code_ownership 1.32.8 → 1.32.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 18e702e6635a7d3d264eaae41f61fb6f83e423e8021ca76e92a6972fe5444848
-  data.tar.gz: 90f34b5a02b21a1f78ff781f7ef3e8220a45549e77c4b93b829381a57f37ba7c
+  metadata.gz: 71768f254f9fbb891679c002c16c20bd70a2125460e3c66995821cfe2784fe56
+  data.tar.gz: ce5698b1505d306026f807a142f648942ea06d3ab65576d05c623f6bdbdc1dfa
 SHA512:
-  metadata.gz: f1deba0493c9ebbff4e39b81c0a282ffd634cef6db756f3f4be2b3b91f8c43e0d0f6d9362e03f2779ac5fe302101ae19fa8bf90022c55f24afbd9cd7c3b1fb13
-  data.tar.gz: 0a1a3da4c7c758b0b64b7d9f817a92924c7a5e54fb4918655af2208a5ba8c65f12d41b37f0f568f443a817dd5893d16e351de745be641b3a56d3479e257add45
+  metadata.gz: 7b2e5da9f4e02b4db10c47be40b1b4bce33f7c44ad18797ff326a6e245f5dc57b2a8534c74fde74a1966970720a9bedf50d59858923e07f64d8e21d9f8411df1
+  data.tar.gz: b08f25b9008d86b19207931cb44c13adea814b998af8700b1fa445ad09e48e6549f86d6e4b75db41e5a783bf080ea5e25371d97db5307de8c5451bcd3d4fe3da

data/lib/code_ownership/cli.rb

@@ -61,7 +61,7 @@ module CodeOwnership
           File.exist?(file)
         end
       else
-        Private.tracked_files
+        nil
       end
 
       CodeOwnership.validate!(

data/lib/code_ownership/private/glob_cache.rb

@@ -18,7 +18,7 @@ module CodeOwnership
       FilesByMapper = T.type_alias do
         T::Hash[
           String,
-          T::Array[MapperDescription]
+          T::Set[MapperDescription]
         ]
       end
 
@@ -32,6 +32,20 @@ module CodeOwnership
         @raw_cache_contents
       end
 
+      sig { params(files: T::Array[String]).returns(FilesByMapper) }
+      def mapper_descriptions_that_map_files(files)
+        if files.count > 100
+          # When looking at many files, expanding the cache out using Dir.glob and checking for intersections is faster
+          files_by_mappers = files.map{ |f| [f, Set.new([]) ]}.to_h
+          files_by_mappers.merge(files_by_mappers_via_expanded_cache)
+        else
+          # When looking at few files, using File.fnmatch is faster
+          files_by_mappers_via_file_fnmatch(files)
+        end
+      end
+
+      private
+
       sig { returns(CacheShape) }
       def expanded_cache
         @expanded_cache = T.let(@expanded_cache, T.nilable(CacheShape))
@@ -52,20 +66,45 @@ module CodeOwnership
       end
 
       sig { returns(FilesByMapper) }
-      def files_by_mapper
-        @files_by_mapper ||= T.let(@files_by_mapper, T.nilable(FilesByMapper))
-        @files_by_mapper ||= begin
-          files_by_mapper = {}
+      def files_by_mappers_via_expanded_cache
+        @files_by_mappers ||= T.let(@files_by_mappers, T.nilable(FilesByMapper))
+        @files_by_mappers ||= begin
+          files_by_mappers = T.let({}, FilesByMapper)
           expanded_cache.each do |mapper_description, file_by_owner|
             file_by_owner.each do |file, owner|
-              files_by_mapper[file] ||= []
-              files_by_mapper[file] << mapper_description
+              files_by_mappers[file] ||= Set.new([])
+              files_by_mappers.fetch(file) << mapper_description
             end
           end
 
-          files_by_mapper
+          files_by_mappers
         end
       end
+
+      sig { params(files: T::Array[String]).returns(FilesByMapper) }
+      def files_by_mappers_via_file_fnmatch(files)
+        files_by_mappers = T.let({}, FilesByMapper)
+
+        files.each do |file|
+          files_by_mappers[file] ||= Set.new([])
+          @raw_cache_contents.each do |mapper_description, globs_by_owner|
+            # As much as I'd like to *not* special case the file annotations mapper, using File.fnmatch? on the thousands of files mapped by the
+            # file annotations mapper is a lot of unnecessary extra work.
+            # Therefore we can just check if the file is in the globs directly for file annotations, otherwise use File.fnmatch
+            if mapper_description == OwnershipMappers::FileAnnotations::DESCRIPTION
+              files_by_mappers.fetch(file) << mapper_description if globs_by_owner[file]
+            else
+              globs_by_owner.each do |glob, owner|
+                if File.fnmatch?(glob, file, File::FNM_PATHNAME | File::FNM_EXTGLOB)
+                  files_by_mappers.fetch(file) << mapper_description
+                end
+              end
+            end
+          end
+        end
+
+        files_by_mappers
+      end
     end
   end
 end

data/lib/code_ownership/private/ownership_mappers/file_annotations.rb

@@ -21,6 +21,7 @@ module CodeOwnership
         @@map_files_to_owners = T.let({}, T.nilable(T::Hash[String, ::CodeTeams::Team])) # rubocop:disable Style/ClassVars
 
         TEAM_PATTERN = T.let(/\A(?:#|\/\/) @team (?<team>.*)\Z/.freeze, Regexp)
+        DESCRIPTION = 'Annotations at the top of file'
 
         sig do
           override.params(file: String).
@@ -51,11 +52,8 @@ module CodeOwnership
         end
         def update_cache(cache, files)
           cache.merge!(globs_to_owner(files))
-
-          # TODO: Make `tracked_files` return a set
-          fileset = Set.new(Private.tracked_files)
           invalid_files = cache.keys.select do |file|
-            !fileset.include?(file)
+            !Private.file_tracked?(file)
           end
           invalid_files.each do |invalid_file|
             cache.delete(invalid_file)
@@ -116,7 +114,7 @@ module CodeOwnership
 
         sig { override.returns(String) }
         def description
-          'Annotations at the top of file'
+          DESCRIPTION
         end
 
         sig { override.void }

data/lib/code_ownership/private/validations/files_have_owners.rb

@@ -10,10 +10,10 @@ module CodeOwnership
 
         sig { override.params(files: T::Array[String], autocorrect: T::Boolean, stage_changes: T::Boolean).returns(T::Array[String]) }
         def validation_errors(files:, autocorrect: true, stage_changes: true)
-          files_by_mapper = Private.glob_cache.files_by_mapper
-
-          files_not_mapped_at_all = files.select do |file|
-            files_by_mapper.fetch(file, []).count == 0
+          cache = Private.glob_cache
+          file_mappings = cache.mapper_descriptions_that_map_files(files)
+          files_not_mapped_at_all = file_mappings.select do |file, mapper_descriptions|
+            mapper_descriptions.count == 0
           end
 
           errors = T.let([], T::Array[String])
@@ -22,7 +22,7 @@ module CodeOwnership
             errors << <<~MSG
               Some files are missing ownership:
 
-              #{files_not_mapped_at_all.map { |file| "- #{file}" }.join("\n")}
+              #{files_not_mapped_at_all.map { |file, mappers| "- #{file}" }.join("\n")}
             MSG
           end
 

data/lib/code_ownership/private/validations/files_have_unique_owners.rb

@@ -10,14 +10,10 @@ module CodeOwnership
 
         sig { override.params(files: T::Array[String], autocorrect: T::Boolean, stage_changes: T::Boolean).returns(T::Array[String]) }
         def validation_errors(files:, autocorrect: true, stage_changes: true)
-          files_by_mapper = Private.glob_cache.files_by_mapper
-
-          files_mapped_by_multiple_mappers = {}
-          files.each do |file|
-            mappers = files_by_mapper.fetch(file, [])
-            if mappers.count > 1
-              files_mapped_by_multiple_mappers[file] = mappers
-            end
+          cache = Private.glob_cache
+          file_mappings = cache.mapper_descriptions_that_map_files(files)
+          files_mapped_by_multiple_mappers = file_mappings.select do |file, mapper_descriptions|
+            mapper_descriptions.count > 1
           end
 
           errors = T.let([], T::Array[String])
@@ -26,7 +22,7 @@ module CodeOwnership
             errors << <<~MSG
               Code ownership should only be defined for each file in one way. The following files have declared ownership in multiple ways.
 
-              #{files_mapped_by_multiple_mappers.map { |file, descriptions| "- #{file} (#{descriptions.join(', ')})" }.join("\n")}
+              #{files_mapped_by_multiple_mappers.map { |file, descriptions| "- #{file} (#{descriptions.to_a.join(', ')})" }.join("\n")}
             MSG
           end
 

data/lib/code_ownership/private.rb

@@ -81,6 +81,24 @@ module CodeOwnership
       @tracked_files ||= Dir.glob(configuration.owned_globs) - Dir.glob(configuration.unowned_globs)
     end
 
+    sig { params(file: String).returns(T::Boolean) }
+    def self.file_tracked?(file)
+      # Another way to accomplish this is
+      # (Dir.glob(configuration.owned_globs) - Dir.glob(configuration.unowned_globs)).include?(file)
+      # However, globbing out can take 5 or more seconds on a large repository, dramatically slowing down
+      # invocations to `bin/codeownership validate --diff`.
+      # Using `File.fnmatch?` is a lot faster!
+      in_owned_globs = configuration.owned_globs.all? do |owned_glob|
+        File.fnmatch?(owned_glob, file, File::FNM_PATHNAME | File::FNM_EXTGLOB)
+      end
+
+      in_unowned_globs = configuration.unowned_globs.all? do |unowned_glob|
+        File.fnmatch?(unowned_glob, file, File::FNM_PATHNAME | File::FNM_EXTGLOB)
+      end
+
+      in_owned_globs && !in_unowned_globs
+    end
+
     sig { params(team_name: String, location_of_reference: String).returns(CodeTeams::Team) }
     def self.find_team!(team_name, location_of_reference)
       found_team = CodeTeams.find(team_name)

data/lib/code_ownership.rb

@@ -82,18 +82,24 @@ module CodeOwnership
 
   sig do
     params(
-      files: T::Array[String],
       autocorrect: T::Boolean,
-      stage_changes: T::Boolean
+      stage_changes: T::Boolean,
+      files: T.nilable(T::Array[String]),
     ).void
   end
   def validate!(
-    files: Private.tracked_files,
     autocorrect: true,
-    stage_changes: true
+    stage_changes: true,
+    files: nil
   )
     Private.load_configuration!
-    tracked_file_subset = Private.tracked_files & files
+
+    tracked_file_subset = if files
+      files.select{|f| Private.file_tracked?(f)}
+    else
+      Private.tracked_files
+    end
+
     Private.validate!(files: tracked_file_subset, autocorrect: autocorrect, stage_changes: stage_changes)
   end
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: code_ownership
 version: !ruby/object:Gem::Version
-  version: 1.32.8
+  version: 1.32.9
 platform: ruby
 authors:
 - Gusto Engineers