rbbt-util 5.31.13 → 5.32.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 42c6838da8c912f444ab59a74bb1cf96fdbe8d20aea2be712a035679cdec41e7
-  data.tar.gz: 77af7a4a5ec48d57637dc66967dca071bfb47563b520e80ba27eb9d6d7573407
+  metadata.gz: 612e1472bae3839b68f143d639025a9bb19294525ecc0e95df99702883af31d3
+  data.tar.gz: bea8c5ec09525363447c5bdc140f899f8a22ccb311d225da061b01d6f68bcf10
 SHA512:
-  metadata.gz: 39328d07cbc07c1be6beaa5149f94a94cac807fba36029415fef57644dc7bc63b90ee70f2e814271ae55033a1321bbb03dbb7d7531f7f033ec0df4401b1a716c
-  data.tar.gz: 9c0cbd95d049b5524072e31bf8f0b47884839bbf792ea8b0d6cd437a0f6aef5f6fa7efa5ef1288e77f1efe0bee50b360966e01c12284ea706d3cf876d0432512
+  metadata.gz: 0aef55d55b97f40011aeaa3f09450dd623ad6d682b4b3ef5dc3f2df603efeeede2e50b3be2f4f586e502adb6a4a0fd898113d3c5c892212170edb4cde7b7f6a7
+  data.tar.gz: 81261ef8e8da6ada5c1d8e672fdbd31ff6dc67a7a08edb4b6a3dcee10cfc6ed3f7abacf0c3628d604839e2743de043793c5169a147bd9b528c2afd614dca62b9

data/lib/rbbt/hpc/batch.rb

@@ -6,6 +6,36 @@ module HPC
     end
   end
 
+  def self.batch_system(batch_system = 'auto')
+    case batch_system.to_s.downcase
+    when 'slurm'
+      HPC::SLURM
+    when 'lsf'
+      HPC::LSF
+    when 'auto'
+      case $previous_commands.last
+      when 'slurm'
+        HPC::SLURM
+      when 'lsf'
+        HPC::LSF
+      else
+        case Rbbt::Config.get(:batch_system, :batch, :batch_system, :hpc, :HPC, :BATCH).to_s.downcase
+        when 'slurm'
+          HPC::SLURM
+        when 'lsf'
+          HPC::LSF
+        else
+          case ENV["BATCH_SYSTEM"].to_s.downcase
+          when 'slurm'
+            HPC::SLURM
+          when 'lsf'
+            HPC::LSF
+          end
+        end
+      end
+    end
+  end
+
   module TemplateGeneration
     def exec_cmd(job, options = {})
       env_cmd     = Misc.process_options options, :env_cmd
@@ -30,12 +60,12 @@ module HPC
 -B "/.singularity_ruby_inline":"#{contain}/.singularity_ruby_inline":rw 
 -B "#{options[:batch_dir]}" \
 -B /scratch/tmp \
-#{ group != user_group ? "-B /gpfs/projects/#{user_group}" : "" } \
+          #{ group != user_group ? "-B /gpfs/projects/#{user_group}" : "" } \
 -B #{scratch_group_dir} \
 -B #{projects_group_dir} \
 -B /apps/ \
 -B ~/git:"#{contain}/git":ro \
-#{Open.exists?('~/.rbbt/software/opt/')? '-B ~/.rbbt/software/opt/:"/opt/":ro' : '' } \
+          #{Open.exists?('~/.rbbt/software/opt/')? '-B ~/.rbbt/software/opt/:"/opt/":ro' : '' } \
 -B ~/.rbbt:"#{contain}/home/":ro)
         end
 
@@ -74,7 +104,7 @@ module HPC
           [name, dep.path] * "="  
         end * ","
 
-        options[:override_deps] = override_deps
+        options[:override_deps] = override_deps unless override_deps.empty?
       end
 
       # Save inputs into inputs_dir
@@ -216,6 +246,7 @@ EOF
         :fexit  => File.join(batch_dir, 'exit.status'),
         :fsync  => File.join(batch_dir, 'sync.log'),
         :fsexit  => File.join(batch_dir, 'sync.status'),
+        :fenv  => File.join(batch_dir, 'env.vars'),
         :fcmd   => File.join(batch_dir, 'command.batch')
 
       batch_options
@@ -441,6 +472,7 @@ exit $exit_status
 
 # #{Log.color :green, "1. Prepare environment"}
 #{prepare_environment}
+env > #{batch_options[:fenv]}
 
 # #{Log.color :green, "2. Execute"}
 #{execute} 

data/lib/rbbt/hpc/lsf.rb

@@ -7,10 +7,13 @@ module HPC
 
     def self.batch_system_variables
       <<-EOF
-[[ -z $LSB_MAX_MEM_RUSAGE ]] || MAX_MEMORY=$LSB_MAX_MEM_RUSAGE 
-[[ -z $MAX_MEMORY ]] && let MAX_MEMORY="$(grep MemTotal /proc/meminfo|grep -o "[[:digit:]]*") / 1024"
-BATCH_JOB_ID=$LSF_JOBID
-BATCH_SYSTEM=LSF
+let TOTAL_PROCESORS="$(cat /proc/cpuinfo|grep ^processor |wc -l)"
+let MAX_MEMORY_DEFAULT="$(grep MemTotal /proc/meminfo|grep -o "[[:digit:]]*") / ( (1024 * $TOTAL_PROCESORS) / $LSB_MAX_NUM_PROCESSORS )"
+[ ! -z $LSB_MAX_MEM_RUSAGE ] && let MAX_MEMORY="$LSB_MAX_MEM_RUSAGE" || MAX_MEMORY="$MAX_MEMORY_DEFAULT"
+export MAX_MEMORY_DEFAULT
+export MAX_MEMORY
+export BATCH_JOB_ID=$LSF_JOBID
+export BATCH_SYSTEM=LSF
       EOF
     end
 

data/lib/rbbt/hpc/orchestrate.rb

@@ -145,6 +145,8 @@ module HPC
       job_rules.delete :workflow
       
 
+      option_config_keys = options[:config_keys]
+
       job_options = IndiferentHash.setup(options.merge(job_rules).merge(:batch_dependencies => dep_ids))
       job_options.delete :orchestration_rules
 
