rbbt-util 5.14.6 → 5.14.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 861750b318943153ffe352fab01ea7cabcd9c3db
-  data.tar.gz: 9ca19a6bae45a91b9fd47df1bacf4316a4bc511f
+  metadata.gz: fc4fe61feb344cc28a196b45ce9a9a3f5f0cf8a1
+  data.tar.gz: 36d5dad6c8f90493f0a998c9078c74a8836084a9
 SHA512:
-  metadata.gz: e952c743e3fc39056eb52152073e387f5b60ddbe363eb85be3266d590f0db89891c86d9b0671f85a11d019e78f69ed7b475da5e83dcc716b677e975bf18d1ccf
-  data.tar.gz: f66789350e2f1f936675b4363971b4c189f6a128b6e5fd90ccf82a09910283969a2cc8307aeeeb6008de120b2c015fc5abd1eb97bc07ba2a0687fc751b946a86
+  metadata.gz: 62bcc281c2dc513b52b67515b8df46b6151b318a6660a5096b7913637779e550a24c8ade84f2847c7af18760526abd5ac30a1f1d912651c44f8db33d16b9acb8
+  data.tar.gz: d89848d0ec11114934bb762fbdde56e5c2d81216925b5450cf223f120798d1378428c707326ac6d3840b00726e24d06c89cd0d733c1e776eaf1700a913a10796

data/lib/rbbt/association.rb

@@ -78,7 +78,7 @@ module Association
       :persist => false,
       :key_field => all_fields.index(source), 
       :fields => fields.collect{|f| String === f ? all_fields.index(f): f },
-      :type => (options[:type] and options[:type].to_sym == :flat) ? :flat : :double,
+      #:type => (options[:type] and options[:type].to_sym == :flat) ? :flat : nil,
       :unnamed => true,
       :merge => (options[:type] and options[:type].to_sym == :flat) ? false : true
     })

data/lib/rbbt/association/index.rb

@@ -20,6 +20,7 @@ module Association
 
                      if File.exists?(reverse_filename)
                        new = Persist.open_tokyocabinet(reverse_filename, false, serializer, TokyoCabinet::BDB)
+                       new
                      else
                        FileUtils.mkdir_p File.basename(reverse_filename) unless File.exists?(File.basename(reverse_filename))
                        new = Persist.open_tokyocabinet(reverse_filename, true, serializer, TokyoCabinet::BDB)
@@ -64,8 +65,8 @@ module Association
       source_type = Entity.formats[source_field] 
       target_type = Entity.formats[target_field]
 
-      source_entities = entities[source_field] || entities[Entity.formats[source_field].to_s]  
-      target_entities = entities[target_field] || entities[Entity.formats[target_field].to_s]
+      source_entities = entities[:source] || entities[source_field] || entities[Entity.formats[source_field].to_s]  
+      target_entities = entities[:target] || entities[target_field] || entities[Entity.formats[target_field].to_s]
 
       [source_entities, target_entities]
     end
@@ -79,7 +80,7 @@ module Association
         matches = source.uniq.inject([]){|acc,e| acc.concat(match(e)) }
       end
 
-      return matches if target == :all or target == "all"
+      return matches if target.nil? or target == :all or target == "all"
 
       target_matches = {}
 
@@ -95,6 +96,10 @@ module Association
 
     def subset_entities(entities)
       source, target = select_entities(entities)
+      raise "Please specify source entities" if source.nil?
+      target = :all if target.nil?
+      return if Array === target and target.empty?
+      return if Array === source and source.empty?
       subset source, target
     end
   end

data/lib/rbbt/association/item.rb

@@ -19,6 +19,14 @@ module AssociationItem
     self.clean_annotations.collect{|p| p[/[^~]+/] }
   end
 
+  property :target_type => :single do
+    reverse ? knowledge_base.source(database) : knowledge_base.target(database)
+  end
+
+  property :source_type => :single do
+    reverse ? knowledge_base.target(database) : knowledge_base.source(database)
+  end
+
   property :target_entity => :array2single do
     type = reverse ? knowledge_base.source(database) : knowledge_base.target(database)
     knowledge_base.annotate self.target, type, database #if self.target.any?

data/lib/rbbt/knowledge_base.rb

@@ -149,10 +149,6 @@ class KnowledgeBase
       end
   end
 
-  #def index(name, file, options = {}, persist_options = {})
-  #  @indices[name] = Association.index(file, open_options.merge(options), persist_options)
-  #end
-
   #{{{ Add manual database
   
   def add_index(name, source_type, target_type, *fields)
@@ -279,19 +275,28 @@ class KnowledgeBase
   end
 
   def subset(name, entities)
-    case entities
-    when AnnotatedArray
-      format = entities.format if entities.respond_to? :format 
-      format ||= entities.base_entity.to_s
-      {format => entities.clean_annotations}
-    when Hash
-    else
-      raise "Entities are not a Hash or an AnnotatedArray: #{Misc.fingerprint entities}"
-    end
+    entities = case entities
+               when AnnotatedArray
+                 format = entities.format if entities.respond_to? :format 
+                 format ||= entities.base_entity.to_s
+                 {format => entities.clean_annotations}
+               when Hash
+                 entities
+               else
+                 raise "Entities are not a Hash or an AnnotatedArray: #{Misc.fingerprint entities}"
+               end
+
     repo = get_index name
+
     begin
       setup(name, repo.subset_entities(entities))
-    rescue 
+    rescue Exception
+      target = entities[:target]
+      source = entities[:source]
+      if target or source
+        entities[:target] = source 
+        entities[:source] = target
+      end
       setup(name, repo.reverse.subset_entities(entities), true)
     end
   end

