scout-gear 10.0.1 → 10.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5af158e978a724dacd722c9584cb4fd3b7f46fff8a4619272b1d2c2258418600
-  data.tar.gz: 6bd5da8eba38542376f4b5566154fc16105f3742ba0758a37139d0ff9679d06e
+  metadata.gz: 3e807e8eea8b8ff3b40be002fcd9cdfa9fbc8d3fdb447f20b1b3c56f1c318de8
+  data.tar.gz: 8838111c9867b63671060c82b7224f2ac310b64d375f0cc62c26b42092da6f1d
 SHA512:
-  metadata.gz: e9f6f4d2a5e223d740332b627c9606907d69536696f477401087b4f72bee09d1b374d060275ecb3fad5e38c27010417d36d661d985c58d71e3cec4bf04bbcbf7
-  data.tar.gz: a7b8e5d1bfd1057bd93cfbd4a2439bccf0e28848f3e08923c81da8bcc2fe1a489ffb2e220587ecb4533ad0cf7edcf506f3c4557955dfa11356d816b87b379968
+  metadata.gz: 2dcc68c9d612b208d6b6b94dd366ef37f2467b364d1da8367b1f6833d305d6ef8c35a1dc0f0e27c6e00ce481344e230e20ac6eb85236dd27c75c59e8ad3f0ec5
+  data.tar.gz: 85d8e7458a1453966e2e123fc8ced3a0c2bc8be884a7570f1bb57f4d260db582e90380d2ea0f8a91c91b68decb8439e8b556d294b9d71258c522fa06628ec22b

data/.vimproject

@@ -8,6 +8,7 @@ scout-gear=/$PWD filter="*.rb *.yaml" {
   alias
   find
   glob
+  log
   doc
   update
   template
@@ -16,7 +17,14 @@ scout-gear=/$PWD filter="*.rb *.yaml" {
    task
    list
    info
+   write_info
    install
+   trace
+   prov
+  }
+  batch=batch{
+   list
+   clean
   }
   resource=resource{
    produce
@@ -28,81 +36,6 @@ scout-gear=/$PWD filter="*.rb *.yaml" {
   rbbt-scout.rb
   scout.rb
   scout=scout{
-   meta_extension.rb
-   exceptions.rb
-
-   misc.rb
-   misc=misc{
-    format.rb
-    insist.rb
-    digest.rb
-    filesystem.rb
-    monitor.rb
-    system.rb
-   }
-   named_array.rb
-   indiferent_hash.rb
-   indiferent_hash=indiferent_hash{
-    case_insensitive.rb
-    options.rb
-   }
-   log.rb
-   log=log{
-    color.rb
-    color_class.rb
-    fingerprint.rb
-    progress.rb
-    progress=progress{
-     report.rb
-     util.rb
-    }
-    trap.rb
-   }
-   tmpfile.rb
-   simple_opt.rb
-   simple_opt=simple_opt{
-    accessor.rb
-    doc.rb
-    parse.rb
-    get.rb
-    setup.rb
-   }
-   path.rb
-   path=path{
-    find.rb
-    util.rb
-    tmpfile.rb
-   }
-   concurrent_stream.rb
-   cmd.rb
-   open.rb
-   open=open{
-    lock.rb
-    remote.rb
-    stream.rb
-    util.rb
-    lock=lock{
-     lockfile.rb
-    }
-   }
-   resource.rb
-   resource=resource{
-    produce.rb
-    produce=produce{
-     rake.rb
-    }
-    software.rb
-    scout.rb
-    util.rb
-    path.rb
-   }
-   config.rb
-   persist.rb
-   persist=persist{
-    serialize.rb
-    open.rb
-    path.rb
-   }
    workflow.rb
    workflow=workflow{
     definition.rb
@@ -146,6 +79,8 @@ scout-gear=/$PWD filter="*.rb *.yaml" {
      process.rb
      unzip.rb
      reorder.rb
+     sort.rb
+     melt.rb
     }
     parser.rb
     dumper.rb
@@ -173,9 +108,6 @@ scout-gear=/$PWD filter="*.rb *.yaml" {
     sync.rb
     step.rb
    }
-   hpc=hpc{
-    slurm.rb
-   }
   }
  }
  test=test {
@@ -199,6 +131,10 @@ scout-gear=/$PWD filter="*.rb *.yaml" {
    test_tsv.rb
    test_work_queue.rb
    test_workflow.rb
+   work_queue=work_queue{
+    test_socket.rb
+    test_worker.rb
+   }
    workflow=workflow{
     test_definition.rb
     test_documentation.rb
@@ -217,10 +153,42 @@ scout-gear=/$PWD filter="*.rb *.yaml" {
      test_dependencies.rb
      test_inputs.rb
     }
+    deployment=deployment{
+     test_orchestrator.rb
+    }
+   }
+   tsv=tsv{
+    test_attach.rb
+    test_change_id.rb
+    test_dumper.rb
+    test_index.rb
+    test_open.rb
+    test_parser.rb
+    test_persist.rb
+    test_stream.rb
+    test_transformer.rb
+    test_traverse.rb
+    test_util.rb
+    util=util{
+     test_filter.rb
+     test_melt.rb
+     test_process.rb
+     test_reorder.rb
+     test_select.rb
+     test_sort.rb
+     test_unzip.rb
+    }
+    persist=persist{
+     test_adapter.rb
+     test_fix_width_table.rb
+     test_tokyocabinet.rb
+    }
    }
-   indiferent_hash=indiferent_hash{
-    test_case_insensitive.rb
-    test_options.rb
+   offsite=offsite{
+    test_ssh.rb
+    test_step.rb
+    test_sync.rb
+    test_task.rb
    }
   }
  }
@@ -601,6 +569,7 @@ scout-gear=/$PWD filter="*.rb *.yaml" {
       test_filter.rb
       test_index.rb
       test_manipulate.rb
+      test_marshal.rb
       test_matrix.rb
       test_parallel.rb
       test_parser.rb