@@ -154,6 +156,11 @@ module HPC
         job_options[:config_keys] = job_options[:config_keys] ? config_keys + "," + job_options[:config_keys] : config_keys
       end
 
+      if option_config_keys
+        option_config_keys = option_config_keys.gsub(/,\s+/,',') 
+        job_options[:config_keys] = job_options[:config_keys] ? job_options[:config_keys] + "," + option_config_keys : option_config_keys
+      end
+
       if options[:piggyback]
         manifest = options[:piggyback].uniq
         manifest += [job]
@@ -165,7 +172,7 @@ module HPC
         new_config_keys = self.job_rules(rules, job)[:config_keys]
         if new_config_keys
           new_config_keys = new_config_keys.gsub(/,\s+/,',') 
-          job_options[:config_keys] = job_options[:config_keys] ? new_config_keys + "," + job_options[:config_keys] : new_config_keys
+          job_options[:config_keys] = job_options[:config_keys] ? job_options[:config_keys] + "," + new_config_keys : new_config_keys
         end
 
         job_options.delete :piggyback

data/lib/rbbt/hpc/slurm.rb

@@ -8,9 +8,13 @@ module HPC
 
     def self.batch_system_variables
       <<-EOF
-let "MAX_MEMORY=$SLURM_MEM_PER_CPU * $SLURM_CPUS_PER_TASK" || let MAX_MEMORY="$(grep MemTotal /proc/meminfo|grep -o "[[:digit:]]*") / 1024"
-BATCH_JOB_ID=$SLURM_JOB_ID
-BATCH_SYSTEM=SLURM
+let TOTAL_PROCESORS="$(cat /proc/cpuinfo|grep ^processor |wc -l)"
+let MAX_MEMORY_DEFAULT="$(grep MemTotal /proc/meminfo|grep -o "[[:digit:]]*") / ( (1024 * $TOTAL_PROCESORS) / $SLURM_CPUS_PER_TASK )"
+[ ! -z $SLURM_MEM_PER_CPU ] && let MAX_MEMORY="$SLURM_MEM_PER_CPU * $SLURM_CPUS_PER_TASK" || MAX_MEMORY="$MAX_MEMORY_DEFAULT"
+export MAX_MEMORY_DEFAULT
+export MAX_MEMORY
+export BATCH_JOB_ID=$SLURM_JOB_ID
+export BATCH_SYSTEM=SLURM
       EOF
     end
 

data/lib/rbbt/persist/tsv/adapter.rb

@@ -132,7 +132,7 @@ module Persist
 
     def read_lock
       read if closed?
-      if read?
+      if read? || write?
         return yield
       end
 

data/lib/rbbt/tsv/csv.rb

@@ -8,6 +8,7 @@ module TSV
     noheaders = ! headers
 
     type = options.delete :type
+    cast = options.delete :cast
     merge = options.delete :merge
     key_field = options.delete :key_field
     fields = options.delete :fields
@@ -46,6 +47,10 @@ module TSV
       else
         key, *values = row
       end
+      
+      if cast
+        values = values.collect{|v| v.send cast }
+      end
 
       case type
       when :double, :flat

data/lib/rbbt/tsv/parser.rb

@@ -40,7 +40,7 @@ module TSV
       # Process fields line
 
       preamble << line if line
-      while line and (TrueClass === @header_hash or (String === @header_hash and Misc.fixutf8(line) =~ /^#{@header_hash}/ ))
+      while line && (TrueClass === @header_hash || (String === @header_hash && Misc.fixutf8(line) =~ /^#{@header_hash}/ ))
         @fields = line.split(@sep, -1)
         @key_field = @fields.shift
         @key_field = @key_field[(0 + header_hash.length)..-1] if String === @header_hash
@@ -49,7 +49,7 @@ module TSV
         line = (@header_hash != "" ?  stream.gets : nil)
         line = Misc.fixutf8 line.chomp if line
         preamble << line if line
-        @header_hash = false if TrueClass === @header_hash
+        @header_hash = false if TrueClass === @header_hash || @header_hash == ""
       end
 
       @preamble = preamble[0..-3] * "\n"

data/lib/rbbt/tsv/stream.rb

@@ -56,6 +56,7 @@ module TSV
       preambles     = []
 
       streams = streams.collect do |stream|
+
         parser = TSV::Parser.new stream, options.dup
         sfields = parser.fields
 

data/lib/rbbt/util/misc/pipes.rb

@@ -43,7 +43,7 @@ module Misc
       File.mkfifo path
       yield path
     ensure
-      FileUtils.rm path if erase
+      FileUtils.rm path if erase && File.exists?(path)
     end
   end
   

data/lib/rbbt/util/open.rb

@@ -11,7 +11,7 @@ module Open
   class OpenGzipError < StandardError; end
 
   REMOTE_CACHEDIR = File.join(ENV["HOME"], "/tmp/open_cache")
-  FileUtils.mkdir_p REMOTE_CACHEDIR unless File.exist? REMOTE_CACHEDIR
+  #FileUtils.mkdir_p REMOTE_CACHEDIR unless File.exist? REMOTE_CACHEDIR
 
