miga-base 1.1.3.6 → 1.2.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 529121bd6f55dfeb21559b77af71a4537db91cda1fe2fdfc9bdee9adfcaaa962
-  data.tar.gz: 8b892eb88019c4c5d0ec654c722142004867665892a38cf96233afc0b1f30aef
+  metadata.gz: afb7c816474f73c547c7f54982293fe72c24d7829184801ab51fc848adf3d9e3
+  data.tar.gz: 06d51ebc47e96c28072c8415321014535ea7576d9e6f1c381e3ca62dc880df15
 SHA512:
-  metadata.gz: ab6397d8b74d163cac4afa9a8e79776b91c1c4175da947d669bcc9ae441d04a0345f9c31feffbbb6349e5d9375ccd61b6396740763f3d96d0606a9f2312d9ad0
-  data.tar.gz: db534b553aca1ecbcee679edaf33b7cdea972e648b507b2a2e7841f8865f7572e3d498714672b8f49b21babda9f5db311bb580a4e8f90e70470a6fff13f51e6e
+  metadata.gz: a4f5279fdba54ad1f8bcbce86cb84a7643b473b7742686e8e9992f8c4af7c982d5776a5a06775b10df67b61de90c4a32f915b51a17a9b11a223474574e7f8159
+  data.tar.gz: bcd547a1dc5337fbf036ba8e2df6de6db4c4fb34979afeb30ce6c1a692e76166b5412457fb3cc778e03cb601a192af0c8d33709b2e5d7233508dfdbd1c5156ed

data/lib/miga/cli/action/browse.rb

@@ -179,7 +179,7 @@ class MiGA::Cli::Action::Browse < MiGA::Cli::Action
     str
       .to_s.unmiga_name
       .sub(/^./, &:upcase)
-      .gsub(/(Aai|Ani|Ogs|Cds|Ssu| db$| ssu )/, &:upcase)
+      .gsub(/(Aai|Ani|Ogs|Cds|Ssu|Rds|ani95|aai90| db$| ssu )/, &:upcase)
       .sub(/Haai/, 'hAAI')
       .sub(/Mytaxa/, 'MyTaxa')
       .sub(/ pvalue$/, ' p-value')

data/lib/miga/cli/action/daemon.rb

@@ -38,9 +38,13 @@ class MiGA::Cli::Action::Daemon < MiGA::Cli::Action
         'Path to the list of execution hostnames'
       ) { |v| cli[:nodelist] = v }
       opt.on(
-        '--ppn INT',
+        '--ppn INT', Integer,
         'Maximum number of cores to use in a single job'
-      ) { |v| cli[:ppn] = v.to_i }
+      ) { |v| cli[:ppn] = v }
+      opt.on(
+        '--ppn-project INT', Integer,
+        'Maximum number of cores to use in project-wide tasks'
+      ) { |v| cli[:ppn_project] = v }
       opt.on(
         '--json PATH',
         'Path to a custom daemon definition in json format'
@@ -81,7 +85,7 @@ class MiGA::Cli::Action::Daemon < MiGA::Cli::Action
     # Configure and run daemon
     p = cli.load_project
     d = MiGA::Daemon.new(p, cli[:json])
-    dopts = %i[latency maxjobs nodelist ppn shutdown_when_done]
+    dopts = %i[latency maxjobs nodelist ppn ppn_project shutdown_when_done]
     dopts.each { |k| d.runopts(k, cli[k]) }
     d.show_log! if cli[:show_log]
     d.daemon(cli.operation, cli[:daemon_opts])

data/lib/miga/cli/action/derep_wf.rb

@@ -73,14 +73,14 @@ class MiGA::Cli::Action::DerepWf < MiGA::Cli::Action
     c_f = r.file_path(:clades_gsp) or raise 'Result incomplete: run failed'
     clades = File.readlines(c_f).map { |i| i.chomp.split("\t") }
     rep = representatives(p)
-    File.open(File.expand_path('genomospecies.tsv', cli[:outdir]), 'w') do |fh|
+    File.open(File.join(cli[:outdir], 'genomospecies.tsv'), 'w') do |fh|
       fh.puts "Clade\tRepresentative\tMembers"
       clades.each_with_index do |i, k|
         fh.puts ["gsp_#{k + 1}", rep[k], i.join(',')].join("\t")
       end
     end
     if cli[:collection]
-      dir = File.expand_path('representatives', cli[:outdir])
+      dir = File.join(cli[:outdir], 'representatives')
       FileUtils.mkdir_p(dir)
       rep.each do |i|
         f = p.dataset(i).result(:assembly).file_path(:largecontigs)
@@ -91,12 +91,12 @@ class MiGA::Cli::Action::DerepWf < MiGA::Cli::Action
 
   def representatives(p)
     cli.say 'Identifying representatives'
-    f = File.expand_path('representatives.txt', cli[:outdir])
+    f = File.join(cli[:outdir], 'representatives.txt')
     if cli[:criterion] == :medoids
       FileUtils.cp(p.result(:clade_finding).file_path(:medoids_gsp), f)
     else
-      src = File.expand_path('utils/representatives.rb', MiGA::MiGA.root_path)
-      `ruby '#{src}' '#{p.path}' | cut -f 2 > '#{f}'`
+      src = File.join(MiGA::MiGA.root_path, 'utils/representatives.rb')
+      MiGA::MiGA.run_cmd("ruby '#{src}' '#{p.path}' | cut -f 2", stdout: f)
     end
     File.readlines(f).map(&:chomp)
   end

