kdl 2.0.0 → 2.0.1

data/lib/kdl/tokenizer.rb

@@ -1,8 +1,10 @@
+# frozen_string_literal: true
+
 require 'bigdecimal'
 
 module KDL
   class Tokenizer
-    class Error < StandardError
+    class Error < ::KDL::Error
       def initialize(message, line, column)
         super("#{message} (#{line}:#{column})")
       end
@@ -40,16 +42,16 @@ module KDL
       '=' => :EQUALS
     }
 
-    WHITESPACE = ["\u0009", "\u000B", "\u0020", "\u00A0",
-                  "\u1680", "\u2000", "\u2001", "\u2002",
-                  "\u2003", "\u2004", "\u2005", "\u2006",
-                  "\u2007", "\u2008", "\u2009", "\u200A",
-                  "\u202F", "\u205F", "\u3000" ]
+    WHITESPACE = ["\u0009", "\u0020", "\u00A0", "\u1680",
+                  "\u2000", "\u2001", "\u2002", "\u2003",
+                  "\u2004", "\u2005", "\u2006", "\u2007",
+                  "\u2008", "\u2009", "\u200A", "\u202F",
+                  "\u205F", "\u3000"]
     WS = "[#{Regexp.escape(WHITESPACE.join)}]"
     WS_STAR = /\A#{WS}*\z/
     WS_PLUS = /\A#{WS}+\z/
 
-    NEWLINES = ["\u000A", "\u0085", "\u000C", "\u2028", "\u2029"]
+    NEWLINES = ["\u000A", "\u0085", "\u000B", "\u000C", "\u2028", "\u2029"]
     NEWLINES_PATTERN = Regexp.new("(#{NEWLINES.map{Regexp.escape(_1)}.join('|')}|\r\n?)", Regexp::MULTILINE)
 
     OTHER_NON_IDENTIFIER_CHARS = ("\x0".."\x20").to_a - WHITESPACE
@@ -68,13 +70,15 @@ module KDL
       "\uFEFF"
     ]
 
+    VERSION_PATTERN = /\A\/-[#{WHITESPACE.join}]*kdl-version[#{WHITESPACE.join}]+(\d+)[#{WHITESPACE.join}]*[#{NEWLINES.join}]/
+
     def initialize(str, start = 0)
       @str = debom(str)
       @context = nil
       @rawstring_hashes = nil
       @start = start
       @index = start
-      @buffer = ""
+      @buffer = +""
       @done = false
       @previous_context = nil
       @line = 1
@@ -84,7 +88,7 @@ module KDL
     end
 
     def version_directive
-      if m = @str.match(/\A\/-[#{WHITESPACE.join}]*kdl-version[#{WHITESPACE.join}]+(\d+)\s*[#{NEWLINES.join}]/)
+      if m = @str.match(VERSION_PATTERN)
         m[1].to_i
       end
     end
@@ -121,11 +125,11 @@ module KDL
             if self[@index + 1] == '"' && self[@index + 2] == '"'
               nl = expect_newline(@index + 3)
               self.context = :multiline_string
-              @buffer = ''
+              @buffer = +''
               traverse(3 + nl.length)
             else
               self.context = :string
-              @buffer = ''
+              @buffer = +''
               traverse(1)
             end
           when '#'
@@ -134,14 +138,14 @@ module KDL
                 nl = expect_newline(@index + 4)
                 self.context = :multiline_rawstring
                 @rawstring_hashes = 1
-                @buffer = ''
+                @buffer = +''
                 traverse(4 + nl.length)
                 next
               else
                 self.context = :rawstring
                 traverse(2)
                 @rawstring_hashes = 1
-                @buffer = ''
+                @buffer = +''
                 next
               end
             elsif self[@index + 1] == '#'
@@ -156,18 +160,18 @@ module KDL
                   nl = expect_newline(i + 3)
                   self.context = :multiline_rawstring
                   traverse(@rawstring_hashes + 3 + nl.length)
-                  @buffer = ''
+                  @buffer = +''
                   next
                 else
                   self.context = :rawstring
                   traverse(@rawstring_hashes + 1)
-                  @buffer = ''
+                  @buffer = +''
                   next
                 end
               end
             end
             self.context = :keyword
-            @buffer = c
+            @buffer = +c
             traverse(1)
           when '-'
             n = self[@index + 1]
@@ -184,16 +188,16 @@ module KDL
               self.context = :ident
               traverse(1)
             end
-            @buffer = c
+            @buffer = +c
           when /[0-9+]/
             n = self[@index + 1]
             if c == '0' && n =~ /[box]/
               traverse(2)
-              @buffer = ''
+              @buffer = +''
               self.context = integer_context(n)
             else
               self.context = :decimal
-              @buffer = c
+              @buffer = +c
               traverse(1)
             end
           when '\\'
@@ -202,20 +206,20 @@ module KDL
             if la[0] == :NEWLINE || la[0] == :EOF || (la[0] == :WS && (lan = t.next_token[0]) == :NEWLINE || lan == :EOF)
               traverse_to(t.index)
               @buffer = "#{c}#{la[1].value}"
-              @buffer += "\n" if lan == :NEWLINE
+              @buffer << "\n" if lan == :NEWLINE
               self.context = :whitespace
             else
               raise_error "Unexpected '\\' (#{la[0]})"
             end
           when '='
             self.context = :equals
-            @buffer = c
+            @buffer = +c
             traverse(1)
           when *SYMBOLS.keys
-            return token(SYMBOLS[c], c).tap { traverse(1) }
+            return token(SYMBOLS[c], -c).tap { traverse(1) }
           when *NEWLINES, "\r"
             nl = expect_newline
-            return token(:NEWLINE, nl).tap do
+            return token(:NEWLINE, -nl).tap do
               traverse(nl.length)
             end
           when "/"
@@ -230,12 +234,12 @@ module KDL
               return token(:SLASHDASH, '/-').tap { traverse(2) }
             else
               self.context = :ident
-              @buffer = c
+              @buffer = +c
               traverse(1)
             end
           when *WHITESPACE
             self.context = :whitespace
-            @buffer = c
+            @buffer = +c
             traverse(1)
           when nil
             return [false, token(:EOF, :EOF)[1]] if @done
@@ -244,14 +248,14 @@ module KDL
             return token(:EOF, :EOF)
           when INITIAL_IDENTIFIER_CHARS
             self.context = :ident
-            @buffer = c
+            @buffer = +c
             traverse(1)
           when '('
             @type_context = true
-            return token(:LPAREN, c).tap { traverse(1) }
+            return token(:LPAREN, -c).tap { traverse(1) }
           when ')'
             @type_context = false
-            return token(:RPAREN, c).tap { traverse(1) }
+            return token(:RPAREN, -c).tap { traverse(1) }
           else
             raise_error "Unexpected character #{c.inspect}"
           end
@@ -259,7 +263,7 @@ module KDL
           case c
           when IDENTIFIER_CHARS
             traverse(1)
-            @buffer += c
+            @buffer << c
           else
             case @buffer
             when 'true', 'false', 'null', 'inf', '-inf', 'nan'
@@ -267,14 +271,14 @@ module KDL
             when /\A\.\d/
               raise_error "Identifier cannot look like an illegal float"
             else
-              return token(:IDENT, @buffer)
+              return token(:IDENT, -@buffer)
             end
           end
         when :keyword
           case c
           when /[a-z\-]/
             traverse(1)
-            @buffer += c
+            @buffer << c
           else
             case @buffer
             when '#true'  then return token(:TRUE, true)
@@ -289,13 +293,13 @@ module KDL
         when :string
           case c
           when '\\'
-            @buffer += c
+            @buffer << c
             c2 = self[@index + 1]
-            @buffer += c2
+            @buffer << c2
             if c2.match?(NEWLINES_PATTERN)
               i = 2
               while self[@index + i]&.match?(NEWLINES_PATTERN)
-                @buffer += self[@index + i]
+                @buffer << self[@index + i]
                 i+=1
               end
               traverse(i)
@@ -303,31 +307,31 @@ module KDL
               traverse(2)
             end
           when '"'
-            return token(:STRING, unescape(@buffer)).tap { traverse(1) }
+            return token(:STRING, -unescape(@buffer)).tap { traverse(1) }
           when *NEWLINES, "\r"
             raise_error "Unexpected NEWLINE in string literal"
           when nil
             raise_error "Unterminated string literal"
           else
-            @buffer += c
+            @buffer << c
             traverse(1)
           end
         when :multiline_string
           case c
           when '\\'
-            @buffer += c
-            @buffer += self[@index + 1]
+            @buffer << c
+            @buffer << self[@index + 1]
             traverse(2)
           when '"'
             if self[@index + 1] == '"' && self[@index + 2] == '"'
-              return token(:STRING, unescape_non_ws(dedent(unescape_ws(@buffer)))).tap { traverse(3) }
+              return token(:STRING, -unescape_non_ws(dedent(unescape_ws(@buffer)))).tap { traverse(3) }
             end
-            @buffer += c
+            @buffer << c
             traverse(1)
           when nil
             raise_error "Unterminated multi-line string literal"
           else
-            @buffer += c
+            @buffer << c
             traverse(1)
           end
         when :rawstring
@@ -338,13 +342,13 @@ module KDL
             h = 0
             h += 1 while self[@index + 1 + h] == '#' && h < @rawstring_hashes
             if h == @rawstring_hashes
-              return token(:RAWSTRING, @buffer).tap { traverse(1 + h) }
+              return token(:RAWSTRING, -@buffer).tap { traverse(1 + h) }
             end
           when *NEWLINES, "\r"
             raise_error "Unexpected NEWLINE in rawstring literal"
           end
 
-          @buffer += c
+          @buffer << c
           traverse(1)
         when :multiline_rawstring
           raise_error "Unterminated multi-line rawstring literal" if c.nil?
@@ -353,17 +357,17 @@ module KDL
             h = 1
             h += 1 while self[@index + 3 + h] == '#' && h < @rawstring_hashes
             if h == @rawstring_hashes
-              return token(:RAWSTRING, dedent(@buffer)).tap { traverse(3 + h) }
+              return token(:RAWSTRING, -dedent(@buffer)).tap { traverse(3 + h) }
             end
           end
 
-          @buffer += c
+          @buffer << c
           traverse(1)
         when :decimal
           case c
           when /[0-9.\-+_eE]/
             traverse(1)
-            @buffer += c
+            @buffer << c
           else
             return parse_decimal(@buffer)
           end
@@ -371,7 +375,7 @@ module KDL
           case c
           when /[0-9a-fA-F_]/
             traverse(1)
-            @buffer += c
+            @buffer << c
           else
             return parse_hexadecimal(@buffer)
           end
@@ -379,7 +383,7 @@ module KDL
           case c
           when /[0-7_]/
             traverse(1)
-            @buffer += c
+            @buffer << c
           else
             return parse_octal(@buffer)
           end
@@ -387,7 +391,7 @@ module KDL
           case c
           when /[01_]/
             traverse(1)
-            @buffer += c
+            @buffer << c
           else
             return parse_binary(@buffer)
           end
@@ -419,10 +423,10 @@ module KDL
         when :whitespace
           if WHITESPACE.include?(c)
             traverse(1)
-            @buffer += c
+            @buffer << c
           elsif c == '='
             self.context = :equals
-            @buffer += c
+            @buffer << c
             traverse(1)
           elsif c == "/" && self[@index + 1] == '*'
             self.context = :multi_line_comment
@@ -433,22 +437,22 @@ module KDL
             la = t.next_token
             if la[0] == :NEWLINE || la[0] == :EOF || (la[0] == :WS && (lan = t.next_token[0]) == :NEWLINE || lan == :EOF)
               traverse_to(t.index)
-              @buffer += "#{c}#{la[1].value}"
-              @buffer += "\n" if lan == :NEWLINE
+              @buffer << "#{c}#{la[1].value}"
+              @buffer << "\n" if lan == :NEWLINE
             else
               raise_error "Unexpected '\\' (#{la[0]})"
             end
           else
-            return token(:WS, @buffer)
+            return token(:WS, -@buffer)
           end
         when :equals
           t = Tokenizer.new(@str, @index)
           la = t.next_token
           if la[0] == :WS
-            @buffer += la[1].value
+            @buffer << la[1].value
             traverse_to(t.index)
           end
-          return token(:EQUALS, @buffer)
+          return token(:EQUALS, -@buffer)
         else
           # :nocov:
           raise_error "Unknown context `#{@context}'"
@@ -482,8 +486,12 @@ module KDL
       traverse(i - @index)
     end
 
-    def raise_error(message)
-      raise Error.new(message, @line, @column)
+    def raise_error(error)
+      case error
+      when String then raise Error.new(error, @line, @column)
+      when Error then raise error
+      else raise Error.new(error.message, @line, @column)
+      end
     end
 
     def context=(val)
@@ -538,11 +546,11 @@ module KDL
       return parse_float(s) if s =~ /[.E]/i
 
       token(:INTEGER, Integer(munch_underscores(s), 10), format: '%d')
-    rescue
+    rescue => e
       if s[0] =~ INITIAL_IDENTIFIER_CHARS && s[1..-1].each_char.all? { |c| c =~ IDENTIFIER_CHARS }
-        token(:IDENT, s)
+        token(:IDENT, -s)
       else
-        raise
+        raise_error(e)
       end
     end
 
@@ -564,14 +572,20 @@ module KDL
 
     def parse_hexadecimal(s)
       token(:INTEGER, Integer(munch_underscores(s), 16))
+    rescue ArgumentError => e
+      raise_error(e)
     end
 
     def parse_octal(s)
       token(:INTEGER, Integer(munch_underscores(s), 8))
+    rescue ArgumentError => e
+      raise_error(e)
     end
 
     def parse_binary(s)
       token(:INTEGER, Integer(munch_underscores(s), 2))
+    rescue ArgumentError => e
+      raise_error(e)
     end
 
     def munch_underscores(s)
@@ -599,8 +613,8 @@ module KDL
         .gsub(rgx) { |m| replace_esc(m) }
         .gsub(/\\u\{[0-9a-fA-F]{0,6}\}/) do |m|
           i = Integer(m[3..-2], 16)
-          if i < 0 || i > 0x10FFFF
-            raise_error "Invalid code point #{u}"
+          if i < 0 || i > 0x10FFFF || (0xD800..0xDFFF).include?(i)
+            raise_error "Invalid code point #{m}"
           end
           i.chr(Encoding::UTF_8)
         end
@@ -631,9 +645,9 @@ module KDL
       end
       return "" if lines.empty?
       raise_error "Invalid multiline string final line" unless indent.match?(WS_STAR)
-      valid = /\A(?:#{Regexp.escape(indent)})(.*)/
+      valid = /\A#{Regexp.escape(indent)}(.*)/
 
