def generate_provenance(jobs)
return if params['tab_pane'] != "Provenance"
- nodes = []
+ nodes = {}
collections = []
+ hashes = []
jobs.each do |j|
- nodes << j
- collections << j[:output]
- collections.concat(ProvenanceHelper::find_collections(j[:script_parameters]))
- nodes << {:uuid => j[:script_version]}
+ nodes[j[:uuid]] = j
+ hashes << j[:output]
+ ProvenanceHelper::find_collections(j[:script_parameters]) do |hash, uuid|
+ collections << uuid if uuid
+ hashes << hash if hash
+ end
+ nodes[j[:script_version]] = {:uuid => j[:script_version]}
end
Collection.where(uuid: collections).each do |c|
- nodes << c
+ nodes[c[:portable_data_hash]] = c
+ end
+
+ Collection.where(portable_data_hash: hashes).each do |c|
+ nodes[c[:portable_data_hash]] = c
end
@svg = ProvenanceHelper::create_provenance_graph nodes, "provenance_svg", {
def graph(pipelines)
return nil, nil if params['tab_pane'] != "Graph"
- count = {}
provenance = {}
pips = {}
n = 1
+ # When comparing more than one pipeline, "pips" stores bit fields that
+ # indicates which objects are part of which pipelines.
+
pipelines.each do |p|
collections = []
+ hashes = []
+ jobs = []
+
+ p[:components].each do |k, v|
+ provenance["component_#{p[:uuid]}_#{k}"] = v
+
+ collections << v[:output_uuid] if v[:output_uuid]
+ jobs << v[:job][:uuid] if v[:job]
+ end
+
+ jobs = jobs.compact.uniq
+ if jobs.any?
+ Job.where(uuid: jobs).each do |j|
+ job_uuid = j.uuid
- p.components.each do |k, v|
- j = v[:job] || next
+ provenance[job_uuid] = j
+ pips[job_uuid] = 0 unless pips[job_uuid] != nil
+ pips[job_uuid] |= n
- uuid = j[:uuid].intern
- provenance[uuid] = j
- pips[uuid] = 0 unless pips[uuid] != nil
- pips[uuid] |= n
+ hashes << j[:output] if j[:output]
+ ProvenanceHelper::find_collections(j) do |hash, uuid|
+ collections << uuid if uuid
+ hashes << hash if hash
+ end
- collections << j[:output]
- ProvenanceHelper::find_collections(j[:script_parameters]).each do |k|
- collections << k
+ if j[:script_version]
+ script_uuid = j[:script_version]
+ provenance[script_uuid] = {:uuid => script_uuid}
+ pips[script_uuid] = 0 unless pips[script_uuid] != nil
+ pips[script_uuid] |= n
+ end
end
+ end
- uuid = j[:script_version].intern
- provenance[uuid] = {:uuid => uuid}
- pips[uuid] = 0 unless pips[uuid] != nil
- pips[uuid] |= n
+ hashes = hashes.compact.uniq
+ if hashes.any?
+ Collection.where(portable_data_hash: hashes).each do |c|
+ hash_uuid = c.portable_data_hash
+ provenance[hash_uuid] = c
+ pips[hash_uuid] = 0 unless pips[hash_uuid] != nil
+ pips[hash_uuid] |= n
+ end
end
- Collection.where(uuid: collections.compact).each do |c|
- uuid = c.uuid.intern
- provenance[uuid] = c
- pips[uuid] = 0 unless pips[uuid] != nil
- pips[uuid] |= n
+ collections = collections.compact.uniq
+ if collections.any?
