Condense gvt paths with identical shas into their common path

[Finish #155382947, #154072994, #155830792]

Signed-off-by: Vikram Yadav <[email protected]>

lib/license_finder/package_managers/gvt.rb

@@ -1,3 +1,4 @@
+require 'license_finder/shared_helpers/common_path'
 module LicenseFinder
   class Gvt < PackageManager
     def possible_package_paths
@@ -47,12 +48,32 @@ def self.takes_priority_over
 
     def packages_from_output(output, path)
       package_lines = output.split("\n")
-      package_lines.map do |package_line|
-        import_path, revision, repo = package_line.split
+      packages_by_sha = {}
+      package_lines.each do |p|
+        package_path, sha, repo = p.split
+        if packages_by_sha[sha].nil?
+          packages_by_sha[sha] = {}
+          packages_by_sha[sha]['paths'] = [package_path]
+          packages_by_sha[sha]['repo'] = repo
+        else
+          packages_by_sha[sha]['paths'] << package_path
+        end
+      end
+
+      result = []
+      packages_by_sha.each do |sha, info|
+        paths = CommonPathHelper.shortest_common_paths(info['paths'])
+
+        paths.each { |p| result << [sha, p, info['repo']] }
+      end
+
+      result.map do |package_info|
+        sha, import_path, repo = package_info
+
         GoPackage.from_dependency({
                                     'ImportPath' => import_path,
                                     'InstallPath' => path.join(import_path),
-                                    'Rev' => revision,
+                                    'Rev' => sha,
                                     'Homepage' => repo
                                   }, nil, true)
       end

lib/license_finder/shared_helpers/common_path.rb

@@ -0,0 +1,25 @@
+module CommonPathHelper
+  def self.shortest_common_paths(paths)
+    [].tap do |common_paths|
+      # organize by matching root paths
+      paths_with_roots = paths.group_by { |path| path.split('/').first }
+      paths_with_roots.each do |common_root, full_paths|
+        # use the shortest path as the 'template'
+        shortest_path = full_paths.sort_by { |path| path.split('/').length }.first
+        shortest_common_path = common_root
+
+        # iterate through each subpath of the 'template'
+        shortest_path.split('/').each_with_index do |subpath, i|
+          potential_path = i.zero? ? shortest_common_path : [shortest_common_path, subpath].join('/')
+
+          # check each for the existence of the subsequent subpath
+          mismatch = full_paths.any? { |path| !path.start_with?(potential_path) }
+          break if mismatch
+
+          shortest_common_path = potential_path
+        end
+        common_paths << shortest_common_path
+      end
+    end
+  end
+end

spec/lib/license_finder/package_managers/gvt_spec.rb

@@ -70,6 +70,61 @@ module LicenseFinder
         end
       end
 
+      context 'when the GVT returns entries with same sha with common base path' do
+        let(:gvt_output_with_common_paths) do
+          <<OUTPUT
+cloud.google.com/go/bigquery abcd gcloud-repo
+cloud.google.com/go/civil abcd gcloud-repo
+cloud.google.com/go/compute/metadata abcd gcloud-repo
+OUTPUT
+        end
+        let(:gvt_output_without_common_paths) do
+          <<OUTPUT
+cloud.google.com/go/bigquery/adsf abcd gcloud-repo
+cloud.google.com/go/civil abcd gcloud-repo
+cloud.aws.com/go/metadata abcd gcloud-repo
+OUTPUT
+        end
+
+        before do
+          FileUtils.mkdir_p '/app/vendor'
+          File.write('/app/vendor/manifest', content)
+        end
+
+        it 'only shows the entry with common base path once' do
+          allow(SharedHelpers::Cmd).to receive(:run).with('gvt list -f "{{.Importpath}} {{.Revision}} {{.Repository}}"') do
+            [gvt_output_with_common_paths, '', cmd_success]
+          end
+          expect(subject.current_packages.length).to eq 1
+
+          package = subject.current_packages.first
+          expect(package.name).to eq 'cloud.google.com/go'
+          expect(package.install_path).to eq Pathname('/app/vendor/cloud.google.com/go')
+          expect(package.version).to eq 'abcd'
+          expect(package.homepage).to eq 'gcloud-repo'
+        end
+
+        it 'shows entries with same sha when they do not have a common base path' do
+          allow(SharedHelpers::Cmd).to receive(:run).with('gvt list -f "{{.Importpath}} {{.Revision}} {{.Repository}}"') do
+            [gvt_output_without_common_paths, '', cmd_success]
+          end
+
+          expect(subject.current_packages.length).to eq 2
+
+          first = subject.current_packages.first
+          expect(first.name).to eq 'cloud.google.com/go'
+          expect(first.install_path).to eq Pathname('/app/vendor/cloud.google.com/go')
+          expect(first.version).to eq 'abcd'
+          expect(first.homepage).to eq 'gcloud-repo'
+
+          last = subject.current_packages.last
+          expect(last.name).to eq 'cloud.aws.com/go/metadata'
+          expect(last.install_path).to eq Pathname('/app/vendor/cloud.aws.com/go/metadata')
+          expect(last.version).to eq 'abcd'
+          expect(last.homepage).to eq 'gcloud-repo'
+        end
+      end
+
       it 'returns empty package list if \'gvt list\' fails' do
         allow(SharedHelpers::Cmd).to receive(:run).with(anything) do
           ["my-package-name 123abc example.com\npackage-name-2 456xyz anotherurl.com", '', cmd_failure]

spec/lib/license_finder/shared_helpers/common_path_spec.rb

@@ -0,0 +1,23 @@
+require 'spec_helper'
+require 'license_finder/shared_helpers/common_path'
+
+describe CommonPathHelper do
+  context 'when the GVT returns entries with same sha with common base path' do
+    let(:gvt_output_with_common_paths) do
+      %w[cloud.google.com/go/bigquery cloud.google.com/go/civil cloud.google.com/go/compute/metadata]
+    end
+    let(:gvt_output_without_common_paths) do
+      %w[cloud.google.com/go/bigquery/adsf cloud.google.com/go/civil cloud.aws.com/go/metadata]
+    end
+
+    it 'only shows the entry with common base path once' do
+      paths = CommonPathHelper.shortest_common_paths gvt_output_with_common_paths
+      expect(paths).to match_array %w[cloud.google.com/go]
+    end
+
+    it 'shows entries with same sha when they do not have a common base path' do
+      paths = CommonPathHelper.shortest_common_paths gvt_output_without_common_paths
+      expect(paths).to match_array %w[cloud.google.com/go cloud.aws.com/go/metadata]
+    end
+  end
+end