aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
-rw-r--r--lib/prism/prism.gemspec4
-rw-r--r--prism/enc/pm_encoding.h3
-rw-r--r--prism/enc/pm_mac_japanese.c57
-rw-r--r--prism/enc/pm_shift_jis.c52
-rw-r--r--prism/enc/pm_windows_31j.c57
-rw-r--r--prism/prism.c3
-rw-r--r--test/prism/encoding_test.rb141
7 files changed, 130 insertions, 187 deletions
diff --git a/lib/prism/prism.gemspec b/lib/prism/prism.gemspec
index 381ecfea5b..cd90b025b2 100644
--- a/lib/prism/prism.gemspec
+++ b/lib/prism/prism.gemspec
@@ -86,16 +86,14 @@ Gem::Specification.new do |spec|
"lib/prism/visitor.rb",
"src/diagnostic.c",
"src/enc/pm_big5.c",
- "src/enc/pm_cp51932.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_mac_japanese.c",
"src/enc/pm_shift_jis.c",
"src/enc/pm_tables.c",
"src/enc/pm_unicode.c",
- "src/enc/pm_windows_31j.c",
"src/node.c",
"src/pack.c",
"src/prettyprint.c",
diff --git a/prism/enc/pm_encoding.h b/prism/enc/pm_encoding.h
index 1985f00f26..7dfc8cd982 100644
--- a/prism/enc/pm_encoding.h
+++ b/prism/enc/pm_encoding.h
@@ -213,6 +213,9 @@ extern pm_encoding_t pm_encoding_mac_thai;
extern pm_encoding_t pm_encoding_mac_turkish;
extern pm_encoding_t pm_encoding_mac_ukraine;
extern pm_encoding_t pm_encoding_shift_jis;
+extern pm_encoding_t pm_encoding_sjis_docomo;
+extern pm_encoding_t pm_encoding_sjis_kddi;
+extern pm_encoding_t pm_encoding_sjis_softbank;
extern pm_encoding_t pm_encoding_tis_620;
extern pm_encoding_t pm_encoding_utf_8;
extern pm_encoding_t pm_encoding_utf8_mac;
diff --git a/prism/enc/pm_mac_japanese.c b/prism/enc/pm_mac_japanese.c
deleted file mode 100644
index a5185f0e55..0000000000
--- a/prism/enc/pm_mac_japanese.c
+++ /dev/null
@@ -1,57 +0,0 @@
-#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/enc/pm_shift_jis.c b/prism/enc/pm_shift_jis.c
index f92956e08b..7833c6653b 100644
--- a/prism/enc/pm_shift_jis.c
+++ b/prism/enc/pm_shift_jis.c
@@ -48,7 +48,57 @@ pm_encoding_shift_jis_isupper_char(const uint8_t *b, ptrdiff_t n) {
/** Shift_JIS encoding */
pm_encoding_t pm_encoding_shift_jis = {
- .name = "shift_jis",
+ .name = "Shift_JIS",
+ .char_width = pm_encoding_shift_jis_char_width,
+ .alnum_char = pm_encoding_shift_jis_alnum_char,
+ .alpha_char = pm_encoding_shift_jis_alpha_char,
+ .isupper_char = pm_encoding_shift_jis_isupper_char,
+ .multibyte = true
+};
+
+/** SJIS-DoCoMo encoding */
+pm_encoding_t pm_encoding_sjis_docomo = {
+ .name = "SJIS-DoCoMo",
+ .char_width = pm_encoding_shift_jis_char_width,
+ .alnum_char = pm_encoding_shift_jis_alnum_char,
+ .alpha_char = pm_encoding_shift_jis_alpha_char,
+ .isupper_char = pm_encoding_shift_jis_isupper_char,
+ .multibyte = true
+};
+
+/** SJIS-KDDI encoding */
+pm_encoding_t pm_encoding_sjis_kddi = {
+ .name = "SJIS-KDDI",
+ .char_width = pm_encoding_shift_jis_char_width,
+ .alnum_char = pm_encoding_shift_jis_alnum_char,
+ .alpha_char = pm_encoding_shift_jis_alpha_char,
+ .isupper_char = pm_encoding_shift_jis_isupper_char,
+ .multibyte = true
+};
+
+/** SJIS-SoftBank encoding */
+pm_encoding_t pm_encoding_sjis_softbank = {
+ .name = "SJIS-SoftBank",
+ .char_width = pm_encoding_shift_jis_char_width,
+ .alnum_char = pm_encoding_shift_jis_alnum_char,
+ .alpha_char = pm_encoding_shift_jis_alpha_char,
+ .isupper_char = pm_encoding_shift_jis_isupper_char,
+ .multibyte = true
+};
+
+/** MacJapanese encoding */
+pm_encoding_t pm_encoding_mac_japanese = {
+ .name = "MacJapanese",
+ .char_width = pm_encoding_shift_jis_char_width,
+ .alnum_char = pm_encoding_shift_jis_alnum_char,
+ .alpha_char = pm_encoding_shift_jis_alpha_char,
+ .isupper_char = pm_encoding_shift_jis_isupper_char,
+ .multibyte = true
+};
+
+/** Windows-31J */
+pm_encoding_t pm_encoding_windows_31j = {
+ .name = "Windows-31J",
.char_width = pm_encoding_shift_jis_char_width,
.alnum_char = pm_encoding_shift_jis_alnum_char,
.alpha_char = pm_encoding_shift_jis_alpha_char,
diff --git a/prism/enc/pm_windows_31j.c b/prism/enc/pm_windows_31j.c
deleted file mode 100644
index 848a9efd36..0000000000
--- a/prism/enc/pm_windows_31j.c
+++ /dev/null
@@ -1,57 +0,0 @@
-#include "prism/enc/pm_encoding.h"
-
-static size_t
-pm_encoding_windows_31j_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_windows_31j_alpha_char(const uint8_t *b, ptrdiff_t n) {
- if (pm_encoding_windows_31j_char_width(b, n) == 1) {
- return pm_encoding_ascii_alpha_char(b, n);
- } else {
- return 0;
- }
-}
-
-static size_t
-pm_encoding_windows_31j_alnum_char(const uint8_t *b, ptrdiff_t n) {
- if (pm_encoding_windows_31j_char_width(b, n) == 1) {
- return pm_encoding_ascii_alnum_char(b, n);
- } else {
- return 0;
- }
-}
-
-static bool
-pm_encoding_windows_31j_isupper_char(const uint8_t *b, ptrdiff_t n) {
- if (pm_encoding_windows_31j_char_width(b, n) == 1) {
- return pm_encoding_ascii_isupper_char(b, n);
- } else {
- return false;
- }
-}
-
-/** Windows-31J */
-pm_encoding_t pm_encoding_windows_31j = {
- .name = "windows-31j",
- .char_width = pm_encoding_windows_31j_char_width,
- .alnum_char = pm_encoding_windows_31j_alnum_char,
- .alpha_char = pm_encoding_windows_31j_alpha_char,
- .isupper_char = pm_encoding_windows_31j_isupper_char,
- .multibyte = true
-};
diff --git a/prism/prism.c b/prism/prism.c
index a0c0e728b6..0cabae6232 100644
--- a/prism/prism.c
+++ b/prism/prism.c
@@ -6317,6 +6317,9 @@ parser_lex_magic_comment_encoding_value(pm_parser_t *parser, const uint8_t *star
case 'S': case 's':
ENCODING1("Shift_JIS", pm_encoding_shift_jis);
ENCODING1("SJIS", pm_encoding_windows_31j);
+ ENCODING1("SJIS-DoCoMo", pm_encoding_sjis_docomo);
+ ENCODING1("SJIS-KDDI", pm_encoding_sjis_kddi);
+ ENCODING1("SJIS-SoftBank", pm_encoding_sjis_softbank);
break;
case 'T': case 't':
ENCODING1("TIS-620", pm_encoding_tis_620);
diff --git a/test/prism/encoding_test.rb b/test/prism/encoding_test.rb
index 13c622e40a..8ab6f323e5 100644
--- a/test/prism/encoding_test.rb
+++ b/test/prism/encoding_test.rb
@@ -7,75 +7,78 @@ require_relative "test_helper"
module Prism
class EncodingTest < TestCase
encodings = {
- Encoding::ASCII => 0x00...0x100,
- Encoding::ASCII_8BIT => 0x00...0x100,
- Encoding::CP850 => 0x00...0x100,
- Encoding::CP852 => 0x00...0x100,
- Encoding::CP855 => 0x00...0x100,
- Encoding::GB1988 => 0x00...0x100,
- Encoding::IBM437 => 0x00...0x100,
- Encoding::IBM720 => 0x00...0x100,
- Encoding::IBM737 => 0x00...0x100,
- Encoding::IBM775 => 0x00...0x100,
- Encoding::IBM852 => 0x00...0x100,
- Encoding::IBM855 => 0x00...0x100,
- Encoding::IBM857 => 0x00...0x100,
- Encoding::IBM860 => 0x00...0x100,
- Encoding::IBM861 => 0x00...0x100,
- Encoding::IBM862 => 0x00...0x100,
- Encoding::IBM863 => 0x00...0x100,
- Encoding::IBM864 => 0x00...0x100,
- Encoding::IBM865 => 0x00...0x100,
- Encoding::IBM866 => 0x00...0x100,
- Encoding::IBM869 => 0x00...0x100,
- Encoding::ISO_8859_1 => 0x00...0x100,
- Encoding::ISO_8859_2 => 0x00...0x100,
- Encoding::ISO_8859_3 => 0x00...0x100,
- Encoding::ISO_8859_4 => 0x00...0x100,
- Encoding::ISO_8859_5 => 0x00...0x100,
- Encoding::ISO_8859_6 => 0x00...0x100,
- Encoding::ISO_8859_7 => 0x00...0x100,
- Encoding::ISO_8859_8 => 0x00...0x100,
- Encoding::ISO_8859_9 => 0x00...0x100,
- Encoding::ISO_8859_10 => 0x00...0x100,
- Encoding::ISO_8859_11 => 0x00...0x100,
- Encoding::ISO_8859_13 => 0x00...0x100,
- Encoding::ISO_8859_14 => 0x00...0x100,
- Encoding::ISO_8859_15 => 0x00...0x100,
- Encoding::ISO_8859_16 => 0x00...0x100,
- Encoding::KOI8_R => 0x00...0x100,
- Encoding::KOI8_U => 0x00...0x100,
- Encoding::MACCENTEURO => 0x00...0x100,
- Encoding::MACCROATIAN => 0x00...0x100,
- Encoding::MACCYRILLIC => 0x00...0x100,
- Encoding::MACGREEK => 0x00...0x100,
- Encoding::MACICELAND => 0x00...0x100,
- Encoding::MACROMAN => 0x00...0x100,
- Encoding::MACROMANIA => 0x00...0x100,
- Encoding::MACTHAI => 0x00...0x100,
- Encoding::MACTURKISH => 0x00...0x100,
- Encoding::MACUKRAINE => 0x00...0x100,
- Encoding::TIS_620 => 0x00...0x100,
- Encoding::Windows_1250 => 0x00...0x100,
- Encoding::Windows_1251 => 0x00...0x100,
- Encoding::Windows_1252 => 0x00...0x100,
- Encoding::Windows_1253 => 0x00...0x100,
- Encoding::Windows_1254 => 0x00...0x100,
- Encoding::Windows_1255 => 0x00...0x100,
- Encoding::Windows_1256 => 0x00...0x100,
- Encoding::Windows_1257 => 0x00...0x100,
- Encoding::Windows_1258 => 0x00...0x100,
- Encoding::Windows_874 => 0x00...0x100,
- Encoding::Big5 => 0x00...0x10000,
- 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::MACJAPANESE => 0x00...0x10000,
- Encoding::Shift_JIS => 0x00...0x10000,
- Encoding::Windows_31J => 0x00...0x10000
+ Encoding::ASCII => 0x00...0x100,
+ Encoding::ASCII_8BIT => 0x00...0x100,
+ Encoding::CP850 => 0x00...0x100,
+ Encoding::CP852 => 0x00...0x100,
+ Encoding::CP855 => 0x00...0x100,
+ Encoding::GB1988 => 0x00...0x100,
+ Encoding::IBM437 => 0x00...0x100,
+ Encoding::IBM720 => 0x00...0x100,
+ Encoding::IBM737 => 0x00...0x100,
+ Encoding::IBM775 => 0x00...0x100,
+ Encoding::IBM852 => 0x00...0x100,
+ Encoding::IBM855 => 0x00...0x100,
+ Encoding::IBM857 => 0x00...0x100,
+ Encoding::IBM860 => 0x00...0x100,
+ Encoding::IBM861 => 0x00...0x100,
+ Encoding::IBM862 => 0x00...0x100,
+ Encoding::IBM863 => 0x00...0x100,
+ Encoding::IBM864 => 0x00...0x100,
+ Encoding::IBM865 => 0x00...0x100,
+ Encoding::IBM866 => 0x00...0x100,
+ Encoding::IBM869 => 0x00...0x100,
+ Encoding::ISO_8859_1 => 0x00...0x100,
+ Encoding::ISO_8859_2 => 0x00...0x100,
+ Encoding::ISO_8859_3 => 0x00...0x100,
+ Encoding::ISO_8859_4 => 0x00...0x100,
+ Encoding::ISO_8859_5 => 0x00...0x100,
+ Encoding::ISO_8859_6 => 0x00...0x100,
+ Encoding::ISO_8859_7 => 0x00...0x100,
+ Encoding::ISO_8859_8 => 0x00...0x100,
+ Encoding::ISO_8859_9 => 0x00...0x100,
+ Encoding::ISO_8859_10 => 0x00...0x100,
+ Encoding::ISO_8859_11 => 0x00...0x100,
+ Encoding::ISO_8859_13 => 0x00...0x100,
+ Encoding::ISO_8859_14 => 0x00...0x100,
+ Encoding::ISO_8859_15 => 0x00...0x100,
+ Encoding::ISO_8859_16 => 0x00...0x100,
+ Encoding::KOI8_R => 0x00...0x100,
+ Encoding::KOI8_U => 0x00...0x100,
+ Encoding::MACCENTEURO => 0x00...0x100,
+ Encoding::MACCROATIAN => 0x00...0x100,
+ Encoding::MACCYRILLIC => 0x00...0x100,
+ Encoding::MACGREEK => 0x00...0x100,
+ Encoding::MACICELAND => 0x00...0x100,
+ Encoding::MACROMAN => 0x00...0x100,
+ Encoding::MACROMANIA => 0x00...0x100,
+ Encoding::MACTHAI => 0x00...0x100,
+ Encoding::MACTURKISH => 0x00...0x100,
+ Encoding::MACUKRAINE => 0x00...0x100,
+ Encoding::TIS_620 => 0x00...0x100,
+ Encoding::Windows_1250 => 0x00...0x100,
+ Encoding::Windows_1251 => 0x00...0x100,
+ Encoding::Windows_1252 => 0x00...0x100,
+ Encoding::Windows_1253 => 0x00...0x100,
+ Encoding::Windows_1254 => 0x00...0x100,
+ Encoding::Windows_1255 => 0x00...0x100,
+ Encoding::Windows_1256 => 0x00...0x100,
+ Encoding::Windows_1257 => 0x00...0x100,
+ Encoding::Windows_1258 => 0x00...0x100,
+ Encoding::Windows_874 => 0x00...0x100,
+ Encoding::Big5 => 0x00...0x10000,
+ 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::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
}
# By default we don't test every codepoint in these encodings because they