Merge pull request #153 from OpenGeoMetadata/repo-warnings

Warn if cloning an empty or archived OGM repo

lib/geo_combine/harvester.rb

@@ -75,10 +75,15 @@ def pull_all
     # If the repository already exists, skip it.
     def clone(repo)
       repo_path = File.join(@ogm_path, repo)
+      repo_info = repository_info(repo)
+
+      # Skip if exists; warn if archived or empty
       if File.directory? repo_path
         puts "Skipping clone to #{repo_path}; directory exists"
         return 0
       end
+      puts "WARNING: repository '#{repo}' is archived" if repo_info['archived']
+      puts "WARNING: repository '#{repo}' is empty" if repo_info['size'].zero?
 
       repo_url = "https://github.com/OpenGeoMetadata/#{repo}.git"
       Git.clone(repo_url, nil, path: ogm_path, depth: 1)
@@ -102,5 +107,9 @@ def repositories
                             .map { |repo| repo['name'] }
                             .reject { |name| self.class.denylist.include? name }
     end
+
+    def repository_info(repo_name)
+      JSON.parse(Net::HTTP.get(URI("https://api.github.com/repos/opengeometadata/#{repo_name}")))
+    end
   end
 end

spec/lib/geo_combine/harvester_spec.rb

@@ -18,11 +18,22 @@
       { name: 'outdated-institution', size: 100, archived: true }, # archived
       { name: 'aardvark', size: 300 },                             # on denylist
       { name: 'empty', size: 0 }                                   # no data
-    ].to_json
+    ]
   end
 
   before do
-    allow(Net::HTTP).to receive(:get).with(described_class.ogm_api_uri).and_return(stub_gh_api)
+    # stub github API requests
+    # use the whole org response, or just a portion for particular repos
+    allow(Net::HTTP).to receive(:get) do |uri|
+      if uri == described_class.ogm_api_uri
+        stub_gh_api.to_json
+      else
+        repo_name = uri.path.split('/').last.gsub('.git', '')
+        stub_gh_api.find { |repo| repo[:name] == repo_name }.to_json
+      end
+    end
+
+    # stub git commands
     allow(Git).to receive(:open).and_return(stub_repo)
     allow(Git).to receive(:clone).and_return(stub_repo)
     allow(stub_repo).to receive(:pull).and_return(stub_repo)
@@ -96,6 +107,20 @@
       harvester.clone(repo_name)
       expect(Git).not_to have_received(:clone)
     end
+
+    it 'warns if a repository is empty' do
+      allow(Net::HTTP).to receive(:get).with('https://api.github.com/repos/opengeometadata/empty').and_return('{"size": 0}')
+      expect do
+        harvester.clone('empty')
+      end.to output(/repository 'empty' is empty/).to_stdout
+    end
+
+    it 'warns if a repository is archived' do
+      allow(Net::HTTP).to receive(:get).with('https://api.github.com/repos/opengeometadata/empty').and_return('{"archived": true}')
+      expect do
+        harvester.clone('outdated-institution')
+      end.to output(/repository 'outdated-institution' is archived/).to_stdout
+    end
   end
 
   describe '#clone_all' do