Merge pull request #304 from yui-knk/better_error_message

Better error message

lib/lrama/grammar/rule_builder.rb

@@ -151,8 +151,15 @@ def numberize_references
               else
                 candidates = rhs.each_with_index.select {|token, i| token.referred_by?(ref_name) }
 
-                raise "Referring symbol `#{ref_name}` is duplicated. #{token}" if candidates.size >= 2
-                raise "Referring symbol `#{ref_name}` is not found. #{token}" unless referring_symbol = candidates.first
+                if candidates.size >= 2
+                  location = token.location.partial_location(ref.first_column, ref.last_column)
+                  raise location.generate_error_message("Referring symbol `#{ref_name}` is duplicated.")
+                end
+
+                unless referring_symbol = candidates.first
+                  location = token.location.partial_location(ref.first_column, ref.last_column)
+                  raise location.generate_error_message("Referring symbol `#{ref_name}` is not found.")
+                end
 
                 ref.index = referring_symbol[1] + 1
               end

lib/lrama/lexer.rb

@@ -31,7 +31,8 @@ class Lexer
       %rule
     )
 
-    def initialize(text)
+    def initialize(path, text)
+      @path = path
       @scanner = StringScanner.new(text)
       @head_column = @head = @scanner.pos
       @head_line = @line = 1
@@ -58,6 +59,7 @@ def column
 
     def location
       Location.new(
+        grammar_file_path: @path,
         first_line: @head_line, first_column: @head_column,
         last_line: line, last_column: column
       )

lib/lrama/lexer/location.rb

@@ -1,9 +1,10 @@
 module Lrama
   class Lexer
     class Location
-      attr_reader :first_line, :first_column, :last_line, :last_column
+      attr_reader :grammar_file_path, :first_line, :first_column, :last_line, :last_column
 
-      def initialize(first_line:, first_column:, last_line:, last_column:)
+      def initialize(grammar_file_path:, first_line:, first_column:, last_line:, last_column:)
+        @grammar_file_path = grammar_file_path
         @first_line = first_line
         @first_column = first_column
         @last_line = last_line
@@ -12,14 +13,82 @@ def initialize(first_line:, first_column:, last_line:, last_column:)
 
       def ==(other)
         self.class == other.class &&
+        self.grammar_file_path == other.grammar_file_path &&
         self.first_line == other.first_line &&
         self.first_column == other.first_column &&
         self.last_line == other.last_line &&
         self.last_column == other.last_column
       end
 
+      def partial_location(left, right)
+        offset = -first_column
+        new_first_line = -1
+        new_first_column = -1
+        new_last_line = -1
+        new_last_column = -1
+
+        _text.each.with_index do |line, index|
+          new_offset = offset + line.length + 1
+
+          if offset <= left && left <= new_offset
+            new_first_line = first_line + index
+            new_first_column = left - offset
+          end
+
+          if offset <= right && right <= new_offset
+            new_last_line = first_line + index
+            new_last_column = right - offset
+          end
+
+          offset = new_offset
+        end
+
+        Location.new(
+          grammar_file_path: grammar_file_path,
+          first_line: new_first_line, first_column: new_first_column,
+          last_line: new_last_line, last_column: new_last_column
+        )
+      end
+
       def to_s
-        "(#{first_line},#{first_column})-(#{last_line},#{last_column})"
+        "#{grammar_file_path} (#{first_line},#{first_column})-(#{last_line},#{last_column})"
+      end
+
+      def generate_error_message(error_message)
+        <<~ERROR.chomp
+          #{grammar_file_path}:#{first_line}:#{first_column}: #{error_message}
+          #{line_with_carrets}
+        ERROR
+      end
+
+      def line_with_carrets
+        <<~TEXT
+          #{text}
+          #{carrets}
+        TEXT
+      end
+
+      private
+
+      def blanks
+        (text[0...first_column] or raise "#{first_column} is invalid").gsub(/[^\t]/, ' ')
+      end
+
+      def carrets
+        blanks + '^' * (last_column - first_column)
+      end
+
+      def text
+        _text.join("\n")
+      end
+
+      def _text
+        return @_text if @_text
+
+        offset = 0
+
+        @_text = File.read(grammar_file_path).split("\n")[(first_line - 1)...last_line]
+        @_text
       end
     end
   end

parser.y

@@ -510,7 +510,7 @@ end
 
 def parse
   report_duration(:parse) do
-    @lexer = Lrama::Lexer.new(@text)
+    @lexer = Lrama::Lexer.new(@path, @text)
     @grammar = Lrama::Grammar.new(@rule_counter)
     @precedence_number = 0
     reset_precs
@@ -527,34 +527,26 @@ end
 
 def on_error(error_token_id, error_value, value_stack)
   if error_value.is_a?(Lrama::Lexer::Token)
-    line = error_value.first_line
-    first_column = error_value.first_column
-    last_column = error_value.last_column
+    location = error_value.location
     value = "'#{error_value.s_value}'"
   else
-    line = @lexer.line
-    first_column = @lexer.head_column
-    last_column = @lexer.column
+    location = @lexer.location
     value = error_value.inspect
   end
 
   error_message = "parse error on value #{value} (#{token_to_str(error_token_id) || '?'})"
 
-  raise_parse_error(error_message, line, first_column, last_column)
+  raise_parse_error(error_message, location)
 end
 
 def on_action_error(error_message, error_value)
   if error_value.is_a?(Lrama::Lexer::Token)
-    line = error_value.first_line
-    first_column = error_value.first_column
-    last_column = error_value.last_column
+    location = error_value.location
   else
-    line = @lexer.line
-    first_column = @lexer.head_column
-    last_column = @lexer.column
+    location = @lexer.location
   end
 
-  raise_parse_error(error_message, line, first_column, last_column)
+  raise_parse_error(error_message, location)
 end
 
 private
@@ -574,20 +566,6 @@ def end_c_declaration
   @lexer.end_symbol = nil
 end
 
-def raise_parse_error(error_message, line, first_column, last_column)
-  text = @text.split("\n")[line - 1]
-
-  raise ParseError, <<~ERROR
-    #{@path}:#{line}:#{first_column}: #{error_message}
-    #{text}
-    #{carrets(text, first_column, last_column)}
-  ERROR
-end
-
-def blanks(text, first_column)
-  text[0...first_column].gsub(/[^\t]/, ' ')
-end
-
-def carrets(text, first_column, last_column)
-  blanks(text, first_column) + '^' * (last_column - first_column)
+def raise_parse_error(error_message, location)
+  raise ParseError, location.generate_error_message(error_message)
 end

sig/lrama/lexer/location.rbs

@@ -1,14 +1,25 @@
 module Lrama
   class Lexer
     class Location
+      attr_reader grammar_file_path: String
       attr_reader first_line: Integer
       attr_reader first_column: Integer
       attr_reader last_line: Integer
       attr_reader last_column: Integer
 
-      def initialize: (first_line: Integer, first_column: Integer, last_line: Integer, last_column: Integer) -> void
+      def initialize: (grammar_file_path: String, first_line: Integer, first_column: Integer, last_line: Integer, last_column: Integer) -> void
 
       def ==: (Location other) -> bool
+      def partial_location: (Integer, Integer) -> Location
+      def generate_error_message: (String) -> String
+      def line_with_carrets: () -> String
+
+      private
+
+      def blanks: () -> String
+      def carrets: () -> String
+      def text: () -> String
+      def _text: () -> Array[String]
     end
   end
 end