major refactoring improving configuration system

it now has a default file with mostly all configuration,
and a user config where it choses

.gitignore

@@ -1,3 +1,6 @@
 output/
 config.yml
 *~
+Gemfile.lock
+vendor/bundle
+.bundle

Gemfile

@@ -0,0 +1,5 @@
+source 'http://rubygems.org'
+
+gem 'bio'
+gem 'deep_merge'
+gem 'configatron', :github => 'markbates/configatron'

blast.rb

@@ -4,26 +4,24 @@
 #
 require_relative 'blast_interface'
 require_relative 'reporting'
+require_relative 'config_blast'
 #
 #
 #
 class Blast
   #
   include BlastInterface
+  include ConfigBlast
   #
   needs_implementation :blast
   #
-  DEF_OUTPUT_DIR  = 'output'
-  DEF_OUTPUT_EXT  = '.out'
-  DEF_CONFIG_PATH = './config.yml'
-
-  attr_reader :logger, :out_dir
-  attr_writer :out_dir, :dbs, :folders
+  attr_reader :logger, :store
 
   #
   #
   # initialize class with all necessary data
-  def initialize(config_path = DEF_CONFIG_PATH)
+  def initialize(config_path = nil)
+    super(config_path)
     # create logger object
     @logger      = Logger.new(STDOUT)
     logger.level = Logger::INFO
@@ -36,13 +34,13 @@ def initialize(config_path = DEF_CONFIG_PATH)
   def blast_folders(folders      = nil,
                     query_parent = nil,
                     db_parent    = nil)
-    query_parent = @query_parent if query_parent.nil?
-    folders      = @folders if folders.nil?
+    query_parent  = @store.query.parent if query_parent.nil?
+    query_folders = @store.query.folders if folders.nil?
     # create new queue to add all operations
     call_queue = Queue.new
     list = []
     # run through each directory
-    folders.each do |query|
+    query_folders.each do |query|
       list = blast_folders_each(query, query_parent, db_parent, call_queue)
     end
 
@@ -73,12 +71,12 @@ def blast_folders_each(query, query_parent, db_parent, call_queue)
       #
       logger.debug "going to blast with query: '#{query_file}'"
       # run query against all databases
-      @dbs.each do |db|
+      @store.db.list.each do |db|
         logger.debug "using db: #{db}"
         new_item = {}
-        new_item[:qfile]    = query_file
-        new_item[:db]       = db
-        new_item[:out_file] = gen_filename(query, query_file, db)
+        new_item[:qfile]        = query_file
+        new_item[:db]           = db
+        new_item[:out_file]     = gen_filename(query, query_file, db)
         new_item[:query_parent] = '' # empty, because it will
         #                             already have the prefix
         new_item[:db_parent] = db_parent
@@ -89,24 +87,10 @@ def blast_folders_each(query, query_parent, db_parent, call_queue)
   end
 
   def cleanup
-    logger.info("removing #{@out_dir}")
-    FileUtils.remove_dir(@out_dir)
-  end
-
-  def db_parent=(new_db_parent)
-    @db_parent    = File.expand_path(new_db_parent)
+    logger.info("removing #{@store.output.dir}")
+    FileUtils.remove_dir(@store.output.dir)
   end
 