data/lib/miga/cli/action/doctor.rb

@@ -236,8 +236,7 @@ class MiGA::Cli::Action::Doctor < MiGA::Cli::Action
         file = res.file_path(f) or next
         if file !~ /\.gz/
           cli.say "  > Gzipping #{d.name} #{f}   "
-          cmdo = `gzip -9 '#{file}'`.chomp
-          warn(cmdo) unless cmdo.empty?
+          run_cmd(['gzip', '-9', file])
           changed = true
         end
       end
@@ -267,8 +266,9 @@ class MiGA::Cli::Action::Doctor < MiGA::Cli::Action
       next if Dir["#{dir}/*.faa"].empty?
 
       cli.say "  > Fixing #{d.name}"
-      cmdo = `cd '#{dir}' && tar -zcf proteins.tar.gz *.faa && rm *.faa`.chomp
-      warn(cmdo) unless cmdo.empty?
+      run_cmd <<~CMD
+        cd #{dir.shellescape} && tar -zcf proteins.tar.gz *.faa && rm *.faa
+      CMD
     end
   end
 
@@ -292,10 +292,11 @@ class MiGA::Cli::Action::Doctor < MiGA::Cli::Action
       fix = false
       unless dir.nil?
         if Dir.exist? dir
-          cmdo = `cd '#{dir}/..' \
+          run_cmd <<~CMD
+            cd #{dir.shellescape}/.. \
                 && tar -zcf '#{d.name}.reg.tar.gz' '#{d.name}.reg' \
-                && rm -r '#{d.name}.reg'`.chomp
-          warn(cmdo) unless cmdo.empty?
+                && rm -r '#{d.name}.reg'
+          CMD
         end
         fix = true
       end
@@ -335,4 +336,12 @@ class MiGA::Cli::Action::Doctor < MiGA::Cli::Action
     # TODO: Find different 95%ANI clusters with genomes from the same species
     # TODO: Find AAI values too high or too low for each LCA rank
   end
+
+  ##
+  # Run command +cmd+ with options +opts+
+  def run_cmd(cmd, opts = {})
+    opts = { return: :output, err2out: true, raise: false }.merge(opts)
+    cmdo = MiGA::MiGA.run_cmd(cmd, opts).chomp
+    warn(cmdo) unless cmdo.empty?
+  end
 end

data/lib/miga/cli/action/get_db.rb

@@ -190,7 +190,11 @@ class MiGA::Cli::Action::GetDb < MiGA::Cli::Action
 
   def unarchive(file)
     cli.say "Unarchiving #{file}"
-    `cd "#{cli[:local]}" && tar -zxf "#{file}" && rm "#{file}"`
+    MiGA::MiGA.run_cmd <<~CMD
+      cd #{cli[:local].shellescape} \
+        && tar -zxf #{file.shellescape} \
+        && rm #{file.shellescape}
+    CMD
   end
 
   def register_database(manif, db, ver)

data/lib/miga/cli/action/init/files_helper.rb

@@ -107,7 +107,12 @@ module MiGA::Cli::Action::Init::FilesHelper
       MiGA::MiGA.download_file_ftp(
         :miga_dist, arch, File.join(miga_db, arch)
       ) { |n, size| cli.advance("#{arch}:", n, size) }
-      `cd '#{miga_db}' && tar zxf '#{arch}' && rm '#{arch}'`
+      cmd = <<~CMD
+        cd #{miga_db.shellescape} \
+          && tar zxf #{arch.shellescape} \
+          && rm #{arch.shellescape}
+      CMD
+      run_cmd(cmd, source: nil)
       cli.puts
     end
   end

data/lib/miga/cli/action/init.rb

@@ -83,14 +83,16 @@ class MiGA::Cli::Action::Init < MiGA::Cli::Action
   def empty_action
   end
 
-  def run_cmd(cli, cmd)
-    `. "#{cli[:config]}" && #{cmd}`
+  def run_cmd(cli, cmd, opts = {})
+    opts = { return: :output, source: cli[:config] }.merge(opts)
+    MiGA::MiGA.run_cmd(cmd, opts)
   end
 
-  def run_r_cmd(cli, paths, cmd)
+  def run_r_cmd(cli, paths, cmd, opts = {})
     run_cmd(
       cli,
-      "echo #{cmd.shellescape} | #{paths['R'].shellescape} --vanilla -q 2>&1"
+      "echo #{cmd.shellescape} | #{paths['R'].shellescape} --vanilla -q",
+      { err2out: true, stdout: '/dev/null' }.merge(opts)
     )
   end
 
