Build a composite license that traverses and expression tree

4 files changed, 119 insertions, 18 deletions

diff --git a/lib/spandx/core/guess.rb b/lib/spandx/core/guess.rb
index 4383374..62967ce 100644
--- a/lib/spandx/core/guess.rb
+++ b/lib/spandx/core/guess.rb
@@ -7,7 +7,6 @@ module Spandx
 
       def initialize(catalogue)
         @catalogue = catalogue
-        @expression_cache = {}
       end
 
       def license_for(raw)
@@ -79,12 +78,8 @@ module Spandx
       end
 
       def from_expression(content)
-        @expression_cache.fetch(content.raw) do
-          tree = Spdx::Expression.new.parse(content.raw)
-          @expression_cache[content.raw] = catalogue[tree[0][:left].to_s]
-        end
-      rescue Parslet::ParseFailed
-        nil
+        Spandx::Spdx::CompositeLicense
+          .from_expression(content.raw, catalogue)
       end
     end
   end
diff --git a/lib/spandx/spdx/composite_license.rb b/lib/spandx/spdx/composite_license.rb
new file mode 100644
index 0000000..e429778
--- /dev/null
+++ b/lib/spandx/spdx/composite_license.rb
@@ -0,0 +1,59 @@
+
+module Spandx
+  module Spdx
+    class CompositeLicense < License
+      def self.from_expression(expression, catalogue)
+        tree = Spdx::Expression.new.parse(expression)
+        new(tree[0], catalogue)
+      rescue Parslet::ParseFailed
+        nil
+      end
+
+      def initialize(tree, catalogue)
+        @catalogue = catalogue
+        @tree = tree
+        super({ })
+      end
+
+      def id
+        if right
+          ["(#{left.id}", operator, "#{right.id})"].compact.join(' ').squeeze(' ').strip
+        else
+          "#{left.id}"
+        end
+      end
+
+      def name
+        if right
+          [left.name, operator, right.name].compact.join(' ').squeeze(' ').strip
+        else
+          left.name
+        end
+      end
+
+      private
+
+      def left
+        node_for(@tree[:left])
+      end
+
+      def operator
+        @tree[:op].to_s.upcase
+      end
+
+      def right
+        node_for(@tree[:right])
+      end
+
+      def node_for(item)
+        return if item.nil?
+
+        if item.is_a?(Hash)
+          self.class.new(item, @catalogue)
+        else
+          @catalogue[item.to_s] || License.unknown(item.to_s)
+        end
+      end
+    end
+  end
+end
diff --git a/spec/unit/core/guess_spec.rb b/spec/unit/core/guess_spec.rb
index f19aa18..63f0bf2 100644
--- a/spec/unit/core/guess_spec.rb
+++ b/spec/unit/core/guess_spec.rb
@@ -44,20 +44,20 @@ RSpec.describe Spandx::Core::Guess do
       specify { expect(subject.license_for(content)&.id).to eql('MIT') }
     end
 
