1 module ProvenanceHelper
4 def initialize(pdata, opts)
12 def self.collection_uuid(uuid)
13 Keep::Locator.parse(uuid).andand.strip_hints.andand.to_s
17 p = { :host => @opts[:request].host,
18 :port => @opts[:request].port,
19 :protocol => @opts[:request].protocol }
21 Rails.application.routes.url_helpers.url_for (p)
24 def determine_fillcolor(n)
25 fillcolor = %w(666666 669966 666699 666666 996666)[n || 0] || '666666'
26 "style=\"filled\",color=\"#ffffff\",fillcolor=\"##{fillcolor}\",fontcolor=\"#ffffff\""
29 def describe_node(uuid, describe_opts={})
30 bgcolor = determine_fillcolor (describe_opts[:pip] || @opts[:pips].andand[uuid])
32 rsc = ArvadosBase::resource_class_for_uuid uuid
34 if GenerateGraph::collection_uuid(uuid) || rsc == Collection
35 if Collection.is_empty_blob_locator? uuid.to_s
37 return "\"#{uuid}\" [label=\"(empty collection)\"];\n"
40 href = url_for ({:controller => Collection.to_s.tableize,
44 return "\"#{uuid}\" [label=\"#{encode_quotes(describe_opts[:label] || (@pdata[uuid] and @pdata[uuid][:name]) || uuid)}\",shape=box,href=\"#{href}\",#{bgcolor}];\n"
47 if describe_opts[:href]
48 href = ",href=\"#{url_for ({:controller => describe_opts[:href][:controller],
50 :id => describe_opts[:href][:id] })}\""
52 return "\"#{uuid}\" [label=\"#{encode_quotes(describe_opts[:label] || uuid)}\",#{bgcolor},shape=#{describe_opts[:shape] || 'box'}#{href}];\n"
57 d = Digest::MD5.hexdigest(job[:script_parameters].to_json)
58 if @opts[:combine_jobs] == :script_only
59 uuid = "#{job[:script]}_#{d}"
60 elsif @opts[:combine_jobs] == :script_and_version
61 uuid = "#{job[:script]}_#{job[:script_version]}_#{d}"
63 uuid = "#{job[:uuid]}"
66 @jobs[uuid] = [] unless @jobs[uuid]
67 @jobs[uuid] << job unless @jobs[uuid].include? job
72 def edge(tail, head, extra)
73 if @opts[:direction] == :bottom_up
74 gr = "\"#{encode_quotes head}\" -> \"#{encode_quotes tail}\""
76 gr = "\"#{encode_quotes tail}\" -> \"#{encode_quotes head}\""
82 gr += "#{k}=\"#{encode_quotes v}\","
90 def script_param_edges(uuid, sp)
94 if @opts[:all_script_parameters]
95 if v.is_a? Array or v.is_a? Hash
96 encv = JSON.pretty_generate(v).gsub("\n", "\\l") + "\\l"
100 gr += "\"#{encode_quotes encv}\" [shape=box];\n"
101 gr += edge(encv, uuid, {:label => k})
107 def col_name_for_project(col_pdh, project_uuid)
109 # Search for collections within the same project first.
110 {portable_data_hash: col_pdh, owner_uuid: project_uuid},
111 # then, earch for collections in any project.
112 {portable_data_hash: col_pdh},
113 ].each do |query_args|
114 col = Collection.where(query_args).limit(1)
116 if col.items_available == 1
117 return col.results.first.name
119 return "#{col.results.first.name} + #{col.items_available - 1} more"
124 # No collections found with this pdh
132 # Search for input mounts
133 input_obj = cr[:mounts].andand[:"/var/lib/cwl/cwl.input.json"].andand[:content] || cr[:mounts] || {}
135 ProvenanceHelper::find_collections input_obj, 'input' do |col_hash, col_uuid, key|
136 # Only include input PDHs
138 gr += describe_node(col_hash,
139 {label: col_name_for_project(col_hash, cr[:owner_uuid])})
140 gr += edge(col_hash, uuid, {:label => key})
145 # Add CR outputs by PDH so they connect with the child CR's inputs.
