rbbt-util 5.17.20 → 5.17.21

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 58cd900e8238a97baf230cbb81502f4c3aca35ba
-  data.tar.gz: fb08bda42782f76c8dd4a8b2b338d5ebc74a2038
+  metadata.gz: 557142d92147d54d45af152135df1fe1743002bd
+  data.tar.gz: 5f56410a030766911e3ca59be65a887a84925636
 SHA512:
-  metadata.gz: bc6c3a7db160f46f55c8cfd7ab266168cc688045077bab418ca5ea0c146e162c89f4b1fdfb41cecb7268f73f5e23e689edeff342b647325ef172f68f4e5d493d
-  data.tar.gz: fa5201d4762db7bba332aa5d0fce2e41c540447793fda061436405e6af902d15ede743197481c2134df90081f889ba7e0636aea169aa0819763e8fdf96e41338
+  metadata.gz: a58ff556d9be4894195d4e317ef7d9f0fb82bca320b7f03804355fb8d7eda0e72307caa50fb432fe1451d4d55ca40d03da5f20d059e105a4cd89044bdbe15ad1
+  data.tar.gz: 79582eaa89d75dccbce3396e71553b4ab3fad049d8787daaa6a6fc93a6fbac0309d0f277639f05faeb6b71ca779d9aeb4d1ec0ad96407ddfd7af30f2ffb06891

data/lib/rbbt/persist/tsv/tokyocabinet.rb

@@ -59,133 +59,8 @@ module Persist
       @closed = false
       self
     end
-
-    #MAX_CHAR = 255.chr
-
-    #def prefix(key)
-    #  range(key, 1, key + MAX_CHAR, 1)
-    #end
-
-    #def get_prefix(key)
-    #  keys = prefix(key)
-    #  select(:key => keys)
-    #end
-
-    #def closed?
-    #  @closed
-    #end
-
-    #def close
-    #  @closed = true
-    #  super
-    #end
-
-    #def read(force = false)
-    #  return if not write? and not closed and not force
-    #  self.close
-    #  if !self.open(@persistence_path, tokyocabinet_class::OREADER)
-    #    ecode = self.ecode
-    #    raise "Open error: #{self.errmsg(ecode)}. Trying to open file #{@persistence_path}"
-    #  end
-    #  @writable = false
-    #  @closed = false
-    #  self
-    #end
-
-    #def write(force = true)
-    #  return if write? and not closed and not force
-    #  self.close
-
-    #  if !self.open(@persistence_path, tokyocabinet_class::OWRITER)
-    #    ecode = self.ecode
-    #    raise "Open error: #{self.errmsg(ecode)}. Trying to open file #{@persistence_path}"
-    #  end
-
-    #  @writable = true
-    #  @closed = false
-    #  self
-    #end
-
-    #def write?
-    #  @writable
-    #end
-
-    #def read?
-    #  ! write?
-    #end
-
-    #def collect
-    #  res = []
-    #  each do |key, value|
-    #    res << if block_given?
-    #             yield key, value
-    #    else
-    #      [key, value]
-    #    end
-    #  end
-    #  res
-    #end
-
-    #def delete(key)
-    #  out(key)
-    #end
-
-    #def write_and_read
-    #  lock_filename = Persist.persistence_path(persistence_path + '.write', {:dir => TSV.lock_dir})
-    #  Misc.lock(lock_filename) do
-    #    @mutex.synchronize do
-    #      write if @closed or not write?
-    #      res = begin
-    #              yield
-    #            ensure
-    #              read
-    #            end
-    #      res
-    #    end
-    #  end
-    #end
-
-    #def write_and_close
-    #  lock_filename = Persist.persistence_path(persistence_path + '.write', {:dir => TSV.lock_dir})
-    #  Misc.lock(lock_filename) do
-    #    @mutex.synchronize do
-    #      write if @closed or not write?
-    #      res = begin
-    #              yield
-    #            ensure
-    #              close
-    #            end
-    #      res
-    #    end
-    #  end
-    #end
-
-    #def read_and_close
-    #  @mutex.synchronize do
-    #    read if @closed or not read?
-    #    res = begin
-    #            yield
-    #          ensure
-    #            close
-    #          end
-    #    res
-    #  end
-    #end
-
-
-    #def merge!(hash)
-    #  hash.each do |key,values|
-    #    self[key] = values
-    #  end
-    #end
-
-
-    #def range(*args)
-    #  super(*args) #- TSV::ENTRY_KEYS.to_a
-    #end
   end
 
