Merge branch 'master' into 2257-inequality-conditions
[arvados.git] / apps / workbench / app / helpers / provenance_helper.rb
index de3401d22e3507ba8c6ad6f3e8c6401a8fbfd881..66754d20b2f209248270d7ca73cedb0b5e758554 100644 (file)
 module ProvenanceHelper
-  def self.describe_node(pdata, uuid)
-    rsc = ArvadosBase::resource_class_for_uuid uuid.to_s
-    if rsc
-      href = "/#{rsc.to_s.underscore.pluralize rsc}/#{uuid}"
-
-      #"\"#{uuid}\" [label=\"#{rsc}\\n#{uuid}\",href=\"#{href}\"];\n"
-      if rsc == Collection
-        if pdata[uuid] 
-          #puts pdata[uuid]
-          if pdata[uuid][:name]
-            return "\"#{uuid}\" [label=\"#{pdata[uuid][:name]}\",href=\"#{href}\",shape=oval];\n"
-          else
-            i = 0
-            label = ""
-            while i < 3 and i < pdata[uuid].files.length
-              label += "\\n" unless label == ""
-              label += pdata[uuid].files[i][1]
-              i += 1
-            end
-            if i < pdata[uuid].files.length
-              label += "\\n&vellip;"
-            end
-            return "\"#{uuid}\" [label=\"#{label}\",href=\"#{href}\",shape=oval];\n"
-          end  
+
+  class GenerateGraph
+    def initialize(pdata, opts)
+      @pdata = pdata
+      @opts = opts
+      @visited = {}
+      @jobs = {}
+    end
+    
+    def self.collection_uuid(uuid)
+      m = CollectionsHelper.match(uuid)
+      if m
+        if m[2]
+          return m[1]+m[2]
+        else
+          return m[1]
         end
-        return "\"#{uuid}\" [label=\"#{rsc}\",href=\"#{href}\"];\n"
+        #  Collection.where(uuid: ['contains', m[1]]).each do |u|
+        #    puts "fixup #{uuid} to #{u.uuid}"
+        #    return u.uuid
+        #  end
+        #end
+      else
+        nil
       end
     end
-    return ""
-  end
 
-  def self.job_uuid(job)
-    # "#{job[:script]}\\n#{job[:script_version]}"
-    "#{job[:script]}"
-  end
+    def url_for u
+      p = { :host => @opts[:request].host, 
+        :port => @opts[:request].port,
+        :protocol => @opts[:request].protocol }
+      p.merge! u
+      Rails.application.routes.url_helpers.url_for (p)      
+    end 
 
-  def self.collection_uuid(uuid)
-    m = /([a-f0-9]{32}(\+[0-9]+)?)(\+.*)?/.match(uuid.to_s)
-    if m
-      m[1]
-    else
-      nil
+    def determine_fillcolor(n)
+      fillcolor = %w(aaaaaa aaffaa aaaaff aaaaaa ffaaaa)[n || 0] || 'aaaaaa'
+      "style=filled,fillcolor=\"##{fillcolor}\""
     end
-  end
 