@@ -159,7 +161,7 @@ class MiGA::Cli::Action::Init < MiGA::Cli::Action
   def find_software(exec)
     path = nil
     loop do
-      d_path = File.dirname(run_cmd(cli, "which #{exec.shellescape}"))
+      d_path = File.dirname(run_cmd(cli, ['which', exec], raise: false))
       if cli[:ask] || d_path == '.'
         path = cli.ask_user('Where can I find it?', d_path, nil, true)
       else
@@ -172,7 +174,7 @@ class MiGA::Cli::Action::Init < MiGA::Cli::Action
         end
         break
       end
-      cli.print "I cannot find #{exec} "
+      cli.print "I cannot find #{exec}. "
     end
     path
   end
@@ -208,19 +210,21 @@ class MiGA::Cli::Action::Init < MiGA::Cli::Action
   end
 
   def test_library(cli, paths, language, pkg)
-    case language
-    when :r
-      run_r_cmd(cli, paths, "library('#{pkg}')")
-    when :ruby
-      x = "#{paths['ruby'].shellescape} -r #{pkg.shellescape} -e '' 2>/dev/null"
-      run_cmd(cli, x)
-    when :python
-      x = "#{paths['python3'].shellescape} -c 'import #{pkg}' 2>/dev/null"
-      run_cmd(cli, x)
-    else
-      raise "Unrecognized language: #{language}"
-    end
-    $?.success?
+    opts = { raise: false, return: :status, stderr: '/dev/null' }
+    status =
+      case language
+      when :r
+        run_r_cmd(cli, paths, "library('#{pkg}')", opts)
+      when :ruby
+        x = [paths['ruby'], '-r', pkg, '-e', '']
+        run_cmd(cli, x, opts)
+      when :python
+        x = [paths['python3'], '-c', "import #{pkg}"]
+        run_cmd(cli, x, opts)
+      else
+        raise "Unrecognized language: #{language}"
+      end
+    status.success?
   end
 
   def install_library(cli, paths, language, pkg)
@@ -232,14 +236,14 @@ class MiGA::Cli::Action::Init < MiGA::Cli::Action
       # This hackey mess is meant to ensure the test and installation are done
       # on the configuration Ruby, not on the Ruby currently executing the
       # init action
-      gem_cmd = "Gem::GemRunner.new.run %w(install --user #{pkg})"
-      x = "#{paths['ruby'].shellescape} -r rubygems -r rubygems/gem_runner \
-            -e #{gem_cmd.shellescape} 2>&1"
-      run_cmd(cli, x)
+      x = [
+        paths['ruby'], '-r', 'rubygems', '-r', 'rubygems/gem_runner',
+        '-e', "Gem::GemRunner.new.run %w(install --user #{pkg})"
+      ]
+      run_cmd(cli, x, err2out: true)
     when :python
-      x = "#{paths['python3'].shellescape} \
-            -m pip install --user #{pkg.shellescape} 2>&1"
-      run_cmd(cli, x)
+      x = [paths['python3'], '-m', 'pip', 'install', '--user', pkg]
+      run_cmd(cli, x, err2out: true)
     else
       raise "Unrecognized language: #{language}"
     end

data/lib/miga/cli/action/ls.rb

@@ -1,5 +1,4 @@
-# @package MiGA
-# @license Artistic-2.0
+# frozen_string_literal: true
 
 require 'miga/cli/action'
 
@@ -42,33 +41,53 @@ class MiGA::Cli::Action::Ls < MiGA::Cli::Action
         '-s', '--silent',
         'No output and exit with non-zero status if the dataset list is empty'
       ) { |v| cli[:silent] = v }
+      opt.on(
+        '--exec CMD',
+        'Command to execute per dataset, with the following token variables:',
+        '~ {{dataset}}: Name of the dataset',
+        '~ {{project}}: Path to the project'
+      ) { |v| cli[:exec] = v }
     end
   end
 
   def perform
     ds = cli.load_and_filter_datasets(cli[:silent])
+    p  = cli.load_project
     exit(ds.empty? ? 1 : 0) if cli[:silent]
-    if !cli[:datum].nil?
+
+    head = nil
+    fun  = nil
+    if cli[:datum]
       cli[:tabular] = true
-      format_table(ds, [nil, nil]) { |d| [d.name, d.metadata[cli[:datum]]] }
-    elsif !cli[:fields].nil?
-      format_table(ds, [:name] + cli[:fields]) do |d|
-        [d.name] + cli[:fields].map { |f| d.metadata[f] }
-      end
+      head = [nil, nil]
+      fun  = proc { |d| [d.name, d.metadata[cli[:datum]]] }
+    elsif cli[:fields]
+      head = [:name] + cli[:fields]
+      fun  = proc { |d| [d.name] + cli[:fields].map { |f| d.metadata[f] } }
     elsif cli[:info]
-      format_table(ds, Dataset.INFO_FIELDS) { |d| d.info }
+      head = Dataset.INFO_FIELDS
+      fun  = proc(&:info)
     elsif cli[:processing]
-      comp = %w[- done queued]
-      format_table(ds, [:name] + MiGA::Dataset.PREPROCESSING_TASKS) do |d|
-        [d.name] + d.profile_advance.map { |i| comp[i] }
+      head = [:name] + MiGA::Dataset.PREPROCESSING_TASKS
+      fun  = proc do |d|
+        [d.name] + d.profile_advance.map { |i| %w[- done queued][i] }
       end
     elsif cli[:taskstatus]
-      format_table(ds, [:name] + MiGA::Dataset.PREPROCESSING_TASKS) do |d|
-        [d.name] + d.results_status.values
-      end
+      head = [:name] + MiGA::Dataset.PREPROCESSING_TASKS
+      fun  = proc { |d| [d.name] + d.results_status.values }
     else
       cli[:tabular] = true
-      format_table(ds, [nil]) { |d| [d.name] }
+      head = [nil]
+      fun  = proc { |d| [d.name] }
+    end
+
+    format_table(ds, head) do |d|
+      if cli[:exec]
+        MiGA::MiGA.run_cmd(
+          cli[:exec].miga_variables(dataset: d.name, project: p.path)
+        )
+      end
+      fun[d]
     end
   end
 

data/lib/miga/cli/action/run.rb

@@ -2,7 +2,6 @@
 # @license Artistic-2.0
 
 require 'miga/cli/action'
-require 'shellwords'
 
 class MiGA::Cli::Action::Run < MiGA::Cli::Action
   def parse_cli
@@ -54,9 +53,13 @@ class MiGA::Cli::Action::Run < MiGA::Cli::Action
 
