Add new file crystalball predictor to enqueue complete re-run

closes OUT-4922

Test-plan:
- Get a copy of crystalball_map.yml, copy it to canvas root
- Available via #crystalball-noisy
- Look at a file that has changes under a given spec
- make changes to it, and add it to a new commit
- verify via crystalball dry-run that specs are predicted and
  printed out to crystalball_spec_list.txt
- create a new .rb file
- run crystalball dry-run again and verify that the output
  warns about a "new file detected" and that the dry-run
  produces only a "." in the crystalball_spec_list.txt file

Change-Id: Ia2a5754c02814d681f1d375f5bc7357774b08926
Reviewed-on: https://gerrit.instructure.com/c/canvas-lms/+/281613
Tested-by: Service Cloud Jenkins <svc.cloudjenkins@instructure.com>
Reviewed-by: Manoel Quirino <manoel.quirino@instructure.com>
Reviewed-by: Michael Hargiss <mhargiss@instructure.com>
QA-Review: Michael Hargiss <mhargiss@instructure.com>
Product-Review: Brian Watson <bwatson@instructure.com>

config/crystalball.yml

@@ -5,7 +5,7 @@ execution_map_path: './crystalball_map.yml'
 # examples_limit: 1
 #
 # Custom prediction builder class to use. Default: 'Crystalball::RSpec::StandardPredictionBuilder'
-prediction_builder_class_name: 'Crystalball::RSpec::StandardPredictionBuilder'
+prediction_builder_class_name: 'Crystalball::RSpec::CanvasPredictionBuilder'
 # #
 # Set of requires. Usually used to require custom predictor class file. Default: []
 requires:

spec/support/crystalball.rb

@@ -22,6 +22,9 @@ require "crystalball/rspec/prediction_builder"
 require "crystalball/rspec/filtering"
 require "crystalball/rspec/prediction_pruning"
 
+require "crystalball/predictor/strategy"
+require "crystalball/predictor/helpers/affected_example_groups_detector"
+
 module Crystalball
   module RSpec
     # Our custom RSpec runner to generate and save predictions to a file, i.e. "dry-run"
@@ -127,6 +130,73 @@ module Crystalball
         @options.configure(@configuration)
       end
     end
+
+    class CanvasPredictionBuilder < Crystalball::RSpec::PredictionBuilder
+      def predictor
+        super do |p|
+          p.use Crystalball::Predictor::ModifiedSpecs.new
+          p.use Crystalball::Predictor::ModifiedExecutionPaths.new
+          p.use Crystalball::Predictor::NewFiles.new
+        end
+      end
+    end
+  end
+
+  class Predictor
+    # Queues a total re-run if any files are added. If no new files, don't add any predictions
+    # Possible git operation types for SourceDiff include: ['new', 'modified', 'moved', 'deleted]
+    class NewFiles
+      include Helpers::AffectedExampleGroupsDetector
+      include Strategy
+
+      # @param [Crystalball::SourceDiff] diff - the diff from which to predict
+      #   which specs should run
+      # @param [Crystalball::ExampleGroupMap] map - the map with the relations of
+      #   examples and used files
+      # @return [Array<String>] the spec paths associated with the changes
+      def call(diff, map)
+        super do
+          file_change_types = diff.map { |source_diff| [source_diff.relative_path, source_diff.type] }
+          # Create a map of git operations to files
+          # Hash["new"] = ["new_file1.rb", "new_file2.rb"]
+          # Hash["modified"] = ["modified_file1.rb", "modified_file2.rb"]
+          # etc...
+          new_files = file_change_types.each_with_object(Hash.new { |h, k| h[k] = [] }) do |arr, change_map|
+            change_path = arr[0]
+            change_type = arr[1]
+            change_map[change_type] << change_path
+          end
+          if new_files["new"].count.positive?
+            Crystalball.log :warn, "Crystalball detected new .git files: #{new_files["new"]}"
+            Crystalball.log :warn, "Crystalball requesting entire suite re-run"
+            ["."]
+          else
+            []
+          end
+        end
+      end
+    end
+  end
+
+  # Override prediction mechanism based on NewFiles predictor. If we requeue an entire suite re-run
+  # ENV["CRYSTALBALL_TEST_SUITE_ROOT"] should point to the root of selenium specs or whatever is deemed
+  #  relevant for a "complete crystalball-predicted run"
+  class Predictor
+    # @return [Crystalball::Prediction] list of examples which may fail
+    def prediction
+      root_suite_path = ENV["CRYSTALBALL_TEST_SUITE_ROOT"] || "."
+      raw_prediction = raw_prediction(diff)
+      prediction_list = includes_root?(raw_prediction) ? [root_suite_path] : raw_prediction
+      Prediction.new(filter(prediction_list))
+    end
+
+    private
+
+    def includes_root?(prediction_list)
+      prediction_list.include?(".") ||
+        prediction_list.include?("./.") ||
+        prediction_list.include?(ENV["CRYSTALBALL_TEST_SUITE_ROOT"])
+    end
   end
 end