[PRISM] Consolidate SJIS encodings

5 files changed, 57 insertions, 115 deletions

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);