-  def self.edge(tail, head, extra, opts)
-    if opts[:direction] == :bottom_up
-      gr = "\"#{tail}\" -> \"#{head}\""
-    else
-      gr = "\"#{head}\" -> \"#{tail}\""
+    def describe_node(uuid)
+      uuid = uuid.to_sym
+      bgcolor = determine_fillcolor @opts[:pips][uuid] if @opts[:pips]
+
+      rsc = ArvadosBase::resource_class_for_uuid uuid.to_s
+      if rsc
+        href = url_for ({:controller => rsc.to_s.tableize, 
+                          :action => :show, 
+                          :id => uuid.to_s })
+      
+        #"\"#{uuid}\" [label=\"#{rsc}\\n#{uuid}\",href=\"#{href}\"];\n"
+        if rsc == Collection
+          #puts uuid
+          if uuid == :"d41d8cd98f00b204e9800998ecf8427e+0"
+            # special case
+            #puts "empty!"
+            return "\"#{uuid}\" [label=\"(empty collection)\"];\n"
+          end
+          puts "#{uuid.class} #{@pdata[uuid]}"
+          if @pdata[uuid] 
+            #puts @pdata[uuid]
+            if @pdata[uuid][:name]
+              return "\"#{uuid}\" [label=\"#{@pdata[uuid][:name]}\",href=\"#{href}\",shape=oval,#{bgcolor}];\n"
+            else              
+              files = nil
+              if @pdata[uuid].respond_to? :files
+                files = @pdata[uuid].files
+              elsif @pdata[uuid][:files]
+                files = @pdata[uuid][:files]
+              end
+              
+              if files
+                i = 0
+                label = ""
+                while i < 3 and i < files.length
+                  label += "\\n" unless label == ""
+                  label += files[i][1]
+                  i += 1
+                end
+                if i < files.length
+                  label += "\\n&vellip;"
+                end
+                #puts "#{uuid} #{label} #{files}"
+                return "\"#{uuid}\" [label=\"#{label}\",href=\"#{href}\",shape=oval,#{bgcolor}];\n"
+              end
+            end  
+          end
+        end
+        return "\"#{uuid}\" [label=\"#{rsc}\",href=\"#{href}\",#{bgcolor}];\n"
+      end
+      "\"#{uuid}\" [#{bgcolor}];\n"
+    end
+
+    def job_uuid(job)
+      if @opts[:combine_jobs] == :script_only
+        uuid = "#{job[:script]}"
+      elsif @opts[:combine_jobs] == :script_and_version
+        uuid = "#{job[:script]}_#{job[:script_version]}"
+      else
+        uuid = "#{job[:uuid]}"
+      end
+
+      @jobs[uuid] = [] unless @jobs[uuid]
+      @jobs[uuid] << job unless @jobs[uuid].include? job
+
+      uuid
     end
-    if extra.length > 0
-      gr += "["
-      extra.each do |k, v|
-        gr += "#{k}=\"#{v}\","
+
+    def edge(tail, head, extra)
+      if @opts[:direction] == :bottom_up
+        gr = "\"#{tail}\" -> \"#{head}\""
+      else
+        gr = "\"#{head}\" -> \"#{tail}\""
+      end
+      if extra.length > 0
+        gr += " ["
+        extra.each do |k, v|
+          gr += "#{k}=\"#{v}\","
+        end
+        gr += "]"
       end
-      gr += "]"
+      gr += ";\n"
+      gr
     end
-    gr += ";\n"
-    gr
-  end
 
-  def self.script_param_edges(pdata, visited, job, prefix, sp, opts)
-    gr = ""
-    if sp and not sp.empty?
+    def script_param_edges(job, prefix, sp)
+      gr = ""
       case sp
       when Hash
         sp.each do |k, v|
           if prefix.size > 0
             k = prefix + "::" + k.to_s
           end
-          gr += ProvenanceHelper::script_param_edges(pdata, visited, job, k.to_s, v, opts)
+          gr += script_param_edges(job, k.to_s, v)
         end
       when Array
         i = 0
         node = ""
         sp.each do |v|
-          if collection_uuid(v)
-            gr += ProvenanceHelper::script_param_edges(pdata, visited, job, "#{prefix}[#{i}]", v, opts)
-          else
+          if GenerateGraph::collection_uuid(v)
+            gr += script_param_edges(job, "#{prefix}[#{i}]", v)
+          elsif @opts[:all_script_parameters]
             node += "', '" unless node == ""
             node = "['" if node == ""
             node += "#{v}"
@@ -88,113 +150,172 @@ module ProvenanceHelper
         unless node == ""
           node += "']"
           #puts node
-          id = "#{job[:uuid]}_#{prefix}"
-          gr += "\"#{id}\" [label=\"#{node}\"];\n"
-          gr += edge(job_uuid(job), id, {:label => prefix}, opts)        
+          #id = "#{job[:uuid]}_#{prefix}"
+          gr += "\"#{node}\" [label=\"#{node}\"];\n"
+          gr += edge(job_uuid(job), node, {:label => prefix})        
         end
-      else
-        m = collection_uuid(sp)
-        if m
-          gr += edge(job_uuid(job), m, {:label => prefix}, opts)
-          gr += ProvenanceHelper::generate_provenance_edges(pdata, visited, m, opts)
-        elsif opts[:all_script_parameters]
-          id = "#{job[:uuid]}_#{prefix}"
-          gr += "\"#{id}\" [label=\"#{sp}\"];\n"
-          gr += edge(job_uuid(job), id, {:label => prefix}, opts)
+      when String
+        return '' if sp.empty?
+        m = GenerateGraph::collection_uuid(sp)
+        #puts "#{m} pdata is #{@pdata[m.intern]}"
+        if m and (@pdata[m.intern] or (not @opts[:pdata_only]))
+          gr += edge(job_uuid(job), m, {:label => prefix})
+          gr += generate_provenance_edges(m)
+        elsif @opts[:all_script_parameters]
+          #id = "#{job[:uuid]}_#{prefix}"
+          gr += "\"#{sp}\" [label=\"#{sp}\"];\n"
+          gr += edge(job_uuid(job), sp, {:label => prefix})
         end
       end
+      gr
     end
-    gr
-  end
 