+ Collection.where(uuid: collections).each do |c|
+ collection_uuid = c.uuid
+ provenance[collection_uuid] = c
+ pips[collection_uuid] = 0 unless pips[collection_uuid] != nil
+ pips[collection_uuid] |= n
+ end
end
n = n << 1
:request => request,
:all_script_parameters => true,
:combine_jobs => :script_and_version,
- :script_version_nodes => true,
- :pips => pips }
+ :pips => pips,
+ :only_components => true,
+ :no_docker => true,
+ :no_log => true}
end
super
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
- else
- nil
- end
+ Keep::Locator.parse(uuid).andand.strip_hints.andand.to_s
end
def url_for u
end
def determine_fillcolor(n)
- fillcolor = %w(aaaaaa aaffaa aaaaff aaaaaa ffaaaa)[n || 0] || 'aaaaaa'
- "style=filled,fillcolor=\"##{fillcolor}\""
+ fillcolor = %w(666666 669966 666699 666666 996666)[n || 0] || '666666'
+ "style=\"filled\",color=\"#ffffff\",fillcolor=\"##{fillcolor}\",fontcolor=\"#ffffff\""
end
- def describe_node(uuid)
- uuid = uuid.to_sym
- bgcolor = determine_fillcolor @opts[:pips].andand[uuid]
+ def describe_node(uuid, describe_opts={})
+ bgcolor = determine_fillcolor (describe_opts[:pip] || @opts[:pips].andand[uuid])
+
+ rsc = ArvadosBase::resource_class_for_uuid uuid
+
+ if GenerateGraph::collection_uuid(uuid) || rsc == Collection
+ if Collection.is_empty_blob_locator? uuid.to_s
+ # special case
+ return "\"#{uuid}\" [label=\"(empty collection)\"];\n"
+ end
- rsc = ArvadosBase::resource_class_for_uuid uuid.to_s
- if rsc
- href = url_for ({:controller => rsc.to_s.tableize,
+ href = url_for ({:controller => Collection.to_s.tableize,
:action => :show,
:id => uuid.to_s })
- #"\"#{uuid}\" [label=\"#{rsc}\\n#{uuid}\",href=\"#{href}\"];\n"
- if rsc == Collection
- if Collection.is_empty_blob_locator? uuid.to_s
- # special case
- return "\"#{uuid}\" [label=\"(empty collection)\"];\n"
- end
- if @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⋮"
- end
- extra_s = @node_extra[uuid].andand.map { |k,v|
- "#{k}=\"#{v}\""
- }.andand.join ","
- return "\"#{uuid}\" [label=\"#{label}\",href=\"#{href}\",shape=oval,#{bgcolor},#{extra_s}];\n"
- end
- end
- end
+ return "\"#{uuid}\" [label=\"#{encode_quotes(describe_opts[:label] || (@pdata[uuid] and @pdata[uuid][:name]) || uuid)}\",shape=box,href=\"#{href}\",#{bgcolor}];\n"
+ else
+ href = ""
+ if describe_opts[:href]
+ href = ",href=\"#{url_for ({:controller => describe_opts[:href][:controller],
+ :action => :show,
+ :id => describe_opts[:href][:id] })}\""
end
- return "\"#{uuid}\" [label=\"#{rsc}\",href=\"#{href}\",#{bgcolor}];\n"
+ return "\"#{uuid}\" [label=\"#{encode_quotes(describe_opts[:label] || uuid)}\",#{bgcolor},shape=#{describe_opts[:shape] || 'box'}#{href}];\n"
end
- "\"#{uuid}\" [#{bgcolor}];\n"
end
def job_uuid(job)
def edge(tail, head, extra)
if @opts[:direction] == :bottom_up
- gr = "\"#{tail}\" -> \"#{head}\""
+ gr = "\"#{encode_quotes head}\" -> \"#{encode_quotes tail}\""
else
- gr = "\"#{head}\" -> \"#{tail}\""
+ gr = "\"#{encode_quotes tail}\" -> \"#{encode_quotes head}\""
end
+
if extra.length > 0
gr += " ["
extra.each do |k, v|
- gr += "#{k}=\"#{v}\","
+ gr += "#{k}=\"#{encode_quotes v}\","
end
gr += "]"
end
gr
end
- def script_param_edges(job, prefix, sp)
+ def script_param_edges(uuid, sp)
gr = ""
- case sp
- when Hash
- sp.each do |k, v|
- if prefix.size > 0
- k = prefix + "::" + k.to_s
- end
- gr += script_param_edges(job, k.to_s, v)
- end
- when Array
- i = 0
- node = ""
- count = 0
- sp.each do |v|
- if GenerateGraph::collection_uuid(v)
- gr += script_param_edges(job, "#{prefix}[#{i}]", v)
- elsif @opts[:all_script_parameters]
- t = "#{v}"
- nl = (if (count+t.length) > 60 then "\\n" else " " end)
- count = 0 if (count+t.length) > 60
- node += "',#{nl}'" unless node == ""
- node = "['" if node == ""
- node += t
- count += t.length
+
+ sp.each do |k, v|
+ if @opts[:all_script_parameters]
+ if v.is_a? Array or v.is_a? Hash
+ encv = JSON.pretty_generate(v).gsub("\n", "\\l") + "\\l"
+ else
+ encv = v.to_json
end
- i += 1
+ gr += "\"#{encode_quotes encv}\" [shape=box];\n"
+ gr += edge(encv, uuid, {:label => k})
end
- unless node == ""
- node += "']"
- node_value = encode_quotes node
- gr += "\"#{node_value}\" [label=\"#{node_value}\"];\n"
- gr += edge(job_uuid(job), node_value, {:label => prefix})
- end
- when String
- return '' if sp.empty?
