[ruby/prism] added CP950 encoding

9c2d1cf4ba

lib/prism/prism.gemspec

@@ -88,6 +88,7 @@ Gem::Specification.new do |spec|
     "src/enc/pm_big5.c",
     "src/enc/pm_cp51932.c",
     "src/enc/pm_cp949.c",
+    "src/enc/pm_cp950.c",
     "src/enc/pm_euc_jp.c",
     "src/enc/pm_gbk.c",
     "src/enc/pm_shift_jis.c",

prism/enc/pm_cp950.c

@@ -0,0 +1,57 @@
+#include "prism/enc/pm_encoding.h"
+
+static size_t
+pm_encoding_cp950_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] >= 0x81 && b[0] <= 0xFE) &&
+        ((b[1] >= 0x40 && b[1] <= 0x7E) || (b[1] >= 0xA1 && b[1] <= 0xFE)))
+    ) {
+        return 2;
+    }
+
+    return 0;
+}
+
+static size_t
+pm_encoding_cp950_alpha_char(const uint8_t *b, ptrdiff_t n) {
+    if (pm_encoding_cp950_char_width(b, n) == 1) {
+        return pm_encoding_ascii_alpha_char(b, n);
+    } else {
+        return 0;
+    }
+}
+
+static size_t
+pm_encoding_cp950_alnum_char(const uint8_t *b, ptrdiff_t n) {
+    if (pm_encoding_cp950_char_width(b, n) == 1) {
+        return pm_encoding_ascii_alnum_char(b, n);
+    } else {
+        return 0;
+    }
+}
+
+static bool
+pm_encoding_cp950_isupper_char(const uint8_t *b, ptrdiff_t n) {
+    if (pm_encoding_cp950_char_width(b, n) == 1) {
+        return pm_encoding_ascii_isupper_char(b, n);
+    } else {
+        return 0;
+    }
+}
+
+/** cp950 encoding */
+pm_encoding_t pm_encoding_cp950 = {
+    .name = "cp950",
+    .char_width = pm_encoding_cp950_char_width,
+    .alnum_char = pm_encoding_cp950_alnum_char,
+    .alpha_char = pm_encoding_cp950_alpha_char,
+    .isupper_char = pm_encoding_cp950_isupper_char,
+    .multibyte = true
+};

prism/enc/pm_encoding.h

@@ -165,6 +165,7 @@ 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_cp950;
 extern pm_encoding_t pm_encoding_euc_jp;
 extern pm_encoding_t pm_encoding_gb1988;
 extern pm_encoding_t pm_encoding_gbk;

prism/prism.c

@@ -6233,6 +6233,7 @@ parser_lex_magic_comment_encoding_value(pm_parser_t *parser, const uint8_t *star
                 ENCODING2("CP932", "csWindows31J", pm_encoding_windows_31j);
                 ENCODING1("CP936", pm_encoding_gbk);
                 ENCODING1("CP949", pm_encoding_cp949);
+                ENCODING1("CP950", pm_encoding_cp950);
                 ENCODING1("CP1250", pm_encoding_windows_1250);
                 ENCODING1("CP1251", pm_encoding_windows_1251);
                 ENCODING1("CP1252", pm_encoding_windows_1252);

test/prism/encoding_test.rb

@@ -69,6 +69,7 @@ module Prism
       Encoding::Big5_HKSCS =>   0x00...0x10000,
       Encoding::Big5_UAO =>     0x00...0x10000,
       Encoding::CP949 =>        0x00...0x10000,
+      Encoding::CP950 =>        0x00...0x10000,
       Encoding::CP51932 =>      0x00...0x10000,
       Encoding::GBK =>          0x00...0x10000,
       Encoding::Shift_JIS =>    0x00...0x10000,