Allow to configure parsers in packwerk (continued)

USAGE.md

@@ -80,6 +80,26 @@ Packwerk reads from the `packwerk.yml` configuration file in the root directory.
 | cache                | false                                     | when true, caches the results of parsing files |
 | cache_directory      | tmp/cache/packwerk                        | the directory that will hold the packwerk cache |
 
+### Using custom parsers
+
+You can specify a custom parser to parse different file formats (e.g. slim or haml)
+
+```ruby
+class SlimParser
+  include Packwerk::FileParser
+
+  REGEX = /\.slim\Z/
+
+  def call
+    # Your parsing logic here
+  end
+
+  def match?(path)
+    REGEX.match?(path)
+  end
+end
+```
+
 ### Using a custom ERB parser
 
 You can specify a custom ERB parser if needed. For example, if you're using `<%graphql>` tags from https://github.com/github/graphql-client in your ERBs, you can use a custom parser subclass to comment them out so that Packwerk can parse the rest of the file:
@@ -99,7 +119,7 @@ class CustomParser < Packwerk::Parsers::Erb
   end
 end
 
-Packwerk::Parsers::Factory.instance.erb_parser_class = CustomParser
+Packwerk::Parsers::Factory.instance.parsers = [Packwerk::Parsers::Ruby, CustomParser]
 ```
 
 ## Using the cache

lib/packwerk.rb

@@ -35,6 +35,7 @@ module Packwerk
   autoload :Package
   autoload :PackageSet
   autoload :ParsedConstantDefinitions
+  autoload :FileParser
   autoload :Parsers
   autoload :ParseRun
   autoload :UnresolvedReference

lib/packwerk/file_parser.rb

@@ -0,0 +1,33 @@
+# typed: strict
+# frozen_string_literal: true
+
+module Packwerk
+  module FileParser
+    extend T::Helpers
+    extend T::Sig
+
+    requires_ancestor { Kernel }
+
+    interface!
+
+    @parsers = T.let([], T::Array[Class])
+
+    sig { params(base: Class).void }
+    def self.included(base)
+      @parsers << base
+    end
+
+    sig { returns(T::Array[FileParser]) }
+    def self.all
+      T.unsafe(@parsers).map(&:new)
+    end
+
+    sig { abstract.params(io: File, file_path: String).returns(T.untyped) }
+    def call(io:, file_path:)
+    end
+
+    sig { abstract.params(path: String).returns(T::Boolean) }
+    def match?(path:)
+    end
+  end
+end

lib/packwerk/file_processor.rb

@@ -40,14 +40,16 @@ def initialize(node_processor_factory:, cache:, parser_factory: nil)
       )
     end
     def call(absolute_file)
-      parser = parser_for(absolute_file)
-      return [UnknownFileTypeResult.new(file: absolute_file)] if T.unsafe(parser).nil?
+      parsers = parsers_for(absolute_file)
+      return [UnknownFileTypeResult.new(file: absolute_file)] if parsers.empty?
 
       @cache.with_cache(absolute_file) do
-        node = parse_into_ast(absolute_file, T.must(parser))
-        return [] unless node
+        parsers.flat_map do |parser|
+          node = parse_into_ast(absolute_file, parser)
+          return [] unless node
 
-        references_from_ast(node, absolute_file)
+          references_from_ast(node, absolute_file)
+        end
       end
     rescue Parsers::ParseError => e
       [e.result]
@@ -56,7 +58,7 @@ def call(absolute_file)
     private
 
     sig do
-      params(node: Parser::AST::Node, absolute_file: String).returns(T::Array[UnresolvedReference])
+      params(node: ::Parser::AST::Node, absolute_file: String).returns(T::Array[UnresolvedReference])
     end
     def references_from_ast(node, absolute_file)
       references = []
@@ -68,15 +70,15 @@ def references_from_ast(node, absolute_file)
       references
     end
 
-    sig { params(absolute_file: String, parser: Parsers::ParserInterface).returns(T.untyped) }
+    sig { params(absolute_file: String, parser: Packwerk::FileParser).returns(T.untyped) }
     def parse_into_ast(absolute_file, parser)
       File.open(absolute_file, "r", nil, external_encoding: Encoding::UTF_8) do |file|
         parser.call(io: file, file_path: absolute_file)
       end
     end
 
-    sig { params(file_path: String).returns(T.nilable(Parsers::ParserInterface)) }
-    def parser_for(file_path)
+    sig { params(file_path: String).returns(T::Array[Packwerk::FileParser]) }
+    def parsers_for(file_path)
       @parser_factory.for_path(file_path)
     end
   end

lib/packwerk/node.rb

@@ -60,7 +60,7 @@ def constant_name(constant_node)
       def each_child(node)
         if block_given?
           node.children.each do |child|
-            yield child if child.is_a?(Parser::AST::Node)
+            yield child if child.is_a?(::Parser::AST::Node)
           end
         else
           enum_for(:each_child, node)

lib/packwerk/node_processor.rb

@@ -19,8 +19,8 @@ def initialize(reference_extractor:, absolute_file:)
 
     sig do
       params(
-        node: Parser::AST::Node,
-        ancestors: T::Array[Parser::AST::Node]
+        node: ::Parser::AST::Node,
+        ancestors: T::Array[::Parser::AST::Node]
       ).returns(T.nilable(UnresolvedReference))
     end
     def call(node, ancestors)

lib/packwerk/parsers/erb.rb

@@ -9,15 +9,18 @@
 module Packwerk
   module Parsers
     class Erb
-      include ParserInterface
+      include Packwerk::FileParser
+
+      ERB_REGEX = /\.erb\Z/
+      private_constant :ERB_REGEX
 
       def initialize(parser_class: BetterHtml::Parser, ruby_parser: Ruby.new)
         @parser_class = parser_class
         @ruby_parser = ruby_parser
       end
 
       def call(io:, file_path: "<unknown>")
-        buffer = Parser::Source::Buffer.new(file_path)
+        buffer = ::Parser::Source::Buffer.new(file_path)
         buffer.source = io.read
         parse_buffer(buffer, file_path: file_path)
       end
@@ -28,11 +31,15 @@ def parse_buffer(buffer, file_path:)
       rescue EncodingError => e
         result = ParseResult.new(file: file_path, message: e.message)
         raise Parsers::ParseError, result
-      rescue Parser::SyntaxError => e
+      rescue ::Parser::SyntaxError => e
         result = ParseResult.new(file: file_path, message: "Syntax error: #{e}")
         raise Parsers::ParseError, result
       end
 
+      def match?(path:)
+        ERB_REGEX.match?(path)
+      end
+
       private
 
       def to_ruby_ast(erb_ast, file_path)

lib/packwerk/parsers/factory.rb

@@ -9,34 +9,9 @@ class Factory
       extend T::Sig
       include Singleton
 
-      RUBY_REGEX = %r{
-        # Although not important for regex, these are ordered from most likely to match to least likely.
-        \.(rb|rake|builder|gemspec|ru)\Z
-        |
-        (Gemfile|Rakefile)\Z
-      }x
-      private_constant :RUBY_REGEX
-
-      ERB_REGEX = /\.erb\Z/
-      private_constant :ERB_REGEX
-
-      sig { params(path: String).returns(T.nilable(ParserInterface)) }
+      sig { params(path: String).returns(T::Array[Packwerk::FileParser]) }
       def for_path(path)
-        case path
-        when RUBY_REGEX
-          @ruby_parser ||= Ruby.new
-        when ERB_REGEX
-          @erb_parser ||= erb_parser_class.new
-        end
-      end
-
-      def erb_parser_class
-        @erb_parser_class ||= Erb
-      end
-
-      def erb_parser_class=(klass)
-        @erb_parser_class = klass
-        @erb_parser = nil
+        Packwerk::FileParser.all.select { |parser| parser.match?(path: path) }
       end
     end
   end

lib/packwerk/parsers/ruby.rb

@@ -7,16 +7,24 @@
 module Packwerk
   module Parsers
     class Ruby
-      include ParserInterface
+      include Packwerk::FileParser
 
-      class RaiseExceptionsParser < Parser::CurrentRuby
+      RUBY_REGEX = %r{
+        # Although not important for regex, these are ordered from most likely to match to least likely.
+        \.(rb|rake|builder|gemspec|ru)\Z
+        |
+        (Gemfile|Rakefile)\Z
+      }x
+      private_constant :RUBY_REGEX
+
+      class RaiseExceptionsParser < ::Parser::CurrentRuby
         def initialize(builder)
           super(builder)
           super.diagnostics.all_errors_are_fatal = true
         end
       end
 
-      class TolerateInvalidUtf8Builder < Parser::Builders::Default
+      class TolerateInvalidUtf8Builder < ::Parser::Builders::Default
         def string_value(token)
           value(token)
         end
@@ -28,17 +36,21 @@ def initialize(parser_class: RaiseExceptionsParser)
       end
 
       def call(io:, file_path: "<unknown>")
-        buffer = Parser::Source::Buffer.new(file_path)
+        buffer = ::Parser::Source::Buffer.new(file_path)
         buffer.source = io.read
         parser = @parser_class.new(@builder)
         parser.parse(buffer)
       rescue EncodingError => e
         result = ParseResult.new(file: file_path, message: e.message)
         raise Parsers::ParseError, result
-      rescue Parser::SyntaxError => e
+      rescue ::Parser::SyntaxError => e
         result = ParseResult.new(file: file_path, message: "Syntax error: #{e}")
         raise Parsers::ParseError, result
       end
+
+      def match?(path:)
+        RUBY_REGEX.match?(path)
+      end
     end
   end
 end

lib/packwerk/reference_extractor.rb

@@ -25,8 +25,8 @@ def initialize(
 
     sig do
       params(
-        node: Parser::AST::Node,
-        ancestors: T::Array[Parser::AST::Node],
+        node: ::Parser::AST::Node,
+        ancestors: T::Array[::Parser::AST::Node],
         absolute_file: String
       ).returns(T.nilable(UnresolvedReference))
     end
@@ -99,8 +99,8 @@ def self.get_fully_qualified_references_and_offenses_from(unresolved_references_
     sig do
       params(
         constant_name: String,
-        node: Parser::AST::Node,
-        ancestors: T::Array[Parser::AST::Node],
+        node: ::Parser::AST::Node,
+        ancestors: T::Array[::Parser::AST::Node],
         absolute_file: String
       ).returns(T.nilable(UnresolvedReference))
     end

test/unit/parsers/erb_test.rb

@@ -19,7 +19,7 @@ class ErbTest < Minitest::Test
 
       test "#call writes parse error to stdout" do
         error_message = "stub error"
-        err = Parser::SyntaxError.new(stub(message: error_message))
+        err = ::Parser::SyntaxError.new(stub(message: error_message))
         parser = stub
         parser.stubs(:ast).raises(err)