scout-gear 7.1.0 → 7.3.0

data/lib/scout/workflow/step.rb

@@ -2,15 +2,26 @@ require_relative '../path'
 require_relative '../persist'
 require_relative 'step/info'
 require_relative 'step/load'
+require_relative 'step/file'
+require_relative 'step/dependencies'
+require_relative 'step/provenance'
+require_relative 'step/config'
+require_relative 'step/progress'
 
 class Step 
 
-  attr_accessor :path, :inputs, :dependencies, :task
+  attr_accessor :path, :inputs, :dependencies, :task, :tee_copies
   def initialize(path, inputs = nil, dependencies = nil, &task) 
     @path = path
     @inputs = inputs
     @dependencies = dependencies
     @task = task
+    @mutex = Mutex.new
+    @tee_copies = 1
+  end
+
+  def synchronize(&block)
+    @mutex.synchronize(&block)
   end
 
   def inputs
@@ -48,15 +59,20 @@ class Step
     @task_name ||= @task.name if @task.respond_to?(:name)
   end
 
+  def workflow
+    @task.workflow if @task
+  end
+
   def exec
-    self.instance_exec(*inputs, &task)
+    @result = self.instance_exec(*inputs, &task)
   end
 
   attr_reader :result
   def run
     return @result || self.load if done?
-    dependencies.each{|dep| dep.run }
-    @result = Persist.persist(name, type, :path => path) do
+    prepare_dependencies
+    run_dependencies
+    @result = Persist.persist(name, type, :path => path, :tee_copies => tee_copies) do
       begin
         merge_info :status => :start, :start => Time.now,
           :pid => Process.pid, :pid_hostname => ENV["HOSTNAME"], 
@@ -64,17 +80,34 @@ class Step
           :dependencies => dependencies.collect{|d| d.path }
 
         @result = exec
+        @result = @result.respond_to?(:stream) ? @result.stream : @result
+        @result
+      rescue Exception => e
+        merge_info :status => :error, :exception => e
+        raise e
       ensure
-        if streaming?
-          ConcurrentStream.setup(@result) do
+        if ! (error? || aborted?)
+          if streaming?
+            ConcurrentStream.setup(@result) do
+              merge_info :status => :done, :end => Time.now
+            end
+
+            @result.abort_callback = proc do |exception|
+              if Aborted === exception || Interrupt === exception
+                merge_info :status => :aborted, :end => Time.now
+              else
+                merge_info :status => :error, :exception => exception, :end => Time.now
+              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
+    @result
   end
 
   def done?
@@ -82,19 +115,35 @@ class Step
   end
 
   def streaming?
-    IO === @result || StringIO === @result
+    @take_stream || IO === @result || StringIO === @result 
   end
 
   def stream
-    join
-    streaming? ? @result : Open.open(path)
+    synchronize do
+      if streaming? && ! @result.nil?
+        if @result.next
+          Log.debug "Taking result #{Log.fingerprint @result} next #{Log.fingerprint @result.next}"
+        else
+          Log.debug "Taking result #{Log.fingerprint @result}"
+        end
+        @take_stream, @result = @result, @result.next
+        @take_stream
+      elsif done?
+        Open.open(self.path)
+      else
+        if running?
+          nil
+        else
+          exec
+        end
+      end
+    end
   end
 
   def join
-    if streaming?
-      Open.consume_stream(@result, false) 
-      @result = nil
-    end
+    io = self.stream if streaming?
+    Open.consume_stream(io, false) if io
+    self
   end
 
   def produce
@@ -109,8 +158,13 @@ class Step
   end
 
   def clean
+    @take_stream = nil 
+    @result = nil
+    @info = nil
+    @info_load_time = nil
     Open.rm path if Open.exist?(path)
     Open.rm info_file if Open.exist?(info_file)
+    Open.rm_rf files_dir if Open.exist?(files_dir)
   end
 
   def recursive_clean
@@ -128,6 +182,6 @@ class Step
   end
 
   def digest_str
-    path
+    path.dup
   end
 end

data/lib/scout/workflow/task.rb

@@ -1,10 +1,11 @@
 require_relative '../meta_extension'
+require_relative '../named_array'
 require_relative 'step'
 require_relative 'task/inputs'
 
 module Task
   extend MetaExtension
-  extension_attr :name, :type, :inputs, :deps, :directory, :description
+  extension_attr :name, :type, :inputs, :deps, :directory, :description, :returns, :extension, :workflow
 