data/Rakefile

@@ -35,6 +35,7 @@ Juwelier::RubygemsDotOrgTasks.new
 
 require 'rake/testtask'
 Rake::TestTask.new(:test) do |test|
+  ENV["SCOUT_LOG"] = "NONE"
   test.libs << 'lib' << 'test'
   test.pattern = 'test/**/test_*.rb'
   test.verbose = true

data/VERSION

@@ -1 +1 @@
-10.0.1
+10.2.0

data/lib/scout/tsv/transformer.rb

@@ -2,15 +2,16 @@ module TSV
   class Transformer
     attr_accessor :unnamed, :parser, :dumper
 
-    def initialize(parser, dumper = nil, unnamed: false)
+    def initialize(parser, dumper = nil, unnamed: nil)
       if TSV::Parser === parser
         @parser = parser
       elsif TSV === parser
         @parser = parser
+        @unnamed = parser.unnamed
       else
         @parser = TSV::Parser.new parser
       end
-      @unnamed = unnamed
+      @unnamed = unnamed unless unnamed.nil?
       if dumper.nil?
         @dumper = TSV::Dumper.new(@parser)
         @dumper.sep = "\t"
@@ -70,7 +71,7 @@ module TSV
       @dumper.init if @dumper.respond_to?(:init) && ! @dumper.initialized
       Log.debug "Transform #{Log.fingerprint @parser} into #{Log.fingerprint @dumper}"
       Open.traverse(@parser, *args, **kwargs) do |k,v|
-        NamedArray.setup(v, @parser.fields, k) unless @unnamed
+        NamedArray.setup(v, @parser.fields, k) unless @unnamed || @parser.fields.nil?
         block.call k, v
       end
     end
@@ -112,16 +113,18 @@ module TSV
 