     # Prepare command
     miga = MiGA.root_path
-    cmd = ["PROJECT=#{p.path.shellescape}",
-           "RUNTYPE=#{cli[:remote] ? 'ssh' : 'bash'}",
-           "MIGA=#{miga.shellescape}", "CORES=#{cli[:thr]}"]
+    opts = {}
+    cmd = [
+      "PROJECT=#{p.path.shellescape}",
+      "RUNTYPE=#{cli[:remote] ? 'ssh' : 'bash'}",
+      "MIGA=#{miga.shellescape}",
+      "CORES=#{cli[:thr]}"
+    ]
     obj = cli.load_project_or_dataset
     klass = obj.class
     virtual_task = %i[p d maintenance].include?(cli[:result])
@@ -67,13 +70,22 @@ class MiGA::Cli::Action::Run < MiGA::Cli::Action
 
     cmd << MiGA.script_path(cli[:result], miga: miga, project: p).shellescape
     if cli[:remote]
-      cmd = ['ssh', '-t', '-t', cli[:remote].shellescape,
-             cmd.join(' ').shellescape]
+      cmd = [
+        'ssh', '-t', '-t',
+        cli[:remote].shellescape,
+        cmd.join(' ').shellescape
+      ]
+    end
+
+    if cli[:log]
+      opts[:stdout] = cli[:log]
+      opts[:err2out] = true
     end
-    cmd << ['>', cli[:log].shellescape, '2>&1'] if cli[:log]
 
     # Launch
-    pid = spawn cmd.join(' ')
-    Process.wait pid
+    # note that all elements were carefully escaped in advace, so this has to be
+    # passed as String to avoid double-escaping or unintentionally escaping
+    # characters such as `=` and `>`
+    MiGA::MiGA.run_cmd(cmd.join(' '), opts)
   end
 end

data/lib/miga/cli/action/tax_dist.rb

@@ -3,7 +3,6 @@
 
 require 'miga/cli/action'
 require 'miga/tax_index'
-require 'zlib'
 require 'tmpdir'
 
 class MiGA::Cli::Action::TaxDist < MiGA::Cli::Action

data/lib/miga/cli/base.rb

@@ -1,5 +1,4 @@
-# @package MiGA
-# @license Artistic-2.0
+# frozen_string_literal: true
 
 module MiGA::Cli::Base
   @@TASK_DESC = {
@@ -20,10 +19,10 @@ module MiGA::Cli::Base
     add: 'Create a dataset in a MiGA project',
     get: 'Download a dataset from public databases into a MiGA project',
     ncbi_get: 'Download all genomes in a taxon from NCBI into a MiGA project',
-    rm: 'Remove a dataset from an MiGA project',
+    rm: 'Remove a dataset from a MiGA project',
     find: 'Find unregistered datasets based on result files',
     ln: 'Link datasets (including results) from one project to another',
-    ls: 'List all registered datasets in an MiGA project',
+    ls: 'List all registered datasets in a MiGA project',
     archive: 'Generate a tar-ball with all files from select datasets',
     # Results
     add_result: 'Register a result',

data/lib/miga/cli/opt_helper.rb

@@ -1,6 +1,7 @@
-# @package MiGA
-# @license Artistic-2.0
+# frozen_string_literal: true
 
+##
+# Helper module including functions for CLI options
 module MiGA::Cli::OptHelper
   ##
   # Send MiGA's banner to OptionParser +opt+
@@ -26,6 +27,10 @@ module MiGA::Cli::OptHelper
         'Accept all defaults as answers'
       ) { |v| self[:auto] = v }
     end
+    opt.on(
+      '--rand-seed INT', Integer,
+      'Set this seed to initialize pseudo-randomness'
+    ) { |v| srand(v) }
     opt.on(
       '-v', '--verbose',
       'Print additional information to STDERR'
@@ -61,7 +66,7 @@ module MiGA::Cli::OptHelper
   # - :result_project To require a type of project result
   # The options :result, :result_opt, :result_dataset, and :result_project
   # are mutually exclusive
-  def opt_object(opt, what = [:project, :dataset])
+  def opt_object(opt, what = %i[project dataset])
     what.each do |w|
       case w
       when :project
@@ -86,7 +91,7 @@ module MiGA::Cli::OptHelper
       when :result, :result_opt
         opt.on(
           '-r', '--result STRING',
-          "#{"(Mandatory) " if w == :result}Name of the result",
+          "#{'(Mandatory) ' if w == :result}Name of the result",
           'Recognized names for dataset-specific results include:',
           *MiGA::Dataset.RESULT_DIRS.keys.map { |n| " ~ #{n}" },
           'Recognized names for project-wide results include:',
@@ -118,7 +123,7 @@ module MiGA::Cli::OptHelper
   # - :active To filter by active (--active) or inactive (--no-active)
   # - :taxonomy To filter by taxonomy (--taxonomy)
   # The "k-th" filter (--dataset-k) is always included
-  def opt_filter_datasets(opt, what = [:ref, :multi, :active, :taxonomy])
+  def opt_filter_datasets(opt, what = %i[ref multi active taxonomy])
     what.each do |w|
       case w
       when :ref
@@ -161,6 +166,6 @@ module MiGA::Cli::OptHelper
   # If +sym+ is nil, +flag+ is used as Symbol
   def opt_flag(opt, flag, description, sym = nil)
     sym = flag.to_sym if sym.nil?
-    opt.on("--#{flag.to_s.gsub('_', '-')}", description) { |v| self[sym] = v }
+    opt.on("--#{flag.to_s.tr('_', '-')}", description) { |v| self[sym] = v }
   end
 end