   DEFAULT_NAME = "Default"
 
@@ -24,7 +25,7 @@ module Task
     return inputs if deps.nil?
     deps.inject(inputs) do |acc,dep|
       workflow, task = dep
-      next if workflow.nil?
+      next acc if workflow.nil? || task.nil?
       acc += workflow.tasks[task].recursive_inputs
     end
   end
@@ -54,7 +55,7 @@ module Task
       resolved_inputs = {}
       inputs.each do |k,v|
         if Symbol === v
-          input_dep = dependencies.select{|d| d.task_name == v}.first
+          input_dep = dependencies.select{|d| d.task_name == v }.first
           resolved_inputs[k] = input_dep || inputs[v] || k
         else
           resolved_inputs[k] = v
@@ -144,12 +145,16 @@ module Task
 
     if non_default_inputs.any?
       hash = Misc.digest(:inputs => input_hash, :dependencies => dependencies)
-      Log.debug "Hash #{name} - #{hash}: #{Misc.digest_str(:inputs => inputs, :non_default_inputs => non_default_inputs, :dependencies => dependencies)}"
+      Log.debug "Hash #{name} - #{hash}: #{Log.fingerprint(:inputs => inputs, :non_default_inputs => non_default_inputs, :dependencies => dependencies)}"
       id = [id, hash] * "_"
     end
 
     path = directory[id]
 
-    Step.new path.find, inputs, dependencies, &self
+    Persist.memory(path) do 
+      Log.debug "Creating job #{path} #{Log.fingerprint inputs} #{Log.fingerprint dependencies}"
+      NamedArray.setup(inputs, @inputs.collect{|i| i[0] }) if @inputs
+      Step.new path.find, inputs, dependencies, &self
+    end
   end
 end

data/lib/scout/workflow/usage.rb

@@ -20,6 +20,7 @@ module Task
       seen = inputs.collect{|name,_| name }
       dep_inputs = {}
       deps.each do |dep_workflow,task_name|
+        next if task_name.nil?
         task = dep_workflow.tasks[task_name]
         next if task.inputs.nil?
         inputs = task.inputs.reject{|name, _| seen.include? name }
@@ -34,7 +35,6 @@ module Task
         str.puts Log.color :yellow, dep + ":"
         str.puts
         str.puts SOPT.input_array_doc(inputs)
-        str.puts
       end
 
       #task_inputs = dep_inputs deps, workflow
@@ -72,6 +72,7 @@ module Task
       str.puts
     end
 
+    str.puts
     str.puts Log.color(:magenta, "Returns: ") << Log.color(:blue, type.to_s) << "\n"
     str.puts
 
@@ -122,6 +123,7 @@ module Workflow
       next if seen.include? dep
       seen << dep
       workflow, task, *rest = dep
+      next if task.nil?
 
       key = [workflow, task]
 

data/lib/scout/workflow.rb

@@ -24,17 +24,25 @@ module Workflow
     base.libdir = Path.setup(libdir).tap{|p| p.resource = base}
   end
 
-  def self.require_workflow(workflow)
-    workflow = Path.setup('workflows')[workflow]["workflow.rb"] unless Open.exists?(workflow)
+  def self.require_workflow(workflow_name)
+    workflow = workflow_name 
+    workflow = Path.setup('workflows')[workflow_name]["workflow.rb"] unless Open.exists?(workflow)
+    workflow = Path.setup('workflows')[Misc.snake_case(workflow_name)]["workflow.rb"] unless Open.exists?(workflow)
+    workflow = Path.setup('workflows')[Misc.camel_case(workflow_name)]["workflow.rb"] unless Open.exists?(workflow)
     if Open.exists?(workflow)
       workflow = workflow.find if Path === workflow
+      $LOAD_PATH.unshift(File.join(File.dirname(workflow), 'lib'))
       load workflow
+    else
+      raise "Workflow #{workflow_name} not found"
     end
     workflows.last
   end
 
