miga-base 0.7.3.0 → 0.7.7.0

data/lib/miga/dataset/status.rb

@@ -4,7 +4,7 @@ module MiGA::Dataset::Status
   ##
   # Returns the status of the dataset. If the status is not yet defined,
   # it recalculates it and, if +save+ is true, saves it in metadata.
-  # Return values are:
+  # Return symbols are:
   # - +:inactive+ The dataset is currently inactive
   # - +:incomplete+ The dataset is not yet fully processed
   # - +:complete+ The dataset is fully processed
@@ -14,12 +14,13 @@ module MiGA::Dataset::Status
   end
 
   ##
-  # Identify the current status instead of relying on metadata, and save
-  # it if +save+ is true. Return codes are the same as +status+.
+  # Identify the current status and save it if +save+ and the status changed.
+  # Return symbols are the same as +status+.
   def recalculate_status(save = true)
+    old_status = metadata[:status]
     metadata[:status] =
-      !active? ? :inactive : done_preprocessing? ? :complete : :incomplete
-    self.save if save
+      !active? ? 'inactive' : done_preprocessing? ? 'complete' : 'incomplete'
+    self.save if save && (old_status.nil? || old_status != metadata[:status])
     metadata[:status].to_sym
   end
 end

data/lib/miga/json.rb

@@ -6,9 +6,7 @@ require 'json'
 ##
 # Taxonomic classifications in MiGA.
 class MiGA::Json < MiGA::MiGA
-
   class << self
-
     ##
     # Default parsing options. Supported +opts+ keys:
     # - +:contents+: If true, the input is assumed to be the contents to parse,
@@ -25,6 +23,7 @@ class MiGA::Json < MiGA::MiGA
       if opts[:additions] and opts[:symbolize]
         raise 'JSON additions are not supported with symbolized names'
       end
+
       opts
     end
 
@@ -37,12 +36,13 @@ class MiGA::Json < MiGA::MiGA
       opts = default_opts(opts)
       cont = opts[:contents] ? path : File.read(path)
       raise "Empty descriptor: #{opts[:contents] ? "''" : path}." if cont.empty?
+
       y = JSON.parse(cont,
-        symbolize_names: opts[:symbolize],
-        create_additions: opts[:additions])
+                     symbolize_names: opts[:symbolize],
+                     create_additions: opts[:additions])
       unless opts[:default].nil?
         opts[:default] = parse(opts[:default]) if opts[:default].is_a? String
-        y.each{ |k, v| opts[:default][k] = v }
+        y.each { |k, v| opts[:default][k] = v }
         y = opts[:default]
       end
       y
@@ -56,7 +56,5 @@ class MiGA::Json < MiGA::MiGA
       File.open(path, 'w') { |fh| fh.print y } unless path.nil?
       y
     end
-
   end
-
 end

data/lib/miga/lair.rb

@@ -70,6 +70,7 @@ class MiGA::Lair < MiGA::MiGA
   def daemon_loop
     check_directories
     return false if options[:dry]
+
     sleep(options[:latency])
     true
   end
@@ -102,10 +103,12 @@ class MiGA::Lair < MiGA::MiGA
   def each_project(dir = path)
     Dir.entries(dir).each do |f|
       next if %w[. ..].include?(f) # Ruby <= 2.3 doesn't have Dir.children
+
       f = File.join(dir, f)
       if MiGA::Project.exist? f
         project = MiGA::Project.load(f)
         raise "Cannot load project: #{f}" if project.nil?
+
         yield(project)
       elsif Dir.exist? f
         each_project(f) { |project| yield(project) }
@@ -127,6 +130,7 @@ class MiGA::Lair < MiGA::MiGA
     each_project do |project|
       d = MiGA::Daemon.new(project)
       next if d.active?
+
       l_alive = d.last_alive
       unless l_alive.nil?
         next if options[:trust_timestamp] && project.metadata.updated < l_alive

data/lib/miga/metadata.rb

@@ -16,6 +16,7 @@ class MiGA::Metadata < MiGA::MiGA
   # exists, or nil otherwise.
   def self.load(path)
     return nil unless Metadata.exist? path
+
     MiGA::Metadata.new(path)
   end
 
@@ -33,11 +34,11 @@ class MiGA::Metadata < MiGA::MiGA
     @path = File.absolute_path(path)
     unless File.exist? path
       @data = {}
-      defaults.each { |k,v| self[k] = v }
+      defaults.each { |k, v| self[k] = v }
       create
     end
   end
-  
+
   ##
   # Parsed data as a Hash
   def data
@@ -73,6 +74,7 @@ class MiGA::Metadata < MiGA::MiGA
     ofh.close
     raise "Lock-racing detected for #{path}" unless
       File.exist?("#{path}.tmp") and File.exist?(lock_file)
+
     File.rename("#{path}.tmp", path)
     File.unlink(lock_file)
   end
@@ -152,5 +154,4 @@ class MiGA::Metadata < MiGA::MiGA
   def to_json
     MiGA::Json.generate(data)
   end
-
 end

data/lib/miga/project.rb

@@ -7,19 +7,18 @@ require 'miga/project/dataset'
 require 'miga/project/hooks'
 