data/lib/miga/common/errors.rb

@@ -0,0 +1,12 @@
+
+module MiGA
+  ##
+  # A generic MiGA error
+  class Error < RuntimeError
+  end
+
+  ##
+  # An error with a system call
+  class SystemCallError < Error
+  end
+end

data/lib/miga/common/format.rb

@@ -1,5 +1,6 @@
+# frozen_string_literal: true
+
 require 'tempfile'
-require 'zlib'
 
 ##
 # General formatting functions shared throughout MiGA.
@@ -38,13 +39,13 @@ module MiGA::Common::Format
         tmp_path = tmp_fh.path
         fh = File.open(file, 'r')
       end
-      buffer = ''
+      buffer = ''.dup
       fh.each_line do |ln|
         ln.chomp!
         if ln =~ /^>\s*(\S+)(.*)/
           id, df = $1, $2
           tmp_fh.print buffer.wrap_width(80)
-          buffer = ''
+          buffer = ''.dup
           tmp_fh.puts ">#{id.gsub(/[^A-Za-z0-9_\|\.]/, '_')}#{df}"
         else
           buffer << ln.gsub(/[^A-Za-z\.\-]/, '')
@@ -166,7 +167,7 @@ class String
   ##
   # Replace {{variables}} using the +vars+ hash
   def miga_variables(vars)
-    o = "#{self}"
+    o = self.dup
     vars.each { |k, v| o.gsub!("{{#{k}}}", v.to_s) }
     o
   end

data/lib/miga/common/system_call.rb

@@ -0,0 +1,93 @@
+require 'shellwords'
+
+##
+# General functions for process (system call) execution
+module MiGA::Common::SystemCall
+  ##
+  # Execute the command +cmd+ with options +opts+ determined by #run_cmd_opts
+  #
+  # The command +cmd+ can be:
+  # - String: The command is processed as is, without changes
+  # - Array: The command is built with +shelljoin+ so each value is escaped
+  def run_cmd(cmd, opts = {})
+    opts = run_cmd_opts(opts)
+    cmd = cmd.shelljoin if cmd.is_a?(Array)
+    spawn_opts = {}
+    spawn_opts[:out] = opts[:stdout] if opts[:stdout]
+    spawn_opts[:err] = opts[:stderr] if opts[:stderr]
+    out_io, spawn_opts[:out] = IO.pipe if opts[:return] == :output
+    spawn_opts[:err] = [:child, :out] if opts[:err2out] && spawn_opts[:out]
+    opts[:source] = MiGA::MiGA.rc_path if opts[:source] == :miga
+    if opts[:source] && File.exist?(opts[:source])
+      cmd = ". #{opts[:source].shellescape} && #{cmd}"
+    end
+
+    DEBUG "CMD: #{cmd}"
+    puts "CMD: #{cmd}" if opts[:show_cmd]
+    return if opts[:dry]
+
+    pid = nil
+    error = nil
+    begin
+      pid = spawn(opts[:env], cmd, spawn_opts)
+      Process.wait(pid)
+    rescue => e
+      error = e
+    end
+    status = $?
+
+    if opts[:raise] && !status&.success?
+      raise MiGA::SystemCallError.new(
+        "Command failed with status " \
+          "#{status&.exitstatus}#{' (core dump)' if status&.coredump?}:\n" \
+          "#{error&.class}: #{error&.message}\n" \
+          "OPT: #{opts}\n" \
+          "CMD: #{cmd}"
+      )
+    end
+
+    case opts[:return]
+    when :status ; status
+    when :pid    ; pid
+    when :error  ; error
+    when :output
+      spawn_opts[:out].close
+      output = out_io.read
+      out_io.close
+      output
+    end
+  end
+
+  ##
+  # Options for #run_cmd using a Hash +opts+ to modify defaults
+  #
+  # Supported keys (as Symbol) include:
+  # - stdout: Redirect STDOUT to this file
+  # - stderr: Redirect STDOUT to this file
+  # - dry: Don't run, just send the command to debug (default: false)
+  # - return: What should the function return, supported values are
+  #   +:status+ (Process::Status, default), +:pid+ (Integer, process ID),
+  #   +:error+ (Error if failed, nil otherwise), +:output+ (String,
+  #   contents sent to STDOUT)
+  # - raise: Raise an exception (MiGA::SystemCallError) in case of failure
+  #   (default: true)
+  # - show_cmd: Print command to the STDOUT (prefixed with CMD: ) to ease
+  #   debugging (default: false)
+  # - err2out: Redirect STDERR to STDOUT
+  # - env: Environmental variables as a Hash, keys and values must be strings
+  # - source: A file to be sourced before running, or the Symbol +:miga+ to
+  #   source the MiGA configuration file
+  def run_cmd_opts(opts = {})
+    {
+      stdout: nil,
+      stderr: nil,
+      dry: false,
+      return: :status,
+      raise: true,
+      show_cmd: false,
+      err2out: false,
+      env: {},
+      source: nil
+    }.merge(opts)
+  end
+end

