rbbt-util 5.14.7 → 5.14.8

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: fc4fe61feb344cc28a196b45ce9a9a3f5f0cf8a1
-  data.tar.gz: 36d5dad6c8f90493f0a998c9078c74a8836084a9
+  metadata.gz: d676e9769e4e13d664c42a65e74ec8c7d70bbe63
+  data.tar.gz: f3a75c5456360ae6e31d713ef964650adcb2edb4
 SHA512:
-  metadata.gz: 62bcc281c2dc513b52b67515b8df46b6151b318a6660a5096b7913637779e550a24c8ade84f2847c7af18760526abd5ac30a1f1d912651c44f8db33d16b9acb8
-  data.tar.gz: d89848d0ec11114934bb762fbdde56e5c2d81216925b5450cf223f120798d1378428c707326ac6d3840b00726e24d06c89cd0d733c1e776eaf1700a913a10796
+  metadata.gz: e2cb4d0761e9b44f8d108888a966b874f572f144b16620754c6551349fc8c8d04ee061393e7032d829a3fa1f0e72a04185e5f6af3d5c264d4582daf7524063e9
+  data.tar.gz: 8402954673614b9b0ff6c4b73237a47d1f077ef795f9c452f96fab732e665428cb470e2ae38c12ed3c818c0c1b2d512d8322444020ba3b1783e528c30ce8a048

data/lib/rbbt/association/index.rb

@@ -96,6 +96,7 @@ module Association
 
     def subset_entities(entities)
       source, target = select_entities(entities)
+      source = :all if source.nil? and target.nil?
       raise "Please specify source entities" if source.nil?
       target = :all if target.nil?
       return if Array === target and target.empty?

data/lib/rbbt/association/item.rb

@@ -38,7 +38,7 @@ module AssociationItem
   end
 
   property :value => :array2single do
-    value = knowledge_base.get_index(database).chunked_values_at self
+    value = (reverse ? knowledge_base.get_index(database).reverse : knowledge_base.get_index(database)).chunked_values_at self
     value.collect{|v| NamedArray.setup(v, knowledge_base.get_index(database).fields)}
   end
 
@@ -46,6 +46,7 @@ module AssociationItem
     fields = knowledge_base.index_fields(database)
     return [{}] * self.length if fields.nil? or fields.empty?
 
+    value = self.value
     value.collect{|v|
       Hash[*fields.zip(v).flatten]
     }

data/lib/rbbt/association.rb

@@ -12,19 +12,22 @@ module Association
 
   def self.add_reciprocal(tsv)
 
-    new = {}
+    new = tsv.dup
     tsv.with_unnamed do
       tsv.through do |key, values|
-        new[key] ||= values
-        Misc.zip_fields(values).each do |fields|
-          target, *rest = fields
-          
-          target_values = new[target] || tsv[target] || [[]] * values.length
-          zipped_target_values = Misc.zip_fields(target_values) 
-
-          zipped_target_values << ([key].concat rest)
-          
-          new_values = Misc.zip_fields zipped_target_values
+        Misc.zip_fields(values).each do |zipped|
+          target, *rest = zipped
+
+          target_values = new[target] || []
+          if target_values and target_values.any?
+            zipped_target_values = Misc.zip_fields(target_values) 
+
+            zipped_target_values << ([key].concat rest)
+
+            new_values = Misc.zip_fields zipped_target_values
+          else
+            new_values = [[key], rest.collect{|v| [v]}]
+          end
 
           new[target] = new_values
         end

data/lib/rbbt/knowledge_base.rb

@@ -196,30 +196,37 @@ class KnowledgeBase
   
   def identify_source(name, entity)
     database = get_database(name, :persist => true)
-    return entity if database.include? entity
+    return entity if Symbol === entity or (String === entity and database.include? entity)
     source = source(name)
     @identifiers[name] ||= {}
     @identifiers[name]['source'] ||= begin
                                        if database.identifier_files.any?
                                          if TSV.parse_header(database.identifier_files.first).all_fields.include? source
-                                           TSV.index(database.identifiers, :target => source, :persist => true)
+                                           TSV.index(database.identifiers, :target => source, :persist => true, :order => true)
                                          else
                                            {}
                                          end
                                        else
                                          if TSV.parse_header(Organism.identifiers(namespace)).all_fields.include? source
-                                           Organism.identifiers(namespace).index(:target => source, :persist => true)
+                                           Organism.identifiers(namespace).index(:target => source, :persist => true, :order => true)
                                          else
                                            {}
                                          end
                                        end
                                      end
 
-    @identifiers[name]['source'][entity]
+    if Array === entity
+      @identifiers[name]['source'].chunked_values_at(entity).zip(entity).collect{|p|
+        p.compact.first
+      }
+    else
+      @identifiers[name]['source'][entity] || entity
+    end
   end
 
   def identify_target(name, entity)
     database = get_database(name, :persist => true)
+    return entity if Symbol === entity
     target = target(name)
 