   GREP_CMD = begin
                if ENV["GREP_CMD"] 

data/lib/rbbt/workflow.rb

@@ -187,17 +187,21 @@ module Workflow
       clean_name = wf_name.sub(/::.*/,'')  
       Log.info{"Looking for '#{wf_name}' in '#{clean_name}'"}
       require_workflow clean_name
-      return Misc.string2const Misc.camel_case(wf_name)
+      workflow = Misc.string2const Misc.camel_case(wf_name)
+      workflow.load_documentation
+      return workflow
     end
 
     Log.high{"Loading workflow #{wf_name}"}
     require_local_workflow(wf_name) or 
     (Workflow.autoinstall and `rbbt workflow install #{Misc.snake_case(wf_name)} || rbbt workflow install #{wf_name}` and require_local_workflow(wf_name)) or raise("Workflow not found or could not be loaded: #{ wf_name }")
-    begin
-      Misc.string2const Misc.camel_case(wf_name)
-    rescue
-      Workflow.workflows.last || true
-    end
+    workflow = begin
+                 Misc.string2const Misc.camel_case(wf_name)
+               rescue
+                 Workflow.workflows.last || true
+               end
+    workflow.load_documentation
+    workflow
   end
 
   attr_accessor :description

data/lib/rbbt/workflow/definition.rb

@@ -74,6 +74,7 @@ module Workflow
   def dep_task(name, workflow, oname, *rest, &block)
     dep(workflow, oname, *rest, &block) 
     extension workflow.tasks[oname].extension if workflow.tasks.include?(oname) unless @extension
+    returns workflow.tasks[oname].result_description if workflow.tasks.include?(oname) unless @result_description 
     task name do
       raise RbbtException, "dependency not found in dep_task" if dependencies.empty?
       dep = dependencies.last

data/lib/rbbt/workflow/doc.rb

@@ -45,10 +45,25 @@ module Workflow
   end
 
   def load_documentation
-    @documentation = Workflow.parse_workflow_doc documentation_markdown
+    return if @documentation
+    @documentation ||= Workflow.parse_workflow_doc documentation_markdown
     @documentation[:tasks].each do |task, description|
-      raise "Documentation for #{ task }, but not a #{ self.to_s } task" unless tasks.include? task.to_sym
-      tasks[task.to_sym].description = description
+      if task.include? "#"
+        workflow, task = task.split("#")
+        workflow = begin
+                     Kernel.const_get workflow
+                   rescue
+                     next
+                   end
+      else
+        workflow = self
+      end
+
+      if workflow.tasks.include? task.to_sym
+        workflow.tasks[task.to_sym].description = description
+      else
+        Log.low "Documentation for #{ task }, but not a #{ workflow.to_s } task" 
+      end
     end
   end
 

data/lib/rbbt/workflow/step.rb

@@ -458,6 +458,10 @@ class Step
   end
 
   def clean
+    if ! Open.exists?(info_file)
+      Log.high "Refusing to clean step with no .info file: #{path}"
+      return self
+    end
     status = []
     status << "dirty" if done? && dirty?
     status << "not running" if ! done? && ! running? 

data/lib/rbbt/workflow/step/accessor.rb

@@ -97,6 +97,7 @@ class Step
         Open.ln_s(value.path, path)
       when type.to_s == "file"
         if String === value && File.exists?(value)
+          value = File.expand_path(value)
           Open.ln_s(value, path)
         else
           value = value.collect{|v| v = "#{v}" if Path === v; v }if Array === value

data/lib/rbbt/workflow/util/archive.rb

@@ -1,6 +1,6 @@
 class Step
 
-  MAIN_RSYNC_ARGS="-avztAXHP"
+  MAIN_RSYNC_ARGS="-avztAXHP --copy-links"
   
   def self.link_job(path, target_dir, task = nil, workflow = nil)
     Path.setup(target_dir)

data/lib/rbbt/workflow/util/trace.rb

@@ -1,12 +1,8 @@
 require 'rbbt/util/R'
 
 module Workflow
-  def self.trace(seed_jobs, options = {})
-
-    jobs = []
-    seed_jobs.each{|j| jobs << j; jobs += j.rec_dependencies}
-
-    data = TSV.setup({}, "Job~Workflow,Task,Start,End#:type=:list")
+  def self.trace_job_times(jobs, fix_gap = false)
+    data = TSV.setup({}, "Job~Code,Workflow,Task,Start,End#:type=:list")
     min_start = nil
     max_done = nil
     jobs.each do |job|
@@ -14,10 +10,10 @@ module Workflow
       started = job.info[:started]
       ddone = job.info[:done]
 
-      code = [job.workflow, job.task_name].compact.collect{|s| s.to_s} * "."
-      code = code + '.' + job.name
+      code = [job.workflow, job.task_name].compact.collect{|s| s.to_s} * " · "
+      code = job.name + " - " + code
 
-      data[code] = [job.workflow.to_s, job.task_name, started, ddone]
+      data[job.path] = [code,job.workflow.to_s, job.task_name, started, ddone]
       if min_start.nil?
         min_start = started
       else
@@ -39,7 +35,7 @@ module Workflow
       value["End"] - min_start
     end
 
-    if options[:fix_gap]
+    if fix_gap
       ranges = []
       data.through do |k,values|
         start, eend = values.values_at "Start.second", "End.second"
@@ -67,115 +63,161 @@ module Workflow
         gap = Misc.sum(gaps.select{|pos,size| pos < values["Start.second"]}.collect{|pos,size| size})
         value - gap
       end
+
+      total_gaps = Misc.sum(gaps.collect{|k,v| v})
+      Log.info "Total gaps: #{total_gaps} seconds"
     end
 