- m = GenerateGraph::collection_uuid(sp)
- 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]
- sp_value = encode_quotes sp
- gr += "\"#{sp_value}\" [label=\"#{sp_value}\"];\n"
- gr += edge(job_uuid(job), sp_value, {:label => prefix})
+ end
+ gr
+ end
+
+ def job_edges job, edge_opts={}
+ uuid = job_uuid(job)
+ gr = ""
+
+ ProvenanceHelper::find_collections job[:script_parameters] do |collection_hash, collection_uuid, key|
+ if collection_uuid
+ gr += describe_node(collection_uuid)
+ gr += edge(collection_uuid, uuid, {:label => key})
+ else
+ gr += describe_node(collection_hash)
+ gr += edge(collection_hash, uuid, {:label => key})
end
end
+
+ if job[:docker_image_locator] and !@opts[:no_docker]
+ gr += describe_node(job[:docker_image_locator], {label: (job[:runtime_constraints].andand[:docker_image] || job[:docker_image_locator])})
+ gr += edge(job[:docker_image_locator], uuid, {label: "docker_image"})
+ end
+
+ if @opts[:script_version_nodes]
+ gr += describe_node(job[:script_version], {:label => "git:#{job[:script_version]}"})
+ gr += edge(job[:script_version], uuid, {:label => "script_version"})
+ end
+
+ if job[:output] and !edge_opts[:no_output]
+ gr += describe_node(job[:output])
+ gr += edge(uuid, job[:output], {label: "output" })
+ end
+
+ gr += edge(uuid, job[:log], {label: "log"}) if job[:log] and !edge_opts[:no_log]
+
gr
end
m = GenerateGraph::collection_uuid(uuid)
uuid = m if m
- uuid = uuid.intern if uuid
-
- if (not uuid) or uuid.empty? or @visited[uuid]
+ if uuid.nil? or uuid.empty? or @visited[uuid]
return ""
end
- if not @pdata[uuid] then
- return describe_node(uuid)
+ if @pdata[uuid].nil?
+ return ""
else
@visited[uuid] = true
end
- if m
- # uuid is a collection
- if not Collection.is_empty_blob_locator? uuid.to_s
- @pdata.each do |k, job|
- if job[:output] == uuid.to_s
- extra = { label: 'output' }
- gr += edge(uuid, job_uuid(job), extra)
- 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
+ if uuid.start_with? "component_"
+ # Pipeline component inputs
+ job = @pdata[@pdata[uuid][:job].andand[:uuid]]
+
+ if job
+ gr += describe_node(job_uuid(job), {label: uuid[38..-1], pip: @opts[:pips].andand[job[:uuid]], shape: "oval",
+ href: {controller: 'jobs', id: job[:uuid]}})
+ gr += job_edges job, {no_output: true, no_log: true}
+ end
+
+ # Pipeline component output
+ outuuid = @pdata[uuid][:output_uuid]
+ if outuuid
+ outcollection = @pdata[outuuid]
+ if outcollection
+ gr += edge(job_uuid(job), outcollection[:portable_data_hash], {label: "output"})
+ gr += describe_node(outcollection[:portable_data_hash], {label: outcollection[:name]})
end
+ elsif job and job[:output]
+ gr += describe_node(job[:output])
+ gr += edge(job_uuid(job), job[:output], {label: "output" })
end
- gr += describe_node(uuid)
else
- # uuid is something else
- rsc = ArvadosBase::resource_class_for_uuid uuid.to_s
+ rsc = ArvadosBase::resource_class_for_uuid uuid
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)
+ gr += job_edges job if job
end
end
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]
+ gr += ";" unless gr.end_with? "?"
