rbbt-util 5.28.9 → 5.29.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 2fa2f71df9aef8810599ab3660c847c084dea75ccb2485d294cfa1962e341162
-  data.tar.gz: 7eb004ddf8bcf30b00325dc5251d9ed73a307de433c18c0847eb4d70a4bea1e2
+  metadata.gz: 169a7a3ce3ebf35313ca6f64d91e98bdb2bb7df79b0367f6888bf0978c67fb01
+  data.tar.gz: 654b5db6fea8f9ab2df8e8b00aaf945bdc9904965df0489ed72feba136ac6fd6
 SHA512:
-  metadata.gz: 6ba23322f9ac768f1782c4e26605c50171b6d35d1d3c2acc0ef195e26ca39af851a232e1581ca2ec220899b7780f35f8d0d36e1f61f0b0e6500c883999c3760b
-  data.tar.gz: afa0952f228a951ea78506ad0776fd96dbfbd2b91bac73310ee6807e4a7d0a74c7ca8f582880c5abd28bbf709c8be3d9dff1797056be756962091b2e97031f30
+  metadata.gz: dd168909a2df9e5931f74ca9b91ee64e5ec6b003561d75d2843736321deb1e62ac5807133822be461d370b80ecc8b64885c9e20486548d949d20c90fe72cbbab
+  data.tar.gz: 377c2c070eb32856de9fb09107cc3ab72de64936590db2bb58ca2c4eaf057e954c258f1345d4b542bb39b2aba756f0c37bfdae999b9aacf01267a1bf327d13fa

data/lib/rbbt/entity.rb

@@ -38,7 +38,7 @@ module Entity
                                                    if value.to_s == k.to_s
                                                      found = k
                                                      break
-                                                   elsif value =~ /\(#{Regexp.quote k}\)/
+                                                   elsif value.to_s =~ /\(#{Regexp.quote k}\)/
                                                      found = k
                                                      break
                                                    end

data/lib/rbbt/fix_width_table.rb

@@ -67,7 +67,7 @@ class FixWidthTable
 
   def format(pos, value)
     padding = value_size - value.length
-    if range
+    if @range
       (pos  + [padding, value + ("\0" * padding)]).pack("llll#{mask}")
     else
       [pos, padding, value + ("\0" * padding)].pack("ll#{mask}")
@@ -105,7 +105,7 @@ class FixWidthTable
 
   def idx_value(index)
     return nil if index < 0 or index >= size
-    @file.seek((range ? 17 : 9 ) + (record_size) * index, IO::SEEK_SET)
+    @file.seek((@range ? 17 : 9 ) + (record_size) * index, IO::SEEK_SET)
     padding = @file.read(4).unpack("l").first+1
     txt = @file.read(value_size)
     str = txt.unpack(mask).first
@@ -277,7 +277,8 @@ class FixWidthTable
 
   def [](pos)
     return [] if size == 0
-    if range
+    self.read
+    if @range
       get_range(pos)
     else
       get_point(pos)
@@ -286,7 +287,7 @@ class FixWidthTable
   
   def overlaps(pos, value = false)
     return [] if size == 0
-    idxs = if range
+    idxs = if @range
       get_range(pos, true)
     else
       get_point(pos, true)

data/lib/rbbt/hpc.rb

@@ -1,551 +1,3 @@
 require 'rbbt-util'
 require 'rbbt/util/cmd'
-
-module Marenostrum
-  SERVER='mn1'
-  class SBATCH < Exception; 
-    attr_accessor :directory
-    def initialize(directory)
-      @directory = directory
-    end
-  end
-
-  module SLURM
-
-    def self.template(args, options = {})
-
-      development      = options.delete :drbbt
-      singularity      = options.delete :singularity
-      contain          = options.delete :contain
-      sync             = options.delete :sync
-      user_group       = options.delete :user_group
-      contain_and_sync = options.delete :contain_and_sync
-      wipe_container   = options.delete :wipe_container
-      copy_image       = options.delete :copy_image
-      exclusive        = options.delete :exclusive
-      highmem          = options.delete :highmem
-
-      queue            = options.delete(:queue) || 'bsc_ls'
-      task_cpus        = options.delete(:task_cpus) || 1
-      nodes            = options.delete(:nodes) || 1
-      time             = options.delete(:time) || "0:00:10"
-
-      inputs_dir       = options.delete :inputs_dir
-      config_keys      = options.delete :config_keys
-
-      user = ENV['USER'] || `whoami`.strip
-      group = File.basename(File.dirname(ENV['HOME']))
-
-      if contain_and_sync
-        contain = "/scratch/tmp/rbbt-#{user}" if contain.nil?
