diff --git a/native/jni/src/defines.h b/native/jni/src/defines.h index 4e3a7b290..ab326169d 100644 --- a/native/jni/src/defines.h +++ b/native/jni/src/defines.h @@ -287,25 +287,6 @@ static inline void prof_out(void) { #define SUGGEST_MULTIPLE_WORDS true #define SUGGEST_INTERFACE_OUTPUT_SCALE 1000000.0f -// The following "rate"s are used as a multiplier before dividing by 100, so they are in percent. -#define WORDS_WITH_MISSING_CHARACTER_DEMOTION_RATE 80 -#define WORDS_WITH_MISSING_CHARACTER_DEMOTION_START_POS_10X 12 -#define WORDS_WITH_MISSING_SPACE_CHARACTER_DEMOTION_RATE 58 -#define WORDS_WITH_MISTYPED_SPACE_DEMOTION_RATE 50 -#define WORDS_WITH_EXCESSIVE_CHARACTER_DEMOTION_RATE 75 -#define WORDS_WITH_EXCESSIVE_CHARACTER_OUT_OF_PROXIMITY_DEMOTION_RATE 75 -#define WORDS_WITH_TRANSPOSED_CHARACTERS_DEMOTION_RATE 70 -#define FULL_MATCHED_WORDS_PROMOTION_RATE 120 -#define WORDS_WITH_PROXIMITY_CHARACTER_DEMOTION_RATE 90 -#define WORDS_WITH_ADDITIONAL_PROXIMITY_CHARACTER_DEMOTION_RATE 70 -#define WORDS_WITH_MATCH_SKIP_PROMOTION_RATE 105 -#define WORDS_WITH_JUST_ONE_CORRECTION_PROMOTION_RATE 148 -#define WORDS_WITH_JUST_ONE_CORRECTION_PROMOTION_MULTIPLIER 3 -#define CORRECTION_COUNT_RATE_DEMOTION_RATE_BASE 45 -#define INPUT_EXCEEDS_OUTPUT_DEMOTION_RATE 70 -#define FIRST_CHAR_DIFFERENT_DEMOTION_RATE 96 -#define TWO_WORDS_CAPITALIZED_DEMOTION_RATE 50 -#define TWO_WORDS_CORRECTION_DEMOTION_BASE 80 #define ZERO_DISTANCE_PROMOTION_RATE 110.0f #define NEUTRAL_SCORE_SQUARED_RADIUS 8.0f #define HALF_SCORE_SQUARED_RADIUS 32.0f @@ -330,6 +311,13 @@ static inline void prof_out(void) { #define MAX_POINTER_COUNT 1 #define MAX_POINTER_COUNT_G 2 +// Queue IDs and size for DicNodesCache +#define DIC_NODES_CACHE_INITIAL_QUEUE_ID_ACTIVE 0 +#define DIC_NODES_CACHE_INITIAL_QUEUE_ID_NEXT_ACTIVE 1 +#define DIC_NODES_CACHE_INITIAL_QUEUE_ID_TERMINAL 2 +#define DIC_NODES_CACHE_INITIAL_QUEUE_ID_CACHE_FOR_CONTINUOUS_SUGGESTION 3 +#define DIC_NODES_CACHE_PRIORITY_QUEUES_SIZE 4 + // Size, in bytes, of the bloom filter index for bigrams // 128 gives us 1024 buckets. The probability of false positive is (1 - e ** (-kn/m))**k, // where k is the number of hash functions, n the number of bigrams, and m the number of diff --git a/native/jni/src/obsolete/correction.cpp b/native/jni/src/obsolete/correction.cpp index e583dfa81..6b80ed8ea 100644 --- a/native/jni/src/obsolete/correction.cpp +++ b/native/jni/src/obsolete/correction.cpp @@ -30,6 +30,27 @@ namespace latinime { class ProximityInfo; +// private static const member variables +// The following "rate"s are used as a multiplier before dividing by 100, so they are in percent. +const int Correction::WORDS_WITH_MISSING_CHARACTER_DEMOTION_RATE = 80; +const int Correction::WORDS_WITH_MISSING_CHARACTER_DEMOTION_START_POS_10X = 12; +const int Correction::WORDS_WITH_MISSING_SPACE_CHARACTER_DEMOTION_RATE = 58; +const int Correction::WORDS_WITH_MISTYPED_SPACE_DEMOTION_RATE = 50; +const int Correction::WORDS_WITH_EXCESSIVE_CHARACTER_DEMOTION_RATE = 75; +const int Correction::WORDS_WITH_EXCESSIVE_CHARACTER_OUT_OF_PROXIMITY_DEMOTION_RATE = 75; +const int Correction::WORDS_WITH_TRANSPOSED_CHARACTERS_DEMOTION_RATE = 70; +const int Correction::FULL_MATCHED_WORDS_PROMOTION_RATE = 120; +const int Correction::WORDS_WITH_PROXIMITY_CHARACTER_DEMOTION_RATE = 90; +const int Correction::WORDS_WITH_ADDITIONAL_PROXIMITY_CHARACTER_DEMOTION_RATE = 70; +const int Correction::WORDS_WITH_MATCH_SKIP_PROMOTION_RATE = 105; +const int Correction::WORDS_WITH_JUST_ONE_CORRECTION_PROMOTION_RATE = 148; +const int Correction::WORDS_WITH_JUST_ONE_CORRECTION_PROMOTION_MULTIPLIER = 3; +const int Correction::CORRECTION_COUNT_RATE_DEMOTION_RATE_BASE = 45; +const int Correction::INPUT_EXCEEDS_OUTPUT_DEMOTION_RATE = 70; +const int Correction::FIRST_CHAR_DIFFERENT_DEMOTION_RATE = 96; +const int Correction::TWO_WORDS_CAPITALIZED_DEMOTION_RATE = 50; +const int Correction::TWO_WORDS_CORRECTION_DEMOTION_BASE = 80; + ///////////////////////////// // edit distance funcitons // ///////////////////////////// diff --git a/native/jni/src/obsolete/correction.h b/native/jni/src/obsolete/correction.h index 710220d66..47dcef2d7 100644 --- a/native/jni/src/obsolete/correction.h +++ b/native/jni/src/obsolete/correction.h @@ -135,6 +135,27 @@ class Correction { private: DISALLOW_COPY_AND_ASSIGN(Correction); + // The following "rate"s are used as a multiplier before dividing by 100, so they are in + // percent. + static const int WORDS_WITH_MISSING_CHARACTER_DEMOTION_RATE; + static const int WORDS_WITH_MISSING_CHARACTER_DEMOTION_START_POS_10X; + static const int WORDS_WITH_MISSING_SPACE_CHARACTER_DEMOTION_RATE; + static const int WORDS_WITH_MISTYPED_SPACE_DEMOTION_RATE; + static const int WORDS_WITH_EXCESSIVE_CHARACTER_DEMOTION_RATE; + static const int WORDS_WITH_EXCESSIVE_CHARACTER_OUT_OF_PROXIMITY_DEMOTION_RATE; + static const int WORDS_WITH_TRANSPOSED_CHARACTERS_DEMOTION_RATE; + static const int FULL_MATCHED_WORDS_PROMOTION_RATE; + static const int WORDS_WITH_PROXIMITY_CHARACTER_DEMOTION_RATE; + static const int WORDS_WITH_ADDITIONAL_PROXIMITY_CHARACTER_DEMOTION_RATE; + static const int WORDS_WITH_MATCH_SKIP_PROMOTION_RATE; + static const int WORDS_WITH_JUST_ONE_CORRECTION_PROMOTION_RATE; + static const int WORDS_WITH_JUST_ONE_CORRECTION_PROMOTION_MULTIPLIER; + static const int