+ gr += "uuid%5b%5d=#{u[:uuid]}"
+ n |= @opts[:pips][u[:uuid]] if @opts[:pips] and @opts[:pips][u[:uuid]]
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]}"
+ label = "#{v[0][:script]}"
+
+ if label == "run-command" and v[0][:script_parameters][:command].is_a? Array
+ label = v[0][:script_parameters][:command].join(' ')
end
+
+ if not @opts[:combine_jobs]
+ label += "\\n#{v[0][:finished_at]}"
+ end
+
+ gr += encode_quotes label
+
gr += "\",#{determine_fillcolor n}];\n"
end
gr
end
def encode_quotes value
- value.andand.gsub("\"", "\\\"")
+ value.to_s.gsub("\"", "\\\"").gsub("\n", "\\n")
end
end
if pdata.is_a? Array or pdata.is_a? ArvadosResourceList
p2 = {}
pdata.each do |k|
- p2[k[:uuid].intern] = k if k[:uuid]
+ p2[k[:uuid]] = k if k[:uuid]
end
pdata = p2
end
end
gr = """strict digraph {
-node [fontsize=10,shape=box];
-edge [fontsize=10];
+node [fontsize=10,fontname=\"Helvetica,Arial,sans-serif\"];
+edge [fontsize=10,fontname=\"Helvetica,Arial,sans-serif\"];
"""
if opts[:direction] == :bottom_up
gr += "edge [dir=back];"
end
- g = GenerateGraph.new(pdata, opts)
+ begin
+ pdata = pdata.stringify_keys
- pdata.each do |k, v|
- gr += g.generate_provenance_edges(k)
- end
+ g = GenerateGraph.new(pdata, opts)
+
+ pdata.each do |k, v|
+ if !opts[:only_components] or k.start_with? "component_"
+ gr += g.generate_provenance_edges(k)
+ else
+ #gr += describe_node(k)
+ end
+ end
- gr += g.describe_jobs
+ if !opts[:only_components]
+ gr += g.describe_jobs
+ end
+
+ rescue => e
+ Rails.logger.warn "#{e.inspect}"
+ Rails.logger.warn "#{e.backtrace.join("\n\t")}"
+ raise
+ end
gr += "}"
svg = ""
svg = svg.sub(/<svg /, "<svg id=\"#{svgId}\" ")
end
- def self.find_collections(sp)
- c = []
+ # yields hash, uuid
+ # Position indicates whether it is a content hash or arvados uuid.
+ # One will hold a value, the other will always be nil.
+ def self.find_collections(sp, key=nil, &b)
case sp
+ when ArvadosBase
+ sp.class.columns.each do |c|
+ find_collections(sp[c.name.to_sym], nil, &b)
+ end
when Hash
sp.each do |k, v|
- c.concat(find_collections(v))
+ find_collections(v, key || k, &b)
end
when Array
sp.each do |v|
- c.concat(find_collections(v))
+ find_collections(v, key, &b)
end
when String
- if !sp.empty?
- m = GenerateGraph::collection_uuid(sp)
- if m
- c << m
- end
+ if m = /[a-f0-9]{32}\+\d+/.match(sp)
+ yield m[0], nil, key
+ elsif m = /[0-9a-z]{5}-4zz18-[0-9a-z]{15}/.match(sp)
+ yield nil, m[0], key
end
end
- c
end
end
--- /dev/null
+require 'test_helper'
+
+class CollectionsControllerTest < ActionController::TestCase
+ include PipelineInstancesHelper
+
+ class RequestDuck
+ def self.host
+ "localhost"
+ end
+
+ def self.port
+ 8080
+ end
+
+ def self.protocol
+ "http"
+ end
+ end
+
+
+ test 'provenance graph' do
+ use_token 'admin'
+
+ obj = find_fixture Collection, "graph_test_collection3"
+
+ provenance = obj.provenance.stringify_keys
+
+ [obj[:portable_data_hash]].each do |k|
+ assert_not_nil provenance[k], "Expected key #{k} in provenance set"
+ end
+
+ prov_svg = ProvenanceHelper::create_provenance_graph(provenance, "provenance_svg",
+ {:request => RequestDuck,
