nyancrimew
/
LatinIME
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Load main dic in native 

Follow up to Id57dce51

bug: 3219819
Change-Id: I00e11ef21d0252ffa88c12dffb9c55b0f2e19a66
main
Ken Wakasa 2011-01-07 15:01:51 +09:00
parent f16028b92e
commit e90b333017
14 changed files with 232 additions and 204 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

69
java/src/com/android/inputmethod/latin/BinaryDictionary.java
View File

@ -17,13 +17,9 @@
package com.android.inputmethod.latin;
import android.content.Context;
import android.content.res.AssetFileDescriptor;
import android.util.Log;
import java.io.IOException;
import java.io.InputStream;
import java.nio.ByteBuffer;
import java.nio.ByteOrder;
import java.nio.channels.Channels;
import java.util.Arrays;
/**
@ -48,15 +44,12 @@ public class BinaryDictionary extends Dictionary {
private int mDicTypeId;
private int mNativeDict;
private int mDictLength;
private long mDictLength;
private final int[] mInputCodes = new int[MAX_WORD_LENGTH * MAX_ALTERNATIVES];
private final char[] mOutputChars = new char[MAX_WORD_LENGTH * MAX_WORDS];
private final char[] mOutputChars_bigrams = new char[MAX_WORD_LENGTH * MAX_BIGRAMS];
private final int[] mFrequencies = new int[MAX_WORDS];
private final int[] mFrequencies_bigrams = new int[MAX_BIGRAMS];
// Keep a reference to the native dict direct buffer in Java to avoid
// unexpected deallocation of the direct buffer.
private ByteBuffer mNativeDictDirectBuffer;
static {
try {
@ -78,30 +71,9 @@ public class BinaryDictionary extends Dictionary {
mDicTypeId = dicTypeId;
}
/**
* Create a dictionary from a byte buffer. This is used for testing.
* @param context application context for reading resources
* @param byteBuffer a ByteBuffer containing the binary dictionary
*/
public BinaryDictionary(Context context, ByteBuffer byteBuffer, int dicTypeId) {
if (byteBuffer != null) {
if (byteBuffer.isDirect()) {
mNativeDictDirectBuffer = byteBuffer;
} else {
mNativeDictDirectBuffer = ByteBuffer.allocateDirect(byteBuffer.capacity());
byteBuffer.rewind();
mNativeDictDirectBuffer.put(byteBuffer);
}
mDictLength = byteBuffer.capacity();
mNativeDict = openNative(mNativeDictDirectBuffer,
TYPED_LETTER_MULTIPLIER, FULL_WORD_FREQ_MULTIPLIER,
MAX_WORD_LENGTH, MAX_WORDS, MAX_ALTERNATIVES);
}
mDicTypeId = dicTypeId;
}
private native int openNative(ByteBuffer bb, int typedLetterMultiplier,
int fullWordMultiplier, int maxWordLength, int maxWords, int maxAlternatives);
private native int openNative(String apkFileName, long dictOffset, long dictSize,
int typedLetterMultiplier, int fullWordMultiplier, int maxWordLength,
int maxWords, int maxAlternatives);
private native void closeNative(int dict);
private native boolean isValidWordNative(int nativeData, char[] word, int wordLength);
private native int getSuggestionsNative(int dict, int[] inputCodes, int codesSize,
@ -112,37 +84,17 @@ public class BinaryDictionary extends Dictionary {
int maxWordLength, int maxBigrams, int maxAlternatives);
private final void loadDictionary(Context context, int resId) {
InputStream is = null;
try {
is = context.getResources().openRawResource(resId);
final int total = is.available();
mNativeDictDirectBuffer =
ByteBuffer.allocateDirect(total).order(ByteOrder.nativeOrder());
final int got = Channels.newChannel(is).read(mNativeDictDirectBuffer);
if (got != total) {
Log.e(TAG, "Read " + got + " bytes, expected " + total);
} else {
mNativeDict = openNative(mNativeDictDirectBuffer,
final AssetFileDescriptor afd = context.getResources().openRawResourceFd(resId);
mNativeDict = openNative(context.getApplicationInfo().sourceDir,
afd.getStartOffset(), afd.getLength(),
TYPED_LETTER_MULTIPLIER, FULL_WORD_FREQ_MULTIPLIER,
MAX_WORD_LENGTH, MAX_WORDS, MAX_ALTERNATIVES);
mDictLength = total;
mDictLength = afd.getLength();
}
} catch (IOException e) {
Log.w(TAG, "No available memory for binary dictionary");
} finally {
try {
if (is != null) is.close();
} catch (IOException e) {
Log.w(TAG, "Failed to close input stream");
}
}
}
@Override
public void getBigrams(final WordComposer codes, final CharSequence previousWord,
final WordCallback callback, int[] nextLettersFrequencies) {
char[] chars = previousWord.toString().toCharArray();
Arrays.fill(mOutputChars_bigrams, (char) 0);
Arrays.fill(mFrequencies_bigrams, 0);
@ -212,7 +164,7 @@ public class BinaryDictionary extends Dictionary {
return isValidWordNative(mNativeDict, chars, chars.length);
}
public int getSize() {
public long getSize() {
return mDictLength; // This value is initialized on the call to openNative()
}
@ -221,6 +173,7 @@ public class BinaryDictionary extends Dictionary {
if (mNativeDict != 0) {
closeNative(mNativeDict);
mNativeDict = 0;
mDictLength = 0;
}
}

28
java/src/com/android/inputmethod/latin/LatinIME.java
View File

@ -371,23 +371,19 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen
int mainDicResId = getMainDictionaryResourceId(res);
mSuggest = new Suggest(this, mainDicResId);
loadAndSetAutoCorrectionThreshold(prefs);
if (mUserDictionary != null) mUserDictionary.close();
mUserDictionary = new UserDictionary(this, locale);
if (mContactsDictionary == null) {
mSuggest.setUserDictionary(mUserDictionary);
mContactsDictionary = new ContactsDictionary(this, Suggest.DIC_CONTACTS);
}
if (mAutoDictionary != null) {
mAutoDictionary.close();
}
mSuggest.setContactsDictionary(mContactsDictionary);
mAutoDictionary = new AutoDictionary(this, this, locale, Suggest.DIC_AUTO);
if (mUserBigramDictionary != null) {
mUserBigramDictionary.close();
}
mSuggest.setAutoDictionary(mAutoDictionary);
mUserBigramDictionary = new UserBigramDictionary(this, this, locale, Suggest.DIC_USER);
mSuggest.setUserBigramDictionary(mUserBigramDictionary);
mSuggest.setUserDictionary(mUserDictionary);
mSuggest.setContactsDictionary(mContactsDictionary);
mSuggest.setAutoDictionary(mAutoDictionary);
updateCorrectionMode();
mWordSeparators = res.getString(R.string.word_separators);
mSentenceSeparators = res.getString(R.string.sentence_separators);
@ -397,11 +393,9 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen
@Override
public void onDestroy() {
if (mUserDictionary != null) {
mUserDictionary.close();
}
if (mContactsDictionary != null) {
mContactsDictionary.close();
if (mSuggest != null) {
mSuggest.close();
mSuggest = null;
}
unregisterReceiver(mReceiver);
mVoiceConnector.destroy();

25
java/src/com/android/inputmethod/latin/Suggest.java
View File

@ -22,7 +22,6 @@ import android.text.TextUtils;
import android.util.Log;
import android.view.View;
import java.nio.ByteBuffer;
import java.util.ArrayList;
import java.util.Arrays;
@ -108,11 +107,6 @@ public class Suggest implements Dictionary.WordCallback {
initPool();
}
public Suggest(Context context, ByteBuffer byteBuffer) {
mMainDict = new BinaryDictionary(context, byteBuffer, DIC_MAIN);
initPool();
}
private void initPool() {
for (int i = 0; i < mPrefMaxSuggestions; i++) {
StringBuilder sb = new StringBuilder(getApproxMaxWordLength());
@ -232,7 +226,7 @@ public class Suggest implements Dictionary.WordCallback {
if (!TextUtils.isEmpty(prevWordForBigram)) {
CharSequence lowerPrevWord = prevWordForBigram.toString().toLowerCase();
if (mMainDict.isValidWord(lowerPrevWord)) {
if (mMainDict != null && mMainDict.isValidWord(lowerPrevWord)) {
prevWordForBigram = lowerPrevWord;
}
if (mUserBigramDictionary != null) {
@ -531,6 +525,23 @@ public class Suggest implements Dictionary.WordCallback {
public void close() {
if (mMainDict != null) {
mMainDict.close();
mMainDict = null;
}
if (mUserDictionary != null) {
mUserDictionary.close();
mUserDictionary = null;
}
if (mUserBigramDictionary != null) {
mUserBigramDictionary.close();
mUserBigramDictionary = null;
}
if (mContactsDictionary != null) {
mContactsDictionary.close();
mContactsDictionary = null;
}
if (mAutoDictionary != null) {
mAutoDictionary.close();
mAutoDictionary = null;
}
}
}

120
native/jni/com_android_inputmethod_latin_BinaryDictionary.cpp
View File

@ -15,12 +15,24 @@
** limitations under the License.
*/
#define LOG_TAG "LatinIME: jni"
#include "dictionary.h"
#include "jni.h"
#include <assert.h>
#include <errno.h>
#include <stdio.h>
#ifdef USE_MMAP_FOR_DICTIONARY
#include <sys/mman.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#else // USE_MMAP_FOR_DICTIONARY
#include <stdlib.h>
#endif // USE_MMAP_FOR_DICTIONARY
// ----------------------------------------------------------------------------
using namespace latinime;
@ -37,16 +49,76 @@ static void throwException(JNIEnv *env, const char* ex, const char* fmt, int dat
}
}
static jint latinime_BinaryDictionary_open(JNIEnv *env, jobject object, jobject dictDirectBuffer,
static jint latinime_BinaryDictionary_open(JNIEnv *env, jobject object,
jstring apkFileName, jlong dictOffset, jlong dictSize,
jint typedLetterMultiplier, jint fullWordMultiplier, jint maxWordLength, jint maxWords,
jint maxAlternatives) {
void *dict = env->GetDirectBufferAddress(dictDirectBuffer);
if (dict == NULL) {
fprintf(stderr, "DICT: Dictionary buffer is null\n");
PROF_OPEN;
PROF_START(66);
const char *apkFileNameChars = env->GetStringUTFChars(apkFileName, NULL);
if (apkFileNameChars == NULL) {
LOGE("DICT: Can't get apk file name");
return 0;
}
Dictionary *dictionary = new Dictionary(dict, typedLetterMultiplier, fullWordMultiplier,
maxWordLength, maxWords, maxAlternatives);
int fd = 0;
void *dictBuf = NULL;
int adjust = 0;
#ifdef USE_MMAP_FOR_DICTIONARY
/* mmap version */
fd = open(apkFileNameChars, O_RDONLY);
if (fd < 0) {
LOGE("DICT: Can't open apk file. errno=%d", errno);
return 0;
}
int pagesize = getpagesize();
adjust = dictOffset % pagesize;
int adjDictOffset = dictOffset - adjust;
int adjDictSize = dictSize + adjust;
dictBuf = mmap(NULL, sizeof(char) * adjDictSize, PROT_READ, MAP_PRIVATE, fd, adjDictOffset);
if (dictBuf == MAP_FAILED) {
LOGE("DICT: Can't mmap dictionary file. errno=%d", errno);
return 0;
}
dictBuf = (void *)((char *)dictBuf + adjust);
#else // USE_MMAP_FOR_DICTIONARY
/* malloc version */
FILE *file = NULL;
file = fopen(apkFileNameChars, "rb");
if (file == NULL) {
LOGE("DICT: Can't fopen apk file. errno=%d", errno);
return 0;
}
dictBuf = malloc(sizeof(char) * dictSize);
if (dictBuf == NULL) {
LOGE("DICT: Can't allocate memory region for dictionary. errno=%d", errno);
return 0;
}
int ret = fseek(file, (long)dictOffset, SEEK_SET);
if (ret != 0) {
LOGE("DICT: Failure in fseek. ret=%d errno=%d", ret, errno);
return 0;
}
ret = fread(dictBuf, sizeof(char) * dictSize, 1, file);
if (ret != 1) {
LOGE("DICT: Failure in fread. ret=%d errno=%d", ret, errno);
return 0;
}
ret = fclose(file);
if (ret != 0) {
LOGE("DICT: Failure in fclose. ret=%d errno=%d", ret, errno);
return 0;
}
#endif // USE_MMAP_FOR_DICTIONARY
env->ReleaseStringUTFChars(apkFileName, apkFileNameChars);
if (!dictBuf) {
LOGE("DICT: dictBuf is null");
return 0;
}
Dictionary *dictionary = new Dictionary(dictBuf, dictSize, fd, adjust, typedLetterMultiplier,
fullWordMultiplier, maxWordLength, maxWords, maxAlternatives);
PROF_END(66);
PROF_CLOSE;
return (jint)dictionary;
}
@ -54,7 +126,7 @@ static int latinime_BinaryDictionary_getSuggestions(JNIEnv *env, jobject object,
jintArray inputArray, jint arraySize, jcharArray outputArray, jintArray frequencyArray,
jintArray nextLettersArray, jint nextLettersSize) {
Dictionary *dictionary = (Dictionary*)dict;
if (dictionary == NULL) return 0;
if (!dictionary) return 0;
int *frequencies = env->GetIntArrayElements(frequencyArray, NULL);
int *inputCodes = env->GetIntArrayElements(inputArray, NULL);
@ -80,7 +152,7 @@ static int latinime_BinaryDictionary_getBigrams(JNIEnv *env, jobject object, jin
jcharArray outputArray, jintArray frequencyArray, jint maxWordLength, jint maxBigrams,
jint maxAlternatives) {
Dictionary *dictionary = (Dictionary*)dict;
if (dictionary == NULL) return 0;
if (!dictionary) return 0;
jchar *prevWord = env->GetCharArrayElements(prevWordArray, NULL);
int *inputCodes = env->GetIntArrayElements(inputArray, NULL);
@ -99,11 +171,10 @@ static int latinime_BinaryDictionary_getBigrams(JNIEnv *env, jobject object, jin
return count;
}
static jboolean latinime_BinaryDictionary_isValidWord(JNIEnv *env, jobject object, jint dict,
jcharArray wordArray, jint wordLength) {
Dictionary *dictionary = (Dictionary*)dict;
if (dictionary == NULL) return (jboolean) false;
if (!dictionary) return (jboolean) false;
jchar *word = env->GetCharArrayElements(wordArray, NULL);
jboolean result = dictionary->isValidWord((unsigned short*) word, wordLength);
@ -113,13 +184,30 @@ static jboolean latinime_BinaryDictionary_isValidWord(JNIEnv *env, jobject objec
}
static void latinime_BinaryDictionary_close(JNIEnv *env, jobject object, jint dict) {
delete (Dictionary*) dict;
Dictionary *dictionary = (Dictionary*)dict;
if (!dictionary) return;
void *dictBuf = dictionary->getDict();
if (!dictBuf) return;
#ifdef USE_MMAP_FOR_DICTIONARY
int ret = munmap((void *)((char *)dictBuf - dictionary->getDictBufAdjust()),
dictionary->getDictSize() + dictionary->getDictBufAdjust());
if (ret != 0) {
LOGE("DICT: Failure in munmap. ret=%d errno=%d", ret, errno);
}
ret = close(dictionary->getMmapFd());
if (ret != 0) {
LOGE("DICT: Failure in close. ret=%d errno=%d", ret, errno);
}
#else // USE_MMAP_FOR_DICTIONARY
free(dictBuf);
#endif // USE_MMAP_FOR_DICTIONARY
delete dictionary;
}
// ----------------------------------------------------------------------------
static JNINativeMethod gMethods[] = {
{"openNative", "(Ljava/nio/ByteBuffer;IIIII)I", (void*)latinime_BinaryDictionary_open},
{"openNative", "(Ljava/lang/String;JJIIIII)I", (void*)latinime_BinaryDictionary_open},
{"closeNative", "(I)V", (void*)latinime_BinaryDictionary_close},
{"getSuggestionsNative", "(I[II[C[I[II)I", (void*)latinime_BinaryDictionary_getSuggestions},
{"isValidWordNative", "(I[CI)Z", (void*)latinime_BinaryDictionary_isValidWord},
@ -132,11 +220,11 @@ static int registerNativeMethods(JNIEnv* env, const char* className, JNINativeMe
clazz = env->FindClass(className);
if (clazz == NULL) {
fprintf(stderr, "Native registration unable to find class '%s'\n", className);
LOGE("Native registration unable to find class '%s'", className);
return JNI_FALSE;
}
if (env->RegisterNatives(clazz, gMethods, numMethods) < 0) {
fprintf(stderr, "RegisterNatives failed for '%s'\n", className);
LOGE("RegisterNatives failed for '%s'", className);
return JNI_FALSE;
}
@ -157,13 +245,13 @@ jint JNI_OnLoad(JavaVM* vm, void* reserved) {
jint result = -1;
if (vm->GetEnv((void**) &env, JNI_VERSION_1_4) != JNI_OK) {
fprintf(stderr, "ERROR: GetEnv failed\n");
LOGE("ERROR: GetEnv failed");
goto bail;
}
assert(env != NULL);
if (!registerNatives(env)) {
fprintf(stderr, "ERROR: BinaryDictionary native registration failed\n");
LOGE("ERROR: BinaryDictionary native registration failed");
goto bail;
}

12
native/src/bigram_dictionary.cpp
View File

@ -31,7 +31,7 @@ BigramDictionary::BigramDictionary(const unsigned char *dict, int maxWordLength,
MAX_ALTERNATIVES(maxAlternatives), IS_LATEST_DICT_VERSION(isLatestDictVersion),
HAS_BIGRAM(hasBigram), mParentDictionary(parentDictionary) {
if (DEBUG_DICT) LOGI("BigramDictionary - constructor");
if (DEBUG_DICT) LOGI("Has Bigram : %d \n", hasBigram);
if (DEBUG_DICT) LOGI("Has Bigram : %d", hasBigram);
}
BigramDictionary::~BigramDictionary() {
@ -42,7 +42,7 @@ bool BigramDictionary::addWordBigram(unsigned short *word, int length, int frequ
if (DEBUG_DICT) {
char s[length + 1];
for (int i = 0; i <= length; i++) s[i] = word[i];
LOGI("Bigram: Found word = %s, freq = %d : \n", s, frequency);
LOGI("Bigram: Found word = %s, freq = %d :", s, frequency);
}
// Find the right insertion point
@ -54,7 +54,7 @@ bool BigramDictionary::addWordBigram(unsigned short *word, int length, int frequ
}
insertAt++;
}
if (DEBUG_DICT) LOGI("Bigram: InsertAt -> %d maxBigrams: %d\n", insertAt, mMaxBigrams);
if (DEBUG_DICT) LOGI("Bigram: InsertAt -> %d maxBigrams: %d", insertAt, mMaxBigrams);
if (insertAt < mMaxBigrams) {
memmove((char*) mBigramFreq + (insertAt + 1) * sizeof(mBigramFreq[0]),
(char*) mBigramFreq + insertAt * sizeof(mBigramFreq[0]),
@ -68,7 +68,7 @@ bool BigramDictionary::addWordBigram(unsigned short *word, int length, int frequ
*dest++ = *word++;
}
*dest = 0; // NULL terminate
if (DEBUG_DICT) LOGI("Bigram: Added word at %d\n", insertAt);
if (DEBUG_DICT) LOGI("Bigram: Added word at %d", insertAt);
return true;
}
return false;
@ -107,7 +107,7 @@ int BigramDictionary::getBigrams(unsigned short *prevWord, int prevWordLength, i
if (HAS_BIGRAM && IS_LATEST_DICT_VERSION) {
int pos = mParentDictionary->isValidWordRec(
DICTIONARY_HEADER_SIZE, prevWord, 0, prevWordLength);
if (DEBUG_DICT) LOGI("Pos -> %d\n", pos);
if (DEBUG_DICT) LOGI("Pos -> %d", pos);
if (pos < 0) {
return 0;
}
@ -151,7 +151,7 @@ void BigramDictionary::searchForTerminalNode(int addressLookingFor, int frequenc
}
pos = followDownBranchAddress; // pos start at count
int count = DICT[pos] & 0xFF;
if (DEBUG_DICT) LOGI("count - %d\n",count);
if (DEBUG_DICT) LOGI("count - %d",count);
pos++;
for (int i = 0; i < count; i++) {
// pos at data

25
native/src/defines.h
View File

@ -36,15 +36,16 @@ static double profile_buf[PROF_BUF_SIZE];
static double profile_old[PROF_BUF_SIZE];
static unsigned int profile_counter[PROF_BUF_SIZE];
#define PROF_RESET prof_reset();
#define PROF_COUNT(prof_buf_id) ++profile_counter[prof_buf_id];
#define PROF_OPEN PROF_RESET;PROF_START(PROF_BUF_SIZE - 1);
#define PROF_START(prof_buf_id) PROF_COUNT(prof_buf_id);profile_old[prof_buf_id] = (clock());
#define PROF_CLOSE PROF_END(PROF_BUF_SIZE - 1);PROF_OUTALL;
#define PROF_END(prof_buf_id) profile_buf[prof_buf_id] += ((clock()) - profile_old[prof_buf_id]);
#define PROF_CLOCKOUT(prof_buf_id) LOGI("%s : clock is %f", __FUNCTION__,\
(clock() - profile_old[prof_buf_id]));
#define PROF_OUTALL LOGI("--- %s ---", __FUNCTION__); prof_out();
#define PROF_RESET prof_reset()
#define PROF_COUNT(prof_buf_id) ++profile_counter[prof_buf_id]
#define PROF_OPEN do { PROF_RESET; PROF_START(PROF_BUF_SIZE - 1); } while(0)
#define PROF_START(prof_buf_id) do { \
PROF_COUNT(prof_buf_id); profile_old[prof_buf_id] = (clock()); } while(0)
#define PROF_CLOSE do { PROF_END(PROF_BUF_SIZE - 1); PROF_OUTALL; } while(0)
#define PROF_END(prof_buf_id) profile_buf[prof_buf_id] += ((clock()) - profile_old[prof_buf_id])
#define PROF_CLOCKOUT(prof_buf_id) \
LOGI("%s : clock is %f", __FUNCTION__, (clock() - profile_old[prof_buf_id]))
#define PROF_OUTALL do { LOGI("--- %s ---", __FUNCTION__); prof_out(); } while(0)
static void prof_reset(void) {
for (int i = 0; i < PROF_BUF_SIZE; ++i) {
@ -75,6 +76,7 @@ static void prof_out(void){
}
#else // FLAG_DBG
#define LOGE
#define LOGI
#define DEBUG_DICT false
#define DEBUG_DICT_FULL false
@ -99,6 +101,11 @@ static void prof_out(void){
#define U_SHORT_MAX 1 << 16
#endif
// 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
// loading time, and acceptable even for several initial lookups which involve page faults.
#define USE_MMAP_FOR_DICTIONARY
// 22-bit address = ~4MB dictionary size limit, which on average would be about 200k-300k words
#define ADDRESS_MASK 0x3FFFFF

24
native/src/dictionary.cpp
View File

@ -23,21 +23,23 @@
namespace latinime {
Dictionary::Dictionary(void *dict, int typedLetterMultiplier, int fullWordMultiplier,
Dictionary::Dictionary(void *dict, int dictSize, int mmapFd, int dictBufAdjust,
int typedLetterMultiplier, int fullWordMultiplier,
int maxWordLength, int maxWords, int maxAlternatives)
: DICT((unsigned char*) dict),
: mDict((unsigned char*) dict), mDictSize(dictSize),
mMmapFd(mmapFd), mDictBufAdjust(dictBufAdjust),
// Checks whether it has the latest dictionary or the old dictionary
IS_LATEST_DICT_VERSION((((unsigned char*) dict)[0] & 0xFF) >= DICTIONARY_VERSION_MIN) {
if (DEBUG_DICT) {
if (MAX_WORD_LENGTH_INTERNAL < maxWordLength) {
LOGI("Max word length (%d) is greater than %d",
maxWordLength, MAX_WORD_LENGTH_INTERNAL);
LOGI("IN NATIVE SUGGEST Version: %d \n", (DICT[0] & 0xFF));
LOGI("IN NATIVE SUGGEST Version: %d", (mDict[0] & 0xFF));
}
}
mUnigramDictionary = new UnigramDictionary(DICT, typedLetterMultiplier, fullWordMultiplier,
mUnigramDictionary = new UnigramDictionary(mDict, typedLetterMultiplier, fullWordMultiplier,
maxWordLength, maxWords, maxAlternatives, IS_LATEST_DICT_VERSION);
mBigramDictionary = new BigramDictionary(DICT, maxWordLength, maxAlternatives,
mBigramDictionary = new BigramDictionary(mDict, maxWordLength, maxAlternatives,
IS_LATEST_DICT_VERSION, hasBigram(), this);
}
@ -47,7 +49,7 @@ Dictionary::~Dictionary() {
}
bool Dictionary::hasBigram() {
return ((DICT[1] & 0xFF) == 1);
return ((mDict[1] & 0xFF) == 1);
}
// TODO: use uint16_t instead of unsigned short
@ -64,12 +66,12 @@ int Dictionary::isValidWordRec(int pos, unsigned short *word, int offset, int le
// returns address of bigram data of that word
// return -99 if not found
int count = Dictionary::getCount(DICT, &pos);
int count = Dictionary::getCount(mDict, &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);
unsigned short c = Dictionary::getChar(mDict, &pos);
int terminal = Dictionary::getTerminal(mDict, &pos);
int childPos = Dictionary::getAddress(mDict, &pos);
if (c == currentChar) {
if (offset == length - 1) {
if (terminal) {
@ -85,7 +87,7 @@ int Dictionary::isValidWordRec(int pos, unsigned short *word, int offset, int le
}
}
if (terminal) {
Dictionary::getFreq(DICT, IS_LATEST_DICT_VERSION, &pos);
Dictionary::getFreq(mDict, IS_LATEST_DICT_VERSION, &pos);
}
// There could be two instances of each alphabet - upper and lower case. So continue
// looking ...

22
native/src/dictionary.h
View File

@ -25,8 +25,8 @@ namespace latinime {
class Dictionary {
public:
Dictionary(void *dict, int typedLetterMultipler, int fullWordMultiplier, int maxWordLength,
int maxWords, int maxAlternatives);
Dictionary(void *dict, int dictSize, int mmapFd, int dictBufAdjust, int typedLetterMultipler,
int fullWordMultiplier, int maxWordLength, int maxWords, int maxAlternatives);
int getSuggestions(int *codes, int codesSize, unsigned short *outWords, int *frequencies,
int *nextLetters, int nextLettersSize) {
return mUnigramDictionary->getSuggestions(codes, codesSize, outWords, frequencies,
@ -42,8 +42,10 @@ public:
}
bool isValidWord(unsigned short *word, int length);
int isValidWordRec(int pos, unsigned short *word, int offset, int length);
void setAsset(void *asset) { mAsset = asset; }
void *getAsset() { return mAsset; }
void *getDict() { return (void *)mDict; }
int getDictSize() { return mDictSize; }
int getMmapFd() { return mMmapFd; }
int getDictBufAdjust() { return mDictBufAdjust; }
~Dictionary();
// public static utility methods
@ -62,11 +64,17 @@ public:
private:
bool hasBigram();
const unsigned char *DICT;
const unsigned char *mDict;
// Used only for the mmap version of dictionary loading, but we use these as dummy variables
// also for the malloc version.
const int mDictSize;
const int mMmapFd;
const int mDictBufAdjust;
const bool IS_LATEST_DICT_VERSION;
void *mAsset;
BigramDictionary *mBigramDictionary;
UnigramDictionary *mUnigramDictionary;
BigramDictionary *mBigramDictionary;
};
// ----------------------------------------------------------------------------

1
native/src/unigram_dictionary.cpp
View File

@ -113,7 +113,6 @@ int UnigramDictionary::getSuggestions(int *codes, int codesSize, unsigned short
LOGI("%c = %d,", k, nextLetters[k]);
}
}
LOGI("\n");
}
PROF_END(6);
PROF_CLOSE;