-        sync = "~/.rbbt/var/jobs" if sync.nil?
-        wipe_container = "post" if wipe_container.nil?
-      end
-
-      contain = nil if contain == "" || contain == "none"
-      sync = nil if sync == "" || sync == "none"
-
-      contain = File.expand_path(contain) if contain
-
-      name = options[:name] ||= Misc.obj2digest({:options => options.collect{|k,v| [k,v]}.sort_by{|k,v| k.to_s }, :args => args})
-      options.delete(:name)
-      slurm_basedir = options[:slurm_basedir] ||= File.expand_path(File.join('~/rbbt-slurm', name)) if slurm_basedir.nil?
-      options.delete(:slurm_basedir)
-
-      rbbt_cmd = args.reject{|e| e == '--' }.collect{|e| e.include?(" ")? '"' + e + '"' : e } * " "
-
-      rbbt_cmd += " "  << options.collect do |o,v|
-        o = o.to_s
-        case v
-        when TrueClass 
-          '--' << o
-        when FalseClass 
-          '--' << o << "=false"
-        else
-          ['--' << o, "'#{v}'"] * " "
-        end
-      end * " "
-
-      rbbt_cmd << " --config_keys='#{config_keys}'" if config_keys and not config_keys.empty?
-
-
-      time = Misc.format_seconds Misc.timespan(time) unless time.include? ":"
-
-
-      #{{{ PREPARE LOCAL LOGFILES
-
-      Open.mkdir slurm_basedir
-
-      fout = File.join(slurm_basedir, 'std.out')
-      ferr = File.join(slurm_basedir, 'std.err')
-      fjob = File.join(slurm_basedir, 'job.id')
-      fexit = File.join(slurm_basedir, 'exit.status')
-      fsync = File.join(slurm_basedir, 'sync.log')
-      fcmd = File.join(slurm_basedir, 'command.slurm')
-
-      #{{{ GENERATE TEMPLATE
-
-      # HEADER
-      header =<<-EOF
-#!/bin/bash
-#SBATCH --qos="#{queue}"
-#SBATCH --job-name="#{name}"
-#SBATCH --workdir="#{Dir.pwd}"
-#SBATCH --output="#{fout}"
-#SBATCH --error="#{ferr}"
-#SBATCH --cpus-per-task="#{task_cpus}"
-#SBATCH --time="#{time}"
-#SBATCH --nodes="#{nodes}"
-      EOF
-
-      if highmem
-        header +=<<-EOF
-#SBATCH --constraint=highmem 
-        EOF
-      end
-
-      if exclusive
-        header +=<<-EOF
-#SBATCH --exclusive 
-        EOF
-      end
-
-      header +=<<-EOF
-#CMD: #{rbbt_cmd}
-      EOF
-
-      # ENV
-      env = ""
-      env +=<<-EOF
-# Prepare env
-[[ -f ~/config/load.sh ]] && source ~/config/load.sh
-module load java
-
-# Calculate max available memory
-let "MAX_MEMORY=$SLURM_MEM_PER_CPU * $SLURM_CPUS_ON_NODE"
-      EOF
-
-
-      # RUN
-      run = ""
-      exec_cmd = %(env _JAVA_OPTIONS="-Xms1g -Xmx${MAX_MEMORY}m")
-
-
-      if singularity
-        #{{{ SINGULARITY
-        
-        singularity_exec = %(singularity exec -e -B $SINGULARITY_OPT_DIR:/singularity_opt/ -B /apps/)
-        
-        env +=<<-EOF
-module load intel/2018.1
-module load singularity
-PROJECTS_ROOT="/gpfs/projects/bsc26/"
-SINGULARITY_IMG="$PROJECTS_ROOT/rbbt.singularity.img"
-SINGULARITY_OPT_DIR="$PROJECTS_ROOT/singularity_opt/"
-SINGULARITY_RUBY_INLINE="$HOME/.singularity_ruby_inline"
-mkdir -p "$SINGULARITY_RUBY_INLINE"
-        EOF
-
-        prep  = ""
-
-        if contain