+ :direction => :bottom_up,
+ :combine_jobs => :script_only})
+
+ stage1 = find_fixture Job, "graph_stage1"
+ stage3 = find_fixture Job, "graph_stage3"
+ previous_job_run = find_fixture Job, "previous_job_run"
+
+ obj_id = obj.portable_data_hash.gsub('+', '\\\+')
+ stage1_out = stage1.output.gsub('+', '\\\+')
+ stage1_id = "#{stage1.script}_#{Digest::MD5.hexdigest(stage1[:script_parameters].to_json)}"
+ stage3_id = "#{stage3.script}_#{Digest::MD5.hexdigest(stage3[:script_parameters].to_json)}"
+
+ assert /#{obj_id}->#{stage3_id}/.match(prov_svg)
+
+ assert /#{stage3_id}->#{stage1_out}/.match(prov_svg)
+
+ assert /#{stage1_out}->#{stage1_id}/.match(prov_svg)
+
+ end
+
+ test 'used_by graph' do
+ use_token 'admin'
+ obj = find_fixture Collection, "graph_test_collection1"
+
+ used_by = obj.used_by.stringify_keys
+
+ used_by_svg = ProvenanceHelper::create_provenance_graph(used_by, "used_by_svg",
+ {:request => RequestDuck,
+ :direction => :top_down,
+ :combine_jobs => :script_only,
+ :pdata_only => true})
+
+ stage2 = find_fixture Job, "graph_stage2"
+ stage3 = find_fixture Job, "graph_stage3"
+
+ stage2_id = "#{stage2.script}_#{Digest::MD5.hexdigest(stage2[:script_parameters].to_json)}"
+ stage3_id = "#{stage3.script}_#{Digest::MD5.hexdigest(stage3[:script_parameters].to_json)}"
+
+ obj_id = obj.portable_data_hash.gsub('+', '\\\+')
+ stage3_out = stage3.output.gsub('+', '\\\+')
+
+ assert /#{obj_id}->#{stage2_id}/.match(used_by_svg)
+
+ assert /#{obj_id}->#{stage3_id}/.match(used_by_svg)
+
+ assert /#{stage3_id}->#{stage3_out}/.match(used_by_svg)
+
+ assert /#{stage3_id}->#{stage3_out}/.match(used_by_svg)
+
+ end
+end
{started_at: 6, finished_at: 8}]
assert_equal 6, determine_wallclock_runtime(r)
end
+
+
+ class RequestDuck
+ def self.host
+ "localhost"
+ end
+
+ def self.port
+ 8080
+ end
+
+ def self.protocol
+ "http"
+ end
+ end
+
+ test "generate graph" do
+
+ use_token 'admin'
+
+ pipeline_for_graph = {
+ state: 'Complete',
+ uuid: 'zzzzz-d1hrv-9fm8l10i9z2kqc9',
+ components: {
+ stage1: {
+ repository: 'foo',
+ script: 'hash',
+ script_version: 'master',
+ job: {uuid: 'zzzzz-8i9sb-graphstage10000'},
+ output_uuid: 'zzzzz-4zz18-bv31uwvy3neko22'
+ },
+ stage2: {
+ repository: 'foo',
+ script: 'hash2',
+ script_version: 'master',
+ script_parameters: {
+ input: 'fa7aeb5140e2848d39b416daeef4ffc5+45'
+ },
+ job: {uuid: 'zzzzz-8i9sb-graphstage20000'},
+ output_uuid: 'zzzzz-4zz18-uukreo9rbgwsujx'
+ }
+ }
+ }
+
+ @controller.params['tab_pane'] = "Graph"
+ provenance, pips = @controller.graph([pipeline_for_graph])
+
+ graph_test_collection1 = find_fixture Collection, "graph_test_collection1"
+ stage1 = find_fixture Job, "graph_stage1"
+ stage2 = find_fixture Job, "graph_stage2"
+
+ ['component_zzzzz-d1hrv-9fm8l10i9z2kqc9_stage1',
+ 'component_zzzzz-d1hrv-9fm8l10i9z2kqc9_stage2',
+ stage1.uuid,
+ stage2.uuid,
+ stage1.output,
+ stage2.output,
+ pipeline_for_graph[:components][:stage1][:output_uuid],
+ pipeline_for_graph[:components][:stage2][:output_uuid]
+ ].each do |k|
+
+ assert_not_nil provenance[k], "Expected key #{k} in provenance set"
+ assert_equal 1, pips[k], "Expected key #{k} in pips set" if !k.start_with? "component_"
