rbbt-util 5.44.1 → 6.0.4

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

@@ -0,0 +1,20 @@
+require_relative '../../refactor'
+require 'scout/misc'
+require 'scout/open'
+require_relative '../log/refactor'
+
+Rbbt.relay_module_method Misc, :sensiblewrite, Open, :sensible_write
+Rbbt.relay_module_method Misc, :file2md5, Misc, :digest_file
+Rbbt.relay_module_method Misc, :lock, Open
+Rbbt.relay_module_method Misc, :consume_stream, Open
+Rbbt.relay_module_method Misc, :sort_stream, Open
+Rbbt.relay_module_method Misc, :sanitize_filename, Path
+Rbbt.relay_module_method Misc, :collapse_stream, Open
+Rbbt.relay_module_method Misc, :open_pipe, Open
+Rbbt.relay_module_method Misc, :pipe, Open
+Rbbt.relay_module_method Misc, :with_fifo, Open
+Rbbt.relay_module_method Misc, :zip2hash, IndiferentHash
+Rbbt.relay_module_method Misc, :obj2md5, Misc, :digest
+Rbbt.relay_module_method Misc, :obj2digest, Misc, :digest
+Rbbt.relay_module_method Misc, :add_defaults, IndiferentHash, :add_defaults
+Rbbt.relay_module_method Misc, :process_options, IndiferentHash, :process_options

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

@@ -238,27 +238,33 @@ end:
 
   end
 
-  def self.align(query_sequence, target_sequence)
-    Log.low { "Aligning #{ Misc.fingerprint query_sequence } to #{ Misc.fingerprint target_sequence }" }
+  def self.align_txt(query_sequence, target_sequence)
+    Log.low { "Aligning #{ Log.fingerprint query_sequence } to #{ Log.fingerprint target_sequence }" }
 
-    begin
-      raise "No query sequence" if query_sequence.nil?
-      raise "No target sequence" if target_sequence.nil?
+    raise "No query sequence" if query_sequence.nil?
+    raise "No target sequence" if target_sequence.nil?
 
-      s_out = Misc.open_pipe do |s_in|
-        SmithWaterman.ssw_aa(query_sequence, target_sequence, query_sequence.length, target_sequence.length, s_in.fileno)
-      end
+    s_out = Open.open_pipe do |s_in|
+      SmithWaterman.ssw_aa(query_sequence, target_sequence, query_sequence.length, target_sequence.length, s_in.fileno)
+    end
+
+    txt = s_out.read
+    s_out.close
+    s_out.join
+
+    txt
+  end
+
+  def self.align(query_sequence, target_sequence)
+    txt = align_txt(query_sequence, target_sequence)
 
-      txt = s_out.read
-      s_out.close
-      s_out.join
-      txt
+    begin
+      ppp txt if Rbbt::Config.get(:print, :ssw_aligmnent, default: false).to_s == 'true'
 
-      target_start, target, target_end = txt.match(/Target:\s+(\d+)\s+([A-Z\-?*]+)\s+(\d+)/).values_at 1, 2, 3
+      target_start, target, target_end = txt.match(/Target:\s+(\d+)\s+([A-Za-z\-?*]+)\s+(\d+)/).values_at 1, 2, 3
 
-      query_start, query, query_end = txt.match(/Query:\s+(\d+)\s+([A-Z\-?*]+)\s+(\d+)/).values_at 1, 2, 3
+      query_start, query, query_end = txt.match(/Query:\s+(\d+)\s+([A-Za-z\-?*]+)\s+(\d+)/).values_at 1, 2, 3
 
-      txt.replace ""
       [("_" * (query_start.to_i - 1)) + query, ("_" * (target_start.to_i - 1)) + target]
     rescue
       Log.warn("Error in aligmnent: #{$!.message}")
@@ -266,7 +272,7 @@ end:
     end
   end
 
-  def self.alignment_map(source, target)
+  def self.alignment_map(source, target, filter_low_quality=true)
     alignment_source, alignment_target = SmithWaterman.align(source, target)
     map = {}
 
@@ -289,10 +295,14 @@ end:
       end
     end
 
-    if miss_match + gaps_source > alignment_source.length.to_f / 2
+    if filter_low_quality && miss_match + gaps_source > alignment_source.length.to_f / 2
       {}
     else
       map
     end
   end
+
+  def self.alignment_percent(source, target, filter_low_quality=true)
+    alignment_map(source, target,filter_low_quality).keys.length.to_f / source.length
+  end
 end

