1 files changed, 57 insertions, 4 deletions

diff --git a/native/src/dictionary.cpp b/native/src/dictionary.cpp
index a21b80a48..05692f7ef 100644
--- a/native/src/dictionary.cpp
+++ b/native/src/dictionary.cpp
@@ -17,17 +17,23 @@
 
 #include <stdio.h>
 
+#define LOG_TAG "LatinIME: dictionary.cpp"
+
 #include "dictionary.h"
 
 namespace latinime {
 
 Dictionary::Dictionary(void *dict, int typedLetterMultiplier, int fullWordMultiplier,
         int maxWordLength, int maxWords, int maxAlternatives)
+    : DICT((unsigned char*) dict),
+    // Checks whether it has the latest dictionary or the old dictionary
+    IS_LATEST_DICT_VERSION((((unsigned char*) dict)[0] & 0xFF) >= DICTIONARY_VERSION_MIN)
 {
-    mUnigramDictionary = new UnigramDictionary(dict, typedLetterMultiplier, fullWordMultiplier,
-            maxWordLength, maxWords, maxAlternatives, this);
-    mBigramDictionary = new BigramDictionary(dict, typedLetterMultiplier, fullWordMultiplier,
-            maxWordLength, maxWords, maxAlternatives, this);
+    LOGI("IN NATIVE SUGGEST Version: %d \n", (DICT[0] & 0xFF));
+    mUnigramDictionary = new UnigramDictionary(DICT, typedLetterMultiplier, fullWordMultiplier,
+            maxWordLength, maxWords, maxAlternatives, IS_LATEST_DICT_VERSION);
+    mBigramDictionary = new BigramDictionary(DICT, maxWordLength, maxAlternatives,
+            IS_LATEST_DICT_VERSION, hasBigram(), this);
 }
 
 Dictionary::~Dictionary()
@@ -35,4 +41,51 @@ Dictionary::~Dictionary()
     delete mUnigramDictionary;
     delete mBigramDictionary;
 }
+
+bool Dictionary::hasBigram() {
+    return ((DICT[1] & 0xFF) == 1);
+}
+
+// TODO: use uint16_t instead of unsigned short
+bool Dictionary::isValidWord(unsigned short *word, int length)
+{
+    if (IS_LATEST_DICT_VERSION) {
+        return (isValidWordRec(DICTIONARY_HEADER_SIZE, word, 0, length) != NOT_VALID_WORD);
+    } else {
+        return (isValidWordRec(0, word, 0, length) != NOT_VALID_WORD);
+    }
+}
+
+int Dictionary::isValidWordRec(int pos, unsigned short *word, int offset, int length) {
+    // returns address of bigram data of that word
+    // return -99 if not found
+
+    int count = Dictionary::getCount(DICT, &pos);
+    unsigned short currentChar = (unsigned short) word[offset];
+    for (int j = 0; j < count; j++) {
+        unsigned short c = Dictionary::getChar(DICT, &pos);
+        int terminal = Dictionary::getTerminal(DICT, &pos);
+        int childPos = Dictionary::getAddress(DICT, &pos);
+        if (c == currentChar) {
+            if (offset == length - 1) {
+                if (terminal) {
+                    return (pos+1);
+                }
+            } else {
+                if (childPos != 0) {
+                    int t = isValidWordRec(childPos, word, offset + 1, length);
+                    if (t > 0) {
+                        return t;
+                    }
+                }
+            }
+        }
+        if (terminal) {
+            Dictionary::getFreq(DICT, IS_LATEST_DICT_VERSION, &pos);
+        }
+        // There could be two instances of each alphabet - upper and lower case. So continue
+        // looking ...
+    }
+    return NOT_VALID_WORD;
+}
 } // namespace latinime