From cd4316a51f73817a04071bbf4e6e56e235e0430d Mon Sep 17 00:00:00 2001
From: Ryan Garver <ragarver@gmail.com>
Date: Fri, 17 Nov 2023 20:35:28 -0800
Subject: [PATCH] [ruby/prism] Big5 HKSCS encoding

https://github.com/ruby/prism/commit/3ca9823eb4
---
 lib/prism/prism.gemspec     |  1 +
 prism/enc/pm_big5_hkscs.c   | 54 +++++++++++++++++++++++++++++++++++++
 prism/enc/pm_encoding.h     |  1 +
 prism/prism.c               |  1 +
 test/prism/encoding_test.rb |  1 +
 5 files changed, 58 insertions(+)
 create mode 100644 prism/enc/pm_big5_hkscs.c

diff --git a/lib/prism/prism.gemspec b/lib/prism/prism.gemspec
index 4b0f87d442..5f3c666e45 100644
--- a/lib/prism/prism.gemspec
+++ b/lib/prism/prism.gemspec
@@ -85,6 +85,7 @@ Gem::Specification.new do |spec|
     "lib/prism/visitor.rb",
     "src/diagnostic.c",
     "src/enc/pm_big5.c",
+    "src/enc/pm_big5_hkscs.c",
     "src/enc/pm_cp51932.c",
     "src/enc/pm_euc_jp.c",
     "src/enc/pm_gbk.c",
diff --git a/prism/enc/pm_big5_hkscs.c b/prism/enc/pm_big5_hkscs.c
new file mode 100644
index 0000000000..e6a13737ed
--- /dev/null
+++ b/prism/enc/pm_big5_hkscs.c
@@ -0,0 +1,54 @@
+#include "prism/enc/pm_encoding.h"
+
+static size_t
+pm_encoding_big5_hkscs_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] >= 0x87 && b[0] <= 0xFE) &&
+                   ((b[1] >= 0x40 && b[1] <= 0x7E) || (b[1] >= 0xA1 && b[1] <= 0xFE))) {
+        return 2;
+    }
+
+    return 0;
+}
+
+static size_t
+pm_encoding_big5_hkscs_alpha_char(const uint8_t *b, ptrdiff_t n) {
+    if (pm_encoding_big5_hkscs_char_width(b, n) == 1) {
+        return pm_encoding_ascii_alpha_char(b, n);
+    } else {
+        return 0;
+    }
+}
+
+static size_t
+pm_encoding_big5_hkscs_alnum_char(const uint8_t *b, ptrdiff_t n) {
+    if (pm_encoding_big5_hkscs_char_width(b, n) == 1) {
+        return pm_encoding_ascii_alnum_char(b, n);
+    } else {
+        return 0;
+    }
+}
+
+static bool
+pm_encoding_big5_hkscs_isupper_char(const uint8_t *b, ptrdiff_t n) {
+    if (pm_encoding_big5_hkscs_char_width(b, n) == 1) {
+        return pm_encoding_ascii_isupper_char(b, n);
+    } else {
+        return false;
+    }
+}
+
+/** Big5 encoding */
+pm_encoding_t pm_encoding_big5_hkscs = {
+    .name = "big5-hkscs",
+    .char_width = pm_encoding_big5_hkscs_char_width,
+    .alnum_char = pm_encoding_big5_hkscs_alnum_char,
+    .alpha_char = pm_encoding_big5_hkscs_alpha_char,
+    .isupper_char = pm_encoding_big5_hkscs_isupper_char,
+    .multibyte = true
+};
diff --git a/prism/enc/pm_encoding.h b/prism/enc/pm_encoding.h
index 957fa794f6..93599fc87f 100644
--- a/prism/enc/pm_encoding.h
+++ b/prism/enc/pm_encoding.h
@@ -158,6 +158,7 @@ extern const uint8_t pm_encoding_unicode_table[256];
 extern pm_encoding_t pm_encoding_ascii;
 extern pm_encoding_t pm_encoding_ascii_8bit;
 extern pm_encoding_t pm_encoding_big5;
+extern pm_encoding_t pm_encoding_big5_hkscs;
 extern pm_encoding_t pm_encoding_cp51932;
 extern pm_encoding_t pm_encoding_cp850;
 extern pm_encoding_t pm_encoding_cp852;
diff --git a/prism/prism.c b/prism/prism.c
index a219ffa52a..45be6cfb23 100644
--- a/prism/prism.c
+++ b/prism/prism.c
@@ -6062,6 +6062,7 @@ parser_lex_magic_comment_encoding_value(pm_parser_t *parser, const uint8_t *star
             case 'B': case 'b':
                 ENCODING1("BINARY", pm_encoding_ascii_8bit);
                 ENCODING1("Big5", pm_encoding_big5);
+                ENCODING1("Big5-HKSCS", pm_encoding_big5_hkscs);
                 break;
             case 'C': case 'c':
                 ENCODING1("CP437", pm_encoding_ibm437);
diff --git a/test/prism/encoding_test.rb b/test/prism/encoding_test.rb
index c2b90aca68..9d8f6bd809 100644
--- a/test/prism/encoding_test.rb
+++ b/test/prism/encoding_test.rb
@@ -57,6 +57,7 @@ module Prism
       Encoding::Windows_1258 => 0x00...0x100,
       Encoding::Windows_874 =>  0x00...0x100,
       Encoding::Big5 =>         0x00...0x10000,
+      Encoding::Big5_HKSCS =>   0x00...0x10000,
       Encoding::CP51932 =>      0x00...0x10000,
       Encoding::GBK =>          0x00...0x10000,
       Encoding::Shift_JIS =>    0x00...0x10000,