rbbt-util 3.2.0 → 3.2.1

data/test/rbbt/util/tsv/test_manipulate.rb

@@ -254,7 +254,7 @@ row3    a2
       tsv = TSV.new(filename + '#:sep=/\s+/#:type=:single')
 
       #TODO: WTF!!!
-      assert_equal "row1", tsv.reorder("LetterValue", :key)[97]
+      assert_equal "row3", tsv.reorder("LetterValue", :key)[97]
     end
  
   end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: rbbt-util
 version: !ruby/object:Gem::Version 
-  hash: 15
-  prerelease: 
+  hash: 13
+  prerelease: false
   segments: 
   - 3
   - 2
-  - 0
-  version: 3.2.0
+  - 1
+  version: 3.2.1
 platform: ruby
 authors: 
 - Miguel Vazquez
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-06-12 00:00:00 +02:00
+date: 2011-07-05 00:00:00 +02:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -107,7 +107,7 @@ email: miguel.vazquez@fdi.ucm.es
 executables: 
 - tsv.rb
 - tchash.rb
-- workflow.rb
+- run_workflow.rb
 - rbbt_query.rb
 - rbbt_exec.rb
 - rbbt_Rutil.rb
@@ -120,8 +120,6 @@ files:
 - lib/rbbt-util.rb
 - lib/rbbt.rb
 - lib/rbbt/util/R.rb
-- lib/rbbt/util/bed.rb
-- lib/rbbt/util/cachehelper.rb
 - lib/rbbt/util/cmd.rb
 - lib/rbbt/util/excel2tsv.rb
 - lib/rbbt/util/filecache.rb
@@ -165,9 +163,8 @@ files:
 - test/rbbt/util/test_excel2tsv.rb
 - test/rbbt/util/test_simpleopt.rb
 - test/rbbt/util/test_task.rb
-- test/rbbt/util/test_bed.rb
-- test/rbbt/util/test_rake.rb
 - test/rbbt/util/test_persistence.rb
+- test/rbbt/util/test_rake.rb
 - test/rbbt/util/test_workflow.rb
 - test/rbbt/util/tsv/test_parse.rb
 - test/rbbt/util/tsv/test_accessor.rb
@@ -180,7 +177,7 @@ files:
 - test/test_rbbt.rb
 - bin/tsv.rb
 - bin/tchash.rb
-- bin/workflow.rb
+- bin/run_workflow.rb
 - bin/rbbt_query.rb
 - bin/rbbt_exec.rb
 - bin/rbbt_Rutil.rb
@@ -214,7 +211,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
 requirements: []
 
 rubyforge_project: 
-rubygems_version: 1.6.2
+rubygems_version: 1.3.7
 signing_key: 
 specification_version: 3
 summary: Utilities for the Ruby Bioinformatics Toolkit (rbbt)
@@ -233,9 +230,8 @@ test_files:
 - test/rbbt/util/test_excel2tsv.rb
 - test/rbbt/util/test_simpleopt.rb
 - test/rbbt/util/test_task.rb
-- test/rbbt/util/test_bed.rb
-- test/rbbt/util/test_rake.rb
 - test/rbbt/util/test_persistence.rb
+- test/rbbt/util/test_rake.rb
 - test/rbbt/util/test_workflow.rb
 - test/rbbt/util/tsv/test_parse.rb
 - test/rbbt/util/tsv/test_accessor.rb

data/bin/workflow.rb

@@ -1,24 +0,0 @@
-#!/usr/bin/env ruby
-
-require 'rbbt-util'
-require 'rbbt/util/simpleopt'
-require 'rbbt/util/workflow'
-
-options = SOPT.get "-h--help:-t--target*:-n--name*"
-
-raise "No target" unless options[:target]
-
-file    = ARGV.shift
-
-if ARGV.empty?
-  data = STDIN.read
-else
-  data = ARGV.shift
-end
-
-job = options[:name] || "Job"
-
-puts "Runing WorkFlow in #{file} for target #{options[:target]}. Job: #{job}"
-WorkFlow.load file, File.join(options[:target], job), data
-puts 
-puts "WorkFlow done. Please find results in: #{File.join(options[:target], job)}"

