rubocop 1.44.1 → 1.78.0

data/lib/rubocop/cop/internal_affairs/node_matcher_directive.rb

@@ -26,10 +26,18 @@ module RuboCop
         MSG = 'Precede `%<method>s` with a `@!method` YARD directive.'
         MSG_WRONG_NAME = '`@!method` YARD directive has invalid method name, ' \
                          'use `%<expected>s` instead of `%<actual>s`.'
+        MSG_MISSING_SCOPE_SELF = 'Follow the `@!method` YARD directive with ' \
+                                 '`@!scope class` if it is a class method.'
+        MSG_WRONG_SCOPE_SELF = 'Do not use the `@!scope class` YARD directive if it ' \
+                               'is not a class method.'
         MSG_TOO_MANY = 'Multiple `@!method` YARD directives found for this matcher.'
 
         RESTRICT_ON_SEND = %i[def_node_matcher def_node_search].to_set.freeze
-        REGEXP = /^\s*#\s*@!method\s+(?<method_name>[a-z0-9_]+[?!]?)(?:\((?<args>.*)\))?/.freeze
+        REGEXP_METHOD = /
+          ^\s*\#\s*
+          @!method\s+(?<receiver>self\.)?(?<method_name>[a-z0-9_]+[?!]?)(?:\((?<args>.*)\))?
+        /x.freeze
+        REGEXP_SCOPE = /^\s*\#\s*@!scope class/.freeze
 
         # @!method pattern_matcher?(node)
         def_node_matcher :pattern_matcher?, <<~PATTERN
@@ -37,69 +45,138 @@ module RuboCop
         PATTERN
 
         def on_send(node)
-          return if node.arguments.none?
+          return unless node.arguments.count == 2
           return unless valid_method_name?(node)
 
-          actual_name = node.arguments.first.value
+          actual_name = node.first_argument.value.to_s
+
+          # Ignore cases where the method has a receiver that isn't self
+          return if actual_name.include?('.') && !actual_name.start_with?('self.')
+
           directives = method_directives(node)
           return too_many_directives(node) if directives.size > 1
 
-          directive = directives.first
-          return if directive_correct?(directive, actual_name)
-
-          register_offense(node, directive, actual_name)
+          process_directive(node, actual_name, directives.first)
         end
 
         private
 
         def valid_method_name?(node)
-          node.arguments.first.str_type? || node.arguments.first.sym_type?
+          node.first_argument.type?(:str, :sym)
         end
 
         def method_directives(node)
           comments = processed_source.ast_with_comments[node]
-
-          comments.map do |comment|
-            match = comment.text.match(REGEXP)
+          group_comments(comments).filter_map do |comment_method, comment_scope|
+            match = comment_method.text.match(REGEXP_METHOD)
             next unless match
 
-            { node: comment, method_name: match[:method_name], args: match[:args] }
-          end.compact
+            {
+              node_method: comment_method,
+              node_scope: comment_scope,
+              method_name: match[:method_name],
+              args: match[:args],
+              receiver: match[:receiver],
+              has_scope_directive: comment_scope&.text&.match?(REGEXP_SCOPE)
+            }
+          end
+        end
+
+        def group_comments(comments)
+          result = []
+          comments.each.with_index do |comment, index|
+            # Grab the scope directive if it is preceded by a method directive
+            if comment.text.include?('@!method')
+              result << if (next_comment = comments[index + 1])&.text&.include?('@!scope')
+                          [comment, next_comment]
+                        else
+                          [comment, nil]
+                        end
+            end
+          end
+          result
         end
 
         def too_many_directives(node)
           add_offense(node, message: MSG_TOO_MANY)
         end
 