-  def query_parent=(new_query_parent)
-    @query_parent = File.expand_path(new_query_parent)
-  end
-
-  def reload_config(config_path = 'config.yml')
-    @config = YAML.load_file(config_path)
-    logger.debug(@config.inspect)
-    set_config
-    logger.debug('loaded config.yml file')
-  end
   #             _            _
   #            (_)          | |
   #  _ __  _ __ ___   ____ _| |_ ___
@@ -131,99 +115,17 @@ def get_nt_seq_from_blastdb(seq_id, db, qstart, qend)
     spliced = seq.splice("#{qstart}..#{qend}")
     spliced
   end
+
   #
   #
   # Generate filenames for each of the query's output
   def gen_filename(prefix, query, db)
     name = query.gsub(%r{[\S]+\/}, '').gsub(/[\.]query/, '').gsub(/[ ]/, '_')
     list = []
-    list << @task
+    list << @store.task
     list << prefix unless prefix.nil?
     list << name
     list << db
-    File.join(@out_dir, list.join('#') + @out_ext)
-  end
-
-  #
-  #
-  # Set config variables
-  def set_config
-    # parent directories for query and blast db
-    self.query_parent = get_config(@config['query_parent'], Dir.pwd)
-    #
-    self.db_parent    = get_config(@config['db_parent'], Dir.pwd)
-    # optional arguments
-    self.dbs     = @config['dbs']
-    self.folders = @config['query_folders']
-    @opts        = @config['opts']
-    @task        = @config['task']
-    @outfmt      = @config['format']['outfmt']
-    #
-    @identity_threshold = @config['identity_threshold']
-    @identity_threshold *= 100
-    #
-    # orf options
-    @orf = {}
-    @orf[:stop]    = @config['orf']['stop_codon']
-    @orf[:start]   = @config['orf']['start_codon']
-    @orf[:reverse] = @config['orf']['reverse']
-    @orf[:direct]  = @config['orf']['direct']
-    @orf[:min]     = @config['orf']['min']
-    #
-    @verbose_out = !get_config(@config['clean_output'], false)
-    #
-    @out_dir = get_config(@config['output']['dir'],    DEF_OUTPUT_DIR)
-    @out_ext = get_config(@config['output']['ext'],    DEF_OUTPUT_EXT)
-
-    @out_dir = File.expand_path(@out_dir)
-    create_out_dir
-    #
-    #
-    logger.debug('query_parent: ' + @query_parent)
-    logger.debug('db_parent: ' + @db_parent)
-    #
-    fail 'Databases must be defined in config.yml.' if @dbs.nil?
-    fail 'Folders must be defined in config.yml.'   if @folders.nil?
-    # set existing dbs
-    logger.info("loads databases (from directory '#{@query_parent}'): " +
-      @dbs.join(', '))
-
-    # outfmt specifiers for the blast query (we choose all)
-    @outfmt_spec    = @config['format']['specifiers'].keys
-    # outfmt specifiers details to add to the report's second line
-    @outfmt_details = @config['format']['specifiers'].values
-  end
-
-  def create_out_dir
-    # create output dir if does not exist
-    begin
-      Dir.mkdir @out_dir unless Dir.exist?(@out_dir)
-    rescue
-      logger.error(msg = 'Could not create output directory')
-      raise msg
-    end
-    # create output dir with timestamp
-    begin
-      if @config['force_folder'].nil?
-        @out_dir = @out_dir +
-                   File::Separator +
-                   Time.now.strftime('%Y_%m_%d-%H_%M_%S') +
-                   '-' + srand.to_s[3..6]
-        Dir.mkdir @out_dir
-      else
-        @out_dir = @out_dir + File::Separator + @config['force_folder']
-        Dir.mkdir(@out_dir) unless Dir.exist?(@out_dir)
-      end
-    rescue
-      logger.error(msg = 'Could not create output directory')
-      raise msg
-    end
-  end
-
-  #
-  #
-  # get default value
-  def get_config(yml_var, default)
-    yml_var.nil? ? default : yml_var
+    File.join(@store.output.dir, list.join('#') + @store.output.extension)
   end
 end # end of class