 ##
-# MiGA representation of a project.
+# MiGA representation of a project
 class MiGA::Project < MiGA::MiGA
-  
   include MiGA::Project::Result
   include MiGA::Project::Dataset
   include MiGA::Project::Hooks
 
   ##
-  # Absolute path to the project folder.
+  # Absolute path to the project folder
   attr_reader :path
 
   ##
-  # Information about the project as MiGA::Metadata.
+  # Information about the project as MiGA::Metadata
   attr_reader :metadata
 
   ##
@@ -29,7 +28,7 @@ class MiGA::Project < MiGA::MiGA
   ##
   # Create a new MiGA::Project at +path+, if it doesn't exist and +update+ is
   # false, or load an existing one.
-  def initialize(path, update=false)
+  def initialize(path, update = false)
     @datasets = {}
     @do_not_save = false
     @path = File.absolute_path(path)
@@ -40,17 +39,19 @@ class MiGA::Project < MiGA::MiGA
   end
 
   ##
-  # Create an empty project.
+  # Create an empty project
   def create
     unless MiGA::MiGA.initialized?
       raise 'Impossible to create project in uninitialized MiGA.'
     end
-    dirs = [path] + @@FOLDERS.map{|d| "#{path}/#{d}" } +
-      @@DATA_FOLDERS.map{ |d| "#{path}/data/#{d}"}
-    dirs.each{ |d| Dir.mkdir(d) unless Dir.exist? d }
+
+    dirs = [path] + @@FOLDERS.map { |d| "#{path}/#{d}" } +
+           @@DATA_FOLDERS.map { |d| "#{path}/data/#{d}" }
+    dirs.each { |d| Dir.mkdir(d) unless Dir.exist? d }
     @metadata = MiGA::Metadata.new(
       File.expand_path('miga.project.json', path),
-      {datasets: [], name: File.basename(path)})
+      { datasets: [], name: File.basename(path) }
+    )
     d_path = File.expand_path('daemon/daemon.json', path)
     File.open(d_path, 'w') { |fh| fh.puts '{}' } unless File.exist? d_path
     pull_hook :on_create
@@ -58,13 +59,13 @@ class MiGA::Project < MiGA::MiGA
   end
 
   ##
-  # Save any changes persistently. Do nothing if +do_not_save+ is true.
+  # Save any changes persistently. Do nothing if +do_not_save+ is true
   def save
     save! unless do_not_save
   end
 
   ##
-  # Save any changes persistently, regardless of +do_not_save+.
+  # Save any changes persistently, regardless of +do_not_save+
   def save!
     metadata.save
     pull_hook :on_save
@@ -72,29 +73,37 @@ class MiGA::Project < MiGA::MiGA
   end
 
   ##
-  # (Re-)load project data and metadata.
+  # (Re-)load project data and metadata
   def load
     @datasets = {}
     @dataset_names_hash = nil
     @metadata = MiGA::Metadata.load "#{path}/miga.project.json"
     raise "Couldn't find project metadata at #{path}" if metadata.nil?
+
     pull_hook :on_load
   end
 
   ##
-  # Name of the project.
-  def name ; metadata[:name] ; end
+  # Name of the project
+  def name
+    metadata[:name]
+  end
 
   ##
-  # Type of project.
-  def type ; metadata[:type] ; end
+  # Type of project
+  def type
+    metadata[:type]
+  end
 
   ##
   # Is this a clade project?
-  def is_clade? ; type==:clade ; end
+  def is_clade?
+    type == :clade
+  end
 
   ##
   # Is this a project for multi-organism datasets?
-  def is_multi? ; @@KNOWN_TYPES[type][:multi] ; end
-
+  def is_multi?
+    @@KNOWN_TYPES[type][:multi]
+  end
 end

data/lib/miga/project/base.rb

@@ -2,7 +2,6 @@
 # @license Artistic-2.0
 
 class MiGA::Project < MiGA::MiGA
-
   class << self
     ##
     # Does the project at +path+ exist?
@@ -15,26 +14,35 @@ class MiGA::Project < MiGA::MiGA
     # otherwise.
     def load(path)
       return nil unless exist? path
+
       new path
     end
 
