[PRISM] Alias CP51932 to EUC-JP

4 files changed, 13 insertions, 61 deletions

diff --git a/lib/prism/prism.gemspec b/lib/prism/prism.gemspec
index cd90b025b2..2e8ebee409 100644
--- a/lib/prism/prism.gemspec
+++ b/lib/prism/prism.gemspec
@@ -88,7 +88,6 @@ Gem::Specification.new do |spec|
     "src/enc/pm_big5.c",
     "src/enc/pm_cp949.c",
     "src/enc/pm_cp950.c",
-    "src/enc/pm_cp51932.c",
     "src/enc/pm_euc_jp.c",
     "src/enc/pm_gbk.c",
     "src/enc/pm_shift_jis.c",
diff --git a/prism/enc/pm_cp51932.c b/prism/enc/pm_cp51932.c
deleted file mode 100644
index 75c9fb824f..0000000000
--- a/prism/enc/pm_cp51932.c
+++ /dev/null
@@ -1,57 +0,0 @@
-#include "prism/enc/pm_encoding.h"
-
-static size_t
-pm_encoding_cp51932_char_width(const uint8_t *b, ptrdiff_t n) {
-    // These are the single byte characters.
-    if (*b < 0x80) {
-        return 1;
-    }
-
-    // These are the double byte characters.
-    if (
-        (n > 1) &&
-        ((b[0] >= 0xa1 && b[0] <= 0xfe) || (b[0] == 0x8e)) &&
-        (b[1] >= 0xa1 && b[1] <= 0xfe)
-    ) {
-        return 2;
-    }
-
-    return 0;
-}
-
-static size_t
-pm_encoding_cp51932_alpha_char(const uint8_t *b, ptrdiff_t n) {
-    if (pm_encoding_cp51932_char_width(b, n) == 1) {
-        return pm_encoding_ascii_alpha_char(b, n);
-    } else {
-        return 0;
-    }
-}
-
-static size_t
-pm_encoding_cp51932_alnum_char(const uint8_t *b, ptrdiff_t n) {
-    if (pm_encoding_cp51932_char_width(b, n) == 1) {
-        return pm_encoding_ascii_alnum_char(b, n);
-    } else {
-        return 0;
-    }
-}
-
-static bool
-pm_encoding_cp51932_isupper_char(const uint8_t *b, ptrdiff_t n) {
-    if (pm_encoding_cp51932_char_width(b, n) == 1) {
-        return pm_encoding_ascii_isupper_char(b, n);
-    } else {
-        return 0;
-    }
-}
-
-/** cp51932 encoding */
-pm_encoding_t pm_encoding_cp51932 = {
-    .name = "cp51932",
-    .char_width = pm_encoding_cp51932_char_width,
-    .alnum_char = pm_encoding_cp51932_alnum_char,
-    .alpha_char = pm_encoding_cp51932_alpha_char,
-    .isupper_char = pm_encoding_cp51932_isupper_char,
-    .multibyte = true
-};
diff --git a/prism/enc/pm_euc_jp.c b/prism/enc/pm_euc_jp.c
index c891326a60..6468712607 100644
--- a/prism/enc/pm_euc_jp.c
+++ b/prism/enc/pm_euc_jp.c
@@ -67,3 +67,13 @@ pm_encoding_t pm_encoding_euc_jp = {
     .isupper_char = pm_encoding_euc_jp_isupper_char,
     .multibyte = true
 };
+
+/** CP51932 encoding */
+pm_encoding_t pm_encoding_cp51932 = {
+    .name = "CP51932",
+    .char_width = pm_encoding_euc_jp_char_width,
+    .alnum_char = pm_encoding_euc_jp_alnum_char,
+    .alpha_char = pm_encoding_euc_jp_alpha_char,
+    .isupper_char = pm_encoding_euc_jp_isupper_char,
+    .multibyte = true
+};
diff --git a/test/prism/encoding_test.rb b/test/prism/encoding_test.rb
index 8ab6f323e5..db51d653f9 100644
--- a/test/prism/encoding_test.rb
+++ b/test/prism/encoding_test.rb
@@ -71,14 +71,13 @@ module Prism
       Encoding::Big5_UAO =>      0x00...0x10000,
       Encoding::CP949 =>         0x00...0x10000,
       Encoding::CP950 =>         0x00...0x10000,
-      Encoding::CP51932 =>       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::Windows_31J =>   0x00...0x10000,
     }
 
     # By default we don't test every codepoint in these encodings because they
@@ -86,7 +85,8 @@ module Prism
     # suite.
     if ENV["PRISM_TEST_ALL_ENCODINGS"]
       encodings.merge!(
-        Encoding::EUC_JP =>        0x00...0x1000000,
+        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,