Merge pull request seek4science#1770 from seek4science/galaxy-link-re…

…based

Store link back to Galaxy instance for workflows

app/controllers/workflows_controller.rb

@@ -355,7 +355,7 @@ def workflow_params
                                      { creator_ids: [] }, { assay_assets_attributes: [:assay_id] },
                                      { publication_ids: [] }, { presentation_ids: [] }, { document_ids: [] }, { data_file_ids: [] }, { sop_ids: [] },
                                      { workflow_data_files_attributes:[:id, :data_file_id, :workflow_data_file_relationship_id, :_destroy] },
-                                     :internals, :maturity_level, :source_link_url,
+                                     :internals, :maturity_level, :source_link_url, :execution_instance_url,
                                      { topic_annotations: [] }, { operation_annotations: [] },
                                      { discussion_links_attributes: [:id, :url, :label, :_destroy] },
                                      { git_version_attributes: [:name, :comment, :ref, :commit, :root_path,

app/helpers/workflows_helper.rb

@@ -77,12 +77,6 @@ def test_status_badge(resource)
     end
   end
 
-  def run_workflow_url(workflow_version)
-    if workflow_version.workflow_class_title == 'Galaxy'
-      "#{Seek::Config.galaxy_instance_trs_import_url}&trs_id=#{workflow_version.parent.id}&trs_version=#{workflow_version.version}"
-    end
-  end
-
   def workflow_class_options_for_select(selected = nil)
     opts = WorkflowClass.order(:title).map do |c|
       extra = {}

app/models/asset_link.rb

@@ -1,6 +1,7 @@
 class AssetLink < ApplicationRecord
   DISCUSSION = 'discussion'.freeze
   SOURCE = 'source'.freeze
+  EXECUTION_INSTANCE = 'execution_instance'.freeze
   MISC_LINKS = 'misc'.freeze
 
   scope :discussion, -> { where(link_type: AssetLink::DISCUSSION) }

app/models/concerns/workflow_extraction.rb

@@ -58,7 +58,23 @@ def structure
   delegate :inputs, :outputs, :steps, to: :structure
 
   def can_run?
-    can_download?(nil) && workflow_class_title == 'Galaxy' && Seek::Config.galaxy_instance_trs_import_url.present?
+    can_download?(nil) && workflow_class&.executable? && run_url.present?
+  end
+
+  def run_url
+    if workflow_class&.key == 'galaxy'
+      base = execution_instance_url || Seek::Config.galaxy_instance_default
+      return if base.nil?
+
+      parent_id = is_a_version? ? parent.id : id
+      url = URI(base) + 'workflows/trs_import'
+      params = {
+        trs_url: Seek::Util.routes.ga4gh_trs_v2_tool_version_url(parent_id, version_id: version),
+        run_form: true
+      }
+      url.query = URI.encode_www_form(params)
+      url.to_s
+    end
   end
 
   def diagram_exists?

app/models/git/blob.rb

@@ -10,12 +10,16 @@ class Blob
     delegate :read, :rewind, to: :file
 
     attr_reader :git_version, :path
+    # Flag to decide if `read`ing this blob should eagerly fetch any remote content pointed to by this blob's `url`.
+    attr_accessor :fetch_remote
+
     alias_method :original_filename, :path
 
     def initialize(git_version, blob, path)
       @git_version = git_version
       @blob = blob
       @path = path
+      @fetch_remote = false
     end
 
     def annotations
@@ -26,15 +30,15 @@ def url
       git_version.remote_sources[path]
     end
 