   def to_list
     res = self.annotate({})
-    transformer = Transformer.new self, res
-    transformer.type = :list
-    transformer.traverse do |k,v|
-      case self.type
-      when :single
-        [k, [v]]
-      when :double
-        [k, v.collect{|v| v.first }]
-      when :flat
-        [k, v.slice(0,1)]
+    self.with_unnamed do
+      transformer = Transformer.new self, res
+      transformer.type = :list
+      transformer.traverse do |k,v|
+        case self.type
+        when :single
+          [k, [v]]
+        when :double
+          [k, v.collect{|v| v.first }]
+        when :flat
+          [k, v.slice(0,1)]
+        end
       end
     end
     res
@@ -131,6 +134,7 @@ module TSV
     res = self.annotate({})
     transformer = Transformer.new self, res
     transformer.type = :single
+    transformer.unnamed = true
     transformer.traverse do |k,v|
       v = v.first while Array === v
       [k, v]

data/lib/scout/tsv/util/melt.rb

@@ -0,0 +1,13 @@
+module TSV
+  def melt_columns(value_field, column_field)
+    target = TSV.setup({}, :key_field => "ID", :fields => [key_field, value_field, column_field], :type => :list, :cast => cast)
+    each do |k,values|
+      i = 0
+      values.zip(fields).each do |v,f|
+        target["#{k}:#{i}"] = [k,v,f]
+        i+=1
+      end
+    end
+    target
+  end
+end

data/lib/scout/tsv/util/sort.rb

@@ -0,0 +1,74 @@
+module TSV
+  def sort_by(field = nil, just_keys = false, &block)
+    field = :all if field.nil?
+
+    if field == :all
+      elems = collect
+    else
+      elems = []
+      case type
+      when :single
+        through :key, field do |key, field|
+          elems << [key, field]
+        end
+      when :list, :flat
+        through :key, field do |key, fields|
+          elems << [key, fields.first]
+        end
+      when :double
+        through :key, field do |key, fields|
+          elems << [key, fields.first]
+        end
+      end
+    end
+
+    if not block_given?
+      if fields == :all
+        if just_keys
+          keys = elems.sort_by{|key, value| key }.collect{|key, values| key}
+          keys = prepare_entity(keys, key_field, entity_options.merge(:dup_array => true))
+        else
+          elems.sort_by{|key, value| key }
+        end
+      else
+        sorted = elems.sort do |a, b| 
+          a_value = a.last
+          b_value = b.last
+          a_empty = a_value.nil? or (a_value.respond_to?(:empty?) and a_value.empty?)
+          b_empty = b_value.nil? or (b_value.respond_to?(:empty?) and b_value.empty?)
+          case
+          when (a_empty and b_empty)
+            0
+          when a_empty
+            -1
+          when b_empty
+            1
+          when Array === a_value
+            if a_value.length == 1 and b_value.length == 1
+              a_value.first <=> b_value.first
+            else
+              a_value.length <=> b_value.length
+            end
+          else
+            a_value <=> b_value
+          end
+        end
+        if just_keys
+          keys = sorted.collect{|key, value| key}
+          keys = prepare_entity(keys, key_field, entity_options.merge(:dup_array => true)) unless @unnamed
+          keys
+        else
+          sorted.collect{|key, value| [key, self[key]]}
+        end
+      end
+    else
+      if just_keys
+        keys = elems.sort_by(&block).collect{|key, value| key}
+        keys = prepare_entity(keys, key_field, entity_options.merge(:dup_array => true)) unless @unnamed
+        keys
+      else
+        elems.sort_by(&block).collect{|key, value| [key, self[key]]}
+      end
+    end
+  end
+end

data/lib/scout/tsv/util.rb

@@ -6,6 +6,8 @@ require_relative 'util/process'
 require_relative 'util/select'
 require_relative 'util/reorder'
 require_relative 'util/unzip'
+require_relative 'util/sort'
+require_relative 'util/melt'
 module TSV
   def self.identify_field(key_field, fields, name, strict: nil)
     return :key if name == :key || (! strict && NamedArray.field_match(key_field, name))