   def job(name, *args)
     task = tasks[name]
-    task.job(*args)
+    step = task.job(*args)
+    step.extend step_module
+    step
   end
 end

data/lib/scout-gear.rb

@@ -7,3 +7,4 @@ require_relative 'scout/simple_opt'
 require_relative 'scout/resource'
 require_relative 'scout/resource/scout'
 require_relative 'scout/persist'
+require_relative 'scout/tsv'

data/lib/scout.rb

@@ -1 +1,2 @@
 require 'workflow-scout'
+require 'rbbt-scout'

data/scout-gear.gemspec

@@ -2,16 +2,16 @@
 # DO NOT EDIT THIS FILE DIRECTLY
 # Instead, edit Juwelier::Tasks in Rakefile, and run 'rake gemspec'
 # -*- encoding: utf-8 -*-
-# stub: scout-gear 7.1.0 ruby lib
+# stub: scout-gear 7.3.0 ruby lib
 
 Gem::Specification.new do |s|
   s.name = "scout-gear".freeze
-  s.version = "7.1.0"
+  s.version = "7.3.0"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0".freeze) if s.respond_to? :required_rubygems_version=
   s.require_paths = ["lib".freeze]
   s.authors = ["Miguel Vazquez".freeze]
-  s.date = "2023-05-01"
+  s.date = "2023-05-12"
   s.description = "Temporary files, logs, etc.".freeze
   s.email = "mikisvaz@gmail.com".freeze
   s.executables = ["scout".freeze]
@@ -28,10 +28,12 @@ Gem::Specification.new do |s|
     "Rakefile",
     "VERSION",
     "bin/scout",
+    "lib/rbbt-scout.rb",
     "lib/scout-gear.rb",
     "lib/scout.rb",
     "lib/scout/cmd.rb",
     "lib/scout/concurrent_stream.rb",
+    "lib/scout/config.rb",
     "lib/scout/exceptions.rb",
     "lib/scout/indiferent_hash.rb",
     "lib/scout/indiferent_hash/case_insensitive.rb",
@@ -48,8 +50,11 @@ Gem::Specification.new do |s|
     "lib/scout/misc/digest.rb",
     "lib/scout/misc/filesystem.rb",
     "lib/scout/misc/format.rb",
+    "lib/scout/misc/helper.rb",
     "lib/scout/misc/insist.rb",
     "lib/scout/misc/monitor.rb",
+    "lib/scout/misc/system.rb",
+    "lib/scout/named_array.rb",
     "lib/scout/open.rb",
     "lib/scout/open/lock.rb",
     "lib/scout/open/remote.rb",
@@ -68,6 +73,7 @@ Gem::Specification.new do |s|
     "lib/scout/resource/produce.rb",
     "lib/scout/resource/produce/rake.rb",
     "lib/scout/resource/scout.rb",
+    "lib/scout/resource/software.rb",
     "lib/scout/resource/util.rb",
     "lib/scout/semaphore.rb",
     "lib/scout/simple_opt.rb",
@@ -78,7 +84,21 @@ Gem::Specification.new do |s|
     "lib/scout/simple_opt/setup.rb",
     "lib/scout/tmpfile.rb",
     "lib/scout/tsv.rb",
+    "lib/scout/tsv/dumper.rb",
+    "lib/scout/tsv/index.rb",
+    "lib/scout/tsv/open.rb",
     "lib/scout/tsv/parser.rb",
+    "lib/scout/tsv/path.rb",
+    "lib/scout/tsv/persist.rb",
+    "lib/scout/tsv/persist/adapter.rb",
+    "lib/scout/tsv/persist/fix_width_table.rb",
+    "lib/scout/tsv/persist/serialize.rb",
+    "lib/scout/tsv/persist/tokyocabinet.rb",
+    "lib/scout/tsv/traverse.rb",
+    "lib/scout/tsv/util.rb",
+    "lib/scout/tsv/util/filter.rb",
+    "lib/scout/tsv/util/process.rb",
+    "lib/scout/tsv/util/select.rb",
     "lib/scout/work_queue.rb",
     "lib/scout/work_queue/socket.rb",
     "lib/scout/work_queue/worker.rb",
@@ -86,8 +106,13 @@ Gem::Specification.new do |s|
     "lib/scout/workflow/definition.rb",
     "lib/scout/workflow/documentation.rb",
     "lib/scout/workflow/step.rb",
+    "lib/scout/workflow/step/config.rb",
