[ruby/prism] Add MacJapanese encoding

MacJapanese (also aliased as MacJapan) is a modified Shift_JIS encoding, but is implemented identically in Ruby ruby/prism@9e0a097699
ruby · Nov 29, 2023 · 9fc40d2 · 9fc40d2
1 parent 86d9a6d
commit 9fc40d2
Show file tree

Hide file tree

Showing 5 changed files with 62 additions and 0 deletions.
diff --git a/lib/prism/prism.gemspec b/lib/prism/prism.gemspec
@@ -91,6 +91,7 @@ Gem::Specification.new do |spec|
     "src/enc/pm_cp950.c",
     "src/enc/pm_euc_jp.c",
     "src/enc/pm_gbk.c",
+    "src/enc/pm_mac_japanese.c",
     "src/enc/pm_shift_jis.c",
     "src/enc/pm_tables.c",
     "src/enc/pm_unicode.c",

diff --git a/prism/enc/pm_encoding.h b/prism/enc/pm_encoding.h
@@ -206,6 +206,7 @@ extern pm_encoding_t pm_encoding_mac_croatian;
 extern pm_encoding_t pm_encoding_mac_cyrillic;
 extern pm_encoding_t pm_encoding_mac_greek;
 extern pm_encoding_t pm_encoding_mac_iceland;
+extern pm_encoding_t pm_encoding_mac_japanese;
 extern pm_encoding_t pm_encoding_mac_roman;
 extern pm_encoding_t pm_encoding_mac_romania;
 extern pm_encoding_t pm_encoding_mac_thai;

diff --git a/prism/enc/pm_mac_japanese.c b/prism/enc/pm_mac_japanese.c
@@ -0,0 +1,57 @@
+#include "prism/enc/pm_encoding.h"
+
+static size_t
+pm_encoding_mac_japanese_char_width(const uint8_t *b, ptrdiff_t n) {
+    // These are the single byte characters.
+    if (*b < 0x80 || (*b >= 0xA1 && *b <= 0xDF)) {
+        return 1;
+    }
+
+    // These are the double byte characters.
+    if (
+        (n > 1) &&
+        ((b[0] >= 0x81 && b[0] <= 0x9F) || (b[0] >= 0xE0 && b[0] <= 0xFC)) &&
+        (b[1] >= 0x40 && b[1] <= 0xFC)
+    ) {
+        return 2;
+    }
+
+    return 0;
+}
+
+static size_t
+pm_encoding_mac_japanese_alpha_char(const uint8_t *b, ptrdiff_t n) {
+    if (pm_encoding_mac_japanese_char_width(b, n) == 1) {
+        return pm_encoding_ascii_alpha_char(b, n);
+    } else {
+        return 0;
+    }
+}
+
+static size_t
+pm_encoding_mac_japanese_alnum_char(const uint8_t *b, ptrdiff_t n) {
+    if (pm_encoding_mac_japanese_char_width(b, n) == 1) {
+        return pm_encoding_ascii_alnum_char(b, n);
+    } else {
+        return 0;
+    }
+}
+
+static bool
+pm_encoding_mac_japanese_isupper_char(const uint8_t *b, ptrdiff_t n) {
+    if (pm_encoding_mac_japanese_char_width(b, n) == 1) {
+        return pm_encoding_ascii_isupper_char(b, n);
+    } else {
+        return 0;
+    }
+}
+
+/** MacJapanese encoding */
+pm_encoding_t pm_encoding_mac_japanese = {
+    .name = "MacJapanese",
+    .char_width = pm_encoding_mac_japanese_char_width,
+    .alnum_char = pm_encoding_mac_japanese_alnum_char,
+    .alpha_char = pm_encoding_mac_japanese_alpha_char,
+    .isupper_char = pm_encoding_mac_japanese_isupper_char,
+    .multibyte = true
+};
diff --git a/prism/prism.c b/prism/prism.c
@@ -6303,6 +6303,8 @@ parser_lex_magic_comment_encoding_value(pm_parser_t *parser, const uint8_t *star
                 ENCODING1("macCyrillic", pm_encoding_mac_cyrillic);
                 ENCODING1("macGreek", pm_encoding_mac_greek);
                 ENCODING1("macIceland", pm_encoding_mac_iceland);
+                ENCODING1("MacJapanese", pm_encoding_mac_japanese);
+                ENCODING1("MacJapan", pm_encoding_mac_japanese);
                 ENCODING1("macRoman", pm_encoding_mac_roman);
                 ENCODING1("macRomania", pm_encoding_mac_romania);
                 ENCODING1("macThai", pm_encoding_mac_thai);

diff --git a/test/prism/encoding_test.rb b/test/prism/encoding_test.rb
@@ -72,6 +72,7 @@ class EncodingTest < TestCase
       Encoding::CP950 =>        0x00...0x10000,
       Encoding::CP51932 =>      0x00...0x10000,
       Encoding::GBK =>          0x00...0x10000,
+      Encoding::MACJAPANESE =>  0x00...0x10000,
       Encoding::Shift_JIS =>    0x00...0x10000,
       Encoding::Windows_31J =>  0x00...0x10000
     }