prism 0.17.1 → 0.18.0

data/src/enc/pm_cp51932.c

@@ -0,0 +1,57 @@
+#include "prism/enc/pm_encoding.h"
+
+static size_t
+pm_encoding_cp51932_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] >= 0xa1 && b[0] <= 0xfe) || (b[0] == 0x8e)) &&
+        (b[1] >= 0xa1 && b[1] <= 0xfe)
+    ) {
+        return 2;
+    }
+
+    return 0;
+}
+
+static size_t
+pm_encoding_cp51932_alpha_char(const uint8_t *b, ptrdiff_t n) {
+    if (pm_encoding_cp51932_char_width(b, n) == 1) {
+        return pm_encoding_ascii_alpha_char(b, n);
+    } else {
+        return 0;
+    }
+}
+
+static size_t
+pm_encoding_cp51932_alnum_char(const uint8_t *b, ptrdiff_t n) {
+    if (pm_encoding_cp51932_char_width(b, n) == 1) {
+        return pm_encoding_ascii_alnum_char(b, n);
+    } else {
+        return 0;
+    }
+}
+
+static bool
+pm_encoding_cp51932_isupper_char(const uint8_t *b, ptrdiff_t n) {
+    if (pm_encoding_cp51932_char_width(b, n) == 1) {
+        return pm_encoding_ascii_isupper_char(b, n);
+    } else {
+        return 0;
+    }
+}
+
+/** cp51932 encoding */
+pm_encoding_t pm_encoding_cp51932 = {
+    .name = "cp51932",
+    .char_width = pm_encoding_cp51932_char_width,
+    .alnum_char = pm_encoding_cp51932_alnum_char,
+    .alpha_char = pm_encoding_cp51932_alpha_char,
+    .isupper_char = pm_encoding_cp51932_isupper_char,
+    .multibyte = true
+};

data/src/enc/pm_euc_jp.c

@@ -18,6 +18,16 @@ pm_encoding_euc_jp_char_width(const uint8_t *b, ptrdiff_t n) {
         return 2;
     }
 
+    // These are the triple byte characters.
+    if (
+        (n > 2) &&
+        (b[0] == 0x8F) &&
+        (b[1] >= 0xA1 && b[2] <= 0xFE) &&
+        (b[2] >= 0xA1 && b[2] <= 0xFE)
+    ) {
+        return 3;
+    }
+
     return 0;
 }
 

data/src/enc/pm_gbk.c

@@ -3,7 +3,7 @@
 static size_t
 pm_encoding_gbk_char_width(const uint8_t *b, ptrdiff_t n) {
     // These are the single byte characters.
-    if (*b < 0x80) {
+    if (*b <= 0x80) {
         return 1;
     }
 
@@ -15,7 +15,10 @@ pm_encoding_gbk_char_width(const uint8_t *b, ptrdiff_t n) {
             ((b[0] >= 0xB0 && b[0] <= 0xF7) && (b[1] >= 0xA1 && b[1] <= 0xFE)) || // GBK/2
             ((b[0] >= 0x81 && b[0] <= 0xA0) && (b[1] >= 0x40 && b[1] <= 0xFE) && (b[1] != 0x7F)) || // GBK/3
             ((b[0] >= 0xAA && b[0] <= 0xFE) && (b[1] >= 0x40 && b[1] <= 0xA0) && (b[1] != 0x7F)) || // GBK/4
-            ((b[0] >= 0xA8 && b[0] <= 0xA9) && (b[1] >= 0x40 && b[1] <= 0xA0) && (b[1] != 0x7F)) // GBK/5
+            ((b[0] >= 0xA8 && b[0] <= 0xA9) && (b[1] >= 0x40 && b[1] <= 0xA0) && (b[1] != 0x7F)) || // GBK/5
+            ((b[0] >= 0xAA && b[0] <= 0xAF) && (b[1] >= 0xA1 && b[1] <= 0xFE)) || // user-defined 1
+            ((b[0] >= 0xF8 && b[0] <= 0xFE) && (b[1] >= 0xA1 && b[1] <= 0xFE)) || // user-defined 2
+            ((b[0] >= 0xA1 && b[0] <= 0xA7) && (b[1] >= 0x40 && b[1] <= 0xA0) && (b[1] != 0x7F)) // user-defined 3
         )
     ) {
         return 2;