data/lib/rbbt/util/bed.rb

@@ -1,325 +0,0 @@
-require 'rbbt/util/tsv'
-require 'rbbt/util/misc'
-require 'rbbt/util/log'
-require 'yaml'
-
-class Bed
-
-  class Entry < Struct.new( :value, :start, :end, :overlap); end
-
-  class FixWidthTable
-    SERIALIZER = Marshal
-    def self.serialize(entry)
-      SERIALIZER.dump(entry)
-    end
-
-    def self.deserialise(entry)
-      SERIALIZER.load(entry)
-    end
-
-    def self.format(entry, index, record_size)
-      data = serialize(entry)
-      padding = record_size - data.length
-      [index, data + "\0" * padding].pack("la#{record_size}")
-    end
-
-    def self.index(format)
-      format.unpack("s").first
-    end
-
-    def self.unformat(format, record_size)
-      index, data = format.unpack("la#{record_size}")
-      deserialise(data)
-    end
-
-    def self.get_record_size(entries)
-      max = 0
-      entries.each do |entry|
-        size = serialize(entry).length
-        max  = size if size > max
-      end
-
-      max
-    end
-
-    attr_accessor :size, :range
-    def initialize(file, record_size = nil, range = false, rewrite = false)
-      @filename = file
-      @range = range
-
-      if rewrite or not File.exists? file
-        Log.debug("Opening FixWidthTable in #{ file } writing. Record size: #{record_size}")
-        @file = File.open(@filename, 'wb')
-        @record_size = record_size
-        @file.write [record_size].pack("L")
-        @file.write [@range ? 1 : 0 ].pack("C")
-        @size = 0
-      else
-        Log.debug("Opening FixWidthTable in #{ file } for reading")
-        @file        = File.open(@filename, 'rb')
-        @record_size = @file.read(4).unpack("L").first
-        @range       = @file.read(1).unpack("C").first == 1
-        @size        = (File.size(@filename) - 5) / (@record_size + 4)
-        Log.debug("Record size #{@record_size}")
-      end
-    end
-
-    def length
-      @size / @record_size
-    end
-
-    def read
-      @file.close 
-      @file = File.open(@filename, 'rb')
-    end
-
-    def add(entry, index)
-      @size += 1
-      format = FixWidthTable.format(entry, index, @record_size)
-      @file.write format
-    end
-
-    def index(index)
-      return nil if index < 0 or index >= size
-      @file.seek(5 + (@record_size + 4) * index, IO::SEEK_SET)
-
-      format = @file.read(@record_size + 4)
-      FixWidthTable.index(format)
-    end
-
-    def [](index)
-      return nil if index < 0 or index >= size
-      @file.seek(5 + (@record_size + 4) * index, IO::SEEK_SET)
-
-      format = @file.read(@record_size + 4)
-      FixWidthTable.unformat(format, @record_size)
-    end
-
-    def closest(pos)
-      upper = size - 1
-      lower = 0
-
-      return -1 if upper < lower
-
-      while(upper >= lower) do
-        idx = lower + (upper - lower) / 2
-        comp = pos <=> index(idx)
-
-        if comp == 0
-          break 
-        elsif comp > 0
-          lower = idx + 1
-        else
-          upper = idx - 1
-        end
-      end
-
-      if index(idx) > pos
-        idx = idx - 1
-      end
-
-      idx
-    end
-
-
-  end
-
-  #{{{ Persistence
-
-  CACHEDIR="/tmp/bed_persistent_cache"
-  FileUtils.mkdir CACHEDIR unless File.exist? CACHEDIR
-
-  def self.cachedir=(cachedir)
-    CACHEDIR.replace cachedir
-    FileUtils.mkdir_p CACHEDIR unless File.exist? CACHEDIR
-  end
-
-  def self.cachedir
-    CACHEDIR
-  end
-
-  def self.get_persistence_file(file, prefix, options = {})
-    File.join(CACHEDIR, prefix.gsub(/\s/,'_').gsub(/\//,'>') + Digest::MD5.hexdigest([file, options].inspect))
-  end
-
-
-  attr_accessor :index, :range, :persistence_file
-  def initialize(tsv, options = {})
-    options = Misc.add_defaults options, :range => nil, :key => 0, :value => 1, :persistence => false, :persistence_file => nil, :tsv => {}
-
-    options[:persistence] = true if options[:persistence].nil? and options[:persistence_file]
-
-    filename = nil
-    case
-    when TSV === tsv
-      filename = tsv.filename
-    when (String === tsv and File.exists? tsv.sub(/#.*/,''))
-      filename = tsv
-    else 
-      filename = "None"
-    end
-
-
-    if options[:range]
-      options[:key]   = options[:range].first
-      options[:value] = [options[:value], options[:range].last]
-      @range = true
-    else
-      @range = false
-    end
-
-    if options[:persistence] and options[:persistence_file].nil?
-      options[:persistence_file] = Bed.get_persistence_file(filename, (options[:range].nil? ? "Point:#{filename}" : "Range:#{filename}"),  options)
-    end
-
-    @persistence_file = options[:persistence_file]
-
-    if @persistence_file and File.exists?(@persistence_file)
-      Log.low("Loading Persistence Bed File: #{ @persistence_file }")
-
-      @index = FixWidthTable.new @persistence_file
-      @range = @index.range
-      return
-    end
-
-    tsv = TSV.new(tsv, options[:tsv]) unless TSV === tsv
-
-    @index = []
-    entry = nil
-    tsv.through options[:key], options[:value] do |key, values|
-      if @range
-        entry = Entry.new(values[0], key.to_i, values[1].to_i,  nil)
-      else
-        entry = Entry.new(values[0], key.to_i, nil, nil)
-      end
-      @index << entry
-    end
-
-    @index.sort!{|a,b| a.start <=> b.start}
-
-    if range
-      latest = []
-      @index.each do |entry|
-        while latest.any? and latest[0] < entry.start
-          latest.shift
-        end
-
-        entry.overlap = latest.length
-        latest << entry.end
-      end
-    end
-
-    if @persistence_file
-      record_size = FixWidthTable.get_record_size(@index)
-
-      table = FixWidthTable.new @persistence_file, record_size, @range
-      @index.each do |entry| table.add entry, entry.start end
-      table.read
-
-      @index = table
-    end
-  end
-
-  def closest(pos)
-    upper = @index.size - 1
-    lower = 0
-
-    return -1 if upper < lower
-
-    while(upper >= lower) do
-      idx = lower + (upper - lower) / 2
-      comp = pos <=> @index[idx].start
-
-      if comp == 0
-        break 
-      elsif comp > 0
-        lower = idx + 1
-      else
-        upper = idx - 1
-      end
-    end
-
-    if @index[idx].start > pos
-      idx = idx - 1
-    end
-
-    idx
-  end
-
-  def get_range(pos)
-    if Range === pos
-      r_start = pos.begin
-      r_end   = pos.end
-    else
-      r_start = pos.to_i
-      r_end   = pos.to_i
-    end
-
-    if FixWidthTable === @index
-      idx = @index.closest(r_start)
-    else
-      idx = closest(r_start)
-    end
-
-    return [] if idx >= @index.size
-    return [] if idx <0 and r_start == r_end
-
-    idx = 0 if idx < 0
-
-    idx -= @index[idx].overlap if @index[idx].overlap
-
-    values = []
-    l = @index[idx]
-    while l.start <= r_end
-      values << l.value if l.end >= r_start 
-      idx += 1
-      break if idx >= @index.size
-      l = @index[idx]
-    end
-
-    values
-  end
-
-  def get_point(pos)
-    if Range === pos
-      r_start = pos.begin
-      r_end   = pos.end
-    else
-      r_start = pos.to_i
-      r_end   = pos.to_i
-    end
-
-    if FixWidthTable === @index
-      idx = @index.closest(r_start)
-    else
-      idx = closest(r_start)
-    end
-
-    return [] if idx >= @index.size
-    return [] if idx <0 and r_start == r_end
-
-    idx = 0 if idx < 0
-
-    idx += 1 unless @index[idx].start >= r_start
-
-    values = []
-    l = @index[idx]
-    while l.start <= r_end
-      values << l.value
-      idx += 1
-      break if idx >= @index.size
-      l = @index[idx]
-    end
-
-    values
-  end
-
-  def [](pos)
-    if range
-      get_range(pos)
-    else
-      get_point(pos)
-    end
-  end
-
-end

