rbbt-util 5.17.77 → 5.17.78

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: a51122d03732e5546ecf0680b478574b177656c4
-  data.tar.gz: 346fb3e137b00d39ff04ae7b7dfe1fcd3f95b237
+  metadata.gz: 449421ec50d68a1e7d718c4c81a5a764be546877
+  data.tar.gz: 43a60da2015d8a802c79c88d051dab6a4dfb59c7
 SHA512:
-  metadata.gz: 3ea134b8da6d0fd0d6600d992de58ec8763c652dfc557774b2564b49fe32ad3e561201ef1d51db40a26b506431d0a42295feca50f151b3756f3e827d3fc5895b
-  data.tar.gz: 97649ad23c304278d1b64985261c17ab933992c9c93eb9fedd0c4f56334aa534ed33df8a56b4811e64c96ec22f5fbfa6cc58e732aaa887ce42f8b76e6e297d25
+  metadata.gz: 41591d6fb40e8402e33a9dad54e5b758b59055b7e7cf062af116933c19428eea3385ff2650bdb10a556f69de719aae4e599cdde77fe672d665f96857234ed585
+  data.tar.gz: 9e68f594e322108c7bf4d6e05fcfa7996d00fa62af40ba4c1b8b578723f8ca995e7f0166a4caa5ec7320cd5fe64afb625ab2c88e80c752931e715e1a698a7d80

data/lib/rbbt/association/database.rb

@@ -140,7 +140,7 @@ module Association
 
   def self.database(file,  options = {})
     database = case file
-               when Step
+               when (defined? Step and Step)
                  file.clean if file.error? or file.aborted? or file.dirty?
                  file.run(true) unless file.done? or file.started?
                  file.join unless file.done?

data/lib/rbbt/tsv/attach.rb

@@ -176,7 +176,7 @@ module TSV
 
   def attach(other, options = {})
     options      = Misc.add_defaults options, :in_namespace => false, :persist_input => true
-    fields, one2one = Misc.process_options options, :fields, :one2one
+    fields, one2one, complete = Misc.process_options options, :fields, :one2one, :complete
     in_namespace = options[:in_namespace]
 
     unless TSV === other
@@ -195,6 +195,29 @@ module TSV
     other_filename = other.respond_to?(:filename) ? other.filename : other.inspect
     Log.low("Attaching fields:#{Misc.fingerprint fields } from #{other_filename}.")
 
+    if complete
+      fill = TrueClass === complete ? nil : complete
+      missing = other.keys - self.keys
+      case type
+      when :single
+        missing.each do |k|
+          self[k] = nil
+        end
+      when :list
+        missing.each do |k|
+          self[k] = [nil] * field_length
+        end
+      when :double
+        missing.each do |k|
+          self[k] = [[]] * field_length
+        end
+      when :flat
+        missing.each do |k|
+          self[k] = []
+        end
+      end
+    end
+
     same_key = true
     begin
       case

data/lib/rbbt/tsv/parser.rb

@@ -25,7 +25,7 @@ module TSV
 
       # Process options line
 
-      if line and line =~ /^#{@header_hash}: (.*)/
+      if line and (String === @header_hash and line =~ /^#{@header_hash}: (.*)/)
         options = Misc.string2hash $1.chomp
         line = stream.gets
         line = Misc.fixutf8 line.chomp if line
@@ -38,15 +38,16 @@ module TSV
       # Process fields line
 
       preamble << line if line
-      while line and Misc.fixutf8(line) =~ /^#{@header_hash}/ 
+      while line and (TrueClass === @header_hash or (String === @header_hash and Misc.fixutf8(line) =~ /^#{@header_hash}/ ))
         @fields = line.split(@sep)
         @key_field = @fields.shift
-        @key_field = @key_field[(0 + header_hash.length)..-1] # Remove initial hash character
+        @key_field = @key_field[(0 + header_hash.length)..-1] if String === @header_hash
 
         #Thread.pass while IO.select([stream], nil, nil, 1).nil? if IO === stream
         line = (@header_hash != "" ?  stream.gets : nil)
         line = Misc.fixutf8 line.chomp if line
         preamble << line if line
+        @header_hash = false if TrueClass === @header_hash
       end
 
       @preamble = preamble[0..-3] * "\n"

data/lib/rbbt/tsv/util.rb

@@ -1,6 +1,11 @@
 require 'rbbt/resource/path'
 module TSV
 
+  def self.guess_id(identifier_file, values, options = {})
+    field_matches = TSV.field_match_counts(identifier_file, values, options)
+    field_matches.sort_by{|field, count| count.to_i}.last
+  end
+
   def self.reorder_stream(stream, positions, sep = "\t")
     Misc.open_pipe do |sin|
       line = stream.gets
@@ -50,7 +55,7 @@ module TSV
 
     filename = TSV === file ? file.filename : file
     path = Persist.persist filename, :string, persist_options.merge(:no_load => true) do
-      tsv = TSV === file ? file : TSV.open(file)
+      tsv = TSV === file ? file : TSV.open(file, options)
 
       text = ""
       fields = nil
@@ -64,8 +69,8 @@ module TSV
       text
     end
 
-    TmpFile.with_file(values.uniq * "\n") do |value_file|
-      cmd = "cat '#{ path }' | sed 's/\\t/\\tHEADERNOMATCH/' | grep -w -F -f '#{ value_file }' |cut -f 2 | sed 's/HEADERNOMATCH//' | sort|uniq -c|sed 's/^ *//;s/ /\t/'"
+    TmpFile.with_file(values.uniq * "\n", false) do |value_file|
+      cmd = "cat '#{ path }' | sed 's/\\t/\\tHEADERNOMATCH/' | grep -w -F -f '#{ value_file }' | sed 's/HEADERNOMATCH//' |sort -u|cut -f 2  |sort|uniq -c|sed 's/^ *//;s/ /\t/'"
       begin
         TSV.open(CMD.cmd(cmd), :key_field => 1, :type => :single, :cast => :to_i)
       rescue

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

@@ -145,8 +145,15 @@ module Misc
       when (ref.length == 1 and m.length == 1)
         m
       else
-        Log.debug{"Cannot understand: #{[ref, m]} (#{ muts })"}
-        '-' * ref.length + m
+        if ref == '-'
+          res = '+' + m
+        else
+          res = '-' * ref.length 
+          res << m unless m == '-'
+        end
+        Log.debug{"Non-standard annotation: #{[ref, m]} (#{ muts }) => #{ res }"}
+        
+        res
       end
     end
 
@@ -173,8 +180,15 @@ module Misc
       when (ref.length == 1 and m.length == 1)
         m
       else
-        Log.debug{"Cannot understand: #{[ref, m]} (#{ muts })"}
-        '-' * ref.length + m
+        if ref == '-'
+          res = '+' + m
+        else
+          res = '-' * ref.length 
+          res << m unless m == '-'
+        end
+        Log.debug{"Non-standard annotation: #{[ref, m]} (#{ muts }) => #{ res }"}
+        
+        res
       end
     end
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-util
 version: !ruby/object:Gem::Version
-  version: 5.17.77
+  version: 5.17.78
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-09-30 00:00:00.000000000 Z
+date: 2015-10-06 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake