Use README contents for description of nfcore workflows

seek4science · Jan 31, 2024 · f1221a5 · f1221a5
1 parent bc4e97d
commit f1221a5
Show file tree

Hide file tree

Showing 7 changed files with 26 additions and 23 deletions.
diff --git a/app/models/concerns/workflow_extraction.rb b/app/models/concerns/workflow_extraction.rb
@@ -13,15 +13,15 @@ def extractor_class
     workflow_class&.extractor_class || Seek::WorkflowExtractors::Base
   end
 
-  def extractor
+  def extractor(opts = {})
     if is_git_ro_crate?
-      Seek::WorkflowExtractors::ROCrate.new(git_version, main_workflow_class: workflow_class)
+      Seek::WorkflowExtractors::ROCrate.new(git_version, main_workflow_class: workflow_class, **opts)
     elsif is_already_ro_crate?
-      Seek::WorkflowExtractors::ROCrate.new(content_blob, main_workflow_class: workflow_class)
+      Seek::WorkflowExtractors::ROCrate.new(content_blob, main_workflow_class: workflow_class, **opts)
     elsif is_git_versioned?
-      Seek::WorkflowExtractors::GitRepo.new(git_version, main_workflow_class: workflow_class)
+      Seek::WorkflowExtractors::GitRepo.new(git_version, main_workflow_class: workflow_class, **opts)
     else
-      extractor_class.new(content_blob)
+      extractor_class.new(content_blob, **opts)
     end
   end
 

diff --git a/app/models/git_workflow_wizard.rb b/app/models/git_workflow_wizard.rb
@@ -10,7 +10,7 @@ class GitWorkflowWizard
 
   attr_reader :next_step, :git_repository
 
-  attr_accessor :params, :workflow, :workflow_class
+  attr_accessor :params, :workflow, :workflow_class, :use_readme_for_description
 
   def run
     if new_version?
@@ -78,7 +78,7 @@ def run
       return workflow
     end
 
-    extractor = workflow.extractor
+    extractor = workflow.extractor(use_readme_for_description: @use_readme_for_description)
     workflow.provide_metadata(extractor.metadata)
 
     @next_step = :provide_metadata

diff --git a/lib/scrapers/nfcore_scraper.rb b/lib/scrapers/nfcore_scraper.rb
@@ -20,15 +20,16 @@ def main_branch(repo)
 
     def workflow_wizard(repo, tag)
       GitWorkflowWizard.new(workflow_class: WorkflowClass.find_by_key('nextflow'),
-        params: {
-        git_version_attributes: {
-          main_workflow_path: 'nextflow.config',
-          git_repository_id: repo.id,
-          ref: "refs/tags/#{tag}",
-          name: tag,
-          comment: "Updated to #{tag}"
-        }
-      })
+                            use_readme_for_description: true,
+                            params: {
+                              git_version_attributes: {
+                                main_workflow_path: 'nextflow.config',
+                                git_repository_id: repo.id,
+                                ref: "refs/tags/#{tag}",
+                                name: tag,
+                                comment: "Updated to #{tag}"
+                              }
+                            })
     end
   end
-end
+end
diff --git a/lib/seek/workflow_extractors/base.rb b/lib/seek/workflow_extractors/base.rb
@@ -7,7 +7,7 @@ class Base
           "name" => "Unrecognized Workflow Type"
       }
 
-      def initialize(io)
+      def initialize(io, opts = {})
         @io = io.is_a?(String) ? StringIO.new(io) : io
       end
 

diff --git a/lib/seek/workflow_extractors/cff.rb b/lib/seek/workflow_extractors/cff.rb
@@ -5,7 +5,7 @@ module WorkflowExtractors
     class CFF
       FILENAME = 'CITATION.cff'
 
-      def initialize(io)
+      def initialize(io, opts = {})
         if io.respond_to?(:path)
           @path = io.path
         else

diff --git a/lib/seek/workflow_extractors/ro_like.rb b/lib/seek/workflow_extractors/ro_like.rb
@@ -7,9 +7,10 @@ module WorkflowExtractors
     # Abstract extractor class for a "Research Object-like" structured bundle of files,
     # e.g. an RO-Crate or an annotated Git repository.
     class ROLike < Base
-      def initialize(obj, main_workflow_class: nil)
+      def initialize(obj, main_workflow_class: nil, use_readme_for_description: false)
         @obj = obj
         @main_workflow_class = main_workflow_class
+        @use_readme_for_description = use_readme_for_description
       end
 
       def can_render_diagram?
@@ -56,7 +57,8 @@ def metadata
             end
 
         if file_exists?('README.md')
-          m[:description] ||= file('README.md').read.force_encoding('utf-8').gsub(/^(---\s*\n.*?\n?)^(---\s*$\n?)/m,'') # Remove "Front matter"
+          readme = file('README.md').read.force_encoding('utf-8').gsub(/^(---\s*\n.*?\n?)^(---\s*$\n?)/m,'') # Remove "Front matter"
+          m[:description] = readme if readme.present? && (m[:description].blank? || @use_readme_for_description)
         end
 
         m[:workflow_class_id] ||= main_workflow_class&.id

diff --git a/test/integration/nfcore_scraper_test.rb b/test/integration/nfcore_scraper_test.rb
@@ -25,7 +25,7 @@ class NfcoreScraperTest < ActionDispatch::IntegrationTest
               assert_equal bot, wf.contributor
               assert_equal [project], wf.projects
               assert_equal 'nf-core/rnaseq', wf.title
-              assert_equal 'Nextflow RNA-Seq analysis pipeline, part of the nf-core community.', wf.description
+              assert_includes wf.description, '**nf-core/rnaseq** is a bioinformatics analysis pipeline'
               assert_equal 'MIT', wf.license
               assert_equal 'nextflow.config', wf.main_workflow_path
               assert_equal '3.0', wf.git_version.name
@@ -66,7 +66,7 @@ class NfcoreScraperTest < ActionDispatch::IntegrationTest
               assert_equal bot, wf.contributor
               assert_equal [project], wf.projects
               assert_equal 'nf-core/rnaseq', wf.title
-              assert_equal 'Nextflow RNA-Seq analysis pipeline, part of the nf-core community.', wf.description
+              assert_includes wf.description, '**nf-core/rnaseq** is a bioinformatics analysis pipeline'
               assert_equal 'MIT', wf.license
               assert_equal 'nextflow.config', wf.main_workflow_path
               assert_equal '3.0', wf.git_version.name