data/lib/scout/tsv.rb

@@ -34,11 +34,14 @@ module TSV
   end
 
   def self.open(file, options = {})
-    persist, type, grep, invert_grep = IndiferentHash.process_options options, :persist, :persist_type, :grep, :invert_grep, :persist => false, :persist_type => "HDB"
-    type = type.to_sym if type
+    grep, invert_grep = IndiferentHash.process_options options, :grep, :invert_grep, :persist => false
+
+    persist_options = IndiferentHash.pull_keys options, :persist
+    persist_options = IndiferentHash.add_defaults persist_options, :prefix => "TSV", :type => "HDB"
+
     file = StringIO.new file if String === file && ! (Path === file) && file.index("\n")
-    Persist.persist(file, type, options.merge(:persist => persist, :prefix => "Tsv", :other_options => options)) do |filename|
-      data = filename ? ScoutCabinet.open(filename, true, type) : nil
+    Persist.persist(file, persist_options[:type], persist_options.merge(:other_options => options)) do |filename|
+      data = filename ? ScoutCabinet.open(filename, true, persist_options[:type]) : nil
       options[:data] = data if data
       options[:filename] = file
 

data/lib/scout/workflow/definition.rb

@@ -84,8 +84,10 @@ module Workflow
     annotate_next_task :deps, [workflow, task, options, block, args]
   end
 
-  def input(*args)
-    annotate_next_task(:inputs, args)
+  def input(name, type = nil, *rest)
+    name = name.to_sym
+    type = type.to_sym if type
+    annotate_next_task(:inputs, [name, type] + rest)
   end
 
   def desc(description)
@@ -102,6 +104,8 @@ module Workflow
 
   def task(name_and_type, &block)
     name, type = name_and_type.collect.first
+    type = type.to_sym if String === type
+    name = name.to_sym if String === name
     @tasks ||= IndiferentHash.setup({})
     block = self.method(name) if block.nil?
     begin
@@ -132,8 +136,10 @@ module Workflow
   def task_alias(name, workflow, oname, *rest, &block)
     dep(workflow, oname, *rest, &block) 
     extension :dep_task unless @extension
-    returns workflow.tasks[oname].returns if @returns.nil?
-    type = workflow.tasks[oname].type 
+    task_proc = workflow.tasks[oname]
+    raise "Task #{name} not found" if task_proc.nil?
+    returns task_proc.returns if @returns.nil?
+    type = task_proc.type 
     task name => type do
       raise RbbtException, "dep_task does not have any dependencies" if dependencies.empty?
       Step.wait_for_jobs dependencies.select{|d| d.streaming? }

data/lib/scout/workflow/step/dependencies.rb

@@ -5,7 +5,7 @@ class Step
       next if seen.include? dep.path
       next if connected && dep.done? && dep.updated?
       direct_deps << dep
-    end
+    end if dependencies
     seen.concat direct_deps.collect{|d| d.path }
     direct_deps.inject(direct_deps){|acc,d| acc.concat(d.rec_dependencies(connected, seen)); acc }
   end

data/lib/scout/workflow/step/file.rb

@@ -12,9 +12,10 @@ class Step
                    end
   end
 
-  def file(file)
+  def file(file = nil)
     dir = files_dir
     Path.setup(dir) unless Path === dir
+    return dir if file.nil?
     dir[file]
   end
 

data/lib/scout/workflow/step/provenance.rb

@@ -68,7 +68,7 @@ class Step
       $inputs.each do |input|
         value = job_inputs[input]
         next if  value.nil?
-        value_str = Misc.fingerprint(value)
+        value_str = Log.fingerprint(value)
         str << "\t#{Log.color :magenta, input}=#{value_str}"
       end
     end
@@ -78,7 +78,7 @@ class Step
         IndiferentHash.setup(info)
         value = info[field]
         next if value.nil?
-        value_str = Misc.fingerprint(value)
+        value_str = Log.fingerprint(value)
         str << "\t#{Log.color :magenta, field}=#{value_str}"
       end
     end