+ end
+
+ prov_svg = ProvenanceHelper::create_provenance_graph provenance, "provenance_svg", {
+ :request => RequestDuck,
+ :all_script_parameters => true,
+ :combine_jobs => :script_and_version,
+ :pips => pips,
+ :only_components => true }
+
+ stage1_id = "#{stage1[:script]}_#{stage1[:script_version]}_#{Digest::MD5.hexdigest(stage1[:script_parameters].to_json)}"
+ stage2_id = "#{stage2[:script]}_#{stage2[:script_version]}_#{Digest::MD5.hexdigest(stage2[:script_parameters].to_json)}"
+
+ stage1_out = stage1[:output].gsub('+','\\\+')
+
+ assert_match /#{stage1_id}->#{stage1_out}/, prov_svg
+
+ assert_match /#{stage1_out}->#{stage2_id}/, prov_svg
+
+ end
+
+ test "generate graph compare" do
+
+ use_token 'admin'
+
+ pipeline_for_graph1 = {
+ state: 'Complete',
+ uuid: 'zzzzz-d1hrv-9fm8l10i9z2kqc9',
+ components: {
+ stage1: {
+ repository: 'foo',
+ script: 'hash',
+ script_version: 'master',
+ job: {uuid: 'zzzzz-8i9sb-graphstage10000'},
+ output_uuid: 'zzzzz-4zz18-bv31uwvy3neko22'
+ },
+ stage2: {
+ repository: 'foo',
+ script: 'hash2',
+ script_version: 'master',
+ script_parameters: {
+ input: 'fa7aeb5140e2848d39b416daeef4ffc5+45'
+ },
+ job: {uuid: 'zzzzz-8i9sb-graphstage20000'},
+ output_uuid: 'zzzzz-4zz18-uukreo9rbgwsujx'
+ }
+ }
+ }
+
+ pipeline_for_graph2 = {
+ state: 'Complete',
+ uuid: 'zzzzz-d1hrv-9fm8l10i9z2kqc0',
+ components: {
+ stage1: {
+ repository: 'foo',
+ script: 'hash',
+ script_version: 'master',
+ job: {uuid: 'zzzzz-8i9sb-graphstage10000'},
+ output_uuid: 'zzzzz-4zz18-bv31uwvy3neko22'
+ },
+ stage2: {
+ repository: 'foo',
+ script: 'hash2',
+ script_version: 'master',
+ script_parameters: {
+ },
+ job: {uuid: 'zzzzz-8i9sb-graphstage30000'},
+ output_uuid: 'zzzzz-4zz18-uukreo9rbgwsujj'
+ }
+ }
+ }
+
+ @controller.params['tab_pane'] = "Graph"
+ provenance, pips = @controller.graph([pipeline_for_graph1, pipeline_for_graph2])
+
+ collection1 = find_fixture Collection, "graph_test_collection1"
+
+ stage1 = find_fixture Job, "graph_stage1"
+ stage2 = find_fixture Job, "graph_stage2"
+ stage3 = find_fixture Job, "graph_stage3"
+
+ [['component_zzzzz-d1hrv-9fm8l10i9z2kqc9_stage1', nil],
+ ['component_zzzzz-d1hrv-9fm8l10i9z2kqc9_stage2', nil],
+ ['component_zzzzz-d1hrv-9fm8l10i9z2kqc0_stage1', nil],
+ ['component_zzzzz-d1hrv-9fm8l10i9z2kqc0_stage2', nil],
+ [stage1.uuid, 3],
+ [stage2.uuid, 1],
+ [stage3.uuid, 2],
+ [stage1.output, 3],
+ [stage2.output, 1],
+ [stage3.output, 2],
+ [pipeline_for_graph1[:components][:stage1][:output_uuid], 3],
+ [pipeline_for_graph1[:components][:stage2][:output_uuid], 1],
+ [pipeline_for_graph2[:components][:stage2][:output_uuid], 2]
+ ].each do |k|
+ assert_not_nil provenance[k[0]], "Expected key #{k[0]} in provenance set"
+ assert_equal k[1], pips[k[0]], "Expected key #{k} in pips" if !k[0].start_with? "component_"
+ end
+
+ prov_svg = ProvenanceHelper::create_provenance_graph provenance, "provenance_svg", {
+ :request => RequestDuck,
+ :all_script_parameters => true,
+ :combine_jobs => :script_and_version,
+ :pips => pips,
+ :only_components => true }
+