-    def file(fetch_remote: false)
+    def file(fetch_remote: @fetch_remote)
       @file ||= to_tempfile(fetch_remote: fetch_remote)
     end
 
     def binread
       file_contents
     end
 
-    def file_contents(as_text: false, fetch_remote: false, &block)
+    def file_contents(as_text: false, fetch_remote: @fetch_remote, &block)
       if fetch_remote && remote? && !fetched?
         if block_given?
           block.call(remote_content)
@@ -92,13 +96,17 @@ def text_contents_for_search
 
     def remote_content
       return unless remote?
-      handler = ContentBlob.remote_content_handler_for(url)
+      handler = remote_content_handler
       return unless handler
       io = handler.fetch
       io.rewind
       io
     end
 
+    def remote_content_handler
+      ContentBlob.remote_content_handler_for(url)
+    end
+
     def cache_key
       "#{git_repository.cache_key}/blobs/#{oid}"
     end
@@ -134,7 +142,7 @@ def is_text?
 
     private
 
-    def to_tempfile(fetch_remote: false)
+    def to_tempfile(fetch_remote: @fetch_remote)
       f = Tempfile.new(path)
       f.binmode if binary?
       f << file_contents(as_text: !binary?, fetch_remote: fetch_remote)

app/models/workflow.rb

@@ -30,6 +30,9 @@ class Workflow < ApplicationRecord
 
   accepts_nested_attributes_for :workflow_data_files
 
+  has_one :execution_instance, -> { where(link_type: AssetLink::EXECUTION_INSTANCE) },
+          class_name: 'AssetLink', as: :asset, dependent: :destroy, inverse_of: :asset, autosave: true
+
   def initialize(*args)
     @extraction_errors = []
     @extraction_warnings = []
@@ -131,6 +134,10 @@ def source_link_url
       parent&.source_link&.url
     end
 
+    def execution_instance_url
+      parent&.execution_instance&.url
+    end
+
     def submit_to_life_monitor
       LifeMonitorSubmissionJob.perform_later(self)
     end
@@ -237,6 +244,18 @@ def update_test_status(status, ver = version)
     v.save!
   end
 
+  def execution_instance_url= url
+    (execution_instance || build_execution_instance).assign_attributes(url: url)
+
+    execution_instance.mark_for_destruction if url.blank?
+
+    url
+  end
+
+  def execution_instance_url
+    execution_instance&.url
+  end
+
   has_filter maturity: Seek::Filtering::Filter.new(
       value_field: 'maturity_level',
       label_mapping: ->(values) {

app/models/workflow_class.rb

@@ -20,6 +20,10 @@ def extractable?
     extractor.present?
   end
 
+  def executable?
+    key == 'galaxy'
+  end
+
   def self.extractable
     where.not(extractor: nil)
   end

app/views/assets/_asset_buttons.html.erb

@@ -36,7 +36,7 @@
         <%= button_link_to('Download RO Crate', 'ro_crate_file', ro_crate_workflow_path(asset, version: version, code: params[:code]),
                            'data-tooltip' => tooltip("The Workflow RO-Crate is a package containing the workflow definition, its metadata and supporting resources like test data")) %>
         <% if asset.can_run? %>
-          <%= button_link_to("Run on #{Seek::Config.galaxy_instance_name || 'Galaxy'}", 'run_galaxy', run_workflow_url(display_asset)) %>
+          <%= button_link_to("Run on Galaxy", 'run_galaxy', display_asset.run_url) %>
         <% end %>
       <% else %>
         <%= button_link_to('Download RO-Crate', 'ro_crate_file', nil,

app/views/workflows/edit.html.erb

@@ -18,6 +18,14 @@
       <%= f.text_area :description, rows: 5, class: 'form-control rich-text-edit' -%>
     </div>
 
+    <% if @workflow.workflow_class&.executable? %>
+      <div class="form-group">
+        <label>Galaxy Instance</label>
+        <%= f.text_field :execution_instance_url, placeholder: 'https://usegalaxy.eu/', class: 'form-control' -%>
+        <p class="help-block">The root URL of the Galaxy instance where this workflow originated from.</p>
+      </div>
+    <% end %>
+
     <div class="form-group">
       <label>Source</label>
       <%= f.text_field :source_link_url, placeholder: 'https://...', class: 'form-control' -%>

app/views/workflows/provide_metadata.html.erb

@@ -45,6 +45,14 @@
       <%= text_area_tag 'workflow[description]', @workflow.description, class: "form-control rich-text-edit" -%>
     </div>
 