-    def INCLADE_TASKS ; @@INCLADE_TASKS ; end
-    def DISTANCE_TASKS ; @@DISTANCE_TASKS ; end
-    def KNOWN_TYPES ; @@KNOWN_TYPES ; end
-    def RESULT_DIRS ; @@RESULT_DIRS ; end
+    def INCLADE_TASKS
+      @@INCLADE_TASKS
+    end
 
-  end
+    def DISTANCE_TASKS
+      @@DISTANCE_TASKS
+    end
+
+    def KNOWN_TYPES
+      @@KNOWN_TYPES
+    end
 
+    def RESULT_DIRS
+      @@RESULT_DIRS
+    end
+  end
 end
 
 module MiGA::Project::Base
-
   ##
-  # Top-level folders inside a project.
+  # Top-level folders inside a project
   @@FOLDERS = %w[data metadata daemon]
 
   ##
-  # Folders for results.
+  # Folders for results
   @@DATA_FOLDERS = %w[
     01.raw_reads 02.trimmed_reads 03.read_quality 04.trimmed_fasta
     05.assembly 06.cds
@@ -54,46 +62,53 @@ module MiGA::Project::Base
   ]
 
   ##
-  # Directories containing the results from project-wide tasks.
+  # Directories containing the results from project-wide tasks
   @@RESULT_DIRS = {
-    project_stats: "90.stats",
+    project_stats: '90.stats',
     # Distances
-    haai_distances: "09.distances/01.haai",
-    aai_distances: "09.distances/02.aai",
-    ani_distances: "09.distances/03.ani",
-    #ssu_distances: "09.distances/04.ssu",
+    haai_distances: '09.distances/01.haai',
+    aai_distances: '09.distances/02.aai',
+    ani_distances: '09.distances/03.ani',
+    # ssu_distances: '09.distances/04.ssu',
     # Clade identification
-    clade_finding: "10.clades/01.find",
+    clade_finding: '10.clades/01.find',
     # Clade analysis
-    subclades: "10.clades/02.ani",
-    ogs: "10.clades/03.ogs"
-    #ess_phylogeny: "10.clades/04.phylogeny/01.essential",
-    #core_phylogeny: "10.clades/04.phylogeny/02.core",
-    #clade_metadata: "10.clades/05.metadata"
+    subclades: '10.clades/02.ani',
+    ogs: '10.clades/03.ogs'
+    # ess_phylogeny: '10.clades/04.phylogeny/01.essential',
+    # core_phylogeny: '10.clades/04.phylogeny/02.core',
+    # clade_metadata: '10.clades/05.metadata'
   }
 
   ##
-  # Supported types of projects.
+  # Supported types of projects
   @@KNOWN_TYPES = {
     mixed: {
-      description: "Mixed collection of genomes, metagenomes, and viromes",
-      single: true, multi: true},
-    genomes: {description: "Collection of genomes",
-      single: true, multi: false},
-    clade: {description: "Collection of closely-related genomes (ANI >= 90%)",
-      single: true, multi: false},
-    metagenomes: {description: "Collection of metagenomes and/or viromes",
-      single: false, multi: true}
+      description: 'Mixed collection of genomes, metagenomes, and viromes',
+      single: true, multi: true
+    },
+    genomes: {
+      description: 'Collection of genomes',
+      single: true, multi: false
+    },
+    clade: {
+      description: 'Collection of closely-related genomes (ANI >= 90%)',
+      single: true, multi: false
+    },
+    metagenomes: {
+      description: 'Collection of metagenomes and/or viromes',
+      single: false, multi: true
+    }
   }
 
   ##
-  # Project-wide distance estimations.
-  @@DISTANCE_TASKS = [:project_stats,
-    :haai_distances, :aai_distances, :ani_distances, :clade_finding]
-  
+  # Project-wide distance estimations
+  @@DISTANCE_TASKS = [
+    :project_stats, :haai_distances, :aai_distances, :ani_distances,
+    :clade_finding
+  ]
+
   ##
-  # Project-wide tasks for :clade projects.
+  # Project-wide tasks for :clade projects
   @@INCLADE_TASKS = [:subclades, :ogs]
-  
 end
-

data/lib/miga/project/dataset.rb

@@ -4,11 +4,10 @@
 ##
 # Helper module including specific functions handle datasets.
 module MiGA::Project::Dataset
-
   ##
   # Returns Array of MiGA::Dataset.
   def datasets
-    metadata[:datasets].map{ |name| dataset(name) }
+    metadata[:datasets].map { |name| dataset(name) }
   end
 
   ##
@@ -21,7 +20,7 @@ module MiGA::Project::Dataset
   # Returns Hash of Strings => true. Similar to +dataset_names+ but as
   # Hash for efficiency.
   def dataset_names_hash
-    @dataset_names_hash ||= Hash[dataset_names.map{ |i| [i,true] }]
+    @dataset_names_hash ||= Hash[dataset_names.map { |i| [i, true] }]
   end
 