+    start = data.column("Start.second").values.flatten.collect{|v| v.to_f}.min
+    eend = data.column("End.second").values.flatten.collect{|v| v.to_f}.max
+    total = eend - start unless eend.nil? || start.nil?
+    Log.info "Total time elapsed: #{total} seconds" if total
+
+    data
+  end
+
+  def self.plot_trace_job_times(data, plot, width=800, height=800)
+    data.R <<-EOF, [:svg]
+rbbt.require('tidyverse')
+rbbt.require('ggplot2')
+
+names(data) <- make.names(names(data))
+data$id = data$Code
+data$content = data$Task
+data$start = data$Start
+data$end = data$End
+data$Project = data$Workflow
+
+tasks = data
+
+#theme_gantt <- function(base_size=11, base_family="Source Sans Pro Light") {
+theme_gantt <- function(base_size=11, base_family="Sans Serif") {
+  ret <- theme_bw(base_size, base_family) %+replace%
+    theme(panel.background = element_rect(fill="#ffffff", colour=NA),
+          axis.title.x=element_text(vjust=-0.2), axis.title.y=element_text(vjust=1.5),
+          title=element_text(vjust=1.2, family="Source Sans Pro Semibold"),
+          panel.border = element_blank(), axis.line=element_blank(),
+          panel.grid.minor=element_blank(),
+          panel.grid.major.y = element_blank(),
+          panel.grid.major.x = element_line(size=0.5, colour="grey80"),
+          axis.ticks=element_blank(),
+          legend.position="bottom", 
+          axis.title=element_text(size=rel(1.2), family="Source Sans Pro Semibold"),
+          strip.text=element_text(size=rel(1.5), family="Source Sans Pro Semibold"),
+          strip.background=element_rect(fill="#ffffff", colour=NA),
+          panel.spacing.y=unit(1.5, "lines"),
+          legend.key = element_blank())
+
+  ret
+}
+
+tasks.long <- tasks %>%
+gather(date.type, task.date, -c(Code,Project, Task, id, Start.second, End.second)) %>%
+arrange(date.type, task.date) %>%
+mutate(id = factor(id, levels=rev(unique(id)), ordered=TRUE))
+
+x.breaks <- seq(length(tasks$Task) + 0.5 - 3, 0, by=-3)
+
+timeline <- ggplot(tasks.long, aes(y=id, yend=id, x=Start.second, xend=End.second, colour=Task)) + 
+  geom_segment() + 
+  geom_vline(xintercept=x.breaks, colour="grey80", linetype="dotted") + 
+  guides(colour=guide_legend(title=NULL)) +
+  labs(x=NULL, y=NULL) + 
+  theme_gantt() + theme(axis.text.x=element_text(angle=45, hjust=1))
+
+rbbt.png_plot('#{plot}', 'plot(timeline)', width=#{width}, height=#{height}, pointsize=6)
+    EOF
+  end
+
+  def self.trace_job_summary(jobs, report_keys = [])
     tasks_info = {}
 
+    report_keys = report_keys.collect{|k| k.to_s}
+
     jobs.each do |dep|
       next unless dep.info[:done]
       task = [dep.workflow, dep.task_name].compact.collect{|s| s.to_s} * "#"
-      info = tasks_info[task] ||= {}
+      info = tasks_info[task] ||= IndiferentHash.setup({})
+      dep_info = IndiferentHash.setup(dep.info)
 
-      time = dep.info[:done] - dep.info[:started]
+      time = dep_info[:done] - dep_info[:started]
       info[:time] ||= []
       info[:time] << time
 
-      cpus = nil
-      spark = false
-      shard = false
+      report_keys.each do |key|
+        info[key] = dep_info[key] 
+      end
+
       dep.info[:config_keys].select do |kinfo| 
         key, value, tokens = kinfo
-        key = key.to_s
-        cpus = value if key.include? 'cpu'
-        spark = value if key == 'spark'
-        shard = value if key == 'shard'
-      end
 
-      info[:cpus] = cpus || 1
-      info[:spark] = spark
-      info[:shard] = shard
+        info[key.to_s] = value if report_keys.include? key.to_s
+      end
     end
 
-    stats = TSV.setup({}, "Task~Calls,Avg. Time,Total Time,Cpus,Spark,Shard#:type=:list")
+    summary = TSV.setup({}, "Task~Calls,Avg. Time,Total Time#:type=:list")
 
     tasks_info.each do |task, info|
-      time_lists, cpus, spark, shard = info.values_at :time, :cpus, :spark, :shard
-      avg_time = Misc.mean(time_lists)
-      total_time = Misc.sum(time_lists)
+      time_lists = info[:time]
+      avg_time = Misc.mean(time_lists).to_i
+      total_time = Misc.sum(time_lists).to_i
       calls = time_lists.length
-      stats[task] = [calls, avg_time, total_time, cpus, spark, shard]
+      summary[task] = [calls, avg_time, total_time]
     end
 
-    raise "No jobs to process" if data.size == 0
+    report_keys.each do |key|
+      summary.add_field Misc.humanize(key) do |task|
+        tasks_info[task][key]
+      end
+    end if Array === report_keys && report_keys.any?
 
-    start = data.column("Start.second").values.flatten.collect{|v| v.to_f}.min
-    eend = data.column("End.second").values.flatten.collect{|v| v.to_f}.max
-    total = eend - start
-    Log.info "Total time elapsed: #{total} seconds"
+    summary
+  end
 
-    if options[:fix_gap]
-      total_gaps = Misc.sum(gaps.collect{|k,v| v})
-      Log.info "Total gaps: #{total_gaps} seconds"
-    end
+  def self.trace(seed_jobs, options = {})
+    jobs = []
+    seed_jobs.each do |step|
+      jobs += step.rec_dependencies + [step]
+      step.info[:archived_info].each do |path,ainfo|
+        archived_step = Step.new path
+
+        archived_step.define_singleton_method :info do
+          ainfo
+        end
+
+        #class << archived_step
+        #  self
+        #end.define_method :info do
+        #  ainfo
+        #end
+
+        jobs << archived_step
+      end if step.info[:archived_info]
 
-    plot, width, height = options.values_at :plot, :width, :height
-    if plot
-      data.R <<-EOF, [:svg]
-    rbbt.require('tidyverse')
-    rbbt.require('ggplot2')
-
-    names(data) <- make.names(names(data))
-    data$id = rownames(data)
-    data$content = data$Task
-    data$start = data$Start
-    data$end = data$End
-    data$Project = data$Workflow
-
-    tasks = data
-
-    #theme_gantt <- function(base_size=11, base_family="Source Sans Pro Light") {
-    theme_gantt <- function(base_size=11, base_family="Sans Serif") {
-      ret <- theme_bw(base_size, base_family) %+replace%
-        theme(panel.background = element_rect(fill="#ffffff", colour=NA),
-              axis.title.x=element_text(vjust=-0.2), axis.title.y=element_text(vjust=1.5),
-              title=element_text(vjust=1.2, family="Source Sans Pro Semibold"),
-              panel.border = element_blank(), axis.line=element_blank(),
-              panel.grid.minor=element_blank(),
-              panel.grid.major.y = element_blank(),
-              panel.grid.major.x = element_line(size=0.5, colour="grey80"),
-              axis.ticks=element_blank(),
-              legend.position="bottom", 
-              axis.title=element_text(size=rel(1.2), family="Source Sans Pro Semibold"),
-              strip.text=element_text(size=rel(1.5), family="Source Sans Pro Semibold"),
-              strip.background=element_rect(fill="#ffffff", colour=NA),
-              panel.spacing.y=unit(1.5, "lines"),
-              legend.key = element_blank())
-
-      ret
-    }
-
-    tasks.long <- tasks %>%
-    gather(date.type, task.date, -c(Project, Task, id, Start.second, End.second)) %>%
-    arrange(date.type, task.date) %>%
-    mutate(id = factor(id, levels=rev(unique(id)), ordered=TRUE))
-
-    x.breaks <- seq(length(tasks$Task) + 0.5 - 3, 0, by=-3)
-
-    timeline <- ggplot(tasks.long, aes(y=id, yend=id, x=Start.second, xend=End.second, colour=Task)) + 
-      geom_segment() + 
-      geom_vline(xintercept=x.breaks, colour="grey80", linetype="dotted") + 
-      guides(colour=guide_legend(title=NULL)) +
-      labs(x=NULL, y=NULL) + 
-      theme_gantt() + theme(axis.text.x=element_text(angle=45, hjust=1))
-
-    rbbt.png_plot('#{plot}', 'plot(timeline)', width=#{width}, height=#{height}, pointsize=6)
-      EOF
     end
 
+    jobs = jobs.uniq.sort_by{|job| t = job.info[:started]; t || Open.mtime(job.path) || Time.now }
+
+    data = trace_job_times(jobs, options[:fix_gap])
+
+    report_keys = options[:report_keys] || ""
+    report_keys = report_keys.split(/,\s*/) if String === report_keys
+    summary = trace_job_summary(jobs, report_keys)
+
+    raise "No jobs to process" if data.size == 0
+
+    plot, size, width, height = options.values_at :plot, :size, :width, :height
+
+    size = 800 if size.nil?
+    width = size.to_i * 2 if width.nil?
+    height = size  if height.nil?
+
+    plot_trace_job_times(data, plot, width, height) if plot
+
     if options[:plot_data]
       data
     else
-      stats
+      summary
     end
 
   end

data/share/rbbt_commands/hpc/clean

@@ -36,33 +36,7 @@ end
 batch_system = options.delete :batch_system
 batch_system ||= 'auto'
 
-HPC::BATCH_MODULE = case batch_system.to_s.downcase
-               when 'slurm'
-                 HPC::SLURM
-               when 'lsf'
-                 HPC::LSF
-               when 'auto'
-                 case $previous_commands.last
-                 when 'slurm'
-                   HPC::SLURM
-                 when 'lsf'
-                   HPC::LSF
-                 else
-                   case Rbbt::Config.get(:batch_system, :batch, :batch_system, :hpc, :HPC, :BATCH).to_s.downcase
-                   when 'slurm'
-                     HPC::SLURM
-                   when 'lsf'
-                     HPC::LSF
-                   else
-                     case ENV["BATCH_SYSTEM"].to_s.downcase
-                     when 'slurm'
-                       HPC::SLURM
-                     when 'lsf'
-                       HPC::LSF
-                     end
-                   end
-                 end
-               end
+HPC::BATCH_MODULE = HPC.batch_system batch_system
 
 raise ParameterException.new("Could not detect batch_system: #{Misc.fingerprint batch_system}") if HPC::BATCH_MODULE.nil?
 

data/share/rbbt_commands/hpc/list

@@ -40,33 +40,7 @@ end
 batch_system = options.delete :batch_system
 batch_system ||= 'auto'
 
-HPC::BATCH_MODULE = case batch_system.to_s.downcase
-               when 'slurm'
-                 HPC::SLURM
-               when 'lsf'
-                 HPC::LSF
-               when 'auto'
-                 case $previous_commands.last
-                 when 'slurm'
-                   HPC::SLURM
-                 when 'lsf'
-                   HPC::LSF
-                 else
-                   case Rbbt::Config.get(:batch_system, :batch, :batch_system, :hpc, :HPC, :BATCH).to_s.downcase
-                   when 'slurm'
-                     HPC::SLURM
-                   when 'lsf'
-                     HPC::LSF
-                   else
-                     case ENV["BATCH_SYSTEM"].to_s.downcase
-                     when 'slurm'
-                       HPC::SLURM
-                     when 'lsf'
-                       HPC::LSF
-                     end
-                   end
-                 end
-               end
+HPC::BATCH_MODULE = HPC.batch_system batch_system
 
 raise ParameterException.new("Could not detect batch_system: #{Misc.fingerprint batch_system}") if HPC::BATCH_MODULE.nil?
 