data/lib/scout/workflow/step/status.rb

@@ -27,10 +27,11 @@ class Step
     @result = nil
     @info = nil
     @info_load_time = nil
-    Open.rm path if Open.exist?(path)
-    Open.rm tmp_path if Open.exist?(tmp_path)
-    Open.rm info_file if Open.exist?(info_file)
-    Open.rm_rf files_dir if Open.exist?(files_dir)
+    Open.rm path if Open.exist_or_link?(path)
+    Open.rm tmp_path if Open.exist_or_link?(tmp_path)
+    Open.rm info_file if Open.exist_or_link?(info_file)
+    Open.rm_rf files_dir if Open.exist_or_link?(files_dir)
+    self
   end
 
   def self.clean(file)

data/lib/scout/workflow/step.rb

@@ -76,10 +76,14 @@ class Step
 
   def task_name
     @task_name ||= @task.name if @task.respond_to?(:name)
+    @task_name ||= info[:task_name] if Open.exist?(info_file)
+    @task_name ||= path.split("/")[-2]
   end
 
   def workflow
-    @task.workflow if @task
+    @workflow ||= @task.workflow if Task === @task
+    @workflow ||= info[:workflow] if Open.exist?(info_file)
+    @workflow ||= path.split("/")[-3]
   end
 
   def exec
@@ -124,62 +128,63 @@ class Step
     return @result || self.load if done?
     prepare_dependencies
     run_dependencies
-    @result = 
-      begin
-        Persist.persist(name, type, :path => path, :tee_copies => tee_copies) do
-          clear_info
-          input_names = (task.respond_to?(:inputs) && task.inputs) ? task.inputs.collect{|name,_| name} : []
-          merge_info :status => :start, :start => Time.now,
-            :pid => Process.pid, :pid_hostname => Misc.hostname, 
-            :inputs => MetaExtension.purge(inputs), :input_names => input_names, :type => type,
-            :dependencies => dependencies.collect{|d| d.path }
-
-          @result = exec
-
-          if @result.nil? && File.exist?(self.tmp_path) && ! File.exist?(self.path)
-            Open.mv self.tmp_path, self.path
-          else
-            @result = @result.stream if @result.respond_to?(:stream)
-          end
+    begin
+      @result = Persist.persist(name, type, :path => path, :tee_copies => tee_copies) do
+        clear_info
+        input_names = (task.respond_to?(:inputs) && task.inputs) ? task.inputs.collect{|name,_| name} : []
+        merge_info :status => :start, :start => Time.now,
+          :pid => Process.pid, :pid_hostname => Misc.hostname, 
+          :task_name => task_name, :workflow => workflow.to_s,
+          :inputs => MetaExtension.purge(inputs), :input_names => input_names, :type => type,
+          :dependencies => dependencies.collect{|d| d.path }
+
+        @exec_result = exec
+
+        if @exec_result.nil? && File.exist?(self.tmp_path) && ! File.exist?(self.path)
+          Open.mv self.tmp_path, self.path
+        else
+          @exec_result = @exec_result.stream if @exec_result.respond_to?(:stream)
+        end
 