   ##
@@ -29,22 +28,19 @@ module MiGA::Project::Dataset
   def dataset(name)
     name = name.miga_name
     return nil unless MiGA::Dataset.exist?(self, name)
+
     @datasets ||= {}
     @datasets[name] ||= MiGA::Dataset.new(self, name)
     @datasets[name]
   end
 
   ##
-  # Iterate through datasets, with one or two variables passed to +blk+.
-  # If one, the dataset MiGA::Dataset object is passed. If two, the name and
-  # the dataset object are passed.
+  # Iterate through datasets (MiGA::Dataset)
   def each_dataset(&blk)
-    metadata[:datasets].each do |name|
-      if blk.arity == 1
-        blk.call(dataset(name))
-      else
-        blk.call(name, dataset(name))
-      end
+    if block_given?
+      metadata[:datasets].each { |name| blk.call(dataset(name)) }
+    else
+      to_enum(:each_dataset)
     end
   end
 
@@ -66,6 +62,7 @@ module MiGA::Project::Dataset
   def unlink_dataset(name)
     d = dataset(name)
     return nil if d.nil?
+
     self.metadata[:datasets].delete(name)
     save
     pull_hook(:on_unlink_dataset, name)
@@ -75,28 +72,40 @@ module MiGA::Project::Dataset
   ##
   # Import the dataset +ds+, a MiGA::Dataset, using +method+ which is any method
   # supported by File#generic_transfer.
-  def import_dataset(ds, method=:hardlink)
+  def import_dataset(ds, method = :hardlink)
     raise "Impossible to import dataset, it already exists: #{ds.name}." if
       MiGA::Dataset.exist?(self, ds.name)
+
     # Import dataset results
     ds.each_result do |task, result|
       # import result files
       result.each_file do |file|
-        File.generic_transfer("#{result.dir}/#{file}",
-          "#{path}/data/#{MiGA::Dataset.RESULT_DIRS[task]}/#{file}", method)
+        File.generic_transfer(
+          File.join(result.dir, file),
+          File.join(path, 'data', MiGA::Dataset.RESULT_DIRS[task], file),
+          method
+        )
       end
       # import result metadata
       %w(json start done).each do |suffix|
-        if File.exist? "#{result.dir}/#{ds.name}.#{suffix}"
-          File.generic_transfer("#{result.dir}/#{ds.name}.#{suffix}",
-            "#{path}/data/#{MiGA::Dataset.RESULT_DIRS[task]}/" +
-	                      "#{ds.name}.#{suffix}", method)
+        if File.exist? File.join(result.dir, "#{ds.name}.#{suffix}")
+          File.generic_transfer(
+            File.join(result.dir, "#{ds.name}.#{suffix}"),
+            File.join(
+              path, 'data', MiGA::Dataset.RESULT_DIRS[task],
+              "#{ds.name}.#{suffix}"
+            ),
+            method
+          )
         end
       end
     end
     # Import dataset metadata
-    File.generic_transfer("#{ds.project.path}/metadata/#{ds.name}.json",
-      "#{self.path}/metadata/#{ds.name}.json", method)
+    File.generic_transfer(
+      File.join(ds.project.path, 'metadata', "#{ds.name}.json"),
+      File.join(self.path, 'metadata', "#{ds.name}.json"),
+      method
+    )
     # Save dataset
     self.add_dataset(ds.name)
   end
@@ -108,11 +117,13 @@ module MiGA::Project::Dataset
     MiGA::Dataset.RESULT_DIRS.values.each do |dir|
       dir_p = "#{path}/data/#{dir}"
       next unless Dir.exist? dir_p
+
       Dir.entries(dir_p).each do |file|
         next unless
           file =~ %r{
             \.(fa(a|sta|stqc?)?|fna|solexaqa|gff[23]?|done|ess)(\.gz)?$
           }x
+
         m = /([^\.]+)/.match(file)
         datasets << m[1] unless m.nil? or m[1] == "miga-project"
       end
@@ -140,9 +151,7 @@ module MiGA::Project::Dataset
   # - 2: To do.
   def profile_datasets_advance
     advance = []
-    self.each_dataset_profile_advance do |ds_adv|
-      advance << ds_adv
-    end
+    each_dataset_profile_advance { |adv| advance << adv }
     advance
   end
 
@@ -152,6 +161,4 @@ module MiGA::Project::Dataset
   def each_dataset_profile_advance(&blk)
     each_dataset { |ds| blk.call(ds.profile_advance) }
   end
-
 end
-