-
   def self.open_tokyocabinet(path, write, serializer = nil, tokyocabinet_class = TokyoCabinet::HDB)
     write = true unless File.exists? path
 

data/lib/rbbt/util/misc/omics.rb

@@ -192,4 +192,63 @@ module Misc
     end
   end
 
+
+  def self.intersect_streams_read(io, sep=":")
+    line = io.gets.strip
+    parts = line.split(sep)
+    chr, start, eend, *rest = parts
+    [line,chr, start.to_i, eend.to_i, rest]
+  end
+
+  def self.intersect_streams(f1, f2, out, sep=":")
+    finish = false
+    line1, chr1, start1, eend1, rest1 = intersect_streams_read(f1,sep)
+    line2, chr2, start2, eend2, rest2 = intersect_streams_read(f2,sep)
+    Misc.profile do
+    while not finish
+      case chr1 <=> chr2
+      when -1
+        move = 1
+      when 1
+        move = 2
+      else
+        if eend1 < start2
+          move = 1
+        elsif eend2 < start1
+          move = 2
+        else
+          pos2 = f2.pos
+
+          sline2, schr2, sstart2, seend2, srest2 = line2, chr2, start2, eend2, rest2
+          while chr1 == chr2 and ((start1 <= eend2 and eend1 >= start2)) 
+            out.puts line1 + "\t" + line2
+            if f2.eof?
+              chr2 = 'next2'
+            else
+              line2, chr2, start2, eend2, rest2 = intersect_streams_read(f2,sep)
+            end
+          end
+          line2, chr2, start2, eend2, rest2 = sline2, schr2, sstart2, seend2, srest2 
+          f2.seek(pos2)
+          move = 1
+        end
+      end
+
+      case move
+      when 1
+        if f1.eof?
+          finish = true
+        else
+          line1, chr1, start1, eend1, rest1 = intersect_streams_read(f1,sep)
+        end
+      when 2
+        if f2.eof?
+          finish = true
+        else
+          line2, chr2, start2, eend2, rest2 = intersect_streams_read(f2,sep)
+        end
+      end
+    end
+    end
+  end
 end

data/lib/rbbt/workflow/accessor.rb

@@ -518,7 +518,7 @@ module Workflow
         workflow, task, options = dependency
 
         #options = dependency.last if Hash === dependency.last
-        _inputs = inputs.dup
+        _inputs = IndiferentHash.setup(inputs.dup)
         options.each{|i,v|
           case v
           when Symbol
@@ -544,9 +544,11 @@ module Workflow
       when Step
         dependency
       when Symbol
-        job(dependency, jobname, inputs)
+        _inputs = IndiferentHash.setup(inputs.dup)
+        job(dependency, jobname, _inputs)
       when Proc
-        dependency.call jobname, inputs, real_dependencies
+        _inputs = IndiferentHash.setup(inputs.dup)
+        dependency.call jobname, _inputs, real_dependencies
       end
     end
     real_dependencies.flatten.compact

data/lib/rbbt/workflow.rb

@@ -279,7 +279,7 @@ module Workflow
     task = tasks[taskname]
     raise "Task not found: #{ taskname }" if task.nil?
 
-    IndiferentHash.setup(inputs)
+    inputs = IndiferentHash.setup(inputs)
 
     Workflow.resolve_locals(inputs)
 

data/test/rbbt/persist/tsv/test_tokyocabinet.rb

@@ -239,4 +239,23 @@ class TestPersistTSVTC < Test::Unit::TestCase
       assert_equal "2", persisted_annotations.sort.last.test_annotation
     end
   end
+
+  def test_fdb
+    TmpFile.with_file do |tmp|
+      repo = Persist.open_tokyocabinet(tmp, true, :list, TokyoCabinet::FDB)
+      repo.write
+      repo.put(1, "hola")
+      repo.put(2, "adios")
+      47000.times do |i|
+        repo.put(i*500, "key: " << i.to_s)
+      end
+      repo.read
+      Misc.profile do
+        repo.values_at *repo.range("[100,1000]")
+      end
+      Misc.benchmark(60_000) do
+        repo.values_at *repo.range("[100,1000]")
+      end
+    end
+  end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-util
 version: !ruby/object:Gem::Version
-  version: 5.17.20
+  version: 5.17.21
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-04-29 00:00:00.000000000 Z
+date: 2015-05-06 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake