1 files changed, 70 insertions, 73 deletions

diff --git a/native/jni/src/defines.h b/native/jni/src/defines.h
index e06ee42b0..7b127a2af 100644
--- a/native/jni/src/defines.h
+++ b/native/jni/src/defines.h
@@ -17,7 +17,14 @@
 #ifndef LATINIME_DEFINES_H
 #define LATINIME_DEFINES_H
 
-#include <stdint.h>
+#ifdef __GNUC__
+#define AK_FORCE_INLINE __attribute__((always_inline)) __inline__
+#else // __GNUC__
+#define AK_FORCE_INLINE inline
+#endif // __GNUC__
+
+// This must be greater than or equal to MAX_WORD_LENGTH defined in BinaryDictionary.java
+#define MAX_WORD_LENGTH_INTERNAL 48
 
 #if defined(FLAG_DO_PROFILE) || defined(FLAG_DBG)
 #include <android/log.h>
@@ -30,31 +37,41 @@
 #define DUMP_RESULT(words, frequencies, maxWordCount, maxWordLength) do { \
         dumpResult(words, frequencies, maxWordCount, maxWordLength); } while (0)
 #define DUMP_WORD(word, length) do { dumpWord(word, length); } while (0)
-#define DUMP_WORD_INT(word, length) do { dumpWordInt(word, length); } while (0)
-// TODO: INTS_TO_CHARS
-#define SHORTS_TO_CHARS(input, length, output) do { \
-        shortArrayToCharArray(input, length, output); } while (0)
-
-static inline void dumpWordInfo(const unsigned short *word, const int length,
-        const int rank, const int frequency) {
-    static char charBuf[50];
-    int i = 0;
-    for (; i < length; ++i) {
-        const unsigned short c = word[i];
-        if (c == 0) {
-            break;
+#define INTS_TO_CHARS(input, length, output) do { \
+        intArrayToCharArray(input, length, output); } while (0)
+
+// TODO: Support full UTF-8 conversion
+AK_FORCE_INLINE static int intArrayToCharArray(const int *source, const int sourceSize,
+        char *dest) {
+    int si = 0;
+    int di = 0;
+    while (si < sourceSize && di < MAX_WORD_LENGTH_INTERNAL - 1 && 0 != source[si]) {
+        const int codePoint = source[si++];
+        if (codePoint < 0x7F) {
+            dest[di++] = codePoint;
+        } else if (codePoint < 0x7FF) {
+            dest[di++] = 0xC0 + (codePoint >> 6);
+            dest[di++] = 0x80 + (codePoint & 0x3F);
+        } else if (codePoint < 0xFFFF) {
+            dest[di++] = 0xE0 + (codePoint >> 12);
+            dest[di++] = 0x80 + ((codePoint & 0xFC0) >> 6);
+            dest[di++] = 0x80 + (codePoint & 0x3F);
         }
-        // static_cast only for debugging
-        charBuf[i] = static_cast<char>(c);
     }
-    charBuf[i] = 0;
-    if (i > 1) {
+    dest[di] = 0;
+    return di;
+}
+
+static inline void dumpWordInfo(const int *word, const int length, const int rank,
+        const int frequency) {
+    static char charBuf[50];
+    const int N = intArrayToCharArray(word, length, charBuf);
+    if (N > 1) {
         AKLOGI("%2d [ %s ] (%d)", rank, charBuf, frequency);
     }
 }
 
-static inline void dumpResult(
-        const unsigned short *outWords, const int *frequencies, const int maxWordCounts,
+static inline void dumpResult(const int *outWords, const int *frequencies, const int maxWordCounts,
         const int maxWordLength) {
     AKLOGI("--- DUMP RESULT ---------");
     for (int i = 0; i < maxWordCounts; ++i) {
@@ -63,48 +80,14 @@ static inline void dumpResult(
     AKLOGI("-------------------------");
 }
 
-static inline void dumpWord(const unsigned short *word, const int length) {
+static AK_FORCE_INLINE void dumpWord(const int *word, const int length) {
     static char charBuf[50];
-    int i = 0;
-    for (; i < length; ++i) {
-        const unsigned short c = word[i];
-        if (c == 0) {
-            break;
-        }
-        // static_cast only for debugging
-        charBuf[i] = static_cast<char>(c);
-    }
-    charBuf[i] = 0;
-    if (i > 1) {
+    const int N = intArrayToCharArray(word, length, charBuf);
+    if (N > 1) {
         AKLOGI("[ %s ]", charBuf);
     }
 }
 
-static inline void dumpWordInt(const int *word, const int length) {
-    static char charBuf[50];
-
-    for (int i = 0; i < length; ++i) {
-        charBuf[i] = word[i];
-    }
-    charBuf[length] = 0;
-    AKLOGI("i[ %s ]", charBuf);
-}
-
-// TODO: Change this to intArrayToCharArray
-static inline void shortArrayToCharArray(
-        const unsigned short *input, const int length, char *output) {
-    int i = 0;
-    for (;i < length; ++i) {
-        const unsigned short c = input[i];
-        if (c == 0) {
-            break;
-        }
-        // static_cast only for debugging
-        output[i] = static_cast<char>(c);
-    }
-    output[i] = 0;
-}
-
 #ifndef __ANDROID__
 #include <cassert>
 #include <execinfo.h>
@@ -137,11 +120,9 @@ static inline void showStackTrace() {
 #define AKLOGI(fmt, ...)
 #define DUMP_RESULT(words, frequencies, maxWordCount, maxWordLength)
 #define DUMP_WORD(word, length)
-#define DUMP_WORD_INT(word, length)
 #define ASSERT(success)
 #define SHOW_STACK_TRACE
-// TODO: INTS_TO_CHARS
-#define SHORTS_TO_CHARS(input, length, output)
+#define INTS_TO_CHARS(input, length, output)
 #endif
 
 #ifdef FLAG_DO_PROFILE
@@ -178,15 +159,15 @@ static inline void prof_out(void) {
     }
     AKLOGI("Total time is %6.3f ms.",
             profile_buf[PROF_BUF_SIZE - 1] * 1000.0f / static_cast<float>(CLOCKS_PER_SEC));
-    float all = 0;
+    float all = 0.0f;
     for (int i = 0; i < PROF_BUF_SIZE - 1; ++i) {
         all += profile_buf[i];
     }
-    if (all == 0) all = 1;
+    if (all < 1.0f) all = 1.0f;
     for (int i = 0; i < PROF_BUF_SIZE - 1; ++i) {
-        if (profile_buf[i]) {
+        if (profile_buf[i] > 0.0f) {
             AKLOGI("(%d): Used %4.2f%%, %8.4f ms. Called %d times.",
-                    i, (profile_buf[i] * 100 / all),
+                    i, (profile_buf[i] * 100.0f / all),
                     profile_buf[i] * 1000.0f / static_cast<float>(CLOCKS_PER_SEC),
                     profile_counter[i]);
         }
@@ -219,6 +200,9 @@ static inline void prof_out(void) {
 #define DEBUG_CORRECTION false
 #define DEBUG_CORRECTION_FREQ false
 #define DEBUG_WORDS_PRIORITY_QUEUE false
+#define DEBUG_SAMPLING_POINTS false
+#define DEBUG_POINTS_PROBABILITY false
+#define DEBUG_DOUBLE_LETTER false
 
 #ifdef FLAG_FULL_DBG
 #define DEBUG_GEO_FULL true
@@ -239,14 +223,14 @@ static inline void prof_out(void) {
 #define DEBUG_CORRECTION false
 #define DEBUG_CORRECTION_FREQ false
 #define DEBUG_WORDS_PRIORITY_QUEUE false
+#define DEBUG_SAMPLING_POINTS false
+#define DEBUG_POINTS_PROBABILITY false
+#define DEBUG_DOUBLE_LETTER false
 
 #define DEBUG_GEO_FULL false
 
 #endif // FLAG_DBG
 
-#ifndef U_SHORT_MAX
-#define U_SHORT_MAX 65535    // ((1 << 16) - 1)
-#endif
 #ifndef S_INT_MAX
 #define S_INT_MAX 2147483647 // ((1 << 31) - 1)
 #endif
@@ -256,6 +240,12 @@ static inline void prof_out(void) {
 // GCC warns about this.
 #define S_INT_MIN (-2147483647 - 1) // -(1 << 31)
 #endif
+// Number of base-10 digits in the largest integer + 1 to leave room for a zero terminator.
+// As such, this is the maximum number of characters will be needed to represent an int as a
+// string, including the terminator; this is used as the size of a string buffer large enough to
+// hold any value that is intended to fit in an integer, e.g. in the code that reads the header
+// of the binary dictionary where a {key,value} string pair scheme is used.
+#define LARGEST_INT_DIGIT_COUNT 11
 
 // Define this to use mmap() for dictionary loading.  Undefine to use malloc() instead of mmap().
 // We measured and compared performance of both, and found mmap() is fairly good in terms of
@@ -288,6 +278,8 @@ static inline void prof_out(void) {
 #define NOT_A_PROBABILITY (-1)
 
 #define KEYCODE_SPACE ' '
+#define KEYCODE_SINGLE_QUOTE '\''
+#define KEYCODE_HYPHEN_MINUS '-'
 
 #define CALIBRATE_SCORE_BY_TOUCH_COORDINATES true
 
@@ -322,10 +314,6 @@ static inline void prof_out(void) {
 #define MAX_FREQ 255
 #define MAX_BIGRAM_FREQ 15
 
-// This must be greater than or equal to MAX_WORD_LENGTH defined in BinaryDictionary.java
-// This is only used for the size of array. Not to be used in c functions.
-#define MAX_WORD_LENGTH_INTERNAL 48
-
 // This must be the same as ProximityInfo#MAX_PROXIMITY_CHARS_SIZE, currently it's 16.
 #define MAX_PROXIMITY_CHARS_SIZE_INTERNAL 16
 
@@ -350,8 +338,8 @@ static inline void prof_out(void) {
 #define MULTIPLE_WORDS_DEMOTION_RATE 80
 #define MIN_INPUT_LENGTH_FOR_THREE_OR_MORE_WORDS_CORRECTION 6
 
-#define TWO_WORDS_CORRECTION_WITH_OTHER_ERROR_THRESHOLD 0.35
-#define START_TWO_WORDS_CORRECTION_THRESHOLD 0.185
+#define TWO_WORDS_CORRECTION_WITH_OTHER_ERROR_THRESHOLD 0.35f
+#define START_TWO_WORDS_CORRECTION_THRESHOLD 0.185f
 /* heuristic... This should be changed if we change the unit of the frequency. */
 #define SUPPRESS_SHORT_MULTIPLE_WORDS_THRESHOLD_FREQ (MAX_FREQ * 58 / 100)
 
@@ -375,6 +363,7 @@ static inline void prof_out(void) {
 #define MIN_USER_TYPED_LENGTH_FOR_EXCESSIVE_CHARACTER_SUGGESTION 3
 
 // TODO: Remove
+#define MAX_POINTER_COUNT 1
 #define MAX_POINTER_COUNT_FOR_G 2
 
 // Size, in bytes, of the bloom filter index for bigrams
@@ -398,6 +387,8 @@ static inline void prof_out(void) {
 template<typename T> inline T min(T a, T b) { return a < b ? a : b; }
 template<typename T> inline T max(T a, T b) { return a > b ? a : b; }
 
+#define NELEMS(x) (sizeof(x) / sizeof((x)[0]))
+
 // The ratio of neutral area radius to sweet spot radius.
 #define NEUTRAL_AREA_RADIUS_RATIO 1.3f
 
@@ -424,4 +415,10 @@ typedef enum {
     // Additional proximity char which can differ by language.
     ADDITIONAL_PROXIMITY_CHAR
 } ProximityType;
+
+typedef enum {
+    NOT_A_DOUBLE_LETTER,
+    A_DOUBLE_LETTER,
+    A_STRONG_DOUBLE_LETTER
+} DoubleLetterLevel;
 #endif // LATINIME_DEFINES_H