data/lib/miga/common.rb

@@ -1,14 +1,16 @@
-# @package MiGA
-# @license Artistic-2.0
+# frozen_string_literal: true
 
+require 'zlib'
+require 'stringio'
 require 'miga/version'
 require 'miga/json'
 require 'miga/parallel'
 require 'miga/common/base'
+require 'miga/common/errors'
 require 'miga/common/path'
 require 'miga/common/format'
 require 'miga/common/net'
-require 'stringio'
+require 'miga/common/system_call'
 
 ##
 # Generic class used to handle system-wide information and methods, and parent
@@ -19,14 +21,21 @@ class MiGA::MiGA
   extend MiGA::Common::Path
   extend MiGA::Common::Format
   extend MiGA::Common::Net
+  extend MiGA::Common::SystemCall
 
   ENV['MIGA_HOME'] ||= ENV['HOME']
 
+  ##
+  # Path to the +.miga_rc+ file
+  def self.rc_path
+    File.join(ENV['MIGA_HOME'], '.miga_rc')
+  end
+
   ##
   # Has MiGA been initialized?
   def self.initialized?
-    File.exist?(File.expand_path('.miga_rc', ENV['MIGA_HOME'])) &&
-      File.exist?(File.expand_path('.miga_daemon.json', ENV['MIGA_HOME']))
+    File.exist?(rc_path) &&
+      File.exist?(File.join(ENV['MIGA_HOME'], '.miga_daemon.json'))
   end
 
   ##

data/lib/miga/daemon.rb

@@ -321,7 +321,9 @@ class MiGA::Daemon < MiGA::MiGA
   def purge!
     say 'Probing running jobs'
     @jobs_running.select! do |job|
-      `#{runopts(:alive).miga_variables(pid: job[:pid])}`.chomp.to_i == 1
+      MiGA::MiGA.run_cmd(
+        runopts(:alive).miga_variables(pid: job[:pid]), return: :output
+      ).chomp.to_i == 1
     end
   end
 
@@ -346,7 +348,7 @@ class MiGA::Daemon < MiGA::MiGA
       Process.detach(job[:pid]) unless [nil, '', 0].include?(job[:pid])
     else
       # Schedule cluster job (qsub, msub, slurm)
-      job[:pid] = `#{job[:cmd]}`.chomp
+      job[:pid] = MiGA::MiGA.run_cmd(job[:cmd], return: :output).chomp
     end
 
     # Check if registered

data/lib/miga/result/stats.rb

@@ -1,4 +1,3 @@
-require 'zlib'
 require 'miga/result/base'
 
 ##
@@ -198,12 +197,12 @@ module MiGA::Result::Stats
        file_path(:raw_report)
 
     MiGA::MiGA.DEBUG "Fixing essential genes by domain"
-    scr = "#{MiGA::MiGA.root_path}/utils/domain-ess-genes.rb"
+    scr = File.join(MiGA::MiGA.root_path, 'utils', 'domain-ess-genes.rb')
     rep = file_path(:report)
-    rc_p = File.expand_path('.miga_rc', ENV['HOME'])
-    rc = File.exist?(rc_p) ? ". '#{rc_p}' && " : ''
-    $stderr.print `#{rc} ruby '#{scr}' \
-      '#{rep}' '#{rep}.domain' '#{tax[:d][0]}'`
+    $stderr.print MiGA::MiGA.run_cmd(
+      ['ruby', scr, rep, "#{rep}.domain", tax[:d][0]],
+      return: :output, err2out: true, source: :miga
+    )
     add_file(:raw_report, "#{source.name}.ess/log")
     add_file(:report, "#{source.name}.ess/log.domain")
   end

data/lib/miga/tax_dist.rb

@@ -3,7 +3,6 @@
 
 require 'miga/common'
 require 'miga/taxonomy'
-require 'zlib'
 
 ##
 # Methods for taxonomy identification based on AAI/ANI values.

data/lib/miga/version.rb

@@ -12,7 +12,7 @@ module MiGA
   # - String indicating release status:
   #   - rc* release candidate, not released as gem
   #   - [0-9]+ stable release, released as gem
-  VERSION = [1.1, 3, 6].freeze
+  VERSION = [1.2, 1, 0].freeze
 