-        def directive_correct?(directive, actual_name)
-          directive && directive[:method_name] == actual_name.to_s
+        def process_directive(node, actual_name, directive)
+          return unless (offense_type = directive_offense_type(directive, actual_name))
+
+          register_offense(offense_type, node, directive, actual_name)
+        end
+
+        def directive_offense_type(directive, actual_name)
+          return :missing_directive unless directive
+
+          return :wrong_scope if wrong_scope?(directive, actual_name)
+          return :no_scope if no_scope?(directive, actual_name)
+
+          # The method directive being prefixed by 'self.' is always an offense.
+          # The matched method_name does not contain the receiver but the
+          # def_node_match method name may so it must be removed.
+          if directive[:method_name] != remove_receiver(actual_name) || directive[:receiver]
+            :wrong_name
+          end
+        end
+
+        def wrong_scope?(directive, actual_name)
+          !actual_name.start_with?('self.') && directive[:has_scope_directive]
         end
 
-        def register_offense(node, directive, actual_name)
-          message = formatted_message(directive, actual_name, node.method_name)
+        def no_scope?(directive, actual_name)
+          actual_name.start_with?('self.') && !directive[:has_scope_directive]
+        end
+
+        def register_offense(offense_type, node, directive, actual_name)
+          message = formatted_message(offense_type, directive, actual_name, node.method_name)
 
           add_offense(node, message: message) do |corrector|
-            if directive
-              correct_directive(corrector, directive, actual_name)
-            else
-              insert_directive(corrector, node, actual_name)
+            case offense_type
+            when :wrong_name
+              correct_method_directive(corrector, directive, actual_name)
+            when :wrong_scope
+              remove_scope_directive(corrector, directive)
+            when :no_scope
+              insert_scope_directive(corrector, directive[:node_method])
+            when :missing_directive
+              insert_method_directive(corrector, node, actual_name)
             end
           end
         end
 
-        def formatted_message(directive, actual_name, method_name)
-          if directive
-            format(MSG_WRONG_NAME, expected: actual_name, actual: directive[:method_name])
-          else
+        # rubocop:disable Metrics/MethodLength
+        def formatted_message(offense_type, directive, actual_name, method_name)
+          case offense_type
+          when :wrong_name
+            # Add the receiver to the name when showing an offense
+            current_name = if directive[:receiver]
+                             directive[:receiver] + directive[:method_name]
+                           else
+                             directive[:method_name]
+                           end
+            # The correct name will never include a receiver, remove it
+            format(MSG_WRONG_NAME, expected: remove_receiver(actual_name), actual: current_name)
+          when :wrong_scope
+            MSG_WRONG_SCOPE_SELF
+          when :no_scope
+            MSG_MISSING_SCOPE_SELF
+          when :missing_directive
             format(MSG, method: method_name)
           end
         end
+        # rubocop:enable Metrics/MethodLength
+
+        def remove_receiver(current)
+          current.delete_prefix('self.')
+        end
 
-        def insert_directive(corrector, node, actual_name)
+        def insert_method_directive(corrector, node, actual_name)
           # If the pattern matcher uses arguments (`%1`, `%2`, etc.), include them in the directive
           arguments = pattern_arguments(node.arguments[1].source)
 
-          range = range_with_surrounding_space(node.loc.expression, side: :left, newlines: false)
+          range = range_with_surrounding_space(node.source_range, side: :left, newlines: false)
           indentation = range.source.match(/^\s*/)[0]
           directive = "#{indentation}# @!method #{actual_name}(#{arguments.join(', ')})\n"
           directive = "\n#{directive}" if add_newline?(node)
@@ -107,6 +184,14 @@ module RuboCop
           corrector.insert_before(range, directive)
         end
 
+        def insert_scope_directive(corrector, node)
+          range = range_with_surrounding_space(node.source_range, side: :left, newlines: false)
+          indentation = range.source.match(/^\s*/)[0]
+          directive = "\n#{indentation}# @!scope class"
+
+          corrector.insert_after(node, directive)
+        end
+
         def pattern_arguments(pattern)
           arguments = %w[node]
           max_pattern_var = pattern.scan(/(?<=%)\d+/).map(&:to_i).max
@@ -117,11 +202,11 @@ module RuboCop
         def add_newline?(node)
           # Determine if a blank line should be inserted before the new directive
           # in order to spread out pattern matchers
-          return if node.sibling_index&.zero?
-          return unless node.parent
+          return false if node.sibling_index&.zero?
+          return false unless node.parent
 
           prev_sibling = node.parent.child_nodes[node.sibling_index - 1]
-          return unless prev_sibling && pattern_matcher?(prev_sibling)
+          return false unless prev_sibling && pattern_matcher?(prev_sibling)
 
           node.loc.line == last_line(prev_sibling) + 1
         end
@@ -134,12 +219,21 @@ module RuboCop
           end
         end
 
-        def correct_directive(corrector, directive, actual_name)
-          correct = "@!method #{actual_name}"
-          regexp = /@!method\s+#{Regexp.escape(directive[:method_name])}/
+        def correct_method_directive(corrector, directive, actual_name)
+          correct = "@!method #{remove_receiver(actual_name)}"
+          current_name = (directive[:receiver] || '') + directive[:method_name]
+          regexp = /@!method\s+#{Regexp.escape(current_name)}/
+
+          replacement = directive[:node_method].text.gsub(regexp, correct)
+          corrector.replace(directive[:node_method], replacement)
+        end
 
-          replacement = directive[:node].text.gsub(regexp, correct)
-          corrector.replace(directive[:node], replacement)
+        def remove_scope_directive(corrector, directive)
+          range = range_by_whole_lines(
+            directive[:node_scope].source_range,
+            include_final_newline: true
+          )
+          corrector.remove(range)
         end
       end
     end

data/lib/rubocop/cop/internal_affairs/node_pattern_groups/ast_processor.rb

@@ -0,0 +1,63 @@
+# frozen_string_literal: true
+
+module RuboCop
+  module Cop
+    module InternalAffairs
+      class NodePatternGroups
+        # AST Processor for NodePattern ASTs, for use with `InternalAffairs/NodePatternGroups`.
+        #
+        # Looks for sequences and subsequences where the first item is a `node_type` node,
+        # and converts them to `node_sequence` nodes (not a true `Rubocop::AST::NodePattern`
+        # node type).
+        #
+        # The resulting AST will be walked by `InternalAffairs::NodePatternGroups::ASTWalker`
+        # in order to find node types in a `union` node that can be rewritten as a node group.
+        #
+        # NOTE: The `on_*` methods in this class relate not to the normal node types but
+        # rather to the Node Pattern node types. Not every node type is handled.
+        #
+        class ASTProcessor
+          include ::AST::Processor::Mixin
+
+          def handler_missing(node)
+            node.updated(nil, process_children(node))
+          end
+
+          # Look for `sequence` and `subsequence` nodes that contain a `node_type` node as
+          # their first child. These are rewritten as `node_sequence` nodes so that it is
+          # possible to compare nodes while looking for replacement candidates for node groups.
+          # This is necessary so that extended patterns can be matched and replaced.
+          # ie. `{(send _ :foo ...) (csend _ :foo ...)}` can become `(call _ :foo ...)`
+          def on_sequence(node)
+            first_child = node.child
+
+            if first_child.type == :node_type
+              children = [first_child.child, *process_children(node, 1..)]
+
+              # The `node_sequence` node contains the `node_type` symbol as its first child,
+              # followed by all the other nodes contained in the `sequence` node.
+              # The location is copied from the sequence, so that the entire sequence can
+              # eventually be corrected in the cop.
+              n(:node_sequence, children, location: node.location)
+            else
+              node.updated(nil, process_children(node))
+            end
+          end
+          alias on_subsequence on_sequence
+
+          private
+
+          def n(type, children = [], properties = {})
+            NodePattern::Node.new(type, children, properties)
+          end
+
+          def process_children(node, range = 0..-1)
+            node.children[range].map do |child|
+              child.is_a?(::AST::Node) ? process(child) : child
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/rubocop/cop/internal_affairs/node_pattern_groups/ast_walker.rb

@@ -0,0 +1,131 @@
+# frozen_string_literal: true
+
+module RuboCop
+  module Cop
+    module InternalAffairs
+      # rubocop:disable InternalAffairs/RedundantSourceRange -- node here is a `NodePattern::Node`
+      class NodePatternGroups
+        # Walks an AST that has been processed by `InternalAffairs::NodePatternGroups::Processor`
+        # in order to find `node_type` and `node_sequence` nodes that can be replaced with a node
+        # group in `InternalAffairs/NodePatternGroups`.
+        #
+        # Calling `ASTWalker#walk` sets `node_groups` with an array of `NodeGroup` structs
+        # that contain metadata about nodes that can be replaced, including location data. That
+        # metadata is used by the cop to register offenses and perform corrections.
+        class ASTWalker
+          # Struct to contain data about parts of a node pattern that can be replaced
+          NodeGroup = Struct.new(
+            :name,             # The name of the node group that will be inserted
+            :union,            # The entire `union` node
+            :node_types,       # An array of `node_type` nodes that will be removed
+            :sequence?,        # The pattern matches a node type with given attributes
+            :start_index,      # The index in the union of the first node type to remove
+            :offense_range,    # The range to mark an offense on
+            :ranges,           # Range of each element to remove, since they may not be adjacent
+            :pipe,             # Is the union delimited by pipes?
+            :other_elements?,  # Does the union have other elements other than those to remove?
+            keyword_init: true
+          )
+
+          def initialize
+            reset!
+          end
+
+          def reset!
+            @node_groups = []
+          end
+
+          attr_reader :node_groups
+
+          # Recursively walk the AST in a depth-first manner.
+          # Only `union` nodes are handled further.
+          def walk(node)
+            return if node.nil?
+
+            on_union(node) if node.type == :union
+
+            node.child_nodes.each do |child|
+              walk(child)
+            end
+          end
+
+          # Search `union` nodes for `node_type` and `node_sequence` nodes that can be
+          # collapsed into a node group.
+          # * `node_type` nodes are nodes with no further configuration (ie. `send`)
+          # * `node_sequence` nodes are nodes with further configuration (ie. `(send ...)`)
+          #
+          # Each group of types that can be collapsed will have a `NodeGroup` record added
+          # to `node_groups`, which is then used by the cop.
+          def on_union(node)
+            all_node_types = each_child_node(node, :node_type, :node_sequence).to_a
+
+            each_node_group(all_node_types) do |group_name, node_types|
+              next unless sequences_match?(node_types)
+
+              node_groups << node_group_data(
+                group_name, node, node_types,
+                all_node_types.index(node_types.first),
+                (node.children - node_types).any?
+              )
+            end
+          end
+
+          private
+
+          def each_child_node(node, *types)
+            return to_enum(__method__, node, *types) unless block_given?
+
+            node.children.each do |child|
+              yield child if types.empty? || types.include?(child.type)
+            end
+
+            self
+          end
+
+          def each_node_group(types_to_check)
+            # Find all node groups where all of the members are present in the union
+            type_names = types_to_check.map(&:child)
+
+            NODE_GROUPS.select { |_, group| group & type_names == group }.each_key do |name|
+              nodes = get_relevant_nodes(types_to_check, name)
+
+              yield name, nodes
+            end
+          end
+
+          def get_relevant_nodes(node_types, group_name)
+            node_types.each_with_object([]) do |node_type, arr|
+              next unless NODE_GROUPS[group_name].include?(node_type.child)
+
+              arr << node_type
+            end
+          end
+
+          def node_group_data(name, union, node_types, start_index, other)
+            NodeGroup.new(
+              name: name,
+              union: union,
+              node_types: node_types,
+              sequence?: node_types.first.type == :node_sequence,
+              start_index: start_index,
+              pipe: union.source_range.source['|'],
+              other_elements?: other
+            )
+          end
+
+          def sequences_match?(types)
+            # Ensure all given types have the same type and the same sequence
+            # ie. `(send ...)` and `(csend ...) is a match
+            #     `(send)` and `(csend ...)` is not a match
+            #     `send` and `(csend ...)` is not a match
+
+            types.each_cons(2).all? do |left, right|
+              left.type == right.type && left.children[1..] == right.children[1..]
+            end
+          end
+        end
+      end
+      # rubocop:enable InternalAffairs/RedundantSourceRange
+    end
+  end
+end