+    "lib/scout/workflow/step/dependencies.rb",
+    "lib/scout/workflow/step/file.rb",
     "lib/scout/workflow/step/info.rb",
     "lib/scout/workflow/step/load.rb",
+    "lib/scout/workflow/step/progress.rb",
+    "lib/scout/workflow/step/provenance.rb",
     "lib/scout/workflow/task.rb",
     "lib/scout/workflow/task/inputs.rb",
     "lib/scout/workflow/usage.rb",
@@ -104,6 +129,7 @@ Gem::Specification.new do |s|
     "scout_commands/workflow/task_old",
     "share/color/color_names",
     "share/color/diverging_colors.hex",
+    "share/software/install_helpers",
     "test/scout/indiferent_hash/test_case_insensitive.rb",
     "test/scout/indiferent_hash/test_options.rb",
     "test/scout/log/test_color.rb",
@@ -111,6 +137,7 @@ Gem::Specification.new do |s|
     "test/scout/misc/test_digest.rb",
     "test/scout/misc/test_filesystem.rb",
     "test/scout/misc/test_insist.rb",
+    "test/scout/misc/test_system.rb",
     "test/scout/open/test_lock.rb",
     "test/scout/open/test_remote.rb",
     "test/scout/open/test_stream.rb",
@@ -122,6 +149,7 @@ Gem::Specification.new do |s|
     "test/scout/persist/test_serialize.rb",
     "test/scout/resource/test_path.rb",
     "test/scout/resource/test_produce.rb",
+    "test/scout/resource/test_software.rb",
     "test/scout/resource/test_util.rb",
     "test/scout/simple_opt/test_doc.rb",
     "test/scout/simple_opt/test_get.rb",
@@ -129,10 +157,12 @@ Gem::Specification.new do |s|
     "test/scout/simple_opt/test_setup.rb",
     "test/scout/test_cmd.rb",
     "test/scout/test_concurrent_stream.rb",
+    "test/scout/test_config.rb",
     "test/scout/test_indiferent_hash.rb",
     "test/scout/test_log.rb",
     "test/scout/test_meta_extension.rb",
     "test/scout/test_misc.rb",
+    "test/scout/test_named_array.rb",
     "test/scout/test_open.rb",
     "test/scout/test_path.rb",
     "test/scout/test_persist.rb",
@@ -142,11 +172,25 @@ Gem::Specification.new do |s|
     "test/scout/test_tsv.rb",
     "test/scout/test_work_queue.rb",
     "test/scout/test_workflow.rb",
+    "test/scout/tsv/persist/test_adapter.rb",
+    "test/scout/tsv/persist/test_fix_width_table.rb",
+    "test/scout/tsv/persist/test_tokyocabinet.rb",
+    "test/scout/tsv/test_dumper.rb",
+    "test/scout/tsv/test_index.rb",
+    "test/scout/tsv/test_open.rb",
     "test/scout/tsv/test_parser.rb",
+    "test/scout/tsv/test_persist.rb",
+    "test/scout/tsv/test_traverse.rb",
+    "test/scout/tsv/test_util.rb",
+    "test/scout/tsv/util/test_filter.rb",
+    "test/scout/tsv/util/test_process.rb",
+    "test/scout/tsv/util/test_select.rb",
     "test/scout/work_queue/test_socket.rb",
     "test/scout/work_queue/test_worker.rb",
+    "test/scout/workflow/step/test_dependencies.rb",
     "test/scout/workflow/step/test_info.rb",
     "test/scout/workflow/step/test_load.rb",
+    "test/scout/workflow/step/test_provenance.rb",
     "test/scout/workflow/task/test_inputs.rb",
     "test/scout/workflow/test_definition.rb",
     "test/scout/workflow/test_documentation.rb",
@@ -159,15 +203,25 @@ Gem::Specification.new do |s|
   ]
   s.homepage = "http://github.com/mikisvaz/scout-gear".freeze
   s.licenses = ["MIT".freeze]
-  s.rubygems_version = "3.4.11".freeze
+  s.rubygems_version = "3.2.15".freeze
   s.summary = "basic gear for scouts".freeze
 