   ##
   # Nickname for the current major.minor version.
@@ -20,7 +20,7 @@ module MiGA
 
   ##
   # Date of the current gem relese.
-  VERSION_DATE = Date.new(2021, 12, 1)
+  VERSION_DATE = Date.new(2021, 12, 12)
 
   ##
   # References of MiGA

data/test/format_test.rb

@@ -58,6 +58,9 @@ class FormatTest < Test::Unit::TestCase
     assert_equal(50.0, o[:gc])
     assert_equal(5, o[:n50])
     assert_equal(4.0, o[:med])
+    o = MiGA::MiGA.seqs_length(f, :fasta, skew: true)
+    assert_equal(-50.0, o[:at_skew])
+    assert_equal(-25.0, o[:gc_skew])
   end
 
   def test_seqs_length_fastq
@@ -77,4 +80,22 @@ class FormatTest < Test::Unit::TestCase
     assert_equal('123  45', tab[2])
     assert_equal('678  90', tab[3])
   end
+
+  def test_miga_name
+    assert_not('a-bad-name'.miga_name?)
+    assert('a_good_one'.miga_name?)
+
+    assert('After_it_s_fixed_', 'After it\'s fixed!'.miga_name)
+
+    assert('A sp.', 'A_sp_'.unmiga_name)
+    assert('B str. C', 'B_sp_C'.unmiga_name)
+    assert('The X content', 'The_x_content'.unmiga_name)
+  end
+
+  def test_miga_variables
+    assert_equal(
+      '1 a box!',
+      '{{n}} {{my}} {{secret}}!'.miga_variables(my: 'a', secret: :box, n: 1)
+    )
+  end
 end

data/test/system_call_test.rb

@@ -0,0 +1,50 @@
+require 'test_helper'
+
+class SystemCallTest < Test::Unit::TestCase
+  include TestHelper
+
+  def test_run_cmd_opts
+    assert_equal(true, MiGA::MiGA.run_cmd_opts[:raise])
+    assert_equal(:status, MiGA::MiGA.run_cmd_opts[:return])
+    assert_equal(:pid, MiGA::MiGA.run_cmd_opts(return: :pid)[:return])
+    assert_nil(MiGA::MiGA.run_cmd_opts[:stdout])
+  end
+
+  def test_run_cmd_redirection
+    f1 = tmpfile('f1')
+    MiGA::MiGA.run_cmd('echo 1', stdout: f1)
+    assert_equal("1\n", File.read(f1))
+
+    MiGA::MiGA.run_cmd('echo 2 >&2', stderr: f1)
+    assert_equal("2\n", File.read(f1))
+
+    f2 = tmpfile('with spaces')
+    MiGA::MiGA.run_cmd('echo 3', stdout: f2)
+    assert_equal("3\n", File.read(f2))
+
+    MiGA::MiGA.run_cmd(['echo', 4], stdout: f2)
+    assert_equal("4\n", File.read(f2))
+  end
+
+  def test_run_cmd_return
+    o = MiGA::MiGA.run_cmd('echo 1', stdout: '/dev/null')
+    assert(o.is_a? Process::Status)
+    assert(o.success?)
+
+    o = MiGA::MiGA.run_cmd('echo 1', stdout: '/dev/null', return: :pid)
+    assert(o.is_a? Integer)
+
+    o = MiGA::MiGA.run_cmd('echo 1', stdout: '/dev/null', return: :error)
+    assert_nil(o)
+  end
+
+  def test_run_cmd_raise
+    assert_raise(MiGA::SystemCallError) { MiGA::MiGA.run_cmd('echoes!!!') }
+
+    o = MiGA::MiGA.run_cmd('echoes!!!', raise: false, return: :status)
+    assert_not(o.success?)
+
+    o = MiGA::MiGA.run_cmd('echoes!!!', raise: false, return: :error)
+    assert(o.is_a? Errno::ENOENT)
+  end
+end

data/utils/distance/base.rb

@@ -1,7 +1,7 @@
 require 'miga'
 require 'miga/tax_dist'
 
-class MiGA::DistanceRunner
+class MiGA::DistanceRunner < MiGA::MiGA
   require_relative 'temporal.rb'
   require_relative 'database.rb'
   require_relative 'commands.rb'

data/utils/distance/commands.rb

@@ -227,7 +227,6 @@ module MiGA::DistanceRunner::Commands
   end
 
   def run_cmd(cmd)
-    puts "CMD: #{cmd}"
-    puts `#{cmd} 2>&1`
+    MiGA::MiGA.run_cmd(cmd, show_cmd: true, err2out: true)
   end
 end

data/utils/distance/pipeline.rb

@@ -58,9 +58,9 @@ module MiGA::DistanceRunner::Pipeline
       end
     end
     ds_matrix_fh.close
-    ref_tree = File.expand_path('utils/ref-tree.R', MiGA::MiGA.root_path)
-    `"#{ref_tree}" "#{ds_matrix}" "#{out_base}" "#{dataset.name}"`
-    File.unlink ds_matrix
+    ref_tree = File.join(MiGA::MiGA.root_path, 'utils', 'ref-tree.R')
+    MiGA::MiGA.run_cmd([ref_tree, ds_matrix, out_base, dataset.name])
+    File.unlink(ds_matrix)
   end
 
   # Tests taxonomy

data/utils/find-medoid.R

@@ -26,7 +26,6 @@ find_medoids <- function (ani.df, out, clades) {
   medoids <- c()
   for(i in cl){
     lab <- strsplit(i, ",")[[1]]
-    cat("Clade of:", lab[1], "\n")
     if(length(lab) == 1) {
       lab.s <- lab
     } else {
@@ -44,7 +43,7 @@ find_medoids <- function (ani.df, out, clades) {
 }
 