CORRECTION_COUNT_RATE_DEMOTION_RATE_BASE; + static const int INPUT_EXCEEDS_OUTPUT_DEMOTION_RATE; + static const int FIRST_CHAR_DIFFERENT_DEMOTION_RATE; + static const int TWO_WORDS_CAPITALIZED_DEMOTION_RATE; + static const int TWO_WORDS_CORRECTION_DEMOTION_BASE; + ///////////////////////// // static inline utils // ///////////////////////// diff --git a/native/jni/src/suggest/core/dicnode/dic_nodes_cache.h b/native/jni/src/suggest/core/dicnode/dic_nodes_cache.h index 7f5bdbcf6..7aab0906e 100644 --- a/native/jni/src/suggest/core/dicnode/dic_nodes_cache.h +++ b/native/jni/src/suggest/core/dicnode/dic_nodes_cache.h @@ -22,12 +22,6 @@ #include "defines.h" #include "suggest/core/dicnode/dic_node_priority_queue.h" -#define INITIAL_QUEUE_ID_ACTIVE 0 -#define INITIAL_QUEUE_ID_NEXT_ACTIVE 1 -#define INITIAL_QUEUE_ID_TERMINAL 2 -#define INITIAL_QUEUE_ID_CACHE_FOR_CONTINUOUS_SUGGESTION 3 -#define PRIORITY_QUEUES_SIZE 4 - namespace latinime { class DicNode; @@ -38,11 +32,12 @@ class DicNode; class DicNodesCache { public: AK_FORCE_INLINE DicNodesCache() - : mActiveDicNodes(&mDicNodePriorityQueues[INITIAL_QUEUE_ID_ACTIVE]), - mNextActiveDicNodes(&mDicNodePriorityQueues[INITIAL_QUEUE_ID_NEXT_ACTIVE]), - mTerminalDicNodes(&mDicNodePriorityQueues[INITIAL_QUEUE_ID_TERMINAL]), - mCachedDicNodesForContinuousSuggestion( - &mDicNodePriorityQueues[INITIAL_QUEUE_ID_CACHE_FOR_CONTINUOUS_SUGGESTION]), + : mActiveDicNodes(&mDicNodePriorityQueues[DIC_NODES_CACHE_INITIAL_QUEUE_ID_ACTIVE]), + mNextActiveDicNodes(&mDicNodePriorityQueues[ + DIC_NODES_CACHE_INITIAL_QUEUE_ID_NEXT_ACTIVE]), + mTerminalDicNodes(&mDicNodePriorityQueues[DIC_NODES_CACHE_INITIAL_QUEUE_ID_TERMINAL]), + mCachedDicNodesForContinuousSuggestion(&mDicNodePriorityQueues[ + DIC_NODES_CACHE_INITIAL_QUEUE_ID_CACHE_FOR_CONTINUOUS_SUGGESTION]), mInputIndex(0), mLastCachedInputIndex(0) { } @@ -147,9 +142,8 @@ class DicNodesCache { mCachedDicNodesForContinuousSuggestion->dump(); } mInputIndex = mLastCachedInputIndex; - mCachedDicNodesForContinuousSuggestion = - moveNodesAndReturnReusableEmptyQueue( - mCachedDicNodesForContinuousSuggestion, &mActiveDicNodes); + mCachedDicNodesForContinuousSuggestion = moveNodesAndReturnReusableEmptyQueue( + mCachedDicNodesForContinuousSuggestion, &mActiveDicNodes); } AK_FORCE_INLINE static DicNodePriorityQueue *moveNodesAndReturnReusableEmptyQueue( @@ -169,7 +163,7 @@ class DicNodesCache { mTerminalDicNodes->clear(); } - DicNodePriorityQueue mDicNodePriorityQueues[PRIORITY_QUEUES_SIZE]; + DicNodePriorityQueue mDicNodePriorityQueues[DIC_NODES_CACHE_PRIORITY_QUEUES_SIZE]; // Active dicNodes currently being expanded. DicNodePriorityQueue *mActiveDicNodes; // Next dicNodes to be expanded.