data/lib/rubocop/cop/internal_affairs/node_pattern_groups.rb

@@ -0,0 +1,231 @@
+# frozen_string_literal: true
+
+module RuboCop
+  module Cop
+    module InternalAffairs
+      # Use node groups (`any_block`, `argument`, `boolean`, `call`, `numeric`, `range`)
+      # in node patterns instead of a union (`{ ... }`) of the member types of the group.
+      #
+      # @example
+      #   # bad
+      #   def_node_matcher :my_matcher, <<~PATTERN
+      #     {send csend}
+      #   PATTERN
+      #
+      #   # good
+      #   def_node_matcher :my_matcher, <<~PATTERN
+      #     call
+      #   PATTERN
+      #
+      class NodePatternGroups < Base
+        require_relative 'node_pattern_groups/ast_processor'
+        require_relative 'node_pattern_groups/ast_walker'
+
+        include RangeHelp
+        extend AutoCorrector
+
+        MSG = 'Replace `%<names>s` in node pattern union with `%<replacement>s`.'
+        RESTRICT_ON_SEND = %i[def_node_matcher def_node_search].freeze
+        NODE_GROUPS = {
+          any_block: %i[block numblock itblock],
+          any_def: %i[def defs],
+          any_match_pattern: %i[match_pattern match_pattern_p],
+          argument: %i[arg optarg restarg kwarg kwoptarg kwrestarg blockarg forward_arg shadowarg],
+          boolean: %i[true false],
+          call: %i[send csend],
+          numeric: %i[int float rational complex],
+          range: %i[irange erange]
+        }.freeze
+
+        def on_new_investigation
+          @walker = ASTWalker.new
+        end
+
+        # When a Node Pattern matcher is defined, investigate the pattern string to search
+        # for node types that can be replaced with a node group (ie. `{send csend}` can be
+        # replaced with `call`).
+        #
+        # In order to deal with node patterns in an efficient and non-brittle way, we will
+        # parse the Node Pattern string given to this `send` node using
+        # `RuboCop::AST::NodePattern::Parser::WithMeta`. `WithMeta` is important! We need
+        # location information so that we can calculate the exact locations within the
+        # pattern to report and correct.
+        #
+        # The resulting AST is processed by `NodePatternGroups::ASTProccessor` which rewrites
+        # the AST slightly to handle node sequences (ie. `(send _ :foo ...)`). See the
+        # documentation of that class for more details.
+        #
+        # Then the processed AST is walked, and metadata is collected for node types that
+        # can be replaced with a node group.
+        #
+        # Finally, the metadata is used to register offenses and make corrections, using
+        # the location data captured earlier. The ranges captured while parsing the Node
+        # Pattern are offset using the string argument to this `send` node to ensure
+        # that offenses are registered at the correct location.
+        #
+        def on_send(node)
+          pattern_node = node.arguments[1]
+          return unless acceptable_heredoc?(pattern_node) || pattern_node.str_type?
+
+          process_pattern(pattern_node)
+          return if node_groups.nil?
+
+          apply_range_offsets(pattern_node)
+
+          node_groups.each_with_index do |group, index|
+            register_offense(group, index)
+          end
+        end
+
+        def after_send(_)
+          @walker.reset!
+        end
+
+        private
+
+        def node_groups
+          @walker.node_groups
+        end
+
+        # rubocop:disable InternalAffairs/RedundantSourceRange -- `node` here is a NodePatternNode
+        def register_offense(group, index)
+          replacement = replacement(group)
+          message = format(
+            MSG,
+            names: group.node_types.map { |node| node.source_range.source }.join('`, `'),
+            replacement: replacement
+          )
+
+          add_offense(group.offense_range, message: message) do |corrector|
+            # Only correct one group at a time to avoid clobbering.
+            # Other offenses will be corrected in the subsequent iterations of the
+            # correction loop.
+            next if index.positive?
+
+            if group.other_elements?
+              replace_types_with_node_group(corrector, group, replacement)
+            else
+              replace_union(corrector, group, replacement)
+            end
+          end
+        end
+
+        def replacement(group)