-          scratch_group_dir = File.join('/gpfs/scratch/', group)
-          projects_group_dir = File.join('/gpfs/projects/', group)
-
-          prep +=<<-EOF
-
-# Prepare container dir
-CONTAINER_DIR="#{contain}"
-mkdir -p $CONTAINER_DIR/.rbbt/etc/
-
-for dir in .ruby_inline git home; do
-    mkdir -p $CONTAINER_DIR/$dir
-done
-
-for tmpd in persist_locks  produce_locks  R_sockets  sensiblewrite  sensiblewrite_locks  step_info_locks  tsv_open_locks; do
-    mkdir -p $CONTAINER_DIR/.rbbt/tmp/$tmpd
-done
-
-# Copy environment 
-cp ~/.rbbt/etc/environment $CONTAINER_DIR/.rbbt/etc/
-
-# Set search_paths
-echo "singularity: /singularity_opt/{PKGDIR}/{TOPLEVEL}/{SUBPATH}" > $CONTAINER_DIR/.rbbt/etc/search_paths
-echo "rbbt_user: /home/rbbt/.rbbt/{TOPLEVEL}/{SUBPATH}" >> $CONTAINER_DIR/.rbbt/etc/search_paths
-echo "home: $CONTAINER_DIR/home/{TOPLEVEL}/{SUBPATH}" >> $CONTAINER_DIR/.rbbt/etc/search_paths
-echo "group_projects: #{projects_group_dir}/{PKGDIR}/{TOPLEVEL}/{SUBPATH}" >> $CONTAINER_DIR/.rbbt/etc/search_paths
-echo "group_scratch: #{scratch_group_dir}/{PKGDIR}/{TOPLEVEL}/{SUBPATH}" >> $CONTAINER_DIR/.rbbt/etc/search_paths
-echo "user_projects: #{projects_group_dir}/#{user}/{PKGDIR}/{TOPLEVEL}/{SUBPATH}" >> $CONTAINER_DIR/.rbbt/etc/search_paths
-echo "user_scratch: #{scratch_group_dir}/#{user}/{PKGDIR}/{TOPLEVEL}/{SUBPATH}" >> $CONTAINER_DIR/.rbbt/etc/search_paths
-          EOF
-
-          if user_group && group != user_group
-            prep +=<<-EOF
-
-# Add user_group search_path
-echo "#{user_group}: /gpfs/projects/#{user_group}/{PKGDIR}/{TOPLEVEL}/{SUBPATH}" >> $CONTAINER_DIR/.rbbt/etc/search_paths
-            EOF
-          end
-
-          if inputs_dir
-            prep +=<<-EOF
-
-# Copy inputs
-[[ -d '#{inputs_dir}' ]] && cp -R '#{inputs_dir}' $CONTAINER_DIR/inputs
-            EOF
-            rbbt_cmd = rbbt_cmd.sub(inputs_dir, "#{contain}/inputs")
-          end
-
-          if copy_image
-            prep +=<<EOF
-
-# Copy image
-rsync -avz "$SINGULARITY_IMG" "$CONTAINER_DIR/rbbt.singularity.img" 1>&2
-SINGULARITY_IMG="$CONTAINER_DIR/rbbt.singularity.img"
-EOF
-          end
-
-          if  wipe_container == "pre" || wipe_container == "both"
-            if singularity
-              prep +=<<-EOF
-
-# Clean container pre
-singularity exec -e -C -H "$CONTAINER_DIR" "$SINGULARITY_IMG" rm -Rfv .rbbt/var/jobs &>> #{fsync}
-singularity exec -e -C -H "$CONTAINER_DIR" "$SINGULARITY_IMG" rbbt system clean -f &>> #{fsync}
-singularity exec -e -C -H "$CONTAINER_DIR" "$SINGULARITY_IMG" rm -Rfv tmp/ &>> #{fsync}
-EOF
-            end
-          end
-        end
-
-        if contain
-          singularity_exec << %( -C -H "$CONTAINER_DIR" \
--B /scratch/tmp \
-#{ group != user_group ? "-B /gpfs/projects/#{user_group}" : "" } \
--B #{scratch_group_dir} \
--B #{projects_group_dir} \
--B "$SINGULARITY_RUBY_INLINE":"$CONTAINER_DIR/.ruby_inline":rw  \
--B ~/git:"$CONTAINER_DIR/git":ro \
-#{Open.exists?('~/.rbbt/software/opt/')? '-B ~/.rbbt/software/opt/:"/opt/":ro' : '' } \