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

@@ -71,13 +71,13 @@ module Misc
     text
   end
 
-  def self.format_definition_list(defs, size = 80, indent = 20, color = :yellow)
+  def self.format_definition_list(defs, size = 80, indent = 20, color = :yellow, sep = "\n\n")
     entries = []
     defs.each do |dt,dd|
       text = format_definition_list_item(dt,dd,size,indent,color)
       entries << text
     end
-    entries * "\n\n"
+    entries * sep 
   end
 
   def self.camel_case(string)

data/lib/rbbt/workflow/accessor.rb

@@ -503,7 +503,8 @@ module Workflow
         options.each{|i,v|
           case v
           when Symbol
-            rec_dependency = real_dependencies.select{|d| d.task.name == v }.first
+            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]
             inputs[i] = rec_dependency
           else
             inputs[i] = v

data/share/rbbt_commands/association/subset

@@ -0,0 +1,41 @@
+#!/usr/bin/env ruby
+
+require 'rbbt-util'
+require 'rbbt/util/simpleopt'
+require 'rbbt/association'
+
+$0 = "rbbt #{$previous_commands*""} #{ File.basename(__FILE__) }" if $previous_commands
+
+options = SOPT.setup <<EOF
+
+Subset of matches between entities
+
+$ rbbt association subset [options] <filename>
+
+Indicate the index file (TokyoCabinet BDB with source~target keys) as the <filename> argument
+
+-h--help Print this help
+-s--source* Source entities
+-t--target* Target entities
+EOF
+rbbt_usage and exit 0 if options[:help] 
+filename = ARGV.shift
+raise ParameterException, "No filename specified" if filename.nil?
+
+file = Persist.open_tokyocabinet(filename,false, nil, TokyoCabinet::BDB)
+
+file.extend Association::Index
+
+source = options[:source] ? options[:source].split(/,\|/) : :all
+target = options[:target] ? options[:target].split(/,\|/) : :all
+
+file.subset(source, target).each do |item|
+  puts Log.color(:magenta, item)
+  info = file.fields.zip(file[item])
+  source,_sep, target = item.partition "~"
+  puts "  " << Misc.format_definition_list_item("source", source)
+  puts "  " << Misc.format_definition_list_item("target", target)
+  info.each do |key,value|
+    puts "  " << Misc.format_definition_list_item(key, value)
+  end
+end

data/share/rbbt_commands/workflow/knowledge_base

@@ -0,0 +1,68 @@
+#!/usr/bin/env ruby
+
+require 'rbbt/util/simpleopt'
+require 'rbbt/workflow'
+require 'rbbt/workflow/usage'
+require 'time'
+
+options = SOPT.setup <<EOF
+Access workflow knowledge_base
+
+$ rbbt workflow knowledge_base <workflow> [<database>] [<action>] [<options>]
+
+Access workflow knowledge base
+
+-h--help Show this help:
+-cl--clean Clean the last step of the job so that it gets recomputed:
+-sf--source_format* Source format
+-se--source_entities* Source entities
+-tf--target_format* Target format
+-te--target_entities* Target entities
+-i--info Print information on matches
+EOF
+
+workflow = ARGV.shift
+usage and exit -1 if workflow.nil? or options[:help]
+
+database     = ARGV.shift
+action       = ARGV.shift
+
+# Get workflow
+
+workflow = Workflow.require_workflow workflow
+
+action = :list_databases if database.nil?
+action ||= :summary 
+
+case action.to_sym
+when :list_databases
+  dbs = workflow.knowledge_base.registry.keys
+  puts dbs * "\n"
+when :summary
+  db = workflow.knowledge_base.get_database(database)
+  puts db.summary
+when :subset
+  source_format = options[:source_format] || :source
+  target_format = options[:target_format] || :target
+  source_entities = options[:source_entities] || :all
+  target_entities = options[:target_entities] || :all
+
+  source_entities = source_entities.split(/,\|/) if String === source_entities
+  target_entities = target_entities.split(/,\|/) if String === target_entities
+
+  entities = {source_format => source_entities, target_format => target_entities}
+
+  matches = workflow.knowledge_base.subset(database, entities)
+  matches.each do |item|
+    puts Log.color :magenta, item
+    if options[:info]
+      source = item.source_entity
+      target = item.target_entity
+      source = source.name if source.respond_to? :name
+      target = target.name if target.respond_to? :name
+
+      info = item.info.merge(:source => source, :target => target)
+      puts Misc.format_definition_list(info, 80, 20, :yellow, "\n") 
+    end
+  end
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-util
 version: !ruby/object:Gem::Version
-  version: 5.14.6
+  version: 5.14.7
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-07-15 00:00:00.000000000 Z
+date: 2014-07-16 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake
@@ -232,6 +232,7 @@ files:
 - share/rbbt_commands/app/install
 - share/rbbt_commands/app/start
 - share/rbbt_commands/app/template
+- share/rbbt_commands/association/subset
 - share/rbbt_commands/benchmark/pthrough
 - share/rbbt_commands/benchmark/throughput
 - share/rbbt_commands/benchmark/tsv
@@ -272,6 +273,7 @@ files:
 - share/rbbt_commands/workflow/info
 - share/rbbt_commands/workflow/install
 - share/rbbt_commands/workflow/jobs
+- share/rbbt_commands/workflow/knowledge_base
 - share/rbbt_commands/workflow/list
 - share/rbbt_commands/workflow/monitor
 - share/rbbt_commands/workflow/prov