+ collection1_id = collection1.portable_data_hash.gsub('+','\\\+')
+
+ stage2_id = "#{stage2[:script]}_#{stage2[:script_version]}_#{Digest::MD5.hexdigest(stage2[:script_parameters].to_json)}"
+ stage3_id = "#{stage3[:script]}_#{stage3[:script_version]}_#{Digest::MD5.hexdigest(stage3[:script_parameters].to_json)}"
+
+ stage2_out = stage2[:output].gsub('+','\\\+')
+ stage3_out = stage3[:output].gsub('+','\\\+')
+
+ assert_match /#{collection1_id}->#{stage2_id}/, prov_svg
+ assert_match /#{collection1_id}->#{stage3_id}/, prov_svg
+
+ assert_match /#{stage2_id}->#{stage2_out}/, prov_svg
+ assert_match /#{stage3_id}->#{stage3_out}/, prov_svg
+
+ end
+
end
# since the pipeline component has a job, expect to see the graph
assert page.has_text? 'Graph'
click_link 'Graph'
- assert page.has_text? 'script_version'
+ page.assert_selector "#provenance_graph"
end
test 'pipeline description' do
super
end
- def script_param_edges(visited, sp)
+ def find_collections(visited, sp, &b)
case sp
+ when ArvadosModel
+ sp.class.columns.each do |c|
+ find_collections(visited, sp[c.name.to_sym], &b) if c.name != "log"
+ end
when Hash
sp.each do |k, v|
- script_param_edges(visited, v)
+ find_collections(visited, v, &b)
end
when Array
sp.each do |v|
- script_param_edges(visited, v)
+ find_collections(visited, v, &b)
end
when String
- return if sp.empty?
- if loc = Keep::Locator.parse(sp)
- search_edges(visited, loc.to_s, :search_up)
+ if m = /[a-f0-9]{32}\+\d+/.match(sp)
+ yield m[0], nil
+ elsif m = /[0-9a-z]{5}-4zz18-[0-9a-z]{15}/.match(sp)
+ yield nil, m[0]
end
end
end
if loc
# uuid is a portable_data_hash
- if c = Collection.readable_by(*@read_users).where(portable_data_hash: loc.to_s).limit(1).first
- visited[loc.to_s] = {
- portable_data_hash: c.portable_data_hash,
- }
+ collections = Collection.readable_by(*@read_users).where(portable_data_hash: loc.to_s)
+ c = collections.limit(2).all
+ if c.size == 1
+ visited[loc.to_s] = c[0]
+ elsif c.size > 1
+ name = collections.limit(1).where("name <> ''").first
+ if name
+ visited[loc.to_s] = {
+ portable_data_hash: c[0].portable_data_hash,
+ name: "#{name.name} + #{collections.count-1} more"
+ }
+ else
+ visited[loc.to_s] = {
+ portable_data_hash: c[0].portable_data_hash,
+ name: loc.to_s
+ }
+ end
end
if direction == :search_up
Job.readable_by(*@read_users).where(["jobs.script_parameters like ?", "%#{loc.to_s}%"]).each do |job|
search_edges(visited, job.uuid, :search_down)
end
+
+ Job.readable_by(*@read_users).where(["jobs.docker_image_locator = ?", "#{loc.to_s}"]).each do |job|
+ search_edges(visited, job.uuid, :search_down)
+ end
end
else
# uuid is a regular Arvados UUID
visited[uuid] = job.as_api_response
if direction == :search_up
# Follow upstream collections referenced in the script parameters
- script_param_edges(visited, job.script_parameters)
+ find_collections(visited, job) do |hash, uuid|
+ search_edges(visited, hash, :search_up) if hash
+ search_edges(visited, uuid, :search_up) if uuid
+ end
elsif direction == :search_down
# Follow downstream job output
search_edges(visited, job.output, direction)
def provenance
visited = {}