-  s.specification_version = 4
+  if s.respond_to? :specification_version then
+    s.specification_version = 4
+  end
 
-  s.add_runtime_dependency(%q<term-ansicolor>.freeze, [">= 0"])
-  s.add_development_dependency(%q<rdoc>.freeze, ["~> 3.12"])
-  s.add_development_dependency(%q<bundler>.freeze, ["~> 1.0"])
-  s.add_development_dependency(%q<juwelier>.freeze, ["~> 2.1.0"])
-  s.add_development_dependency(%q<simplecov>.freeze, [">= 0"])
+  if s.respond_to? :add_runtime_dependency then
+    s.add_runtime_dependency(%q<term-ansicolor>.freeze, [">= 0"])
+    s.add_development_dependency(%q<rdoc>.freeze, ["~> 3.12"])
+    s.add_development_dependency(%q<bundler>.freeze, ["~> 1.0"])
+    s.add_development_dependency(%q<juwelier>.freeze, ["~> 2.1.0"])
+    s.add_development_dependency(%q<simplecov>.freeze, [">= 0"])
+  else
+    s.add_dependency(%q<term-ansicolor>.freeze, [">= 0"])
+    s.add_dependency(%q<rdoc>.freeze, ["~> 3.12"])
+    s.add_dependency(%q<bundler>.freeze, ["~> 1.0"])
+    s.add_dependency(%q<juwelier>.freeze, ["~> 2.1.0"])
+    s.add_dependency(%q<simplecov>.freeze, [">= 0"])
+  end
 end
 

data/scout_commands/find

@@ -69,7 +69,7 @@ end if resource
 
 path = (resource || Scout)[path.dup]
 
-if where.nil? || where == 'all' || path.search_paths.include?(where.to_sym)
+if where.nil? || where == 'all' || path.path_maps.include?(where.to_sym)
   location = path.find(where)
 
   if Array === location

data/scout_commands/workflow/task

@@ -13,6 +13,7 @@ $ #{$0} [<options>] <workflow> <task>
 -h--help Print this help
 -jn--job_name* Name to use as job identifier
 -pf--print_filepath Print the file path
+-prov--provenance Print the step provenance
 -cl--clean Clean the last step
 -rcl--recursive_clean Clean all steps
 EOF
@@ -22,6 +23,7 @@ workflow_name, task_name = ARGV
 raise MissingParameterException.new :workflow if workflow_name.nil?
 
 workflow = Workflow.require_workflow workflow_name
+task_name = task_name.to_sym if task_name
 task = workflow.tasks[task_name.to_sym] if task_name
 
 options[:help] = true if task.nil?
@@ -38,20 +40,25 @@ if options[:help]
 end
 
 job_options = task.get_SOPT(task)
-job = task.job(options[:job_name], job_options)
+job = workflow.job(task_name, options[:job_name], job_options)
 
 job.recursive_clean if options[:recursive_clean]
 job.clean if options[:clean]
 
-job.run
-
-if options[:print_filepath]
-  path = job.path
-  path = path.find if Path === path
-  puts path
+if options[:provenance]
+  puts Step.prov_report(job)
 else
-  if ! Open.consume_stream(job.stream, false, STDOUT, false).end_with? "\n"
-    puts
+  job.run unless job.done?
+
+  if options[:print_filepath]
+    job.join
+    path = job.path
+    path = path.find if Path === path
+    puts path
+  else
+    if ! ((c = Open.consume_stream(job.stream, false, STDOUT, false)) && c.end_with?("\n"))
+      puts
+    end
   end
 end
 

data/scout_commands/workflow/task_old

@@ -656,12 +656,12 @@ when (defined?(RemoteStep) and RemoteStep)
   end
 when Step
   if res.streaming?
-    io = TSV.get_stream res
+    io = TSV.stream res
     Misc.consume_stream(io, false, out)
     io.join if io.respond_to? :join
   elsif IO === res.result
     begin
-      io = res.get_stream
+      io = res.stream
       Misc.consume_stream(io, false, out)
       io.join if io.respond_to? :join
     rescue Aborted, Interrupt