--B ~/.rbbt:"$CONTAINER_DIR/home/":ro \
-"$SINGULARITY_IMG")
-          exec_cmd << ' TMPDIR="$CONTAINER_DIR/.rbbt/tmp" '
-        else
-          singularity_exec += %( -B "$SINGULARITY_RUBY_INLINE":"$HOME/.ruby_inline":rw "$SINGULARITY_IMG" )
-        end
-
-        if development
-          exec_cmd += " rbbt --dev='#{development}'"
-        else
-          exec_cmd += ' rbbt'
-        end
-
-        exec_cmd = singularity_exec + " " + exec_cmd
-      else
-        if development
-          exec_cmd << " " << %(~/git/rbbt-util/bin/rbbt --dev=#{development})
-        else
-          exec_cmd << " " << 'rbbt'
-        end
-
-        if contain
-          rbbt_cmd << " " << %(--workdir_all='#{contain}')
-        end
-      end
-
-
-      cmd =<<-EOF
-#{exec_cmd} \\
-#{rbbt_cmd}
-EOF
-
-      run +=<<-EOF
-
-# Run command
-#{cmd}
-
-# Save exit status
-exit_status=$?
-
-EOF
-
-      # CODA
-      coda = ""
-      if sync
-        if singularity
-          coda +=<<-EOF
-singularity exec -e -C -H "$CONTAINER_DIR" "$SINGULARITY_IMG" rbbt system clean all -q &>> #{fsync}
-EOF
-        else
-          coda +=<<-EOF
-rbbt system clean all -q &>> #{fsync}
-EOF
-        end
-
-        if sync.include?("=>")
-          source, _sep, sync = sync.partition("=>")
-          source = source.strip
-          sync = sync.strip
-          source = File.join(File.expand_path(contain), source)
-        else
-          source = File.join(File.expand_path(contain), '.rbbt/var/jobs')
-        end
-
-        target = File.expand_path(sync)
-        coda +=<<-EOF
-
-# Sync data to target location
-mkdir -p "$(dirname '#{target}')"
-rsync -avztAXHP --copy-unsafe-links "#{source}/" "#{target}/" &>> #{fsync} 
-sync_es="$?" 
-find '#{target}' -type l -ls | awk '$13 ~ /^#{target.gsub('/','\/')}/ { sub("#{source}", "#{target}", $13); print $11, $13 }' | while read A B; do rm $A; ln -s $B $A; done
-EOF
-
-        if  contain && (wipe_container == "post" || wipe_container == "both")
-          prep =<<-EOF + prep
-if ls -A '#{contain}' &> /dev/null ; then
-    echo "ERROR: Container directory not empty, refusing to wipe. #{contain}" &>> #{fsync}
-fi
-          EOF
-          if singularity
-            coda +=<<-EOF
-singularity exec -e -C -H "$CONTAINER_DIR" "$SINGULARITY_IMG" rm -v /dev/shm/sem.*.{in,out,process} /dev/shm/sem.Session-PID.*.sem 2> /dev/null >> #{fsync}
-
-
-# Clean container directory
-#if [ $exit_status == '0' -a $sync_es == '0' ]; then 
-    singularity exec -e -C -H "$CONTAINER_DIR" "$SINGULARITY_IMG" rbbt system clean -f &>> #{fsync}
-    singularity exec -e -C -H "$CONTAINER_DIR" "$SINGULARITY_IMG" rm -Rfv .rbbt/var/jobs &>> #{fsync}
-    singularity exec -e -C -H "$CONTAINER_DIR" "$SINGULARITY_IMG" rm -Rfv tmp/ &>> #{fsync}
-#else
-#    echo "ERROR: Process failed or results could not sync correctly. Contain directory not purged" &>> #{fsync}
-#fi
-EOF
-          else
-            coda +=<<-EOF
-#{exec_cmd} system clean
-if [ $exit_status == '0' -a $sync_es == '0' ]; then 
-    rm -Rfv #{contain} &>> #{fsync}