-          @result
+        @exec_result
 
-          if (IO === @result || StringIO === @result) && (ENV["SCOUT_NO_STREAM"] == "true" || ! stream)
-            Open.sensible_write(self.path, @result)
-            @result = nil
-          else
-            @result
-          end
+        if (IO === @exec_result || StringIO === @exec_result) && (ENV["SCOUT_NO_STREAM"] == "true" || ! stream)
+          Open.sensible_write(self.path, @exec_result)
+          @exec_result = nil
+        else
+          @exec_result
         end
-      rescue Exception => e
-        merge_info :status => :error, :exception => e, :end => Time.now
-        abort_dependencies
-        raise e
-      ensure
-        if ! (error? || aborted?)
-          if streaming?
-            ConcurrentStream.setup(@result) do
-              merge_info :status => :done, :end => Time.now
-            end
+      end
+    rescue Exception => e
+      merge_info :status => :error, :exception => e, :end => Time.now
+      abort_dependencies
+      raise e
+    ensure
+      if ! (error? || aborted?)
+        if streaming?
+          ConcurrentStream.setup(@result) do
+            merge_info :status => :done, :end => Time.now
+          end
 
-            @result.abort_callback = proc do |exception|
-              if exception.nil? || Aborted === exception || Interrupt === exception
-                merge_info :status => :aborted, :end => Time.now
-              else
-                begin
-                  merge_info :status => :error, :exception => exception, :end => Time.now
-                rescue Exception
-                  Log.exception $!
-                end
+          @result.abort_callback = proc do |exception|
+            if exception.nil? || Aborted === exception || Interrupt === exception
+              merge_info :status => :aborted, :end => Time.now
+            else
+              begin
+                merge_info :status => :error, :exception => exception, :end => Time.now
+              rescue Exception
+                Log.exception $!
               end
             end
-
-            log :streaming
-          else
-            merge_info :status => :done, :end => Time.now
           end
+
+
+          log :streaming
+        else
+          merge_info :status => :done, :end => Time.now
         end
       end
+    end
   end
 
   def fork
@@ -232,11 +237,14 @@ class Step
     while @result && streaming? && stream = self.stream
       threads << Open.consume_stream(stream, true)
     end
+
+    threads.compact!
+
     threads.each do |t| 
       begin
         t.join 
       rescue Exception
-        threads.each{|t| t.raise(Aborted); t.join }
+        threads.compact.each{|t| t.raise(Aborted); t.join }
         raise $!
       end
     end
@@ -252,6 +260,7 @@ class Step
     while ! present?
       sleep 0.1
     end
+    self
   end
 
   def terminated?

data/lib/scout/workflow/task/inputs.rb

@@ -85,8 +85,12 @@ module Task
       input_file = File.join(directory, name.to_s)
 
       if Path.is_filename?(value) 
-        relative_file = save_file_input(value, directory)
-        Open.write(input_file + ".as_file", relative_file)
+        if type == :path
+          Open.write(input_file + ".as_path", value)
+        else
+          relative_file = save_file_input(value, directory)
+          Open.write(input_file + ".as_file", relative_file)
+        end
       elsif Step === value
         Open.write(input_file + ".as_step", value.short_path)
       elsif type == :file

data/lib/scout/workflow/usage.rb

@@ -52,8 +52,9 @@ module Task
         inputs = task.inputs.reject{|name, _| seen.include? name }
         inputs = task.inputs.reject{|name, _| options.include? name }
         next unless inputs.any?
-        task.inputs.select{|name, _| inputs.include? name }.each do |name,_,_,_,options|
-          selects << [i, options[:select_options]] if options[:select_options]
+        input_names = inputs.collect{|name,_| name }
+        task.inputs.select{|name,_| input_names.include? name }.each do |name,_,_,_,options|
+          selects << [name, options[:select_options]] if options && options[:select_options]
         end
 
         dep = workflow.nil? || dep_workflow.name != workflow.name ? ["#{dep_workflow.name}", task_name.to_s] *"#" : task_name.to_s
@@ -291,6 +292,7 @@ module Workflow
       prov_tree = prov_tree(dep_tree)
       if prov_tree && ! prov_tree.empty? && prov_tree.split("\n").length > 2
 
+        str.puts
         str.puts Log.color :magenta, "## DEPENDENCY GRAPH (abridged)"
         str.puts
         prov_tree.split("\n").each do |line|