[ruby/prism] Support other EUC-JP encodings

https://github.com/ruby/prism/commit/d040337ce9

1 files changed, 87 insertions, 78 deletions

diff --git a/test/prism/encoding_test.rb b/test/prism/encoding_test.rb
index db51d653f9..b206ab20e5 100644
--- a/test/prism/encoding_test.rb
+++ b/test/prism/encoding_test.rb
@@ -6,92 +6,101 @@ require_relative "test_helper"
 
 module Prism
   class EncodingTest < TestCase
+    codepoints_1byte = 0x00...0x100
+    codepoints_2bytes = 0x00...0x10000
+
     encodings = {
-      Encoding::ASCII =>         0x00...0x100,
-      Encoding::ASCII_8BIT =>    0x00...0x100,
-      Encoding::CP850 =>         0x00...0x100,
-      Encoding::CP852 =>         0x00...0x100,
-      Encoding::CP855 =>         0x00...0x100,
-      Encoding::GB1988 =>        0x00...0x100,
-      Encoding::IBM437 =>        0x00...0x100,
-      Encoding::IBM720 =>        0x00...0x100,
-      Encoding::IBM737 =>        0x00...0x100,
-      Encoding::IBM775 =>        0x00...0x100,
-      Encoding::IBM852 =>        0x00...0x100,
-      Encoding::IBM855 =>        0x00...0x100,
-      Encoding::IBM857 =>        0x00...0x100,
-      Encoding::IBM860 =>        0x00...0x100,
-      Encoding::IBM861 =>        0x00...0x100,
-      Encoding::IBM862 =>        0x00...0x100,
-      Encoding::IBM863 =>        0x00...0x100,
-      Encoding::IBM864 =>        0x00...0x100,
-      Encoding::IBM865 =>        0x00...0x100,
-      Encoding::IBM866 =>        0x00...0x100,
-      Encoding::IBM869 =>        0x00...0x100,
-      Encoding::ISO_8859_1 =>    0x00...0x100,
-      Encoding::ISO_8859_2 =>    0x00...0x100,
-      Encoding::ISO_8859_3 =>    0x00...0x100,
-      Encoding::ISO_8859_4 =>    0x00...0x100,
-      Encoding::ISO_8859_5 =>    0x00...0x100,
-      Encoding::ISO_8859_6 =>    0x00...0x100,
-      Encoding::ISO_8859_7 =>    0x00...0x100,
-      Encoding::ISO_8859_8 =>    0x00...0x100,
-      Encoding::ISO_8859_9 =>    0x00...0x100,
-      Encoding::ISO_8859_10 =>   0x00...0x100,
-      Encoding::ISO_8859_11 =>   0x00...0x100,
-      Encoding::ISO_8859_13 =>   0x00...0x100,
-      Encoding::ISO_8859_14 =>   0x00...0x100,
-      Encoding::ISO_8859_15 =>   0x00...0x100,
-      Encoding::ISO_8859_16 =>   0x00...0x100,
-      Encoding::KOI8_R =>        0x00...0x100,
-      Encoding::KOI8_U =>        0x00...0x100,
-      Encoding::MACCENTEURO =>   0x00...0x100,
-      Encoding::MACCROATIAN =>   0x00...0x100,
-      Encoding::MACCYRILLIC =>   0x00...0x100,
-      Encoding::MACGREEK =>      0x00...0x100,
-      Encoding::MACICELAND =>    0x00...0x100,
-      Encoding::MACROMAN =>      0x00...0x100,
-      Encoding::MACROMANIA =>    0x00...0x100,
-      Encoding::MACTHAI =>       0x00...0x100,
-      Encoding::MACTURKISH =>    0x00...0x100,
-      Encoding::MACUKRAINE =>    0x00...0x100,
-      Encoding::TIS_620 =>       0x00...0x100,
-      Encoding::Windows_1250 =>  0x00...0x100,
-      Encoding::Windows_1251 =>  0x00...0x100,
-      Encoding::Windows_1252 =>  0x00...0x100,
-      Encoding::Windows_1253 =>  0x00...0x100,
-      Encoding::Windows_1254 =>  0x00...0x100,
-      Encoding::Windows_1255 =>  0x00...0x100,
-      Encoding::Windows_1256 =>  0x00...0x100,
-      Encoding::Windows_1257 =>  0x00...0x100,
-      Encoding::Windows_1258 =>  0x00...0x100,
-      Encoding::Windows_874 =>   0x00...0x100,
-      Encoding::Big5 =>          0x00...0x10000,
-      Encoding::Big5_HKSCS =>    0x00...0x10000,
-      Encoding::Big5_UAO =>      0x00...0x10000,
-      Encoding::CP949 =>         0x00...0x10000,
-      Encoding::CP950 =>         0x00...0x10000,
-      Encoding::GBK =>           0x00...0x10000,
-      Encoding::MACJAPANESE =>   0x00...0x10000,
-      Encoding::Shift_JIS =>     0x00...0x10000,
-      Encoding::SJIS_DoCoMo =>   0x00...0x10000,
-      Encoding::SJIS_KDDI =>     0x00...0x10000,
-      Encoding::SJIS_SoftBank => 0x00...0x10000,
-      Encoding::Windows_31J =>   0x00...0x10000,
+      Encoding::ASCII =>         codepoints_1byte,
+      Encoding::ASCII_8BIT =>    codepoints_1byte,
+      Encoding::CP850 =>         codepoints_1byte,
+      Encoding::CP852 =>         codepoints_1byte,
+      Encoding::CP855 =>         codepoints_1byte,
+      Encoding::GB1988 =>        codepoints_1byte,
+      Encoding::IBM437 =>        codepoints_1byte,
+      Encoding::IBM720 =>        codepoints_1byte,
+      Encoding::IBM737 =>        codepoints_1byte,
+      Encoding::IBM775 =>        codepoints_1byte,
+      Encoding::IBM852 =>        codepoints_1byte,
+      Encoding::IBM855 =>        codepoints_1byte,
+      Encoding::IBM857 =>        codepoints_1byte,
+      Encoding::IBM860 =>        codepoints_1byte,
+      Encoding::IBM861 =>        codepoints_1byte,
+      Encoding::IBM862 =>        codepoints_1byte,
+      Encoding::IBM863 =>        codepoints_1byte,
+      Encoding::IBM864 =>        codepoints_1byte,
+      Encoding::IBM865 =>        codepoints_1byte,
+      Encoding::IBM866 =>        codepoints_1byte,
+      Encoding::IBM869 =>        codepoints_1byte,
+      Encoding::ISO_8859_1 =>    codepoints_1byte,
+      Encoding::ISO_8859_2 =>    codepoints_1byte,
+      Encoding::ISO_8859_3 =>    codepoints_1byte,
+      Encoding::ISO_8859_4 =>    codepoints_1byte,
+      Encoding::ISO_8859_5 =>    codepoints_1byte,
+      Encoding::ISO_8859_6 =>    codepoints_1byte,
+      Encoding::ISO_8859_7 =>    codepoints_1byte,
+      Encoding::ISO_8859_8 =>    codepoints_1byte,
+      Encoding::ISO_8859_9 =>    codepoints_1byte,
+      Encoding::ISO_8859_10 =>   codepoints_1byte,
+      Encoding::ISO_8859_11 =>   codepoints_1byte,
+      Encoding::ISO_8859_13 =>   codepoints_1byte,
+      Encoding::ISO_8859_14 =>   codepoints_1byte,
+      Encoding::ISO_8859_15 =>   codepoints_1byte,
+      Encoding::ISO_8859_16 =>   codepoints_1byte,
+      Encoding::KOI8_R =>        codepoints_1byte,
+      Encoding::KOI8_U =>        codepoints_1byte,
+      Encoding::MACCENTEURO =>   codepoints_1byte,
+      Encoding::MACCROATIAN =>   codepoints_1byte,
+      Encoding::MACCYRILLIC =>   codepoints_1byte,
+      Encoding::MACGREEK =>      codepoints_1byte,
+      Encoding::MACICELAND =>    codepoints_1byte,
+      Encoding::MACROMAN =>      codepoints_1byte,
+      Encoding::MACROMANIA =>    codepoints_1byte,
+      Encoding::MACTHAI =>       codepoints_1byte,
+      Encoding::MACTURKISH =>    codepoints_1byte,
+      Encoding::MACUKRAINE =>    codepoints_1byte,
+      Encoding::TIS_620 =>       codepoints_1byte,
+      Encoding::Windows_1250 =>  codepoints_1byte,
+      Encoding::Windows_1251 =>  codepoints_1byte,
+      Encoding::Windows_1252 =>  codepoints_1byte,
+      Encoding::Windows_1253 =>  codepoints_1byte,
+      Encoding::Windows_1254 =>  codepoints_1byte,
+      Encoding::Windows_1255 =>  codepoints_1byte,
+      Encoding::Windows_1256 =>  codepoints_1byte,
+      Encoding::Windows_1257 =>  codepoints_1byte,
+      Encoding::Windows_1258 =>  codepoints_1byte,
+      Encoding::Windows_874 =>   codepoints_1byte,
+      Encoding::Big5 =>          codepoints_2bytes,
+      Encoding::Big5_HKSCS =>    codepoints_2bytes,
+      Encoding::Big5_UAO =>      codepoints_2bytes,
+      Encoding::CP949 =>         codepoints_2bytes,
+      Encoding::CP950 =>         codepoints_2bytes,
+      Encoding::GBK =>           codepoints_2bytes,
+      Encoding::MACJAPANESE =>   codepoints_2bytes,
+      Encoding::Shift_JIS =>     codepoints_2bytes,
+      Encoding::SJIS_DoCoMo =>   codepoints_2bytes,
+      Encoding::SJIS_KDDI =>     codepoints_2bytes,
+      Encoding::SJIS_SoftBank => codepoints_2bytes,
+      Encoding::Windows_31J =>   codepoints_2bytes,
     }
 