blastn.rb

@@ -2,32 +2,20 @@
 #
 #
 class Blastn < Blast
-  #
-  DEF_OPTIONS = '-dust no -max_target_seqs 500 -evalue 1E-100'
-  DEF_FORMAT  = '6'
-  DEF_TASK    = 'blastn'
-
-  def initialize(*args)
-    super(*args)
-    @opts    = get_config(@config['opts'], DEF_OPTIONS)
-    @task    = get_config(@config['task'], DEF_TASK)
-    @outfmt  = get_config(@config['format']['outfmt'], DEF_FORMAT)
-  end
-
   #
   # blastn blast
   def blast(qfile, db, out_file, query_parent = nil, db_parent = nil)
-    query_parent = @query_parent if query_parent.nil?
-    db_parent    = @db_parent if db_parent.nil?
+    query_parent = @store.query.parent if query_parent.nil?
+    db_parent    = @store.db.parent if db_parent.nil?
     # create command for this call
-    cmd = "blastn -query \"#{File.join(query_parent, qfile)}\" -db \"#{db}\""
-    cmd += " #{@opts} -out #{out_file}"
-    cmd += " -outfmt \"#{@outfmt} #{@outfmt_spec.join(' ')}\""
-    logger.info "running '#{qfile}' with database '#{db}' that will \
-      store in '#{out_file}'"
+    cmd = "blastn -query \"#{File.join(query_parent, qfile)}\" -db \"#{db}\"" \
+          " #{@store.opts} -out #{out_file}" \
+          " -outfmt \"#{@store.format.outfmt}" \
+          " #{@store.format.specifiers.keys.join(' ')}\""
+    logger.info "running '#{qfile}' with database '#{db}' that will " \
+      "store in '#{out_file}'"
     logger.debug cmd
     output = `BLASTDB="#{db_parent}" #{cmd}` # actual call to blast
     logger.debug output
   end
-
 end

config_blast.rb

@@ -0,0 +1,78 @@
+require 'configatron/core'
+
+require 'yaml'
+
+#
+#
+#
+module ConfigBlast
+  #
+  def initialize(*args)
+    # setup config defaults
+    @store = Configatron::RootStore.new
+    @store.config.default = File.expand_path('default.yml')
+    @store.config.user    = File.expand_path('config.yml')
+  end
+
+  def reload_config(config_path = nil)
+    config_path = @store.config.user if config_path.nil?
+    # get configuration from default yml file
+    logger.info("loads configuration from defaults: #{@store.config.default}")
+    @store.configure_from_hash(YAML.load_file(@store.config.default))
+    logger.info("loads configuration from user: #{config_path}")
+    @store.configure_from_hash(YAML.load_file(config_path))
+    # process the configuration to adjust paths and values
+    process_config
+    logger.debug('loaded and processed configuration files')
+  end
+
+  private
+
+  # create output dir if does not exist
+  def create_output_dir
+    begin
+      Dir.mkdir @store.output.dir unless Dir.exist?(@store.output.dir)
+    rescue
+      logger.error(msg = 'Could not create output directory')
+      raise msg
+    end
+    # create output dir with timestamp
+    begin
+      if !@store.key?(:force_folder)
+        @store.output.dir += File::Separator +
+                             Time.now.strftime('%Y_%m_%d-%H_%M_%S') +
+                             '-' + srand.to_s[3..6]
+        Dir.mkdir @store.output.dir
+      else
+        @store.output.dir += File::Separator + @store.force_folder
+        Dir.mkdir(@store.output.dir) unless Dir.exist?(@store.output.dir)
+      end
+    rescue StandardError => e
+      logger.error msg = "Could not create output directory (why: #{e.message})"
+      raise msg
+    end
+  end
+
+  #
+  #
+  # Set config variables
+  def process_config
+    # optional arguments
+    @store.identity_threshold *= 100
+    # convert paths to an absolutes
+    @store.output.dir   = File.expand_path(@store.output.dir)
+    @store.db.parent    = File.expand_path(@store.db.parent)
+    @store.query.parent = File.expand_path(@store.query.parent)
+    # create the output directory
+    create_output_dir
+    #
+    logger.debug('query_parent: ' + @store.query.parent)
+    logger.debug('db_parent: ' + @store.db.parent)
+    #
+    fail 'Databases must be defined in config.yml.' if @store.db.list.nil?
+    fail 'Folders must be defined in config.yml.'   if @store.query.folders.nil?
+    # set existing dbs
+    logger.info("loads databases (from directory '#{@store.db.parent}'): " +
+      @store.db.list.join(', '))
+  end
+end