@@ -108,7 +82,7 @@ workdir.glob("**/command.batch").sort_by{|f| File.mtime(f)}.each do |fcmd|
   cmd = nil
  end
 
- if m = command_txt.match(/^BATCH_SYSTEM=(.*)/)
+ if m = command_txt.match(/^export BATCH_SYSTEM=(.*)/)
   job_batch_system = m[1].downcase
  else
   job_batch_system = nil
@@ -235,6 +209,8 @@ workdir.glob("**/command.batch").sort_by{|f| File.mtime(f)}.each do |fcmd|
      text = CMD.cmd('grep "^#SBATCH" |tail -n +5', :in => Open.read(fcmd)).read.strip
    when 'lsf'
      text = CMD.cmd('grep "^#BSUB" |tail -n +5', :in => Open.read(fcmd)).read.strip
+   else
+     text = ""
    end
    lines = text.split("\n").collect{|line| header, _sep, value = line.partition(/\s+/); Log.color(:yellow, header + ": ") + value}
    puts Log.color :yellow, lines * "\n"

data/share/rbbt_commands/hpc/orchestrate

@@ -34,33 +34,7 @@ EOF
 batch_system = $slurm_options.delete :batch_system
 batch_system ||= 'auto'
 
-HPC::BATCH_MODULE = case batch_system.to_s.downcase
-               when 'slurm'
-                 HPC::SLURM
-               when 'lsf'
-                 HPC::LSF
-               when 'auto'
-                 case $previous_commands.last
-                 when 'slurm'
-                   HPC::SLURM
-                 when 'lsf'
-                   HPC::LSF
-                 else
-                   case Rbbt::Config.get(:batch_system, :batch, :batch_system, :hpc, :HPC, :BATCH).to_s.downcase
-                   when 'slurm'
-                     HPC::SLURM
-                   when 'lsf'
-                     HPC::LSF
-                   else
-                     case ENV["BATCH_SYSTEM"].to_s.downcase
-                     when 'slurm'
-                       HPC::SLURM
-                     when 'lsf'
-                       HPC::LSF
-                     end
-                   end
-                 end
-               end
+HPC::BATCH_MODULE = HPC.batch_system batch_system
 
 raise ParameterException.new("Could not detect batch_system: #{Misc.fingerprint batch_system}") if HPC::BATCH_MODULE.nil?
 

data/share/rbbt_commands/hpc/tail

@@ -40,33 +40,7 @@ end
 batch_system = options.delete :batch_system
 batch_system ||= 'auto'
 
-HPC::BATCH_MODULE = case batch_system.to_s.downcase
-               when 'slurm'
-                 HPC::SLURM
-               when 'lsf'
-                 HPC::LSF
-               when 'auto'
-                 case $previous_commands.last
-                 when 'slurm'
-                   HPC::SLURM
-                 when 'lsf'
-                   HPC::LSF
-                 else
-                   case Rbbt::Config.get(:batch_system, :batch, :batch_system, :hpc, :HPC, :BATCH).to_s.downcase
-                   when 'slurm'
-                     HPC::SLURM
-                   when 'lsf'
-                     HPC::LSF
-                   else
-                     case ENV["BATCH_SYSTEM"].to_s.downcase
-                     when 'slurm'
-                       HPC::SLURM
-                     when 'lsf'
-                       HPC::LSF
-                     end
-                   end
-                 end
-               end
+HPC::BATCH_MODULE = HPC.batch_system batch_system
 
 raise ParameterException.new("Could not detect batch_system: #{Misc.fingerprint batch_system}") if HPC::BATCH_MODULE.nil?
 

data/share/rbbt_commands/hpc/task

@@ -33,33 +33,7 @@ EOF
 batch_system = $slurm_options.delete :batch_system
 batch_system ||= 'auto'
 
-HPC::BATCH_MODULE = case batch_system.to_s.downcase
-               when 'slurm'
-                 HPC::SLURM
-               when 'lsf'
-                 HPC::LSF
-               when 'auto'
-                 case $previous_commands.last
-                 when 'slurm'
-                   HPC::SLURM
-                 when 'lsf'
-                   HPC::LSF
-                 else
-                   case Rbbt::Config.get(:batch_system, :batch, :batch_system, :hpc, :HPC, :BATCH).to_s.downcase
-                   when 'slurm'
-                     HPC::SLURM
-                   when 'lsf'
-                     HPC::LSF
-                   else
-                     case ENV["BATCH_SYSTEM"].to_s.downcase
-                     when 'slurm'
-                       HPC::SLURM
-                     when 'lsf'
-                       HPC::LSF
-                     end
-                   end
-                 end
-               end
+HPC::BATCH_MODULE = HPC.batch_system batch_system
 
 raise ParameterException.new("Could not detect batch_system: #{Misc.fingerprint batch_system}") if HPC::BATCH_MODULE.nil?
 

data/share/rbbt_commands/workflow/forget_deps

@@ -8,9 +8,11 @@ $0 = "rbbt #{$previous_commands*" "} #{ File.basename(__FILE__) }" if $previous_
 
 options = SOPT.setup <<EOF
 
-Make a job forget all its dependencies and archive their meta-data
+Make a job forget its dependencies and archive their meta-data
 
-$ #{$0} [options] <job_path>
+$ #{$0} [options] <job_path> [<task_name>|<workflow>#<task_name>] ...
+
+Specific dependencies can be specified as . Otherwise, all are forgoten.
 
 -h--help Print this help
 -p--purge Purge dependencies
@@ -26,7 +28,7 @@ if options[:help]
   exit 0
 end
 
-path = ARGV[0]
+path, *remove = ARGV
 
 raise ParameterException, "No path given" if path.nil?
 step = Workflow.load_step path
@@ -34,10 +36,20 @@ step = Workflow.load_step path
 step.archive_deps
 step.copy_files_dir
 dependencies = step.dependencies
-step.set_info :dependencies, []
 
-if options[:purge]
+if remove && remove.any?
+ remove_paths = dependencies.select do |dep|
+  remove.include?(dep.task_name) || remove.include?([dep.workflow.to_s, dep.task_name] * "#")
+ end.collect{|dep| dep.path } 
+else
+ remove_paths = dependencies.collect{|dep| dep.path }
+end
+
+step.set_info :dependencies, step.info[:dependencies].reject{|info| remove_paths.include? info.last}
+
+if options[:purge] || options[:recursive_purge]
   dependencies.each do |dependency|
+    next unless remove_paths.include? dependency.path
     Step.purge(dependency.path, options[:recursive_purge])
   end
 end

data/share/rbbt_commands/workflow/trace

@@ -1,6 +1,7 @@
 #!/usr/bin/env ruby
 
 require 'rbbt/workflow'
+require 'rbbt/workflow/util/trace'
 
 require 'rbbt-util'
 require 'fileutils'
@@ -16,30 +17,26 @@ require 'rbbt/util/R'
 $0 = "rbbt #{$previous_commands*""} #{ File.basename(__FILE__) }" if $previous_commands
 
 options = SOPT.setup <<EOF
-Examine the provenance of a job result
+Examine the execution trace of a job or set of jobs
 
 $ rbbt workflow trace <job-result>
 
 -h--help Help
+-fg--fix_gap Remove execution gaps
+-rk--report_keys* Config keys and info fields to report
+-p--plot* Plot file
 -w--width* Image Width
 -h--height* Image Height
--p--plot* Plot file
 -s--size* Image Size (Height and Width)
--fg--fix_gap Remove execution gaps
 -pd--plot_data Print plot data
 EOF
 
 SOPT.usage if options[:help]
 
+
 files = ARGV
 plot = options[:plot]
 
-width, height, size = options.values_at :width, :height, :size
-
-size = 800 if size.nil?
-width = size if width.nil?
-height = size if height.nil?
-
 def get_step(file)
   file = File.expand_path(file)
   file = file.sub(/\.(info|files)/,'')
@@ -47,191 +44,8 @@ def get_step(file)
 end
 
 jobs = []
-files.each do |file|
-  step = get_step file
-
-  jobs += step.rec_dependencies + [step]
-
-  step.info[:archived_info].each do |path,ainfo|
-    archived_step = Step.new path
-    class << archived_step
-      self
-    end.define_method :info do
-      ainfo
-    end
-    jobs << archived_step
-  end if step.info[:archived_info]
-end
-
-jobs = jobs.select{|job| job.info[:done]}.sort_by{|job| job.info[:started]}
-
-data = TSV.setup({}, "Job~Workflow,Task,Start,End#:type=:list")
-min_start = nil
-max_done = nil
-jobs.each do |job|
-  next unless job.info[:done]
-  started = job.info[:started]
-  ddone = job.info[:done]
-
-  code = [job.workflow, job.task_name].compact.collect{|s| s.to_s} * "."
-  code = code + '.' + job.name
-
-  data[code] = [job.workflow.to_s, job.task_name, started, ddone]
-  if min_start.nil?
-    min_start = started
-  else
-    min_start = started if started < min_start
-  end
-
-  if max_done.nil?
-    max_done = ddone
-  else
-    max_done = ddone if ddone > max_done
-  end
-end
-
-data.add_field "Start.second" do |k,value|
-  value["Start"] - min_start
-end
-
-data.add_field "End.second" do |k,value|
-  value["End"] - min_start
-end
-
-if options[:fix_gap]
-  ranges = []
-  data.through do |k,values|
-    start, eend = values.values_at "Start.second", "End.second"
-
-    ranges << (start..eend)
-  end
-
-  gaps = {}
-  last = nil
-  Misc.collapse_ranges(ranges).each do |range|
-    start = range.begin
-    eend = range.end
-    if last
-      gaps[last] = start - last
-    end
-    last = eend
-  end
-
-  data.process "End.second" do |value,k,values|
-    gap = Misc.sum(gaps.select{|pos,size| pos < values["Start.second"]}.collect{|pos,size| size})
-    value - gap
-  end
-
-  data.process "Start.second" do |value,k,values|
-    gap = Misc.sum(gaps.select{|pos,size| pos < values["Start.second"]}.collect{|pos,size| size})
-    value - gap
-  end
+jobs = files.collect do |file|
+  get_step file
 end
 
