1 files changed, 12 insertions, 10 deletions
diff --git a/native/src/binary_format.h b/native/src/binary_format.h
index 6f65088db..1d74998f6 100644
--- a/native/src/binary_format.h
+++ b/native/src/binary_format.h
@@ -22,12 +22,12 @@
 namespace latinime {
 
 class BinaryFormat {
-private:
+ private:
     const static int32_t MINIMAL_ONE_BYTE_CHARACTER_VALUE = 0x20;
     const static int32_t CHARACTER_ARRAY_TERMINATOR = 0x1F;
     const static int MULTIPLE_BYTE_CHARACTER_ADDITIONAL_SIZE = 2;
 
-public:
+ public:
     const static int UNKNOWN_FORMAT = -1;
     const static int FORMAT_VERSION_1 = 1;
     const static uint16_t FORMAT_VERSION_1_MAGIC_NUMBER = 0x78B1;
@@ -61,7 +61,9 @@ inline int BinaryFormat::detectFormat(const uint8_t* const dict) {
 }
 
 inline int BinaryFormat::getGroupCountAndForwardPointer(const uint8_t* const dict, int* pos) {
-    return dict[(*pos)++];
+    const int msb = dict[(*pos)++];
+    if (msb < 0x80) return msb;
+    return ((msb & 0x7F) << 8) | dict[(*pos)++];
 }
 
 inline uint8_t BinaryFormat::getFlagsAndForwardPointer(const uint8_t* const dict, int* pos) {
@@ -145,15 +147,15 @@ inline int BinaryFormat::skipFrequency(const uint8_t flags, const int pos) {
 
 inline int BinaryFormat::skipAllAttributes(const uint8_t* const dict, const uint8_t flags,
         const int pos) {
-    // This function skips all attributes. The format makes provision for future extension
-    // with other attributes (notably shortcuts) but for the time being, bigrams are the
-    // only attributes that may be found in a character group, so we only look at bigrams
-    // in this version.
+    // This function skips all attributes: shortcuts and bigrams.
+    int newPos = pos;
+    if (UnigramDictionary::FLAG_HAS_SHORTCUT_TARGETS & flags) {
+        newPos = skipAttributes(dict, newPos);
+    }
     if (UnigramDictionary::FLAG_HAS_BIGRAMS & flags) {
-        return skipAttributes(dict, pos);
-    } else {
-        return pos;
+        newPos = skipAttributes(dict, newPos);
     }
+    return newPos;
 }
 
 inline int BinaryFormat::skipChildrenPosAndAttributes(const uint8_t* const dict,