-      lines.map! do |line|
+      lines.map do |line|
         case line
         when WS_STAR then ""
         when valid then $1

data/lib/kdl/types/base64.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'base64'
 
 module KDL

data/lib/kdl/types/country/iso3166_countries.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module KDL
   module Types
     class Country < Value::Custom

data/lib/kdl/types/country/iso3166_subdivisions.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module KDL
   module Types
     class CountrySubdivision < Value::Custom

data/lib/kdl/types/country.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'kdl/types/country/iso3166_countries'
 require 'kdl/types/country/iso3166_subdivisions'
 

data/lib/kdl/types/currency/iso4217_currencies.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module KDL
   module Types
     class Currency < Value::Custom

data/lib/kdl/types/currency.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'kdl/types/currency/iso4217_currencies'
 
 module KDL

data/lib/kdl/types/date_time.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'time'
 
 module KDL

data/lib/kdl/types/decimal.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module KDL
   module Types
     class Decimal < Value::Custom

data/lib/kdl/types/duration/iso8601_parser.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 # Shamelessly borrowed from https://github.com/rails/rails/tree/main/activesupport
 #
 # Copyright (c) 2005-2021 David Heinemeier Hansson

data/lib/kdl/types/duration.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'kdl/types/duration/iso8601_parser'
 
 module KDL

data/lib/kdl/types/email/parser.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative '../hostname/validator'
 
 module KDL
@@ -11,7 +13,7 @@ module KDL
         end
 
         def parse
-          local = ''
+          local = +''
           unicode_domain = nil
           domain = nil
           context = :start
@@ -23,7 +25,7 @@ module KDL
             when :part
               case context
               when :start, :after_dot
-                local += value
+                local << value
                 context = :after_part
               else
                 raise ArgumentError, "invalid email #{@string} (unexpected part #{value} at #{context})"
@@ -31,7 +33,7 @@ module KDL
             when :dot
               case context
               when :after_part
-                local += value
+                local << value
                 context = :after_dot
               else
                 raise ArgumentError, "invalid email #{@string} (unexpected dot at #{context})"
@@ -91,7 +93,7 @@ module KDL
             end
           end
           @context = nil
-          @buffer = ''
+          @buffer = +''
           loop do
             c = @string[@index]
             return [:end, nil] if c.nil?
@@ -111,7 +113,7 @@ module KDL
                 @index += 1
               when local_part_chars
                 @context = :part
-                @buffer += c
+                @buffer << c
                 @index += 1
               else
                 raise ArgumentError, "invalid email #{@string} (unexpected #{c})"
@@ -119,7 +121,7 @@ module KDL
             when :part
               case c
               when local_part_chars
-                @buffer += c
+                @buffer << c
                 @index += 1
               when '.', '@'
                 return [:part, @buffer]
@@ -135,7 +137,7 @@ module KDL
                 @index += 1
                 return [:part, @buffer]
               else
-                @buffer += c
+                @buffer << c
                 @index += 1
               end
             end

data/lib/kdl/types/email.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative './email/parser'
 
 module KDL

data/lib/kdl/types/hostname/validator.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'simpleidn'
 
 module KDL

data/lib/kdl/types/hostname.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative './hostname/validator'
 
 module KDL

data/lib/kdl/types/ip.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module KDL
   module Types
     class IP < Value::Custom

data/lib/kdl/types/irl/parser.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module KDL
   module Types
     class IRLReference < Value::Custom
@@ -96,13 +98,13 @@ module KDL
         end
 
         def self.build_uri_string(scheme, auth, domain, path, search, hash)
-          string = ''
-          string += "#{scheme}://" if scheme
-          string += auth if auth
-          string += domain if domain
-          string += path if path
-          string += "?#{search}" if search
-          string += "##{hash}" if hash
+          string = +''
+          string << "#{scheme}://" if scheme
+          string << auth if auth
+          string << domain if domain
+          string << path if path
+          string << "?#{search}" if search
+          string << "##{hash}" if hash
           string
         end
       end

data/lib/kdl/types/irl.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative './irl/parser'
 
 module KDL

data/lib/kdl/types/regex.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module KDL
   module Types
     class Regex < Value::Custom

data/lib/kdl/types/url.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module KDL
   module Types
     class URLReference < Value::Custom