-  def self.generate_provenance_edges(pdata, visited, uuid, opts)
-    gr = ""
-    m = ProvenanceHelper::collection_uuid(uuid)
-    uuid = m if m
+    def generate_provenance_edges(uuid)
+      gr = ""
+      m = GenerateGraph::collection_uuid(uuid)
+      uuid = m if m
 
-    uuid = uuid.intern if uuid
+      uuid = uuid.intern if uuid
 
-    if (not uuid) or uuid.empty? or visited[uuid]
+      if (not uuid) or uuid.empty? or @visited[uuid]
 
-      #puts "already visited #{uuid}"
-      return ""
-    end
+        #puts "already @visited #{uuid}"
+        return ""
+      end
 
-    if not pdata[uuid] then 
-      return ProvenanceHelper::describe_node(pdata, uuid)
-    else
-      visited[uuid] = true
-    end
+      if not @pdata[uuid] then 
+        return describe_node(uuid)
+      else
+        @visited[uuid] = true
+      end
+
+      #puts "visiting #{uuid}"
 
-    #puts "visiting #{uuid}"
+      if m  
+        # uuid is a collection
+        gr += describe_node(uuid)
 
-    if m  
-      # uuid is a collection
-      gr += ProvenanceHelper::describe_node(pdata, uuid)
+        if m == :"d41d8cd98f00b204e9800998ecf8427e+0"
+          # empty collection, don't follow any further
+          return gr
+        end
 
-      pdata.each do |k, job|
-        if job[:output] == uuid.to_s
-          gr += self.edge(uuid, job_uuid(job), {:label => "output"}, opts)
-          gr += ProvenanceHelper::generate_provenance_edges(pdata, visited, job[:uuid], opts)
+        @pdata.each do |k, job|
+          if job[:output] == uuid.to_s
+            gr += edge(uuid, job_uuid(job), {:label => "output"})
+            gr += generate_provenance_edges(job[:uuid])
+          end
+          if job[:log] == uuid.to_s
+            gr += edge(uuid, job_uuid(job), {:label => "log"})
+            gr += generate_provenance_edges(job[:uuid])
+          end
         end
-        if job[:log] == uuid.to_s
-          gr += edge(uuid, job_uuid(job), {:label => "log"}, opts)
-          gr += ProvenanceHelper::generate_provenance_edges(pdata, visited, job[:uuid], opts)
+      else
+        # uuid is something else
+        rsc = ArvadosBase::resource_class_for_uuid uuid.to_s
+
+        if rsc == Job
+          job = @pdata[uuid]
+          if job
+            gr += script_param_edges(job, "", job[:script_parameters])
+
+            if @opts[:script_version_nodes]
+              gr += describe_node(job[:script_version])
+              gr += edge(job_uuid(job), job[:script_version], {:label => "script_version"})
+            end
+          end
+        elsif rsc == Link
+          # do nothing
+        else
+          gr += describe_node(uuid)
         end
       end
-    else
-      # uuid is something else
-      rsc = ArvadosBase::resource_class_for_uuid uuid.to_s
 
-      if rsc == Job
-        job = pdata[uuid]
-        if job
-          gr += ProvenanceHelper::script_param_edges(pdata, visited, job, "", job[:script_parameters], opts)
+      @pdata.each do |k, link|
+        if link[:head_uuid] == uuid.to_s and link[:link_class] == "provenance"
+          href = url_for ({:controller => Link.to_s.tableize, 
+                            :action => :show, 
+                            :id => link[:uuid] })
+
+          gr += describe_node(link[:tail_uuid])
+          gr += edge(link[:head_uuid], link[:tail_uuid], {:label => link[:name], :href => href}) 
+          gr += generate_provenance_edges(link[:tail_uuid])
         end
-      else
-        gr += ProvenanceHelper::describe_node(pdata, uuid)
       end
+
+      #puts "finished #{uuid}"
+
+      gr
     end
 