4
native/src/unigram_dictionary.h
View File

@ -80,13 +80,13 @@ private:
bool existsAdjacentProximityChars(const int inputIndex, const int inputLength);
int* getInputCharsAt(const int index) {return mInputCodes + (index * MAX_PROXIMITY_CHARS);}
const unsigned char *DICT;
const int MAX_WORDS;
const int MAX_WORD_LENGTH;
const int MAX_WORDS;
const int MAX_PROXIMITY_CHARS;
const bool IS_LATEST_DICT_VERSION;
const int ROOT_POS;
const int TYPED_LETTER_MULTIPLIER;
const int FULL_WORD_MULTIPLIER;
const int ROOT_POS;
int *mFrequencies;
unsigned short *mOutputChars;

40
tests/src/com/android/inputmethod/latin/SuggestHelper.java
View File

@ -38,49 +38,15 @@ public class SuggestHelper {
private final String TAG;
/** Uses main dictionary only **/
public SuggestHelper(String tag, Context context, int[] resId) {
public SuggestHelper(String tag, Context context, int resId) {
TAG = tag;
InputStream[] is = null;
try {
// merging separated dictionary into one if dictionary is separated
int total = 0;
is = new InputStream[resId.length];
for (int i = 0; i < resId.length; i++) {
is[i] = context.getResources().openRawResource(resId[i]);
total += is[i].available();
}
ByteBuffer byteBuffer =
ByteBuffer.allocateDirect(total).order(ByteOrder.nativeOrder());
int got = 0;
for (int i = 0; i < resId.length; i++) {
got += Channels.newChannel(is[i]).read(byteBuffer);
}
if (got != total) {
Log.w(TAG, "Read " + got + " bytes, expected " + total);
} else {
mSuggest = new Suggest(context, byteBuffer);
Log.i(TAG, "Created mSuggest " + total + " bytes");
}
} catch (IOException e) {
Log.w(TAG, "No available memory for binary dictionary");
} finally {
try {
if (is != null) {
for (int i = 0; i < is.length; i++) {
is[i].close();
}
}
} catch (IOException e) {
Log.w(TAG, "Failed to close input stream");
}
}
mSuggest = new Suggest(context, resId);
mSuggest.setAutoTextEnabled(false);
mSuggest.setCorrectionMode(Suggest.CORRECTION_FULL_BIGRAM);
}
/** Uses both main dictionary and user-bigram dictionary **/
public SuggestHelper(String tag, Context context, int[] resId, int userBigramMax,
public SuggestHelper(String tag, Context context, int resId, int userBigramMax,
int userBigramDelete) {
this(tag, context, resId);
mUserBigram = new UserBigramDictionary(context, null, Locale.US.toString(),

4
tests/src/com/android/inputmethod/latin/SuggestPerformanceTests.java
View File

@ -36,9 +36,9 @@ public class SuggestPerformanceTests extends AndroidTestCase {
// For testing with real dictionary, TEMPORARILY COPY main dictionary into test directory.
// DO NOT SUBMIT real dictionary under test directory.
//int[] resId = new int[] { R.raw.main0, R.raw.main1, R.raw.main2 };
//int resId = R.raw.main;
int[] resId = new int[] { R.raw.test };
int resId = R.raw.test;
sh = new SuggestHelper(TAG, getTestContext(), resId);
loadString();

2
tests/src/com/android/inputmethod/latin/SuggestTests.java
View File

@ -26,7 +26,7 @@ public class SuggestTests extends AndroidTestCase {
@Override
protected void setUp() {
int[] resId = new int[] { R.raw.test };
int resId = R.raw.test;
sh = new SuggestHelper(TAG, getTestContext(), resId);
}

2
tests/src/com/android/inputmethod/latin/UserBigramTests.java
View File

@ -31,7 +31,7 @@ public class UserBigramTests extends AndroidTestCase {
@Override
protected void setUp() {
int[] resId = new int[] { R.raw.test };
int resId = R.raw.test;
sh = new SuggestHelper(TAG, getTestContext(), resId, MAX_DATA, DELETE_DATA);
}