     @identifiers[name] ||= {}
@@ -238,7 +245,13 @@ class KnowledgeBase
                                          end
                                        end
                                      end
-    @identifiers[name]['target'][entity]
+    if Array === entity
+      @identifiers[name]['target'].chunked_values_at(entity).zip(entity).collect{|p|
+        p.compact.first
+      }
+    else
+      @identifiers[name]['target'][entity] || entity
+    end
   end
 
   def identify(name, entity)

data/lib/rbbt/tsv/accessor.rb

@@ -593,6 +593,7 @@ module TSV
     end
     with_unnamed do
       <<-EOF
+Filename = #{Path === filename ? filename.find : (filename || "No filename")}
 Key field = #{key_field || "*No key field*"}
 Fields = #{fields ? Misc.fingerprint(fields) : "*No field info*"}
 Type = #{type}

data/lib/rbbt/tsv/change_id.rb

@@ -7,7 +7,13 @@ module TSV
     identifiers, persist_input = Misc.process_options options, :identifiers, :persist_input
 
     if not tsv.fields.include? format
-      tsv = tsv.annotate(Hash[*tsv.keys.zip(tsv.values.collect{|l| l.dup}).flatten(1)]) 
+      new = {}
+      tsv.each do |k,v|
+        new[k] = v.dup
+      end
+      orig_fields = tsv.fields
+      tsv = tsv.annotate new
+      new.fields = new.fields.collect{|f| "TMP-" << f }
 
       orig_type = tsv.type 
       tsv = tsv.to_double if orig_type != :double
@@ -18,12 +24,14 @@ module TSV
         tsv = tsv.attach identifiers, :fields => [format], :persist_input => true
       end
 
-      tsv = tsv.reorder(format, tsv.fields - [format])
+      tsv = tsv.reorder(format, tsv.fields[0..-2])
 
       tsv = tsv.to_flat  if orig_type == :flat
 
       tsv = tsv.to_list(&block)  if orig_type == :list
 
+      tsv.fields = orig_fields
+
       tsv
     else
       tsv.reorder(format)

data/lib/rbbt/util/named_array.rb

@@ -15,8 +15,9 @@ module NamedArray
   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 = fields
+    array.fields = Annotated.purge fields
     array.key = key
     array.entity_options = entity_options unless entity_options.nil?
     array.entity_templates = entity_templates unless entity_templates.nil?
@@ -112,9 +113,12 @@ module NamedArray
 
   def each(&block)
     if defined?(Entity) and not @fields.nil? and not @fields.empty?
-      @fields.zip(self).each do |field,elem|
+      i = 0
+      super do |elem|
+        field = @fields[i]
         elem = prepare_entity(elem, field, entity_options)
         yield(elem)
+        i += 1
         elem
       end
     else

data/lib/rbbt/workflow/accessor.rb

@@ -499,12 +499,13 @@ module Workflow
       real_dependencies << case dependency
       when Array
         inputs = inputs.dup
+        workflow, task, options = dependency
         options = dependency.last if Hash === dependency.last
         options.each{|i,v|
           case v
           when Symbol
             rec_dependency = real_dependencies.collect{|d| [d, d.dependencies].flatten}.flatten.select{|d| d.task.name == v }.first
-            rec_dependency = rec_dependency.run unless (dependency.first.tasks[dependency[1]].input_options[i] || {})[:stream]
+            rec_dependency = rec_dependency.run(true).grace.join.load unless (dependency.first.tasks[dependency[1]].input_options[i] || {})[:stream]
             inputs[i] = rec_dependency
           else
             inputs[i] = v

data/share/rbbt_commands/workflow/knowledge_base

@@ -14,11 +14,12 @@ Access workflow knowledge base
 
 -h--help Show this help:
 -cl--clean Clean the last step of the job so that it gets recomputed:
+-s--source_entities* Source entities
 -sf--source_format* Source format
--se--source_entities* Source entities
+-t--target_entities* Target entities
 -tf--target_format* Target format
--te--target_entities* Target entities
--i--info Print information on matches
+-d--details Print details on matches
+-i--identify Attempt to identify entities by identifier translation
 EOF
 
 workflow = ARGV.shift
@@ -34,12 +35,14 @@ workflow = Workflow.require_workflow workflow
 action = :list_databases if database.nil?
 action ||= :summary 
 
+knowledge_base = workflow.knowledge_base
+
 case action.to_sym
 when :list_databases
-  dbs = workflow.knowledge_base.registry.keys
+  dbs = knowledge_base.registry.keys
   puts dbs * "\n"
 when :summary
-  db = workflow.knowledge_base.get_database(database)
+  db = knowledge_base.get_database(database)
   puts db.summary
 when :subset
   source_format = options[:source_format] || :source
@@ -50,12 +53,16 @@ when :subset
   source_entities = source_entities.split(/,\|/) if String === source_entities
   target_entities = target_entities.split(/,\|/) if String === target_entities
 
+  if options[:identify]
+    source_entities = knowledge_base.identify(database, source_entities)
+    target_entities = knowledge_base.identify(database, target_entities)
+  end
   entities = {source_format => source_entities, target_format => target_entities}
 
-  matches = workflow.knowledge_base.subset(database, entities)
+  matches = knowledge_base.subset(database, entities)
   matches.each do |item|
     puts Log.color :magenta, item
-    if options[:info]
+    if options[:details]
       source = item.source_entity
       target = item.target_entity
       source = source.name if source.respond_to? :name

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-util
 version: !ruby/object:Gem::Version
-  version: 5.14.7
+  version: 5.14.8
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-07-16 00:00:00.000000000 Z
+date: 2014-07-17 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake