prism 2.0.0.rc1-x86_64-linux-gnu

data/lib/prism/relocation.rb

@@ -0,0 +1,567 @@
+# frozen_string_literal: true
+# :markup: markdown
+#--
+# rbs_inline: enabled
+
+module Prism
+  # Prism parses deterministically for the same input. This provides a nice
+  # property that is exposed through the #node_id API on nodes. Effectively this
+  # means that for the same input, these values will remain consistent every
+  # time the source is parsed. This means we can reparse the source same with a
+  # #node_id value and find the exact same node again.
+  #
+  # The Relocation module provides an API around this property. It allows you to
+  # "save" nodes and locations using a minimal amount of memory (just the
+  # node_id and a field identifier) and then reify them later.
+  module Relocation
+    # @rbs!
+    #    type entry_value = untyped
+    #    type entry_values = Hash[Symbol, entry_value]
+    #
+    #    interface _Value
+    #      def start_line: () -> Integer
+    #      def end_line: () -> Integer
+    #      def start_offset: () -> Integer
+    #      def end_offset: () -> Integer
+    #      def start_character_offset: () -> Integer
+    #      def end_character_offset: () -> Integer
+    #      def cached_start_code_units_offset: (_CodeUnitsCache cache) -> Integer
+    #      def cached_end_code_units_offset: (_CodeUnitsCache cache) -> Integer
+    #      def start_column: () -> Integer
+    #      def end_column: () -> Integer
+    #      def start_character_column: () -> Integer
+    #      def end_character_column: () -> Integer
+    #      def cached_start_code_units_column: (_CodeUnitsCache cache) -> Integer
+    #      def cached_end_code_units_column: (_CodeUnitsCache cache) -> Integer
+    #    end
+    #
+    #    interface _Field
+    #      def fields: (_Value value) -> entry_values
+    #    end
+
+    # An entry in a repository that will lazily reify its values when they are
+    # first accessed.
+    class Entry
+      # Raised if a value that could potentially be on an entry is missing
+      # because it was either not configured on the repository or it has not yet
+      # been fetched.
+      class MissingValueError < StandardError
+      end
+
+      # @rbs @repository: Repository?
+      # @rbs @values: Hash[Symbol, untyped]?
+
+      # Initialize a new entry with the given repository.
+      #
+      #: (Repository repository) -> void
+      def initialize(repository)
+        @repository = repository
+        @values = nil
+      end
+
+      # Fetch the filepath of the value.
+      #
+      #: () -> String
+      def filepath
+        fetch_value(:filepath)
+      end
+
+      # Fetch the start line of the value.
+      #
+      #: () -> Integer
+      def start_line
+        fetch_value(:start_line)
+      end
+
+      # Fetch the end line of the value.
+      #
+      #: () -> Integer
+      def end_line
+        fetch_value(:end_line)
+      end
+
+      # Fetch the start byte offset of the value.
+      #
+      #: () -> Integer
+      def start_offset
+        fetch_value(:start_offset)
+      end
+
+      # Fetch the end byte offset of the value.
+      #
+      #: () -> Integer
+      def end_offset
+        fetch_value(:end_offset)
+      end
+
+      # Fetch the start character offset of the value.
+      #
+      #: () -> Integer
+      def start_character_offset
+        fetch_value(:start_character_offset)
+      end
+
+      # Fetch the end character offset of the value.
+      #
+      #: () -> Integer
+      def end_character_offset
+        fetch_value(:end_character_offset)
+      end
+
+      # Fetch the start code units offset of the value, for the encoding that
+      # was configured on the repository.
+      #
+      #: () -> Integer
+      def start_code_units_offset
+        fetch_value(:start_code_units_offset)
+      end
+
+      # Fetch the end code units offset of the value, for the encoding that was
+      # configured on the repository.
+      #
+      #: () -> Integer
+      def end_code_units_offset
+        fetch_value(:end_code_units_offset)
+      end
+
+      # Fetch the start byte column of the value.
+      #
+      #: () -> Integer
+      def start_column
+        fetch_value(:start_column)
+      end
+
+      # Fetch the end byte column of the value.
+      #
+      #: () -> Integer
+      def end_column
+        fetch_value(:end_column)
+      end
+
+      # Fetch the start character column of the value.
+      #
+      #: () -> Integer
+      def start_character_column
+        fetch_value(:start_character_column)
+      end
+
+      # Fetch the end character column of the value.
+      #
+      #: () -> Integer
+      def end_character_column
+        fetch_value(:end_character_column)
+      end
+
+      # Fetch the start code units column of the value, for the encoding that
+      # was configured on the repository.
+      #
+      #: () -> Integer
+      def start_code_units_column
+        fetch_value(:start_code_units_column)
+      end
+
+      # Fetch the end code units column of the value, for the encoding that was
+      # configured on the repository.
+      #
+      #: () -> Integer
+      def end_code_units_column
+        fetch_value(:end_code_units_column)
+      end
+
+      # Reify the values on this entry with the given values. This is an
+      # internal-only API that is called from the repository when it is time to
+      # reify the values.
+      #
+      #: (entry_values values) -> void
+      def reify!(values) # :nodoc:
+        @repository = nil
+        @values = values
+      end
+
+      private
+
+      # Fetch a value from the entry, raising an error if it is missing.
+      #
+      #: (Symbol name) -> entry_value
+      def fetch_value(name)
+        values.fetch(name) do
+          raise MissingValueError, "No value for #{name}, make sure the " \
+            "repository has been properly configured"
+        end
+      end
+
+      # Return the values from the repository, reifying them if necessary.
+      #
+      #: () -> entry_values
+      def values
+        @values || (@repository&.reify!; @values) #: entry_values
+      end
+    end
+
+    # Represents the source of a repository that will be reparsed.
+    class Source
+      # The value that will need to be reparsed.
+      attr_reader :value #: untyped
+
+      # Initialize the source with the given value.
+      #
+      #: (untyped value) -> void
+      def initialize(value)
+        @value = value
+      end
+
+      # Reparse the value and return the parse result.
+      #
+      #: () -> ParseResult
+      def result
+        raise NotImplementedError, "Subclasses must implement #result"
+      end
+
+      # Create a code units cache for the given encoding.
+      #
+      #: (Encoding encoding) -> _CodeUnitsCache
+      def code_units_cache(encoding)
+        result.code_units_cache(encoding)
+      end
+    end
+
+    # A source that is represented by a file path.
+    class SourceFilepath < Source
+      # Reparse the file and return the parse result.
+      #
+      #: () -> ParseResult
+      def result
+        Prism.parse_file(value)
+      end
+    end
+
+    # A source that is represented by a string.
+    class SourceString < Source
+      # Reparse the string and return the parse result.
+      #
+      #: () -> ParseResult
+      def result
+        Prism.parse(value)
+      end
+    end
+
+    # A field that represents the file path.
+    class FilepathField
+      # The file path that this field represents.
+      attr_reader :value #: String
+
+      # Initialize a new field with the given file path.
+      #
+      #: (String value) -> void
+      def initialize(value)
+        @value = value
+      end
+
+      # Fetch the file path.
+      #
+      #: (_Value _value) -> entry_values
+      def fields(_value)
+        { filepath: value }
+      end
+    end
+
+    # A field representing the start and end lines.
+    class LinesField
+      # Fetches the start and end line of a value.
+      #
+      #: (_Value value) -> entry_values
+      def fields(value)
+        { start_line: value.start_line, end_line: value.end_line }
+      end
+    end
+
+    # A field representing the start and end byte offsets.
+    class OffsetsField
+      # Fetches the start and end byte offset of a value.
+      #
+      #: (_Value value) -> entry_values
+      def fields(value)
+        { start_offset: value.start_offset, end_offset: value.end_offset }
+      end
+    end
+
+    # A field representing the start and end character offsets.
+    class CharacterOffsetsField
+      # Fetches the start and end character offset of a value.
+      #
+      #: (_Value value) -> entry_values
+      def fields(value)
+        {
+          start_character_offset: value.start_character_offset,
+          end_character_offset: value.end_character_offset
+        }
+      end
+    end
+
+    # A field representing the start and end code unit offsets.
+    class CodeUnitOffsetsField
+      # A pointer to the repository object that is used for lazily creating a
+      # code units cache.
+      attr_reader :repository #: Repository
+
+      # The associated encoding for the code units.
+      attr_reader :encoding #: Encoding
+
+      # @rbs @cache: _CodeUnitsCache?
+
+      # Initialize a new field with the associated repository and encoding.
+      #
+      #: (Repository repository, Encoding encoding) -> void
+      def initialize(repository, encoding)
+        @repository = repository
+        @encoding = encoding
+        @cache = nil
+      end
+
+      # Fetches the start and end code units offset of a value for a particular
+      # encoding.
+      #
+      #: (_Value value) -> entry_values
+      def fields(value)
+        {
+          start_code_units_offset: value.cached_start_code_units_offset(cache),
+          end_code_units_offset: value.cached_end_code_units_offset(cache)
+        }
+      end
+
+      private
+
+      # Lazily create a code units cache for the associated encoding.
+      #
+      #: () -> _CodeUnitsCache
+      def cache
+        @cache ||= repository.code_units_cache(encoding)
+      end
+    end
+
+    # A field representing the start and end byte columns.
+    class ColumnsField
+      # Fetches the start and end byte column of a value.
+      #
+      #: (_Value value) -> entry_values
+      def fields(value)
+        { start_column: value.start_column, end_column: value.end_column }
+      end
+    end
+
+    # A field representing the start and end character columns.
+    class CharacterColumnsField
+      # Fetches the start and end character column of a value.
+      #
+      #: (_Value value) -> entry_values
+      def fields(value)
+        {
+          start_character_column: value.start_character_column,
+          end_character_column: value.end_character_column
+        }
+      end
+    end
+
+    # A field representing the start and end code unit columns for a specific
+    # encoding.
+    class CodeUnitColumnsField
+      # The repository object that is used for lazily creating a code units
+      # cache.
+      attr_reader :repository #: Repository
+
+      # The associated encoding for the code units.
+      attr_reader :encoding #: Encoding
+
+      # @rbs @cache: _CodeUnitsCache?
+
+      # Initialize a new field with the associated repository and encoding.
+      #
+      #: (Repository repository, Encoding encoding) -> void
+      def initialize(repository, encoding)
+        @repository = repository
+        @encoding = encoding
+        @cache = nil
+      end
+
+      # Fetches the start and end code units column of a value for a particular
+      # encoding.
+      #
+      #: (_Value value) -> entry_values
+      def fields(value)
+        {
+          start_code_units_column: value.cached_start_code_units_column(cache),
+          end_code_units_column: value.cached_end_code_units_column(cache)
+        }
+      end
+
+      private
+
+      # Lazily create a code units cache for the associated encoding.
+      #
+      #: () -> _CodeUnitsCache
+      def cache
+        @cache ||= repository.code_units_cache(encoding)
+      end
+    end
+
+    # A repository is a configured collection of fields and a set of entries
+    # that knows how to reparse a source and reify the values.
+    class Repository
+      # Raised when multiple fields of the same type are configured on the same
+      # repository.
+      class ConfigurationError < StandardError
+      end
+
+      # The source associated with this repository. This will be either a
+      # SourceFilepath (the most common use case) or a SourceString.
+      attr_reader :source #: Source
+
+      # The fields that have been configured on this repository.
+      attr_reader :fields #: Hash[Symbol, _Field]
+
+      # The entries that have been saved on this repository.
+      attr_reader :entries #: Hash[Integer, Hash[Symbol, Entry]]
+
+      # Initialize a new repository with the given source.
+      #
+      #: (Source source) -> void
+      def initialize(source)
+        @source = source
+        @fields = {}
+        @entries = Hash.new { |hash, node_id| hash[node_id] = {} }
+      end
+
+      # Create a code units cache for the given encoding from the source.
+      #
+      #: (Encoding encoding) -> _CodeUnitsCache
+      def code_units_cache(encoding)
+        source.code_units_cache(encoding)
+      end
+
+      # Configure the filepath field for this repository and return self.
+      #
+      #: () -> self
+      def filepath
+        raise ConfigurationError, "Can only specify filepath for a filepath source" unless source.is_a?(SourceFilepath)
+        field(:filepath, FilepathField.new(source.value))
+      end
+
+      # Configure the lines field for this repository and return self.
+      #
+      #: () -> self
+      def lines
+        field(:lines, LinesField.new)
+      end
+
+      # Configure the offsets field for this repository and return self.
+      #
+      #: () -> self
+      def offsets
+        field(:offsets, OffsetsField.new)
+      end
+
+      # Configure the character offsets field for this repository and return
+      # self.
+      #
+      #: () -> self
+      def character_offsets
+        field(:character_offsets, CharacterOffsetsField.new)
+      end
+
+      # Configure the code unit offsets field for this repository for a specific
+      # encoding and return self.
+      #
+      #: (Encoding encoding) -> self
+      def code_unit_offsets(encoding)
+        field(:code_unit_offsets, CodeUnitOffsetsField.new(self, encoding))
+      end
+
+      # Configure the columns field for this repository and return self.
+      #
+      #: () -> self
+      def columns
+        field(:columns, ColumnsField.new)
+      end
+
+      # Configure the character columns field for this repository and return
+      # self.
+      #
+      #: () -> self
+      def character_columns
+        field(:character_columns, CharacterColumnsField.new)
+      end
+
+      # Configure the code unit columns field for this repository for a specific
+      # encoding and return self.
+      #
+      #: (Encoding encoding) -> self
+      def code_unit_columns(encoding)
+        field(:code_unit_columns, CodeUnitColumnsField.new(self, encoding))
+      end
+
+      # This method is called from nodes and locations when they want to enter
+      # themselves into the repository. It it internal-only and meant to be
+      # called from the #save* APIs.
+      #
+      #: (Integer node_id, Symbol field_name) -> Entry
+      def enter(node_id, field_name) # :nodoc:
+        entry = Entry.new(self)
+        @entries[node_id][field_name] = entry
+        entry
+      end
+
+      # This method is called from the entries in the repository when they need
+      # to reify their values. It is internal-only and meant to be called from
+      # the various value APIs.
+      #
+      #: () -> void
+      def reify! # :nodoc:
+        result = source.result
+
+        queue = [result.value] #: Array[Prism::node]
+        while (node = queue.shift)
+          @entries[node.node_id].each do |field_name, entry|
+            value = node.public_send(field_name)
+            values = {} #: entry_values
+
+            fields.each_value do |field|
+              values.merge!(field.fields(value))
+            end
+
+            entry.reify!(values)
+          end
+
+          queue.concat(node.compact_child_nodes)
+        end
+
+        @entries.clear
+      end
+
+      private
+
+      # Append the given field to the repository and return the repository so
+      # that these calls can be chained.
+      #
+      #: (Symbol name, _Field) -> self
+      def field(name, value)
+        raise ConfigurationError, "Cannot specify multiple #{name} fields" if @fields.key?(name)
+        @fields[name] = value
+        self
+      end
+    end
+
+    # Create a new repository for the given filepath.
+    #
+    #: (String value) -> Repository
+    def self.filepath(value)
+      Repository.new(SourceFilepath.new(value))
+    end
+
+    # Create a new repository for the given string.
+    #
+    #: (String value) -> Repository
+    def self.string(value)
+      Repository.new(SourceString.new(value))
+    end
+  end
+end