- search_edges(visited, @object[:uuid] || @object[:portable_data_hash], :search_up)
+ search_edges(visited, @object[:portable_data_hash], :search_up)
+ search_edges(visited, @object[:uuid], :search_up)
render json: visited
end
def used_by
visited = {}
- search_edges(visited, @object[:uuid] || @object[:portable_data_hash], :search_down)
+ search_edges(visited, @object[:uuid], :search_down)
+ search_edges(visited, @object[:portable_data_hash], :search_down)
render json: visited
end
updated_at: 2014-02-03T17:22:54Z
manifest_text: ". 85877ca2d7e05498dd3d109baf2df106+95+A3a4e26a366ee7e4ed3e476ccf05354761be2e4ae@545a9920 0:95:file_in_subdir1\n./subdir2/subdir3 2bbc341c702df4d8f42ec31f16c10120+64+A315d7e7bad2ce937e711fc454fae2d1194d14d64@545a9920 0:32:file1_in_subdir3.txt 32:32:file2_in_subdir3.txt\n./subdir2/subdir3/subdir4 2bbc341c702df4d8f42ec31f16c10120+64+A315d7e7bad2ce937e711fc454fae2d1194d14d64@545a9920 0:32:file1_in_subdir4.txt 32:32:file2_in_subdir4.txt"
+graph_test_collection1:
+ uuid: zzzzz-4zz18-bv31uwvy3neko22
+ portable_data_hash: fa7aeb5140e2848d39b416daeef4ffc5+45
+ manifest_text: ". 37b51d194a7513e45b56f6524f2d51f2+3 0:3:bar\n"
+ name: bar_file
+
+graph_test_collection2:
+ uuid: zzzzz-4zz18-uukreo9rbgwsujx
+ portable_data_hash: b519d9cb706a29fc7ea24dbea2f05851+93
+ manifest_text: ". 6a4ff0499484c6c79c95cd8c566bd25f+249025 0:249025:GNU_General_Public_License,_version_3.pdf\n"
+ name: "GNU General Public License, version 3"
+
+graph_test_collection3:
+ uuid: zzzzz-4zz18-uukreo9rbgwsujj
+ portable_data_hash: ea10d51bcf88862dbcc36eb292017dfd+45
+ manifest_text: ". 73feffa4b7f6bb68e44cf984c85f6e88+3 0:3:baz\n"
+ name: "baz file"
+
collection_1_owned_by_fuse:
uuid: zzzzz-4zz18-ovx05bfzormx3bg
portable_data_hash: fa7aeb5140e2848d39b416daeef4ffc5+45
script_parameters:
input: fa7aeb5140e2848d39b416daeef4ffc5+45
an_integer: "1"
+ runtime_constraints:
+ docker_image: arvados/test
success: true
output: ea10d51bcf88862dbcc36eb292017dfd+45
docker_image_locator: fa3c1a9cb6783f85f2ecda037e07b8c3+167
script_version: 4fe459abe02d9b365932b8f5dc419439ab4e2577
state: Complete
+graph_stage1:
+ uuid: zzzzz-8i9sb-graphstage10000
+ owner_uuid: zzzzz-j7d0g-v955i6s2oi1cbso
+ repository: foo
+ script: hash
+ script_version: 4fe459abe02d9b365932b8f5dc419439ab4e2577
+ state: Complete
+ output: fa7aeb5140e2848d39b416daeef4ffc5+45
+
+graph_stage2:
+ uuid: zzzzz-8i9sb-graphstage20000
+ owner_uuid: zzzzz-j7d0g-v955i6s2oi1cbso
+ repository: foo
+ script: hash2
+ script_version: 4fe459abe02d9b365932b8f5dc419439ab4e2577
+ state: Complete
+ script_parameters:
+ input: fa7aeb5140e2848d39b416daeef4ffc5+45
+ input2: "stuff"
+ output: b519d9cb706a29fc7ea24dbea2f05851+93
+
+graph_stage3:
+ uuid: zzzzz-8i9sb-graphstage30000
+ owner_uuid: zzzzz-j7d0g-v955i6s2oi1cbso
+ repository: foo
+ script: hash2
+ script_version: 4fe459abe02d9b365932b8f5dc419439ab4e2577
+ state: Complete
+ script_parameters:
+ input: fa7aeb5140e2848d39b416daeef4ffc5+45
+ input2: "stuff2"
+ output: ea10d51bcf88862dbcc36eb292017dfd+45
+
+
# Test Helper trims the rest of the file
# Do not add your fixtures below this line as the rest of this file will be trimmed by test_helper