147 output_pdh = Collection.find(cr[:output_uuid])[:portable_data_hash]
149 gr += describe_node(output_pdh,
150 {label: col_name_for_project(output_pdh, cr[:owner_uuid])})
151 gr += edge(uuid, output_pdh, {label: 'output'})
158 def job_edges job, edge_opts={}
162 ProvenanceHelper::find_collections job[:script_parameters] do |collection_hash, collection_uuid, key|
164 gr += describe_node(collection_uuid)
165 gr += edge(collection_uuid, uuid, {:label => key})
167 gr += describe_node(collection_hash)
168 gr += edge(collection_hash, uuid, {:label => key})
172 if job[:docker_image_locator] and !@opts[:no_docker]
173 gr += describe_node(job[:docker_image_locator], {label: (job[:runtime_constraints].andand[:docker_image] || job[:docker_image_locator])})
174 gr += edge(job[:docker_image_locator], uuid, {label: "docker_image"})
177 if @opts[:script_version_nodes]
178 gr += describe_node(job[:script_version], {:label => "git:#{job[:script_version]}"})
179 gr += edge(job[:script_version], uuid, {:label => "script_version"})
182 if job[:output] and !edge_opts[:no_output]
183 gr += describe_node(job[:output])
184 gr += edge(uuid, job[:output], {label: "output" })
187 if job[:log] and !edge_opts[:no_log]
188 gr += describe_node(job[:log])
189 gr += edge(uuid, job[:log], {label: "log"})
195 def generate_provenance_edges(uuid)
197 m = GenerateGraph::collection_uuid(uuid)
200 if uuid.nil? or uuid.empty? or @visited[uuid]
207 @visited[uuid] = true
210 if uuid.start_with? "component_"
211 # Pipeline component inputs
212 job = @pdata[@pdata[uuid][:job].andand[:uuid]]
215 gr += describe_node(job_uuid(job), {label: uuid[38..-1], pip: @opts[:pips].andand[job[:uuid]], shape: "oval",
216 href: {controller: 'jobs', id: job[:uuid]}})
217 gr += job_edges job, {no_output: true, no_log: true}
220 # Pipeline component output
221 outuuid = @pdata[uuid][:output_uuid]
223 outcollection = @pdata[outuuid]
225 gr += edge(job_uuid(job), outcollection[:portable_data_hash], {label: "output"})
226 gr += describe_node(outcollection[:portable_data_hash], {label: outcollection[:name]})
228 elsif job and job[:output]
229 gr += describe_node(job[:output])
230 gr += edge(job_uuid(job), job[:output], {label: "output" })
233 rsc = ArvadosBase::resource_class_for_uuid uuid
237 gr += job_edges job if job
238 elsif rsc == ContainerRequest
242 gr += describe_node(uuid, {href: {controller: 'container_requests',
244 label: @pdata[uuid][:name],
246 # Search for child CRs
247 if cr[:container_uuid]
248 child_crs = ContainerRequest.where(requesting_container_uuid: cr[:container_uuid])
249 child_crs.each do |child|
250 gr += generate_provenance_edges(child[:uuid])
251 gr += edge(child[:uuid], uuid, {label: 'child'})
258 @pdata.each do |k, link|
259 if link[:head_uuid] == uuid.to_s and link[:link_class] == "provenance"
260 href = url_for ({:controller => Link.to_s.tableize,
262 :id => link[:uuid] })
264 gr += describe_node(link[:tail_uuid])
265 gr += edge(link[:head_uuid], link[:tail_uuid], {:label => link[:name], :href => href})
266 gr += generate_provenance_edges(link[:tail_uuid])
276 href = url_for ({:controller => Job.to_s.tableize,
279 gr += "\"#{k}\" [href=\"#{href}?"
283 gr += ";" unless gr.end_with? "?"
284 gr += "uuid%5b%5d=#{u[:uuid]}"
285 n |= @opts[:pips][u[:uuid]] if @opts[:pips] and @opts[:pips][u[:uuid]]
290 label = "#{v[0][:script]}"
292 if label == "run-command" and v[0][:script_parameters][:command].is_a? Array
293 label = v[0][:script_parameters][:command].join(' ')
296 if not @opts[:combine_jobs]
297 label += "\\n#{v[0][:finished_at]}"
300 gr += encode_quotes label
302 gr += "\",#{determine_fillcolor n}];\n"
307 def encode_quotes value
308 value.to_s.gsub("\"", "\\\"").gsub("\n", "\\n")
312 def self.create_provenance_graph(pdata, svgId, opts={})
313 if pdata.is_a? Array or pdata.is_a? ArvadosResourceList
316 p2[k[:uuid]] = k if k[:uuid]
321 unless pdata.is_a? Hash
322 raise "create_provenance_graph accepts Array or Hash for pdata only, pdata is #{pdata.class}"
325 gr = """strict digraph {
326 node [fontsize=10,fontname=\"Helvetica,Arial,sans-serif\"];
327 edge [fontsize=10,fontname=\"Helvetica,Arial,sans-serif\"];
330 if opts[:direction] == :bottom_up
331 gr += "edge [dir=back];"
335 pdata = pdata.stringify_keys
337 g = GenerateGraph.new(pdata, opts)
340 if !opts[:only_components] or k.start_with? "component_"
341 gr += g.generate_provenance_edges(k)
343 #gr += describe_node(k)
347 if !opts[:only_components]
348 gr += g.describe_jobs
352 Rails.logger.warn "#{e.inspect}"
353 Rails.logger.warn "#{e.backtrace.join("\n\t")}"
362 Open3.popen2("dot", "-Tsvg") do |stdin, stdout, wait_thr|
370 svg = svg.sub(/<\?xml.*?\?>/m, "")
371 svg = svg.sub(/<!DOCTYPE.*?>/m, "")
372 svg = svg.sub(/<svg /, "<svg id=\"#{svgId}\" ")
376 # Position indicates whether it is a content hash or arvados uuid.
377 # One will hold a value, the other will always be nil.
378 def self.find_collections(sp, key=nil, &b)
381 sp.class.columns.each do |c|
382 find_collections(sp[c.name.to_sym], nil, &b)
386 find_collections(v, key || k, &b)
390 find_collections(v, key, &b)
393 if m = /[a-f0-9]{32}\+\d+/.match(sp)
395 elsif m = /[0-9a-z]{5}-4zz18-[0-9a-z]{15}/.match(sp)