     # By default we don't test every codepoint in these encodings because they
     # are 3 and 4 byte representations so it can drastically slow down the test
     # suite.
     if ENV["PRISM_TEST_ALL_ENCODINGS"]
+      codepoints_eucjp = [*(0x00...0x10000), *(0x00...0x10000).map { |bytes| bytes | 0x8F0000 }]
+      codepoints_unicode = 0x00...0x110000
+
+      encodings.clear
       encodings.merge!(
-        Encoding::CP51932 =>       [*(0x00...0x10000), *(0x00...0x10000).map { |bytes| bytes | 0x8F0000 }],
-        Encoding::EUC_JP =>        [*(0x00...0x10000), *(0x00...0x10000).map { |bytes| bytes | 0x8F0000 }],
-        Encoding::UTF_8 =>         0x00...0x110000,
-        Encoding::UTF8_MAC =>      0x00...0x110000,
-        Encoding::UTF8_DoCoMo =>   0x00...0x110000,
-        Encoding::UTF8_KDDI =>     0x00...0x110000,
-        Encoding::UTF8_SoftBank => 0x00...0x110000
+        Encoding::CP51932 =>       codepoints_eucjp,
+        Encoding::EUC_JP =>        codepoints_eucjp,
+        Encoding::EUCJP_MS =>      codepoints_eucjp,
+        Encoding::EUC_JIS_2004 =>  codepoints_eucjp,
+        Encoding::UTF_8 =>         codepoints_unicode,
+        Encoding::UTF8_MAC =>      codepoints_unicode,
+        Encoding::UTF8_DoCoMo =>   codepoints_unicode,
+        Encoding::UTF8_KDDI =>     codepoints_unicode,
+        Encoding::UTF8_SoftBank => codepoints_unicode
       )
     end