 #= Main
-cat("Finding Medoids")
+cat("Finding Medoids\n")
 ani <- readRDS(argv[1])
 find_medoids(ani.df = ani, out = argv[2], clades = argv[3])
 

data/utils/subclade/base.rb

@@ -1,5 +1,10 @@
 require 'zlib'
 require 'miga'
 
-class MiGA::SubcladeRunner
+class MiGA::SubcladeRunner < MiGA::MiGA
+  require_relative 'temporal.rb'
+  require_relative 'pipeline.rb'
+
+  include MiGA::SubcladeRunner::Temporal
+  include MiGA::SubcladeRunner::Pipeline
 end

data/utils/subclade/pipeline.rb

@@ -29,7 +29,10 @@ module MiGA::SubcladeRunner::Pipeline
       ofh.close
       # Cluster genomes
       if File.size? abc_path
-        `ogs.mcl.rb -o '#{ogs_file}.tmp' --abc '#{abc_path}' -t '#{opts[:thr]}'`
+        run_cmd([
+          'ogs.mcl.rb',
+          '-o', "#{ogs_file}.tmp", '--abc', abc_path, '-t', opts[:thr]
+        ])
         File.open(ogs_file, 'w') do |fh|
           File.foreach("#{ogs_file}.tmp").with_index do |ln, lno|
             fh.puts(ln) if lno > 0
@@ -45,8 +48,10 @@ module MiGA::SubcladeRunner::Pipeline
     # Find genomospecies medoids
     src = File.expand_path('utils/find-medoid.R', MiGA::MiGA.root_path)
     dir = opts[:gsp_metric] == 'aai' ? '02.aai' : '03.ani'
-    `Rscript '#{src}' '../../09.distances/#{dir}/miga-project.rds' \
-      miga-project.gsp-medoids miga-project.gsp-clades`
+    run_cmd([
+      'Rscript', src, "../../09.distances/#{dir}/miga-project.rds",
+      'miga-project.gsp-medoids', 'miga-project.gsp-clades'
+    ])
     if File.exist? 'miga-project.gsp-clades.sorted'
       File.rename 'miga-project.gsp-clades.sorted', 'miga-project.gsp-clades'
     end
@@ -62,13 +67,15 @@ module MiGA::SubcladeRunner::Pipeline
     ofh.close
   end
 
-  def subclades metric
+  def subclades(metric)
     src = File.expand_path('utils/subclades.R', MiGA::MiGA.root_path)
     step = :"#{metric}_distances"
     metric_res = project.result(step) or raise "Incomplete step #{step}"
     matrix = metric_res.file_path(:matrix)
-    `Rscript '#{src}' '#{matrix}' miga-project '#{opts[:thr]}' \
-      miga-project.gsp-medoids '#{opts[:run_clades] ? 'cluster' : 'empty'}'`
+    run_cmd([
+      'Rscript', src, matrix, 'miga-project', opts[:thr],
+      'miga-project.gsp-medoids', opts[:run_clades] ? 'cluster' : 'empty'
+    ])
     if File.exist? 'miga-project.nwk'
       File.rename('miga-project.nwk', "miga-project.#{metric}.nwk")
     end
@@ -76,6 +83,10 @@ module MiGA::SubcladeRunner::Pipeline
 
   def compile
     src = File.expand_path('utils/subclades-compile.rb', MiGA::MiGA.root_path)
-    `ruby '#{src}' '.' 'miga-project.class'`
+    run_cmd(['ruby', src, '.', 'miga-project.class'])
+  end
+
+  def run_cmd(cmd)
+    MiGA::MiGA.run_cmd(cmd, show_cmd: true, err2out: true)
   end
 end

data/utils/subclade/runner.rb

@@ -1,12 +1,8 @@
+
 require_relative 'base.rb'
-require_relative 'temporal.rb'
-require_relative 'pipeline.rb'
 
 class MiGA::SubcladeRunner
-  include MiGA::SubcladeRunner::Temporal
-  include MiGA::SubcladeRunner::Pipeline
-
-  attr_reader :project, :step, :opts, :home, :tmp
+  attr_reader(:project, :step, :opts, :home, :tmp)
 
   def initialize(project_path, step, opts_hash = {})
     @opts = opts_hash

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: miga-base
 version: !ruby/object:Gem::Version
-  version: 1.1.3.6
+  version: 1.2.1.0
 platform: ruby
 authors:
 - Luis M. Rodriguez-R
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2021-12-01 00:00:00.000000000 Z
+date: 2021-12-12 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: daemons
@@ -176,10 +176,12 @@ files:
 - lib/miga/cli/opt_helper.rb
 - lib/miga/common.rb
 - lib/miga/common/base.rb
+- lib/miga/common/errors.rb
 - lib/miga/common/format.rb
 - lib/miga/common/hooks.rb
 - lib/miga/common/net.rb
 - lib/miga/common/path.rb
+- lib/miga/common/system_call.rb
 - lib/miga/common/with_daemon.rb
 - lib/miga/common/with_daemon_class.rb
 - lib/miga/common/with_option.rb
@@ -253,6 +255,7 @@ files:
 - test/remote_dataset_test.rb
 - test/result_stats_test.rb
 - test/result_test.rb
+- test/system_call_test.rb
 - test/tax_dist_test.rb
 - test/tax_index_test.rb
 - test/taxonomy_test.rb