-else
-    echo "ERROR: Process failed or results could not sync correctly. Contain directory not purged" &>> #{fsync}
-fi
-unset sync_es
-EOF
-
-          end
-        end
-      end
-      coda +=<<-EOF
-
-# Write exit status to file
-echo $exit_status > #{fexit}
-EOF
-      if sync
-        coda +=<<-EOF 
-if [ "$sync_es" == '0' ]; then 
-  unset sync_es
-  exit $exit_status
-else
-  exit $sync_es
-fi
-EOF
-      else
-        coda +=<<-EOF 
-exit $exit_status
-EOF
-      end
-
-      template = [header, env, prep, run, coda] * "\n"
-
-      template
-    end
-    
-    def self.issue_template(template, options = {})
-
-      slurm_basedir = options[:slurm_basedir]
-      Open.mkdir slurm_basedir
-
-      dry_run = options.delete :dry_run
-
-      fout  = File.join(slurm_basedir, 'std.out')
-      ferr  = File.join(slurm_basedir, 'std.err')
-      fjob  = File.join(slurm_basedir, 'job.id')
-      fexit = File.join(slurm_basedir, 'exit.status')
-      fsync = File.join(slurm_basedir, 'sync.log')
-      fcmd  = File.join(slurm_basedir, 'command.slurm')
-
-      job = nil
-      if options[:clean_job]
-        [fcmd, fjob, fout, ferr, fsync, fexit].each do |file|
-          Open.rm file if Open.exists? file
-        end
-      end
-
-      return if Open.exists?(fexit)
-
-      STDERR.puts Log.color(:magenta, "Issuing SLURM file: #{fcmd}")
-      STDERR.puts template
-
-      Open.write(fcmd, template) unless File.exists? fcmd
-      if File.exists?(fjob)
-        job = Open.read(fjob).to_i
-      else
-        if File.exists?(fout)
-          return
-        elsif dry_run
-          STDERR.puts Log.color(:magenta, "To execute run: ") + Log.color(:blue, "sbatch '#{slurm_basedir}/command.slurm'")
-          STDERR.puts Log.color(:magenta, "To monitor progress run (needs local rbbt): ") + Log.color(:blue, "rbbt mn --tail -w '#{slurm_basedir}'")
-          raise Marenostrum::SBATCH, slurm_basedir
-        else
-          Open.rm fsync
-          Open.rm fexit
-          Open.rm fout
-          Open.rm ferr
-          job = CMD.cmd("sbatch '#{fcmd}'").read.scan(/\d+/).first.to_i
-          Open.write(fjob, job.to_s)
-        end
-      end
-    end
-
-    def self.follow_job(slurm_basedir, tail = true)
-      fjob = File.join(slurm_basedir, 'job.id')
-      fout = File.join(slurm_basedir, 'std.out')
-      ferr = File.join(slurm_basedir, 'std.err')
-      fstatus = File.join(slurm_basedir, 'job.status')
-
-      job = Open.read(fjob).strip if Open.exists?(fjob)
-
-      if job
-        status_txt = CMD.cmd("squeue --job #{job}").read
-        STDERR.puts Log.color(:magenta, "Status [#{job.to_i}]:")
-        STDERR.puts status_txt
-        lines = status_txt.split("\n").length
-      end
-
-      if tail
-        Log.severity = 10
-        while ! File.exists? fout
-          if job
-            STDERR.puts
-            Log.clear_line(STDERR)
-            STDERR.write Log.color(:magenta, "Waiting for Output")
-            3.times do
-              STDERR.write Log.color(:magenta, ".")
