[ruby/prism] Add `CP949` encoding

https://github.com/ruby/prism/commit/9e78dfdf69

5 files changed, 61 insertions, 0 deletions

diff --git a/lib/prism/prism.gemspec b/lib/prism/prism.gemspec
index 6fbd6c2945..43ec005452 100644
--- a/lib/prism/prism.gemspec
+++ b/lib/prism/prism.gemspec
@@ -86,6 +86,7 @@ Gem::Specification.new do |spec|
     "src/diagnostic.c",
     "src/enc/pm_big5.c",
     "src/enc/pm_cp51932.c",
+    "src/enc/pm_cp949.c",
     "src/enc/pm_euc_jp.c",
     "src/enc/pm_gbk.c",
     "src/enc/pm_shift_jis.c",
diff --git a/prism/enc/pm_cp949.c b/prism/enc/pm_cp949.c
new file mode 100644
index 0000000000..f3b5a50fde
--- /dev/null
+++ b/prism/enc/pm_cp949.c
@@ -0,0 +1,57 @@
+#include "prism/enc/pm_encoding.h"
+
+static size_t
+pm_encoding_cp949_char_width(const uint8_t *b, ptrdiff_t n) {
+    // These are the single byte characters
+    if (*b < 0x81) {
+        return 1;
+    }
+
+    // These are the double byte characters
+    if (
+        (n > 1) &&
+        (b[0] >= 0x81 && b[0] <= 0xfe) &&
+        (b[1] >= 0x41 && b[1] <= 0xfe)
+    ) {
+        return 2;
+    }
+
+    return 0;
+}
+
+static size_t
+pm_encoding_cp949_alpha_char(const uint8_t *b, ptrdiff_t n) {
+    if (pm_encoding_cp949_char_width(b, n) == 1) {
+        return pm_encoding_ascii_alpha_char(b, n);
+    } else {
+        return 0;
+    }
+}
+
+static size_t
+pm_encoding_cp949_alnum_char(const uint8_t *b, ptrdiff_t n) {
+    if (pm_encoding_cp949_char_width(b, n) == 1) {
+        return pm_encoding_ascii_alnum_char(b, n);
+    } else {
+        return 0;
+    }
+}
+
+static bool
+pm_encoding_cp949_isupper_char(const uint8_t *b, ptrdiff_t n) {
+    if (pm_encoding_cp949_char_width(b, n) == 1) {
+        return pm_encoding_ascii_isupper_char(b, n);
+    } else {
+        return 0;
+    }
+}
+
+/** cp949 encoding */
+pm_encoding_t pm_encoding_cp949 = {
+    .name = "cp949",
+    .char_width = pm_encoding_cp949_char_width,
+    .alnum_char = pm_encoding_cp949_alnum_char,
+    .alpha_char = pm_encoding_cp949_alpha_char,
+    .isupper_char = pm_encoding_cp949_isupper_char,
+    .multibyte = true
+};
diff --git a/prism/enc/pm_encoding.h b/prism/enc/pm_encoding.h
index 51227b9c96..698abc8be9 100644
--- a/prism/enc/pm_encoding.h
+++ b/prism/enc/pm_encoding.h
@@ -164,6 +164,7 @@ extern pm_encoding_t pm_encoding_cp51932;
 extern pm_encoding_t pm_encoding_cp850;
 extern pm_encoding_t pm_encoding_cp852;
 extern pm_encoding_t pm_encoding_cp855;
+extern pm_encoding_t pm_encoding_cp949;
 extern pm_encoding_t pm_encoding_euc_jp;
 extern pm_encoding_t pm_encoding_gb1988;
 extern pm_encoding_t pm_encoding_gbk;
diff --git a/prism/prism.c b/prism/prism.c
index e7848f0e04..99ece901f5 100644
--- a/prism/prism.c
+++ b/prism/prism.c
@@ -6165,6 +6165,7 @@ parser_lex_magic_comment_encoding_value(pm_parser_t *parser, const uint8_t *star
                 ENCODING1("CP863", pm_encoding_ibm863);
                 ENCODING2("CP932", "csWindows31J", pm_encoding_windows_31j);
                 ENCODING1("CP936", pm_encoding_gbk);
+                ENCODING1("CP949", pm_encoding_cp949);
                 ENCODING1("CP1250", pm_encoding_windows_1250);
                 ENCODING1("CP1251", pm_encoding_windows_1251);
                 ENCODING1("CP1252", pm_encoding_windows_1252);
diff --git a/test/prism/encoding_test.rb b/test/prism/encoding_test.rb
index 5a0a671b5b..463cb95121 100644
--- a/test/prism/encoding_test.rb
+++ b/test/prism/encoding_test.rb
@@ -68,6 +68,7 @@ module Prism
       Encoding::Big5 =>         0x00...0x10000,
       Encoding::Big5_HKSCS =>   0x00...0x10000,
       Encoding::Big5_UAO =>     0x00...0x10000,
+      Encoding::CP949 =>        0x00...0x10000,
       Encoding::CP51932 =>      0x00...0x10000,
       Encoding::GBK =>          0x00...0x10000,
       Encoding::Shift_JIS =>    0x00...0x10000,