data/lib/rbbt/util/cachehelper.rb

@@ -1,100 +0,0 @@
-require 'digest'
-
-module CacheHelper
-  CACHE_DIR = '/tmp/cachehelper'
-  FileUtils.mkdir_p(CACHE_DIR) unless File.exist?(CACHE_DIR)
-
-  LOG_TIME = false
-  class CacheLocked < Exception; end
-
-  def self.time(id)
-    t = Time.now
-    data = block.call
-    STDERR.puts "#{ id } time: #{Time.now - t}"
-    data
-  end
-
-  def self.cachedir=(dir)
-    @@cachedir=dir
-    FileUtils.mkdir_p(dir) unless File.exist?(dir)
-  end
-
-  def self.cachedir
-    @@cachedir ||= CACHE_DIR
-  end
-
-
-  def self.reset
-    FileUtils.rm Dir.glob(cachedir + '*')
-  end
- 
-  def self.reset_locks
-    FileUtils.rm Dir.glob(cachedir + '*.lock')
-  end
- 
- 
-  def self.build_filename(name, key)
-    File.join(cachedir, name + ": " + Digest::MD5.hexdigest(key.to_s))
-  end
- 
-  def self.do(filename, block)
-    FileUtils.touch(filename + '.lock')
-
-    if LOG_TIME
-      data = time do
-        block.call
-      end
-    else
-      data = block.call
-    end
-    
-    File.open(filename, 'w'){|f| f.write data}
-    FileUtils.rm(filename + '.lock')
-    return data
-  end
- 
-  def self.clean(name)
-    FileUtils.rm Dir.glob(File.join(cachedir, "#{ name }*"))
-  end
- 
-  def self.cache_ready?(name, key)
-    filename = CacheHelper.build_filename(name, key)
-    File.exist?(filename)
-  end
- 
-  def self.cache(name, key = [], wait = nil, &block)
-    filename = CacheHelper.build_filename(name, key)
-    begin
-      case
-      when File.exist?(filename)
-        return File.open(filename){|f| f.read}
-      when File.exist?(filename + '.lock')
-        raise CacheLocked
-      else
-        if wait.nil?
-          CacheHelper.do(filename, block)
-        else
-          Thread.new{CacheHelper.do(filename, block)}
-          return wait
-        end
- 
-      end
-    rescue CacheLocked
-      if wait.nil?
-        sleep 30
-        retry
-      else
-        return wait
-      end
-    rescue Exception
-      FileUtils.rm(filename + '.lock') if File.exist?(filename + '.lock')
-      raise $!
-    end
-  end
- 
-  def self.marshal_cache(name, key = [])
-    Marshal::load( cache(name, key) do
-      Marshal::dump(yield)
-    end)
-  end
-end