-              sleep 1
-            end
-            status_txt = CMD.cmd("squeue --job #{job}").read
-            lines.times do
-              Log.clear_line(STDERR)
-            end
-            Log.clear_line(STDERR)
-            STDERR.puts Log.color(:magenta, "Status [#{job.to_i}]:")
-            STDERR.puts status_txt
-            lines = status_txt.split("\n").length
-          end
-        end
-        STDERR.puts
-        Log.clear_line(STDERR)
-        STDERR.puts Log.color(:magenta, "Output:")
-        begin
-          CMD.cmd("squeue --job #{job} > #{fstatus}")
-          out = CMD.cmd("tail -f '#{fout}'", :pipe => true) if File.exists?(fout) and not tail == :STDERR
-          err = CMD.cmd("tail -f '#{ferr}'", :pipe => true) if File.exists?(ferr)
-
-          terr = Misc.consume_stream(err, true, STDERR) if err
-          tout = Misc.consume_stream(out, true, STDOUT) if out
-
-          sleep 3 while CMD.cmd("squeue --job #{job}").read.include? job.to_s
-        rescue Aborted
-        ensure
-          begin
-            terr.exit if terr
-            tout.exit if tout
-            err.close if err
-            err.join if err
-          rescue Exception
-          end
-
-          begin
-            out.close if out
-            out.join if out
-          rescue Exception
-          end
-        end
-      end
-    end
-
-    def self.wait_for_job(slurm_basedir, time = 1)
-      fexit = File.join(slurm_basedir, 'exit.status')
-      fjob = File.join(slurm_basedir, 'job.id')
-      job = Open.read(fjob) if Open.exists?(fjob)
-
-
-      while ! Open.exists?(fexit)
-        sleep time
-      end
-    end
-
-    def self.run_job(job, options = {})
-      options = IndiferentHash.setup(options.dup)
-        
-      dry_run          = options.delete :dry_run
-      tail             = options.delete :tail
-
-      workflow = job.workflow
-      task = job.task_name
-
-      keep_slurm_basedir = options.delete :keep_SLURM_slurm_basedir 
-      slurm_basedir = options.delete :SLURM_basedir
-      slurm_basedir = "~/rbbt-slurm" if slurm_basedir.nil?
-      TmpFile.with_file(nil, !keep_slurm_basedir, :tmpdir => slurm_basedir, :prefix => "SLURM_rbbt_job-") do |tmp_directory|
-        options[:slurm_basedir] ||= tmp_directory
-        slurm_basedir = options[:slurm_basedir]
-        inputs_dir = File.join(tmp_directory, 'inputs_dir')
-        saved = Step.save_job_inputs(job, inputs_dir, options)
-        if saved
-          options[:inputs_dir] = inputs_dir
-          cmd = ['workflow', 'task', workflow.to_s, task.to_s, '-pf', '--load_inputs', inputs_dir, '--log', (options[:log] || Log.severity).to_s]
-        else
-          cmd = ['workflow', 'task', workflow.to_s, task.to_s, '-pf', '--log', (options[:log] || Log.severity).to_s]
-        end
-
-
-        template = self.template(cmd, options)
-        self.issue_template(template, options.merge(:slurm_basedir => slurm_basedir, :dry_run => dry_run))
-
-        return unless tail
-
-        t_monitor = Thread.new do
-          self.follow_job(slurm_basedir, :STDERR)
-        end
-        self.wait_for_job(slurm_basedir)
-        t_monitor.raise Aborted
-        return unless Open.read(File.join(slurm_basedir, 'exit.status')).strip == '0'
-        path = Open.read(File.join(slurm_basedir, 'std.out')).strip
-        if Open.exists?(path) && job.path != path
-          Log.info "Path of SLURM job #{path} is different from original job #{job.path}. Stablishing link."
-          Open.ln path, job.path
-          Open.ln path + '.info', job.path + '.info'  if Open.exists?(path + '.info')
-          Open.ln path + '.files', job.path + '.files' if Open.exists?(path + '.files')
-        end
-      end
-    end
-  end
-
-  def self.relay(job, options={})
-    options = Misc.add_defaults options, :target => 'mn1', :search_path => 'user'
-    done_deps = job.dependencies.select do |dep|
-      dep.done? 
-    end
-
-    error_deps = job.dependencies.select do |dep|
-      dep.error? && ! dep.recoverable_error?
-    end
-
-    (done_deps + error_deps).each do |dep|
-      Step.migrate(dep.path, options[:search_path], options)
-    end
-
-  end
-end
-
+require 'rbbt/hpc/slurm'