character_set 1.6.0-java → 1.7.0-java

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5ccdb4b8c43fdf0e3ed297415810a92805ee61324d91c464ccc7bee2575cb14b
-  data.tar.gz: cc36d5c8d65c036981bd368577bf49eb59d2d91f5fb779cae2e42aa991c94610
+  metadata.gz: 36050dd00f44b6efc26567bfd867ff21535fe1e35c9a8018d00f2145b27bfd37
+  data.tar.gz: d88d01cae2f5650271d73c654877b6cf62cf87acba9b8e699677b569c514b0e9
 SHA512:
-  metadata.gz: 35db3bcae78967f9d74beebccb3412a08ff505ead29fb3ef2c1bef05a5ebc5815c16ea948a3c831bb92041493f12502e86c3e01b64d19bfe8baba6dfb89c88e4
-  data.tar.gz: de08614dabac51cc9c71b5ed8a96d2865cab747fcb27ca3b00b0d75efcedf7f7f0f81bdd0472c275901d09258af137ddaf5150d9fea067a79f5fd6b526aa2ac5
+  metadata.gz: 646450cc07172ffdbceaf6cf215c03a60487ced2fdf578c4467f08374b77f6a8d4e043cbb92c89ea2ebc39c5b5adf38f8d74502632e033dbc8982928d6002f99
+  data.tar.gz: 1d77ccb0abef9c591189a77ed862657a04020f6bf9b3f31b7760ced20cdbee962411e29375b7d3883e95d8c97cac992afc89b17dbdaafbe99f0af02cfa22a0e1

data/BENCHMARK.md

@@ -3,88 +3,88 @@ Results of `rake:benchmark` on ruby 3.2.0dev (2022-02-14T14:35:54Z master 26187a
 ```
 Counting non-letters
 
-CharacterSet#count_in: 14794607.9 i/s
-        String#count:  3875939.3 i/s - 3.82x slower
+CharacterSet#count_in: 14627506.2 i/s
+        String#count:  3859777.0 i/s - 3.79x slower
 ```
 ```
 Detecting non-whitespace
 
- CharacterSet#cover?: 17448329.0 i/s
-       Regexp#match?: 13089358.1 i/s - 1.33x slower
+ CharacterSet#cover?: 17241902.8 i/s
+       Regexp#match?: 12971122.6 i/s - 1.33x slower
 ```
 ```
 Detecting non-letters
 
- CharacterSet#cover?: 17565596.9 i/s
-       Regexp#match?:  7951108.0 i/s - 2.21x slower
+ CharacterSet#cover?: 17243472.3 i/s
+       Regexp#match?:  7957626.9 i/s - 2.17x slower
 ```
 ```
 Removing ASCII whitespace
 
-CharacterSet#delete_in:  6306078.2 i/s
-           String#tr:  4734401.0 i/s - 1.33x slower
-         String#gsub:   211631.8 i/s - 29.80x slower
+CharacterSet#delete_in:  6190975.7 i/s
+           String#tr:  4722716.6 i/s - 1.31x slower
+         String#gsub:   214239.5 i/s - 28.90x slower
 ```
 ```
 Removing whitespace, emoji and umlauts
 
-CharacterSet#delete_in:  5984149.6 i/s
-           String#tr:   363643.1 i/s - 16.46x slower
-         String#gsub:   317201.7 i/s - 18.87x slower
+CharacterSet#delete_in:  5890471.8 i/s
+           String#tr:   348506.8 i/s - 16.90x slower
+         String#gsub:   318268.3 i/s - 18.51x slower
 ```
 ```
 Removing non-whitespace
 
-CharacterSet#keep_in:  7650925.6 i/s
-         String#gsub:   207374.6 i/s - 36.89x slower
-           String#tr:       12.3 i/s - 619745.60x slower
+CharacterSet#keep_in:  7396898.0 i/s
+         String#gsub:   208809.7 i/s - 35.42x slower
+           String#tr:       13.1 i/s - 564682.50x slower
 ```
 ```
 Keeping only emoji
 
-CharacterSet#keep_in:  7272940.1 i/s
-         String#gsub:   177993.8 i/s - 40.86x slower
-           String#tr:       12.3 i/s - 590222.71x slower
+CharacterSet#keep_in:  7022741.1 i/s
+         String#gsub:   180939.6 i/s - 38.81x slower
+           String#tr:       13.1 i/s - 536724.50x slower
 ```
 ```
 Extracting emoji to an Array
 
-   CharacterSet#scan:  2978285.0 i/s
-         String#scan:   865793.8 i/s - 3.44x slower
+   CharacterSet#scan:  3023176.8 i/s
+         String#scan:   893225.8 i/s - 3.38x slower
 ```
 ```
 Detecting whitespace
 
-CharacterSet#used_by?: 17292338.4 i/s
-       Regexp#match?: 11705563.9 i/s - 1.48x slower
+CharacterSet#used_by?: 17284025.9 i/s
+       Regexp#match?: 11847064.5 i/s - 1.46x slower
 ```
 ```
 Detecting emoji in a large string
 
-CharacterSet#used_by?:   340444.1 i/s
-       Regexp#match?:   180549.8 i/s - 1.89x slower
+CharacterSet#used_by?:   341386.1 i/s
+       Regexp#match?:   183121.6 i/s - 1.86x slower
 ```
 ```
 Adding entries
 
-    CharacterSet#add:  4951781.4 i/s
-       SortedSet#add:  1019637.9 i/s - 4.86x slower
+    CharacterSet#add:  4989762.3 i/s
+       SortedSet#add:  1157911.7 i/s - 4.31x slower
 ```
 ```
 Removing entries
 
- CharacterSet#delete:  5006337.6 i/s
-    SortedSet#delete:  3922752.2 i/s - same-ish
+ CharacterSet#delete:  4996703.6 i/s
+    SortedSet#delete:  4177401.5 i/s - same-ish
 ```
 ```
 Merging entries
 
-  CharacterSet#merge:      661.8 i/s
-     SortedSet#merge:        3.9 i/s - 167.82x slower
+  CharacterSet#merge:      666.7 i/s
+     SortedSet#merge:        4.0 i/s - 167.84x slower
 ```
 ```
 Getting the min and max
 
- CharacterSet#minmax:  1212462.2 i/s
-    SortedSet#minmax:      844.4 i/s - 1435.93x slower
+ CharacterSet#minmax:  1596470.9 i/s
+    SortedSet#minmax:      866.4 i/s - 1842.74x slower
 ```

data/CHANGELOG.md

@@ -4,6 +4,20 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/)
 and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.html).
 
+## [Unreleased]
+
+## [1.7.0] - 2023-05-12
+
+### Added
+
+- new codepoints for `::assigned` and `::emoji` predefined sets, as in Ruby 3.2.0
+
+### Fixed
+
+- fixed processing of Strings that are not ASCII- or UTF8-encoded
+- removed dependency on `set` and `sorted_set`
+  - thanks to https://github.com/mikebaldry for reporting a related issue (#2)
+
 ## [1.6.0] - 2022-02-16
 
 ### Added
@@ -63,7 +77,7 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
 - reduced memory consumption by > 90% for most use cases via dynamic resizing
   - before, every set instance required 136 KB for codepoints
   - now, 16 bytes for a CharacterSet in ASCII space, 8 KB for one in BMP space etc.
-- `#count_in` and `#scan_in` methods for `String` interaction
+- `#count_in` and `#scan` methods for `String` interaction
 - new predefined sets `::any`/`::all`, `::assigned`, `::surrogate`
 - conversion methods `#assigned_part`, `#valid_part`
 - sectioning methods `#ascii_part`, `#plane(n)`

data/README.md

@@ -96,7 +96,7 @@ string # => ''
 
 ```ruby
 CharacterSet.non_ascii.count_in('Tüür') # => 2
-CharacterSet.non_ascii.scan_in('Tüür') # => ['ü', 'ü']
+CharacterSet.non_ascii.scan('Tüür') # => ['ü', 'ü']
 ```
 
 There is also a core extension for String interaction.

data/Rakefile

@@ -3,6 +3,8 @@ require 'rspec/core/rake_task'
 require 'rubygems/package_task'
 require 'rake/extensiontask'
 
+Dir['tasks/**/*.rake'].each { |file| load(file) }
+
 RSpec::Core::RakeTask.new(:spec)
 
 task default: :spec
@@ -34,129 +36,6 @@ end
 
 task package: 'java:gem'
 
-desc 'Download relevant ruby/spec tests, adapt to CharacterSet and its variants'
-task :sync_ruby_spec do
-  require 'fileutils'
-
-  variants = {
-    'CharacterSet'       => './spec/ruby-spec/library/character_set',
-    'CharacterSet::Pure' => './spec/ruby-spec/library/character_set_pure',
-  }
-
-  # download fresh specs from ruby/spec repository
-  variants.each do |_, dir|
-    FileUtils.rm_rf(dir) if File.exist?(dir)
-    `svn export https://github.com/ruby/spec/trunk/library/set/sortedset #{dir}`
-  end
-
-  # make copies for each CharacterSet variant
-  base = variants.first[1]
-  variants.each_value { |dir| FileUtils.copy_entry(base, dir) unless dir == base }
-
-  # adapt specs to work with CharacterSet
-  variants.each do |class_name, dir|
-    Dir["#{dir}/**/*.rb"].each do |spec|
-      # ignore some tests that do not apply or are covered otherwise
-      if spec =~ %r{/(classify|divide|flatten|initialize|pretty_print)}
-        File.delete(spec)
-        next
-      end
-
-      adapted_content =
-        File.read(spec).
-        # adapt class name
-        gsub('SortedSet', (spec['/shared/'] ? 'variant' : class_name)).
-        gsub(/(it_behaves_like :[^,\n]+), (:[^,\n]+)/, "\\1, #{class_name}, \\2").
-        # get shared specs from a single shared dir at the parent level
-        gsub(/(require_relative ['"])(shared\/)/, '\1../\2').
-        # make 'mspec' syntax rspec-compatible
-        gsub(/describe (.*), shared.*$/, 'shared_examples \1 do |variant, method|').
-        gsub(/be_(false|true)/, 'be \1').
-        gsub('stub!', 'stub').
-        gsub('mock', 'double').
-        gsub('@method', 'method').
-        # remove unneeded requires
-        gsub(/require 'set'\n/, '').
-        gsub(/require.*spec_helper.*\n/, '').
-        gsub(/\A\n+/, '').
-        # make examples use Integers/codepoints
-        gsub(/1\.0|"cat"|"dog"|"hello"|"test"/, '0').
-        gsub('"one"', '1').
-        gsub('"two"', '2').
-        gsub('"three"', '3').
-        gsub('"four"', '4').
-        gsub('"five"', '5').
-        gsub(/x.(size|length) == 3/, 'x != 3').
-        gsub(/x.(size|length) != 3/, 'x == 3').
-        gsub(/(add)\(\d\)(\.to_a \}.should raise)/, '\1(:foo)\2')
-
-      File.open(spec, 'w') { |f| f.puts adapted_content }
-    end
-  end
-
-  # keep only one copy of the shared specs, at the parent level
-  FileUtils.rm_rf(base + '/../shared')
-  FileUtils.mv(base + '/shared', base + '/../')
-  variants.each_value { |dir| FileUtils.rm_rf(dir + '/shared') }
-end
-
-desc 'Download unicode casefold data and write new C header file'
-task :sync_casefold_data do
-  src_path = './CaseFolding.txt'
-  dst_path = './ext/character_set/unicode_casefold_table.h'
-
-  `wget http://www.unicode.org/Public/UNIDATA/CaseFolding.txt`
-
-  mapping = File.foreach(src_path).each_with_object({}) do |line, hash|
-    from, type, to = line.split(/\s*;\s*/).first(3)
-    # type 'C' stands for 'common', excludes mappings to multiple chars
-    hash[from] = to if type == 'C'
-  end.sort
-
-  content = File.read(dst_path + '.tmpl')
-    .sub(/(CASEFOLD_COUNT )0/, "\\1#{mapping.count}")
-    .sub('{}', ['{', mapping.map { |a, b| "{0x#{a},0x#{b}}," }, '}'].join("\n"))
-
-  File.write(dst_path, content)
-  File.unlink(src_path)
-end
-
-desc 'Update codepoint data for predefined sets, based on Onigmo'
-task :sync_predefined_sets do
-  %w[assigned emoji whitespace].each do |prop|
-    require 'regexp_property_values'
-    ranges = RegexpPropertyValues[prop].matched_ranges
-    str = ranges.map { |r| "#{r.min.to_s(16)},#{r.max.to_s(16)}\n" }.join.upcase
-    File.write("./lib/character_set/predefined_sets/#{prop}.cps", str, mode: 'w')
-  end
-end
-
-desc 'Run all IPS benchmarks'
-task :benchmark do
-  Dir['./benchmarks/*.rb'].sort.each { |file| require file }
-end
-
-namespace :benchmark do
-  desc 'Run all IPS benchmarks and store the comparison results in BENCHMARK.md'
-  task :write_to_file do
-    $store_comparison_results = {}
-
-    Rake.application[:benchmark].invoke
-
-    File.open('BENCHMARK.md', 'w') do |f|
-      f.puts "Results of `rake:benchmark` on #{RUBY_DESCRIPTION}", ''
-
-      $store_comparison_results.each do |caption, result|
-        f.puts '```',
-               caption,
-               '',
-               result.strip.gsub(/ \(±[^)]+\) /, '').gsub(/(same-ish).*$/, '\1').lines[1..-1],
-               '```'
-      end
-    end
-  end
-end
-
 unless RUBY_PLATFORM =~ /java/
   # recompile before benchmarking or running specs
   task(:benchmark).enhance([:compile])

data/character_set.gemspec

@@ -21,11 +21,4 @@ Gem::Specification.new do |s|
   s.extensions  = %w[ext/character_set/extconf.rb]
 
   s.required_ruby_version = '>= 2.1.0'
-
-  # SortedSet, needed for RubyFallback, was moved to a gem in Ruby 3.
-  # This dependency is only used if the C extension is unavailable.
-  # JRuby has it in the stdlib.
-  if RUBY_VERSION.to_f >= 3.0 && !RUBY_PLATFORM[/java/i]
-    s.add_dependency 'sorted_set', '~> 1.0'
-  end
 end

data/ext/character_set/character_set.c

@@ -376,22 +376,20 @@ cs_toggle_codepoint(VALUE cs, VALUE cp_num, int on, int return_nil_if_noop)
   cps = data->cps;
   len = data->len;
   cp = FIX2ULONG(cp_num);
-  if (return_nil_if_noop && (!tst_cp(cps, len, cp) == !on))
+  if (return_nil_if_noop && tst_cp(cps, len, cp) == on)
   {
     return Qnil;
   }
+
+  if (on)
+  {
+    set_cp(data, cp);
+  }
   else
   {
-    if (on)
-    {
-      set_cp(data, cp);
-    }
-    else
-    {
-      clr_cp(cps, len, cp);
-    }
-    return cs;
+    clr_cp(cps, len, cp);
   }
+  return cs;
 }
 
 static VALUE
@@ -575,7 +573,7 @@ cs_method_merge(VALUE self, VALUE other)
   {
     return cs_merge_cs(self, other);
   }
-  else if (TYPE(other) == T_ARRAY)
+  if (TYPE(other) == T_ARRAY)
   {
     return cs_merge_rb_array(self, other);
   }
@@ -917,10 +915,10 @@ cs_method_ext_inversion(int argc, VALUE *argv, VALUE self)
   return new_cs;
 }
 
-typedef int (*str_cp_handler)(unsigned int, cs_ar *, cs_cp len, struct cs_data *data, VALUE *memo);
+typedef int (*str_cp_handler)(unsigned int, cs_ar *, cs_cp len, struct cs_data *data, VALUE memo);
 
 static inline int
-add_str_cp_to_arr(unsigned int str_cp, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE *memo)
+add_str_cp_to_arr(unsigned int str_cp, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE memo)
 {
   set_cp(data, str_cp);
   return 1;
@@ -967,7 +965,7 @@ cs_method_case_insensitive(VALUE self)
 }
 
 static inline VALUE
-each_sb_cp(VALUE str, str_cp_handler func, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE *memo)
+each_sb_cp(VALUE str, str_cp_handler func, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE memo)
 {
   long i, str_len;
   unsigned int str_cp;
@@ -986,21 +984,29 @@ each_sb_cp(VALUE str, str_cp_handler func, cs_ar *cp_arr, cs_cp len, struct cs_d
 }
 
 static inline VALUE
-each_mb_cp(VALUE str, str_cp_handler func, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE *memo)
+each_mb_cp(VALUE str, str_cp_handler func, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE memo)
 {
   int n;
   unsigned int str_cp;
   const char *ptr, *end;
-  rb_encoding *enc;
+  rb_encoding *utf8;
+
+  utf8 = rb_utf8_encoding();
+  if (rb_enc_get(str) == utf8)
+  {
+    str = rb_str_new_frozen(str);
+  }
+  else
+  {
+    str = rb_str_encode(str, rb_enc_from_encoding(utf8), 0, Qnil);
+  }
 
-  str = rb_str_new_frozen(str);
   ptr = RSTRING_PTR(str);
   end = RSTRING_END(str);
-  enc = rb_enc_get(str);
 
   while (ptr < end)
   {
-    str_cp = rb_enc_codepoint_len(ptr, end, &n, enc);
+    str_cp = rb_enc_codepoint_len(ptr, end, &n, utf8);
     if (!(*func)(str_cp, cp_arr, len, data, memo))
     {
       return Qfalse;
@@ -1031,12 +1037,13 @@ single_byte_optimizable(VALUE str)
 }
 
 static inline VALUE
-each_cp(VALUE str, str_cp_handler func, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE *memo)
+each_cp(VALUE str, str_cp_handler func, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE memo)
 {
   if (single_byte_optimizable(str))
   {
     return each_sb_cp(str, func, cp_arr, len, data, memo);
   }
+
   return each_mb_cp(str, func, cp_arr, len, data, memo);
 }
 
@@ -1062,11 +1069,11 @@ cs_class_method_of_string(VALUE self, VALUE string)
 }
 
 static inline int
-count_str_cp(unsigned int str_cp, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE *memo)
+count_str_cp(unsigned int str_cp, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE memo)
 {
   if (tst_cp(cp_arr, len, str_cp))
   {
-    *memo += 1;
+    *((VALUE *)memo) += 1;
   }
   return 1;
 }
@@ -1074,17 +1081,17 @@ count_str_cp(unsigned int str_cp, cs_ar *cp_arr, cs_cp len, struct cs_data *data
 static VALUE
 cs_method_count_in(VALUE self, VALUE str)
 {
-  VALUE count;
+  long count;
   struct cs_data *data;
   raise_arg_err_unless_string(str);
   data = cs_fetch_data(self);
   count = 0;
-  each_cp(str, count_str_cp, data->cps, data->len, data, &count);
-  return INT2NUM((int)count);
+  each_cp(str, count_str_cp, data->cps, data->len, data, (VALUE)&count);
+  return LONG2FIX(count);
 }
 
 static inline int
-str_cp_in_arr(unsigned int str_cp, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE *memo)
+str_cp_in_arr(unsigned int str_cp, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE memo)
 {
   return tst_cp(cp_arr, len, str_cp);
 }
@@ -1099,11 +1106,11 @@ cs_method_cover_p(VALUE self, VALUE str)
 }
 
 static inline int
-add_str_cp_to_str_arr(unsigned int str_cp, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE *memo)
+add_str_cp_to_str_arr(unsigned int str_cp, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE memo)
 {
   if (tst_cp(cp_arr, len, str_cp))
   {
-    rb_ary_push(memo[0], rb_enc_uint_chr((int)str_cp, (rb_encoding *)memo[1]));
+    rb_ary_push(memo, rb_enc_uint_chr((int)str_cp, rb_utf8_encoding()));
   }
   return 1;
 }
@@ -1111,18 +1118,17 @@ add_str_cp_to_str_arr(unsigned int str_cp, cs_ar *cp_arr, cs_cp len, struct cs_d
 static VALUE
 cs_method_scan(VALUE self, VALUE str)
 {
-  VALUE memo[2];
+  VALUE memo;
   struct cs_data *data;
   raise_arg_err_unless_string(str);
   data = cs_fetch_data(self);
-  memo[0] = rb_ary_new();
-  memo[1] = (VALUE)rb_enc_get(str);
+  memo = rb_ary_new();
   each_cp(str, add_str_cp_to_str_arr, data->cps, data->len, data, memo);
-  return memo[0];
+  return memo;
 }
 
 static inline int
-str_cp_not_in_arr(unsigned int str_cp, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE *memo)
+str_cp_not_in_arr(unsigned int str_cp, cs_ar *cp_arr, cs_cp len, struct cs_data *data, VALUE memo)
 {
   return !tst_cp(cp_arr, len, str_cp);
 }
@@ -1146,9 +1152,9 @@ cs_apply_to_str(VALUE set, VALUE str, int delete, int bang)
   cs_cp cs_len;
   VALUE orig_str_len;
 
-  rb_encoding *enc;
+  rb_encoding *orig_enc, *utf8;
   char *s, *send, *t;
-  int ascompat, cr;
+  int orig_was_utf8, cr;
 
   raise_arg_err_unless_string(str);
 
@@ -1159,24 +1165,34 @@ cs_apply_to_str(VALUE set, VALUE str, int delete, int bang)
     return bang ? Qnil : str;
   }
 
-  if (!bang)
+  orig_enc = rb_enc_get(str);
+  utf8 = rb_utf8_encoding();
+  orig_was_utf8 = orig_enc == utf8;
+
+  if (!orig_was_utf8 && orig_enc != rb_usascii_encoding())
+  {
+    str = rb_str_encode(str, rb_enc_from_encoding(utf8), 0, Qnil);
+  }
+  else
   {
-    str = rb_str_dup(str);
+    if (!bang)
+    {
+      str = rb_str_dup(str);
+    }
   }
 
   cps = cs_fetch_cps(set, &cs_len);
   rb_str_modify(str);
-  enc = rb_enc_get(str);
-  ascompat = rb_enc_asciicompat(enc);
   s = t = RSTRING_PTR(str);
   send = RSTRING_END(str);
-  cr = ascompat ? ENC_CODERANGE_7BIT : ENC_CODERANGE_VALID;
+  cr = ENC_CODERANGE_7BIT;
+
   while (s < send)
   {
     unsigned int c;
     int clen;
 
-    if (ascompat && (c = *(unsigned char *)s) < 0x80)
+    if ((c = *(unsigned char *)s) < 0x80)
     {
       if (tst_cp(cps, cs_len, c) != delete)
       {
@@ -1188,12 +1204,12 @@ cs_apply_to_str(VALUE set, VALUE str, int delete, int bang)
     }
     else
     {
-      c = rb_enc_codepoint_len(s, send, &clen, enc);
+      c = rb_enc_codepoint_len(s, send, &clen, utf8);
 
       if (tst_cp(cps, cs_len, c) != delete)
       {
         if (t != s)
-          rb_enc_mbcput(c, t, enc);
+          rb_enc_mbcput(c, t, utf8);
         t += clen;
         if (cr == ENC_CODERANGE_7BIT)
           cr = ENC_CODERANGE_VALID;
@@ -1210,6 +1226,11 @@ cs_apply_to_str(VALUE set, VALUE str, int delete, int bang)
     return Qnil;
   }
 
+  if (!orig_was_utf8)
+  {
+    return rb_str_encode(str, rb_enc_from_encoding(orig_enc), 0, Qnil);
+  }
+
   return str;
 }
 

data/lib/character_set/parser.rb

@@ -4,11 +4,15 @@ class CharacterSet
 
     def codepoints_from_enumerable(object)
       raise ArgumentError, 'pass an Enumerable' unless object.respond_to?(:each)
+
       # Use #each to check first element (only this works for all Enumerables)
-      object.each do |e| # rubocop:disable Lint/UnreachableLoop
-        return object            if e.is_a?(Integer) && e >= 0 && e < 0x110000
-        return object.map(&:ord) if e.is_a?(String)  && e.length == 1
-        raise ArgumentError, "#{e.inspect} is not valid as a codepoint"
+      object.each do |el| # rubocop:disable Lint/UnreachableLoop
+        if el.is_a?(Integer) && el >= 0 && el < 0x110000
+          return object
+        elsif el.is_a?(String) && el.length == 1
+          return object.to_a.join.encode('utf-8').codepoints
+        end
+        raise ArgumentError, "#{el.inspect} is not valid as a codepoint"
       end
     end