+    <% if @workflow.workflow_class&.executable? %>
+      <div class="form-group">
+        <label>Galaxy Instance</label>
+        <%= text_field_tag 'workflow[execution_instance_url]', @workflow.execution_instance_url, placeholder: 'https://usegalaxy.eu/', class: 'form-control' -%>
+        <p class="help-block">The root URL of the Galaxy instance where this workflow originated from.</p>
+      </div>
+    <% end %>
+
     <div class="form-group">
       <label>Source</label>
       <%= text_field_tag 'workflow[source_link_url]', @workflow.source_link_url, placeholder: 'https://...', class: "form-control" -%>

config/initializers/seek_testing.rb

@@ -139,5 +139,6 @@ def load_seek_testing_defaults!
       Settings.defaults[:git_support_enabled] = true
       Settings.defaults[:fair_signposting_enabled] = true
       Settings.defaults[:bio_tools_enabled] = true
+      Settings.defaults[:galaxy_instance_default] = 'https://usegalaxy.eu'
     end
 end

config/routes.rb

@@ -16,7 +16,7 @@
         get 'tools' => 'tools#index'
         get 'tools/:id' => 'tools#show'
         get 'tools/:id/versions' => 'tool_versions#index'
-        get 'tools/:id/versions/:version_id' => 'tool_versions#show'
+        get 'tools/:id/versions/:version_id' => 'tool_versions#show', as: :tool_version
         get 'tools/:id/versions/:version_id/containerfile' => 'tool_versions#containerfile'
         get 'tools/:id/versions/:version_id/:type/descriptor(/*relative_path)' => 'tool_versions#descriptor', constraints: { relative_path: /.+/ }, format: false, as: :tool_versions_descriptor
         get 'tools/:id/versions/:version_id/:type/files' => 'tool_versions#files', format: false

lib/seek/config_setting_attributes.yml

@@ -240,8 +240,8 @@ life_monitor_url:
 life_monitor_client_id:
 life_monitor_client_secret:
 life_monitor_ui_url:
-galaxy_instance_name:
-galaxy_instance_trs_import_url:
+galaxy_instance_default:
+galaxy_instance_trs_server:
 # Controlled vocabs
 cv_dropdown_limit:
   convert: :to_i

lib/seek/download_handling/galaxy_http_handler.rb

@@ -10,11 +10,20 @@ module DownloadHandling
     class GalaxyHTTPHandler < Seek::DownloadHandling::HTTPHandler
       attr_reader :galaxy_host, :workflow_id
 
-      def initialize(url, fallback_to_get: true)
-        uri = URI(url)
+      URL_PATTERNS = [
+        /(.+)\/api\/workflows\/([^\/]+)\/download\?format=json-download/, # Download
+        /(.+)\/workflows\/run\?id=([^&]+)/, # Run
+        /(.+)\/published\/workflow\?id=([^&]+)/, # View
+      ].freeze
 
-        @galaxy_host = URI(url.split(/\/workflows?\//).first + '/')
-        @workflow_id = CGI.parse(uri.query)['id'].first
+      def initialize(url, fallback_to_get: true)
+        URL_PATTERNS.each do |pattern|
+          matches = url.match(pattern)
+          if matches
+            @galaxy_host = matches[1].chomp('/') + '/'
+            @workflow_id = matches[2]
+          end
+        end
 
         super(download_url, fallback_to_get: fallback_to_get)
       end
@@ -26,21 +35,24 @@ def info
       end
 
       def display_url
-        URI.join(galaxy_host, "workflow/display_by_id?id=#{workflow_id}").to_s
+        URI.join(galaxy_host, "published/workflow?id=#{workflow_id}").to_s
       end
 
       def download_url
-        URI.join(galaxy_host, "workflow/export_to_file?id=#{workflow_id}").to_s
+        URI.join(galaxy_host, "api/workflows/#{workflow_id}/download?format=json-download").to_s
       end
 