-    specify { expect(subject.license_for('(0BSD OR MIT)')&.id).to eql('0BSD') }
-    specify { expect(subject.license_for('(BSD-2-Clause OR MIT OR Apache-2.0)')&.id).to eql('BSD-2-Clause') }
-    specify { expect(subject.license_for('(BSD-3-Clause OR GPL-2.0)')&.id).to eql('BSD-3-Clause') }
-    specify { expect(subject.license_for('(MIT AND CC-BY-3.0)')&.id).to eql('MIT') }
-    specify { expect(subject.license_for('(MIT AND Zlib)')&.id).to eql('MIT') }
-    specify { expect(subject.license_for('(MIT OR Apache-2.0)')&.id).to eql('MIT') }
-    specify { expect(subject.license_for('(MIT OR CC0-1.0)')&.id).to eql('MIT') }
-    specify { expect(subject.license_for('(MIT OR GPL-3.0)')&.id).to eql('MIT') }
-    specify { expect(subject.license_for('(WTFPL OR MIT)')&.id).to eql('WTFPL') }
+    specify { expect(subject.license_for('(0BSD OR MIT)')&.id).to eql('(0BSD OR MIT)') }
+    specify { expect(subject.license_for('(BSD-2-Clause OR MIT OR Apache-2.0)')&.id).to eql('(BSD-2-Clause OR MIT OR Apache-2.0)') }
+    specify { expect(subject.license_for('(BSD-3-Clause OR GPL-2.0)')&.id).to eql('(BSD-3-Clause OR GPL-2.0)') }
+    specify { expect(subject.license_for('(MIT AND CC-BY-3.0)')&.id).to eql('(MIT AND CC-BY-3.0)') }
+    specify { expect(subject.license_for('(MIT AND Zlib)')&.id).to eql('(MIT AND Zlib)') }
+    specify { expect(subject.license_for('(MIT OR Apache-2.0)')&.id).to eql('(MIT OR Apache-2.0)') }
+    specify { expect(subject.license_for('(MIT OR CC0-1.0)')&.id).to eql('(MIT OR CC0-1.0)') }
+    specify { expect(subject.license_for('(MIT OR GPL-3.0)')&.id).to eql('(MIT OR GPL-3.0)') }
+    specify { expect(subject.license_for('(WTFPL OR MIT)')&.id).to eql('(WTFPL OR MIT)') }
     specify { expect(subject.license_for('Apache 2.0')&.id).to eql('Apache-2.0') }
-    specify { expect(subject.license_for('BSD-3-Clause OR MIT')&.id).to eql('BSD-3-Clause') }
+    specify { expect(subject.license_for('BSD-3-Clause OR MIT')&.id).to eql('(BSD-3-Clause OR MIT)') }
     specify { expect(subject.license_for('BSD-like')&.id).to eql('Nonstandard') }
     specify { expect(subject.license_for('Common Public License Version 1.0')&.id).to eql('CPL-1.0') }
-    specify { expect(subject.license_for('MIT or GPLv3')&.id).to eql('MIT') }
+    specify { expect(subject.license_for('MIT or GPLv3')&.id).to eql('(MIT OR Nonstandard)') }
     pending { expect(subject.license_for('MIT/X11')&.id).to eql('X11') }
   end
 end
diff --git a/spec/unit/spdx/composite_license_spec.rb b/spec/unit/spdx/composite_license_spec.rb
new file mode 100644
index 0000000..84dca0c
--- /dev/null
+++ b/spec/unit/spdx/composite_license_spec.rb
@@ -0,0 +1,47 @@
+RSpec.describe Spandx::Spdx::CompositeLicense do
+  context ".from_expression" do
+    subject { described_class.from_expression(expression, catalogue) }
+    let(:catalogue) { ::Spandx::Spdx::Catalogue.from_file(fixture_file('spdx/json/licenses.json')) }
+
+    before do
+      puts subject.instance_variable_get(:@tree).inspect
+    end
+
+    context 'when parsing a simple binary expression' do
+      let(:expression) { '(0BSD OR MIT)' }
+
+      specify { expect(subject.id).to eql(expression) }
+      specify { expect(subject.name).to eql("#{catalogue['0BSD'].name} OR #{catalogue['MIT'].name}") }
+      specify { expect(subject).to be_kind_of(::Spandx::Spdx::License) }
+    end
+
+    context 'when parsing an expression with a valid and an invalid license id' do
+      let(:expression) { 'MIT or GPLv3' }
+
+      specify { expect(subject.id).to eql("(MIT OR Nonstandard)") }
+      specify { expect(subject.name).to eql("#{catalogue['MIT'].name} OR GPLv3") }
+      specify { expect(subject).to be_kind_of(::Spandx::Spdx::License) }
+    end
+
+    context 'when parsing a license name' do
+      let(:expression) { 'Common Public License Version 1.0' }
+
+      specify { expect(subject).to be_nil }
+    end
+
+    context 'when parsing an invalid expression' do
+      let(:expression) { 'BSD-like' }
+
+      specify { expect(subject.id).to eql('Nonstandard') }
+      specify { expect(subject.name).to eql('BSD-like') }
+    end
+
+    context 'when parsing a binary expression with an id that is similar to another' do
+      let(:expression) { '(MIT OR CC0-1.0)' }
+
+      specify { expect(subject.id).to eql(expression) }
+      specify { expect(subject.name).to eql("#{catalogue['MIT'].name} OR #{catalogue['CC0-1.0'].name}") }
+      specify { expect(subject).to be_kind_of(::Spandx::Spdx::License) }
+    end
+  end
+end