_equal object, Marshal.load(File.open(f))
+      assert_equal Marshal.dump(object), Open.read(f)
+      
+      rm f
+    end
+  end
+
+  def test_tsv
+    object = {:a => 1, :b => 2}
+    TmpFile.with_file do |f|
+      Persistence.persist("token_file", :Test, :tsv, :persistence_file => f) do 
+        [object, {:fields => ["Number"], :key_field => "Letter", :type => :list, :filename => "foo"}]
+      end
+
+      assert File.exists? f
+      new, extra = Persistence.persist("token_file", :Test, :tsv, :persistence_file => f)
+
+      assert_equal 1, new["a"]
+      assert_equal "Letter", new.key_field
+      
+      rm f
+    end
+  end
+
+
+end
+

data/test/rbbt/util/test_pkg_data.rb

@@ -0,0 +1,113 @@
+require File.join(File.expand_path(File.dirname(__FILE__)), '../..', 'test_helper.rb')
+require 'rbbt'
+require 'rbbt/util/pkg_data'
+
+class TestPKGData < Test::Unit::TestCase
+  def _test_claims
+    begin
+      assert Rbbt.claims.empty?
+      Rbbt.claim :foo, "bar"
+      assert_equal 1, Rbbt.claims.length
+    rescue
+      Rbbt.declaim Rbbt.files.foo
+    end
+  end
+
+  def _test_path
+    assert_equal File.join(Rbbt.datadir, 'Organism/Hsa'), Rbbt.files.Organism.Hsa
+    Rbbt.files.Organism.Hsa.identifiers.produce
+  end
+
+  def _test_claim_proc
+    begin
+      assert_nil Rbbt.reclaim(Rbbt.files.foo)
+
+      Rbbt.claim :foo, proc{"bar"}
+      assert_not_nil Rbbt.reclaim Rbbt.files.foo
+
+      assert Hash === Rbbt.reclaim(Rbbt.files.foo).last
+      assert_equal "bar", Rbbt.files.foo.read
+    ensure
+      Rbbt.declaim Rbbt.files.foo
+      FileUtils.rm Rbbt.files.foo
+    end
+  end
+
+  def _test_claim_cp
+    begin
+      Open.write File.join(Rbbt.rootdir, 'share', 'foo'), "bar"
+      Rbbt.claim :foo
+      assert_equal "bar", Rbbt.files.foo.read
+    ensure
+      Rbbt.declaim Rbbt.files.foo
+      FileUtils.rm Rbbt.files.foo if File.exists? Rbbt.files.foo
+      FileUtils.rm File.join(Rbbt.rootdir, 'share', 'foo') if File.exists? File.join(Rbbt.rootdir, 'share', 'foo')
+    end
+  end
+
+  def _test_claim_tsv
+    begin
+      Rbbt.claim :foo, TSV.new({:a => 1, :b => 2})
+      assert File.exists? Rbbt.files.foo
+      assert_equal "1", Rbbt.files.foo.tsv(:single => true)["a"]
+    ensure
+      FileUtils.rm Rbbt.files.foo if File.exists? Rbbt.files.foo
+    end
+  end
+
+  def _test_claim_rakefile
+    begin
+      FileUtils.mkdir_p File.join(PKGData.sharedir_for_file(__FILE__), 'test/Rake/')
+      Open.write(File.join(PKGData.sharedir_for_file(__FILE__), 'test/Rake/Rakefile'), "file :foo do |t| Open.write(t.name, 'bar') end")
+      Rbbt.claim :foo, :Rakefile, 'test/Rake' 
+      assert_equal "bar", Rbbt.files.test.Rake.foo.read
+    ensure
+      begin
+        FileUtils.rm File.join(PKGData.sharedir_for_file(__FILE__), 'test/Rake/Rakefile')
+        FileUtils.rmdir File.join(PKGData.sharedir_for_file(__FILE__), 'test/Rake')
+        FileUtils.rmdir File.join(PKGData.sharedir_for_file(__FILE__), 'test')
+        FileUtils.rm Rbbt.files.test.Rake.foo 
+        FileUtils.rm_r Rbbt.files.test.Rake 
+        FileUtils.rm_r Rbbt.files.test
+      rescue
+      end
+    end
+  end
+
+  def test_claim_rakefile2
+    begin
+      FileUtils.mkdir_p File.join(PKGData.sharedir_for_file(__FILE__), 'test/Rake/')
+      Open.write(File.join(PKGData.sharedir_for_file(__FILE__), 'test/Rake/Rakefile'), "file :foo do |t| Open.write(t.name, 'bar') end")
+      Rbbt.claim :foo, "test/Rake/Rakefile", 'test' 
+      assert_equal "bar", Rbbt.files.test.foo.read
+    ensure
+      begin
+        FileUtils.rm File.join(PKGData.sharedir_for_file(__FILE__), 'test/Rake/Rakefile')
+        FileUtils.rmdir File.join(PKGData.sharedir_for_file(__FILE__), 'test/Rake')
+        FileUtils.rmdir File.join(PKGData.sharedir_for_file(__FILE__), 'test')
+        FileUtils.rm Rbbt.files.test.foo 
+        FileUtils.rm_r Rbbt.files.test
+      rescue
+      end
+    end
+  end
+
+  def _test_claim_rakefile3
+    begin
+      FileUtils.mkdir_p File.join(PKGData.sharedir_for_file(__FILE__), 'test/Rake/')
+      Open.write(File.join(PKGData.sharedir_for_file(__FILE__), 'test/Rake/Rakefile'), "file :foo do |t| Open.write(t.name, 'bar') end")
+      Rbbt.claim :all, "test/Rake/Rakefile", 'test' 
+      assert_equal "bar", Rbbt.files.test.foo.read
+    ensure
+      begin
+        FileUtils.rm File.join(PKGData.sharedir_for_file(__FILE__), 'test/Rake/Rakefile')
+        FileUtils.rmdir File.join(PKGData.sharedir_for_file(__FILE__), 'test/Rake')
+        FileUtils.rmdir File.join(PKGData.sharedir_for_file(__FILE__), 'test')
+        FileUtils.rm Rbbt.files.test.foo 
+        FileUtils.rm_r Rbbt.files.test
+      rescue
+      end
+    end
+  end
+end
+