-      # Note that the path is `/workflows/` (plural) here for some reason.
       def run_url
         URI.join(galaxy_host, "workflows/run?id=#{workflow_id}").to_s
       end
 
+      def execution_instance_url
+        galaxy_host.to_s
+      end
+
       def self.is_galaxy_workflow_url?(uri)
-        uri.hostname.include?('galaxy') && (uri.path.include?('/workflow/') || uri.path.include?('/workflows/')) &&
-          uri.query.present? && CGI.parse(uri.query)&.key?('id')
+        string_uri = uri.to_s
+        uri.hostname.include?('galaxy') && URL_PATTERNS.any? { |pattern| string_uri.match?(pattern) }
       end
     end
   end

lib/seek/workflow_extractors/base.rb

@@ -12,7 +12,13 @@ def initialize(io)
       end
 
       def metadata
-        { }
+        m = {}
+
+        if @io.respond_to?(:remote_content_handler)
+          m.merge!(extract_source_metadata(@io.remote_content_handler))
+        end
+
+        m
       end
 
       def has_tests?
@@ -84,6 +90,24 @@ def extract_author(obj)
 
         author
       end
+
+      def extract_source_metadata(handler)
+        m = {}
+        return m unless handler
+        source_url = nil
+        if handler.respond_to?(:repository_url)
+          source_url = handler.repository_url
+        elsif handler.respond_to?(:display_url)
+          source_url = handler.display_url
+        end
+        m[:source_link_url] = source_url
+
+        if handler.respond_to?(:execution_instance_url)
+          m[:execution_instance_url] = handler.execution_instance_url
+        end
+
+        m
+      end
     end
   end
 end

lib/seek/workflow_extractors/cff.rb

@@ -6,7 +6,7 @@ class CFF
       FILENAME = 'CITATION.cff'
 
       def initialize(io)
-        if io.respond_to?(:path)
+        if io.respond_to?(:path) && !io.is_a?(Git::Blob)
           @path = io.path
         else
           f = Tempfile.new('cff')

lib/seek/workflow_extractors/galaxy.rb

@@ -8,6 +8,7 @@ def self.file_extensions
       end
 
       def metadata
+        metadata = super
         galaxy_string = @io.read
         f = Tempfile.new('ga')
         f.binmode
@@ -21,13 +22,13 @@ def metadata
         end
         cf.rewind
         if status.success?
-          metadata = Seek::WorkflowExtractors::CWL.new(cf).metadata
+          metadata.merge!(Seek::WorkflowExtractors::CWL.new(cf).metadata)
         else
-          metadata = super
           metadata[:warnings] ||= []
           metadata[:warnings] << 'Unable to convert workflow to CWL, some metadata may be missing.'
           Rails.logger.error("Galaxy -> CWL conversion failed. Error was: #{err}")
         end
+
         galaxy = JSON.parse(galaxy_string)
 
         if galaxy.has_key?('name')

lib/seek/workflow_extractors/git_repo.rb

@@ -32,7 +32,7 @@ def file_exists?(path)
       end
 
       def file(path)
-        @obj.get_blob(path)&.file(fetch_remote: true)
+        @obj.get_blob(path)&.tap { |blob| blob.fetch_remote = true }
       end
 
       def licensee_project

lib/seek/workflow_extractors/knime.rb

@@ -8,7 +8,7 @@ def self.file_extensions
       def metadata
         metadata = super
 
-        metadata.merge(parse_internal_workflow(extract_workflow))
+        metadata.merge!(parse_internal_workflow(extract_workflow))
 
         metadata
       end