-    pdata.each do |k, link|
-      if link[:head_uuid] == uuid.to_s and link[:link_class] == "provenance"
-        gr += ProvenanceHelper::describe_node(pdata, link[:tail_uuid])
-        gr += edge(link[:head_uuid], link[:tail_uuid], {:label => link[:name], :href => "/links/#{link[:uuid]}"}, opts) 
-        gr += ProvenanceHelper::generate_provenance_edges(pdata, visited, link[:tail_uuid], opts)
+    def describe_jobs
+      gr = ""
+      @jobs.each do |k, v|
+        href = url_for ({:controller => Job.to_s.tableize, 
+                          :action => :index })
+
+        gr += "\"#{k}\" [href=\"#{href}?"
+        
+        n = 0
+        v.each do |u|
+          gr += "uuid%5b%5d=#{u[:uuid]}&"
+          n |= @opts[:pips][u[:uuid].intern] if @opts[:pips] and @opts[:pips][u[:uuid].intern]
+        end
+
+        gr += "\",label=\""
+        
+        if @opts[:combine_jobs] == :script_only
+          gr += "#{v[0][:script]}"
+        elsif @opts[:combine_jobs] == :script_and_version
+          gr += "#{v[0][:script]}" # Just show the name but the nodes will be distinct
+        else
+          gr += "#{v[0][:script]}\\n#{v[0][:finished_at]}"
+        end
+        gr += "\",#{determine_fillcolor n}];\n"
       end
+      gr
     end
 
-    #puts "finished #{uuid}"
-
-    gr
   end
 
-  def self.create_provenance_graph(pdata, uuid, opts={})
-    require 'open3'
+  def self.create_provenance_graph(pdata, svgId, opts={})
+    if pdata.is_a? Array or pdata.is_a? ArvadosResourceList
+      p2 = {}
+      pdata.each do |k|
+        p2[k[:uuid].intern] = k if k[:uuid]
+      end
+      pdata = p2
+    end
+
+    unless pdata.is_a? Hash
+      raise "create_provenance_graph accepts Array or Hash for pdata only, pdata is #{pdata.class}"
+    end
     
     gr = """strict digraph {
-node [fontsize=8,shape=box];
-edge [fontsize=8];"""
+node [fontsize=10,shape=box];
+edge [fontsize=10];
+"""
 
     if opts[:direction] == :bottom_up
       gr += "edge [dir=back];"
     end
 
-    #puts "pdata is #{pdata}"
+    #puts "@pdata is #{pdata}"
 
-    visited = {}
-    if uuid.respond_to? :each
-      uuid.each do |u|
-        gr += ProvenanceHelper::generate_provenance_edges(pdata, visited, u, opts)
-      end
-    else
-      gr += ProvenanceHelper::generate_provenance_edges(pdata, visited, uuid, opts)
+    g = GenerateGraph.new(pdata, opts)
+
+    pdata.each do |k, v|
+      gr += g.generate_provenance_edges(k)
     end
 
+    gr += g.describe_jobs
+
     gr += "}"
     svg = ""
 
-    #puts gr
+    puts gr
+    
+    require 'open3'
 
     Open3.popen2("dot", "-Tsvg") do |stdin, stdout, wait_thr|
       stdin.print(gr)
@@ -206,5 +327,28 @@ edge [fontsize=8];"""
 
     svg = svg.sub(/<\?xml.*?\?>/m, "")
     svg = svg.sub(/<!DOCTYPE.*?>/m, "")
+    svg = svg.sub(/<svg /, "<svg id=\"#{svgId}\" ")
+  end
+
+  def self.find_collections(sp)
+    c = []
+    case sp
+    when Hash
+      sp.each do |k, v|
+        c.concat(find_collections(v))
+      end
+    when Array
+      sp.each do |v|
+        c.concat(find_collections(v))
+      end
+    when String
+      if !sp.empty?
+        m = GenerateGraph::collection_uuid(sp)
+        if m
+          c << m
+        end
+      end
+    end
+    c
   end
 end