-tasks_info = {}
-
-jobs.each do |dep|
-  next unless dep.info[:done]
-  task = [dep.workflow, dep.task_name].compact.collect{|s| s.to_s} * "#"
-  info = tasks_info[task] ||= {}
-
-  time = dep.info[:done] - dep.info[:started]
-  info[:time] ||= []
-  info[:time] << time
-
-  cpus = nil
-  spark = false
-  shard = false
-  dep.info[:config_keys].select do |kinfo| 
-    key, value, tokens = kinfo
-    key = key.to_s
-    cpus = value if key.include? 'cpu'
-    spark = value if key == 'spark'
-    shard = value if key == 'shard'
-  end
-
-  info[:cpus] = cpus || 1
-  info[:spark] = spark
-  info[:shard] = shard
-end
-
-stats = TSV.setup({}, "Task~Calls,Avg. Time,Total Time,Cpus,Spark,Shard#:type=:list")
-
-tasks_info.each do |task, info|
-  time_lists, cpus, spark, shard = info.values_at :time, :cpus, :spark, :shard
-  avg_time = Misc.mean(time_lists).to_i
-  total_time = Misc.sum(time_lists)
-  calls = time_lists.length
-  stats[task] = [calls, avg_time, total_time, cpus, spark, shard]
-end
-
-raise "No jobs to process" if data.size == 0
-
-start = data.column("Start.second").values.flatten.collect{|v| v.to_i}.min
-eend = data.column("End.second").values.flatten.collect{|v| v.to_i}.max
-total = eend - start
-Log.info "Total time elapsed: #{total} seconds"
-
-if options[:fix_gap]
-  total_gaps = Misc.sum(gaps.collect{|k,v| v})
-  Log.info "Total gaps: #{total_gaps} seconds"
-end
-
-if options[:plot_data]
-  puts data.to_s
-else
-  puts stats.to_s
-end
-
-if plot
-  data.R <<-EOF, [:svg]
-    rbbt.require('tidyverse')
-    rbbt.require('ggplot2')
-
-    names(data) <- make.names(names(data))
-    data$id = rownames(data)
-    data$content = data$Task
-    data$start = data$Start
-    data$end = data$End
-    data$Project = data$Workflow
-
-    tasks = data
-
-    #theme_gantt <- function(base_size=11, base_family="Source Sans Pro Light") {
-    theme_gantt <- function(base_size=11, base_family="Sans Serif") {
-      ret <- theme_bw(base_size, base_family) %+replace%
-        theme(panel.background = element_rect(fill="#ffffff", colour=NA),
-              axis.title.x=element_text(vjust=-0.2), axis.title.y=element_text(vjust=1.5),
-              title=element_text(vjust=1.2, family="Source Sans Pro Semibold"),
-              panel.border = element_blank(), axis.line=element_blank(),
-              panel.grid.minor=element_blank(),
-              panel.grid.major.y = element_blank(),
-              panel.grid.major.x = element_line(size=0.5, colour="grey80"),
-              axis.ticks=element_blank(),
-              legend.position="bottom", 
-              axis.title=element_text(size=rel(1.2), family="Source Sans Pro Semibold"),
-              strip.text=element_text(size=rel(1.5), family="Source Sans Pro Semibold"),
-              strip.background=element_rect(fill="#ffffff", colour=NA),
-              panel.spacing.y=unit(1.5, "lines"),
-              legend.key = element_blank())
-
-      ret
-    }
-
-    tasks.long <- tasks %>%
-    gather(date.type, task.date, -c(Project, Task, id, Start.second, End.second)) %>%
-    arrange(date.type, task.date) %>%
-    mutate(id = factor(id, levels=rev(unique(id)), ordered=TRUE))
-
-    x.breaks <- seq(length(tasks$Task) + 0.5 - 3, 0, by=-3)
-
-    timeline <- ggplot(tasks.long, aes(y=id, yend=id, x=Start.second, xend=End.second, colour=Task)) + 
-      geom_segment() + 
-      geom_vline(xintercept=x.breaks, colour="grey80", linetype="dotted") + 
-      guides(colour=guide_legend(title=NULL)) +
-      labs(x=NULL, y=NULL) + 
-      theme_gantt() + theme(axis.text.x=element_text(angle=45, hjust=1))
-
-    rbbt.png_plot('#{plot}', 'plot(timeline)', width=#{width}, height=#{height}, pointsize=6)
-  EOF
-end
+puts Workflow.trace(jobs, options)

data/share/rbbt_commands/workflow/write_info

@@ -46,7 +46,21 @@ pid = step.info[:pid]
 host = step.info[:pid_hostname]
 
 step.rec_dependencies.each do |dep|
-  dep.set_info key, value if (force || ! dep.info.include?(key)) && (!check_pid || dep.info[:pid].to_s == pid and dep.info[:pid_hostname] == host)
-rescue
-  Log.warn "Could no set info #{key} for #{dep.path}: #{$!.message}"
+  begin
+    dep.set_info key, value if (force || ! dep.info.include?(key)) && (!check_pid || dep.info[:pid].to_s == pid and dep.info[:pid_hostname] == host)
+  rescue
+    Log.warn "Could no set info #{key} for #{dep.path}: #{$!.message}"
+  end
 end if recursive
+
+if recursive && step.info[:archived_info]
+  ad = step.info[:archived_info]
+  ad.each do |d,info|
+    begin
+      info[key] = value if (force || ! info.include?(key)) && (!check_pid || info[:pid].to_s == pid and info[:pid_hostname] == host)
+    rescue
+      Log.warn "Could no set info #{key} for archived_dep #{info[:path]}: #{$!.message}"
+    end 
+  end
+  step.set_info :archived_info, ad
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-util
 version: !ruby/object:Gem::Version
-  version: 5.31.13
+  version: 5.32.2
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2021-04-08 00:00:00.000000000 Z
+date: 2021-04-26 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake