diff options
Diffstat (limited to 'native/jni/src/binary_format.h')
| -rw-r--r-- | native/jni/src/binary_format.h | 15 |
1 files changed, 12 insertions, 3 deletions
diff --git a/native/jni/src/binary_format.h b/native/jni/src/binary_format.h index f59302460..d5d67c108 100644 --- a/native/jni/src/binary_format.h +++ b/native/jni/src/binary_format.h @@ -62,10 +62,11 @@ class BinaryFormat { static bool hasChildrenInFlags(const uint8_t flags); static int getAttributeAddressAndForwardPointer(const uint8_t* const dict, const uint8_t flags, int *pos); - static int getTerminalPosition(const uint8_t* const root, const uint16_t* const inWord, + static int getTerminalPosition(const uint8_t* const root, const int32_t* const inWord, const int length); static int getWordAtAddress(const uint8_t* const root, const int address, const int maxDepth, uint16_t* outWord); + static int getProbability(const int bigramListPosition, const int unigramFreq); // Flags for special processing // Those *must* match the flags in makedict (BinaryDictInputOutput#*_PROCESSING_FLAG) or @@ -304,7 +305,7 @@ inline int BinaryFormat::getAttributeAddressAndForwardPointer(const uint8_t* con // This function gets the byte position of the last chargroup of the exact matching word in the // dictionary. If no match is found, it returns NOT_VALID_WORD. inline int BinaryFormat::getTerminalPosition(const uint8_t* const root, - const uint16_t* const inWord, const int length) { + const int32_t* const inWord, const int length) { int pos = 0; int wordPos = 0; @@ -313,7 +314,7 @@ inline int BinaryFormat::getTerminalPosition(const uint8_t* const root, // there was no match (or we would have found it). if (wordPos > length) return NOT_VALID_WORD; int charGroupCount = BinaryFormat::getGroupCountAndForwardPointer(root, &pos); - const uint16_t wChar = inWord[wordPos]; + const int32_t wChar = inWord[wordPos]; while (true) { // If there are no more character groups in this node, it means we could not // find a matching character for this depth, therefore there is no match. @@ -517,6 +518,14 @@ inline int BinaryFormat::getWordAtAddress(const uint8_t* const root, const int a return 0; } +// This should probably return a probability in log space. +inline int BinaryFormat::getProbability(const int bigramListPosition, const int unigramFreq) { + // TODO: use the bigram list position to get the bigram probability. If the bigram + // is not found, use the unigram frequency. + // TODO: if the unigram frequency is used, compute the actual probability + return unigramFreq; +} + } // namespace latinime #endif // LATINIME_BINARY_FORMAT_H |