+          if group.sequence?
+            # If the original nodes were in a sequence (ie. wrapped in parentheses),
+            # use it to generate the resulting NodePattern syntax.
+            first_node_type = group.node_types.first
+            template = first_node_type.source_range.source
+            template.sub(first_node_type.child.to_s, group.name.to_s)
+          else
+            group.name
+          end
+        end
+        # rubocop:enable InternalAffairs/RedundantSourceRange
+
+        # When there are other elements in the union, remove the node types that can be replaced.
+        def replace_types_with_node_group(corrector, group, replacement)
+          ranges = group.ranges.map.with_index do |range, index|
+            # Collect whitespace and pipes preceding each element
+            range_for_full_union_element(range, index, group.pipe)
+          end
+
+          ranges.each { |range| corrector.remove(range) }
+
+          corrector.insert_before(ranges.first, replacement)
+        end
+
+        # If the union contains pipes, remove the pipe character as well.
+        # Unfortunately we don't get the location of the pipe in `loc` object, so we have
+        # to find it.
+        def range_for_full_union_element(range, index, pipe)
+          if index.positive?
+            range = if pipe
+                      range_with_preceding_pipe(range)
+                    else
+                      range_with_surrounding_space(range: range, side: :left, newlines: true)
+                    end
+          end
+
+          range
+        end
+
+        # Collect a preceding pipe and any whitespace left of the pipe
+        def range_with_preceding_pipe(range)
+          pos = range.begin_pos - 1
+
+          while pos
+            unless processed_source.buffer.source[pos].match?(/[\s|]/)
+              return range.with(begin_pos: pos + 1)
+            end
+
+            pos -= 1
+          end
+
+          range
+        end
+
+        # When there are no other elements, the entire union can be replaced
+        def replace_union(corrector, group, replacement)
+          corrector.replace(group.ranges.first, replacement)
+        end
+
+        # rubocop:disable Metrics/AbcSize
+        # Calculate the ranges for each node within the pattern string that will
+        # be replaced or removed. Takes the offset of the string node into account.
+        def apply_range_offsets(pattern_node)
+          range, offset = range_with_offset(pattern_node)
+
+          node_groups.each do |node_group|
+            node_group.ranges ||= []
+            node_group.offense_range = pattern_range(range, node_group.union, offset)
+
+            if node_group.other_elements?
+              node_group.node_types.each do |node_type|
+                node_group.ranges << pattern_range(range, node_type, offset)
+              end
+            else
+              node_group.ranges << node_group.offense_range
+            end
+          end
+        end
+        # rubocop:enable Metrics/AbcSize
+
+        def pattern_range(range, node, offset)
+          begin_pos = node.source_range.begin_pos
+          end_pos = node.source_range.end_pos
+          size = end_pos - begin_pos
+
+          range.adjust(begin_pos: begin_pos + offset).resize(size)
+        end
+
+        def range_with_offset(pattern_node)
+          if pattern_node.heredoc?
+            [pattern_node.loc.heredoc_body, 0]
+          else
+            [pattern_node.source_range, pattern_node.loc.begin.size]
+          end
+        end
+
+        # A heredoc can be a `dstr` without interpolation, but if there is interpolation
+        # there'll be a `begin` node, in which case, we cannot evaluate the pattern.
+        def acceptable_heredoc?(node)
+          node.type?(:str, :dstr) && node.heredoc? && node.each_child_node(:begin).none?
+        end
+
+        def process_pattern(pattern_node)
+          parser = RuboCop::AST::NodePattern::Parser::WithMeta.new
+          ast = parser.parse(pattern_value(pattern_node))
+          ast = ASTProcessor.new.process(ast)
+          @walker.walk(ast)
+        rescue RuboCop::AST::NodePattern::Invalid
+          # if the pattern is invalid, no offenses will be registered
+        end
+
+        def pattern_value(pattern_node)
+          pattern_node.heredoc? ? pattern_node.loc.heredoc_body.source : pattern_node.value
+        end
+      end
+    end
+  end
+end