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

@@ -1,105 +1,92 @@
-module Misc
-
-  def self.hostname
-    @hostanem ||= `hostname`.strip
-  end
-
-  def self.pid_exists?(pid)
-    return false if pid.nil?
-    begin
-      Process.getpgid(pid.to_i)
-      true
-    rescue Errno::ESRCH
-      false
-    end
-  end
-
-  def self.env_add(var, value, sep = ":", prepend = true)
-    ENV[var] ||= ""
-    return if ENV[var] =~ /(#{sep}|^)#{Regexp.quote value}(#{sep}|$)/
-      if prepend
-        ENV[var] = value + sep + ENV[var]
-      else
-        ENV[var] += sep + ENV[var]
-      end
-  end
-
-  def self.with_env(var, value, &block)
-    var = var.to_s
-    value = value.to_s
-    current = ENV[var]
-    begin
-      ENV[var] = value
-      yield
-    ensure
-      ENV[var] = current
-    end
-  end
-
-  def self.path_relative_to(basedir, path)
-    path = File.expand_path(path) unless path[0] == "/"
-    basedir = File.expand_path(basedir) unless basedir[0] == "/"
-
-    if path.index(basedir) == 0
-      if basedir[-1] == "/"
-        return path[basedir.length..-1]
-      else
-        return path[basedir.length+1..-1]
-      end
-    else
-      return nil
-    end
-  end
-
-  def self.common_path(dir, file)
-    file = File.expand_path file
-    dir = File.expand_path dir
-
-    return true if file == dir
-    while File.dirname(file) != file
-      file = File.dirname(file)
-      return true if file == dir
-    end
-
-    return false
-  end
-
-
-  def self.relative_link(source, target_dir)
-    path = "."
-    current = target_dir
-    while ! Misc.common_path current, source
-      current = File.dirname(current)
-      path = File.join(path, '..')
-      return nil if current == "/"
-    end
-
-    File.join(path, Misc.path_relative_to(current, source))
-  end
-
-  # WARN: probably not thread safe...
-  def self.in_dir(dir)
-    old_pwd = FileUtils.pwd
-    res = nil
-    begin
-      FileUtils.mkdir_p dir unless File.exist?(dir)
-      FileUtils.cd dir
-      res = yield
-    ensure
-      FileUtils.cd old_pwd
-    end
-    res
-  end
-
-  def self.is_filename?(string, need_to_exists = true)
-    return false if string.nil?
-    return true if defined? Path and Path === string
-    return true if string.respond_to? :exists
-    return true if String === string and ! string.include?("\n") and string.split("/").select{|p| p.length > 265}.empty? and (! need_to_exists || File.exist?(string))
-    return false
-  end
-
-  class << self
-    alias filename? is_filename?
-  end
-end
+require_relative 'refactor'
+Rbbt.require_instead 'scout/misc/format'
+#module Misc
+#
+#  def self.hostname
+#    @hostanem ||= `hostname`.strip
+#  end
+#
+#  def self.pid_exists?(pid)
+#    return false if pid.nil?
+#    begin
+#      Process.getpgid(pid.to_i)
+#      true
+#    rescue Errno::ESRCH
+#      false
+#    end
+#  end
+#
+#  def self.env_add(var, value, sep = ":", prepend = true)
+#    ENV[var] ||= ""
+#    return if ENV[var] =~ /(#{sep}|^)#{Regexp.quote value}(#{sep}|$)/
+#      if prepend
+#        ENV[var] = value + sep + ENV[var]
+#      else
+#        ENV[var] += sep + ENV[var]
+#      end
+#  end
+#
+#  def self.with_env(var, value, &block)
+#    var = var.to_s
+#    value = value.to_s
+#    current = ENV[var]
+#    begin
+#      ENV[var] = value
+#      yield
+#    ensure
+#      ENV[var] = current
+#    end
+#  end
+#
+#  def self.common_path(dir, file)
+#    file = File.expand_path file
+#    dir = File.expand_path dir
+#
+#    return true if file == dir
+#    while File.dirname(file) != file
+#      file = File.dirname(file)
+#      return true if file == dir
+#    end
+#
+#    return false
+#  end
+#
+#
+#  def self.relative_link(source, target_dir)
+#    path = "."
+#    current = target_dir
+#    while ! Misc.common_path current, source
+#      current = File.dirname(current)
+#      path = File.join(path, '..')
+#      return nil if current == "/"
+#    end
+#
+#    File.join(path, Misc.path_relative_to(current, source))
+#  end
+#
+#  # WARN: probably not thread safe...
+#  def self.in_dir(dir)
+#    old_pwd = FileUtils.pwd
+#    res = nil
+#    begin
+#      FileUtils.mkdir_p dir unless File.exist?(dir)
+#      FileUtils.cd dir
+#      res = yield
+#    ensure
+#      FileUtils.cd old_pwd
+#    end
+#    res
+#  end
+#
+#  def self.is_filename?(string, need_to_exists = true)
+#    return false if string.nil?
+#    return true if defined? Path and Path === string
+#    return true if string.respond_to? :exists
+#    return true if String === string and ! string.include?("\n") and string.split("/").select{|p| p.length > 265}.empty? and (! need_to_exists || File.exist?(string))
+#    return false
+#  end
+#
+#  class << self
+#    alias filename? is_filename?
+#  end
+#end

data/lib/rbbt/util/misc.rb

@@ -1,28 +1,28 @@
-require 'lockfile'
+require_relative '../../rbbt'
 require 'digest/md5'
 require 'cgi'
 require 'zlib'
 require 'etc'
 require 'rubygems/package'
 
-require 'rbbt/util/tar'
-require 'rbbt/util/misc/exceptions'
-require 'rbbt/util/misc/concurrent_stream'
-require 'rbbt/util/misc/indiferent_hash'
-require 'rbbt/util/misc/pipes'
-require 'rbbt/util/misc/format'
-require 'rbbt/util/misc/omics'
-require 'rbbt/util/misc/inspect'
-require 'rbbt/util/misc/math'
-require 'rbbt/util/misc/development'
-require 'rbbt/util/misc/lock'
-require 'rbbt/util/misc/options'
-require 'rbbt/util/misc/system'
-require 'rbbt/util/misc/objects'
-require 'rbbt/util/misc/manipulation'
-require 'rbbt/util/misc/communication'
-
-require 'rbbt/util/misc/serialize'
+require_relative 'tar'
+require_relative 'misc/exceptions'
+require_relative 'misc/concurrent_stream'
+require_relative 'misc/indiferent_hash'
+require_relative 'misc/pipes'
+require_relative 'misc/format'
+require_relative 'misc/omics'
+require_relative 'misc/inspect'
+#require_relative 'misc/math'
+require_relative 'misc/development'
+require_relative 'misc/lock'
+require_relative 'misc/options'
+require_relative 'misc/system'
+require_relative 'misc/objects'
+require_relative 'misc/manipulation'
+require_relative 'misc/communication'
+
+require_relative 'misc/serialize'
 
 require 'to_regexp'
 
@@ -104,7 +104,26 @@ module Misc
     time
   end
 
-  
+  def self.name2basename(file)
+    sanitize_filename(file.gsub("/",'·').gsub("~", '-'))
+  end
+
+  def self.sanitize_filename(filename, length = 254)
+    if filename.length > length
+      if filename =~ /(\..{2,9})$/
+        extension = $1
+      else
+        extension = ''
+      end
+
+      post_fix = "--#{filename.length}@#{length}_#{Misc.digest(filename)[0..4]}" + extension
+
+      filename = filename[0..(length - post_fix.length - 1)] << post_fix
+    else
+      filename
+    end
+    filename
+  end
 end
 
 module PDF2Text

data/lib/rbbt/util/named_array/refactor.rb

@@ -0,0 +1,4 @@
+require 'scout/named_array'
+module NamedArray
+  annotation :entity_options, :entity_templates
+end

data/lib/rbbt/util/named_array.rb

@@ -1,220 +1,3 @@
-require 'rbbt/util/misc'
-
-module NamedArray
-  #extend ChainMethods
-  #self.chain_prefix = :named_array
-
-  attr_accessor :fields
-  attr_accessor :key
-  attr_accessor :entity_options
-  attr_accessor :entity_templates
-
-  def shift
-    fields.shift
-    super
-  end
-
-  def entity_templates
-    @entity_templates ||= {}
-  end
-
-  def self.setup(array, fields, key = nil, entity_options = nil, entity_templates = nil)
-    return array if array.nil?
-    array.extend NamedArray unless NamedArray === array
-    array.fields = Annotated === fields ? Annotated.purge(fields) : fields
-    array.key = key
-    array.entity_options = entity_options unless entity_options.nil?
-    array.entity_templates = entity_templates unless entity_templates.nil?
-    array
-  end
-
-  def prepare_entity(entity, field, options = {})
-    return entity if entity.nil?
-    return entity unless defined? Entity
-    template = entity_templates[field]
-    entity_templates ||= {}
-    if template and template.respond_to?(:annotate)
-      begin entity = entity.dup if entity.frozen?;  rescue; end
-      entity = template.annotate(entity)
-      entity.extend AnnotatedArray if Array === entity
-      entity
-    else
-      if entity_templates.include? field
-        entity
-      else
-        template = Misc.prepare_entity("ENTITY_TEMPLATE", field, options)
-        if template.respond_to?(:annotate)
-          entity_templates[field] = template
-          begin entity = entity.dup if entity.frozen?;  rescue; end
-          entity = template.annotate(entity)
-          entity.extend AnnotatedArray if Array === entity
-          entity
-        else
-          entity_templates[field] = nil
-          entity
-        end
-      end
-    end
-  end
-
-  def merge(array)
-    double = Array === array.first 
-    new = self.dup
-    (0..length - 1).each do |i|
-      if double
-        new[i] = new[i] + array[i]
-      else
-        new[i] << array[i]
-      end
-    end
-    new
-  end
-
-  def positions(fields)
-    if Array ==  fields
-      fields.collect{|field|
-        Misc.field_position(@fields, field)
-      }
-    else
-      Misc.field_position(@fields, fields)
-    end
-  end
-
-  #def named_array_get_brackets(key)
-  #  pos = Misc.field_position(fields, key)
-  #  elem = named_array_clean_get_brackets(pos)
-
-  #  return elem if @fields.nil? or @fields.empty?
-
-  #  field = NamedArray === @fields ? @fields.named_array_clean_get_brackets(pos) : @fields[pos]
-  #  elem = prepare_entity(elem, field, entity_options)
-  #  elem
-  #end
-
-  def [](key, clean = false)
-    pos = Misc.field_position(fields, key)
-    elem = super(pos)
-    return elem if clean
-
-    return elem if @fields.nil? or @fields.empty?
-
-    field = NamedArray === @fields ? @fields[pos, true] : @fields[pos]
-    elem = prepare_entity(elem, field, entity_options)
-    elem
-  end
-
-  #def named_array_each(&block)
-  #  if defined?(Entity) and not @fields.nil? and not @fields.empty?
-  #    @fields.zip(self).each do |field,elem|
-  #      elem = prepare_entity(elem, field, entity_options)
-  #      yield(elem)
-  #      elem
-  #    end
-  #  else
-  #    named_array_clean_each &block
-  #  end
-  #end
-
-  def each(&block)
-    if defined?(Entity) && ! (@fields.nil? || @fields.empty?)
-      i = 0
-      super do |elem|
-        field = @fields[i]
-        elem = prepare_entity(elem, field, entity_options)
-        yield(elem)
-        i += 1
-        elem
-      end
-    else
-      super &block
-    end
-
-  end
-
-  #def named_array_collect
-  #  res = []
-
-  #  each do |elem|
-  #    if block_given?
-  #      res << yield(elem)
-  #    else
-  #      res << elem
-  #    end
-  #  end
-
-  #  res
-  #end
-
-
-  def collect
-    res = []
-
-    each do |elem|
-      if block_given?
-        res << yield(elem)
-      else
-        res << elem
-      end
-    end
-
-    res
-  end
-
-  #def named_array_set_brackets(key,value)
-  #  named_array_clean_set_brackets(Misc.field_position(fields, key), value)
-  #end
-  
-  def []=(key, value)
-    super(Misc.field_position(fields, key), value)
-  end
-
-  #def named_array_values_at(*keys)
-  #  keys = keys.collect{|k| Misc.field_position(fields, k, true) }
-  #  keys.collect{|k|
-  #    named_array_get_brackets(k) unless k.nil?
-  #  }
-  #end
-
-
-
-  def values_at(*keys)
-    keys = keys.collect{|k| Misc.field_position(fields, k, true) }
-    keys.collect{|k|
-      self[k] unless k.nil?
-    }
-  end
-
-  def zip_fields
-    return [] if self.empty?
-    zipped = Misc.zip_fields(self)
-    zipped = zipped.collect{|v| NamedArray.setup(v, fields)}
-    zipped 
-  end
-
-  def detach(file)
-    file_fields = file.fields.collect{|field| field.fullname}
-    detached_fields = []
-    self.fields.each_with_index{|field,i| detached_fields << i if file_fields.include? field.fullname}
-    fields = self.fields.values_at *detached_fields
-    values = self.values_at *detached_fields
-    values = NamedArray.name(values, fields)
-    values.zip_fields
-  end
-
-  def report
-    fields.zip(self).collect do |field,value|
-      "#{Log.color(:magenta, field) }: #{ Array === value ? value * "|" : value }"
-    end * "\n"
-  end
-
-  def to_hash
-    hash = {}
-    self.fields.zip(self) do |field,value|
-      hash[field] = value
-    end
-    IndiferentHash.setup hash
-    hash
-  end
-
-end
-
+require_relative '../refactor'
+Rbbt.require_instead 'scout/named_array'
+require_relative 'named_array/refactor'

data/lib/rbbt/util/open/refactor.rb

@@ -0,0 +1,7 @@
+require 'rbbt/util/misc/bgzf'
+module Open
+  class << self
+    attr_accessor :repository_dirs
+  end
+  self.repository_dirs = []
+end