aboutsummaryrefslogtreecommitdiffstats
path: root/native/jni/src/utils
diff options
context:
space:
mode:
Diffstat (limited to 'native/jni/src/utils')
-rw-r--r--native/jni/src/utils/byte_array_view.h15
-rw-r--r--native/jni/src/utils/char_utils.cpp4
-rw-r--r--native/jni/src/utils/char_utils.h37
-rw-r--r--native/jni/src/utils/int_array_view.h59
-rw-r--r--native/jni/src/utils/jni_data_utils.h19
5 files changed, 99 insertions, 35 deletions
diff --git a/native/jni/src/utils/byte_array_view.h b/native/jni/src/utils/byte_array_view.h
index 2c97c6d58..2b778af6f 100644
--- a/native/jni/src/utils/byte_array_view.h
+++ b/native/jni/src/utils/byte_array_view.h
@@ -42,6 +42,13 @@ class ReadOnlyByteArrayView {
return mPtr;
}
+ AK_FORCE_INLINE const ReadOnlyByteArrayView skip(const size_t n) const {
+ if (mSize <= n) {
+ return ReadOnlyByteArrayView();
+ }
+ return ReadOnlyByteArrayView(mPtr + n, mSize - n);
+ }
+
private:
DISALLOW_ASSIGNMENT_OPERATOR(ReadOnlyByteArrayView);
@@ -77,10 +84,12 @@ class ReadWriteByteArrayView {
}
private:
- DISALLOW_ASSIGNMENT_OPERATOR(ReadWriteByteArrayView);
+ // Default copy constructor and assignment operator are used for using this class with STL
+ // containers.
- uint8_t *const mPtr;
- const size_t mSize;
+ // These members cannot be const to have the assignment operator.
+ uint8_t *mPtr;
+ size_t mSize;
};
} // namespace latinime
diff --git a/native/jni/src/utils/char_utils.cpp b/native/jni/src/utils/char_utils.cpp
index b17e0847d..3bb9055b2 100644
--- a/native/jni/src/utils/char_utils.cpp
+++ b/native/jni/src/utils/char_utils.cpp
@@ -1057,11 +1057,11 @@ static int compare_pair_capital(const void *a, const void *b) {
- static_cast<int>((static_cast<const struct LatinCapitalSmallPair *>(b))->capital);
}
-/* static */ unsigned short CharUtils::latin_tolower(const unsigned short c) {
+/* static */ int CharUtils::latin_tolower(const int c) {
struct LatinCapitalSmallPair *p =
static_cast<struct LatinCapitalSmallPair *>(bsearch(&c, SORTED_CHAR_MAP,
NELEMS(SORTED_CHAR_MAP), sizeof(SORTED_CHAR_MAP[0]), compare_pair_capital));
- return p ? p->small : c;
+ return p ? static_cast<int>(p->small) : c;
}
/*
diff --git a/native/jni/src/utils/char_utils.h b/native/jni/src/utils/char_utils.h
index 63786502b..7871c26ef 100644
--- a/native/jni/src/utils/char_utils.h
+++ b/native/jni/src/utils/char_utils.h
@@ -27,20 +27,14 @@ namespace latinime {
class CharUtils {
public:
+ static const std::vector<int> EMPTY_STRING;
+
static AK_FORCE_INLINE bool isAsciiUpper(int c) {
// Note: isupper(...) reports false positives for some Cyrillic characters, causing them to
// be incorrectly lower-cased using toAsciiLower(...) rather than latin_tolower(...).
return (c >= 'A' && c <= 'Z');
}
- static AK_FORCE_INLINE int toAsciiLower(int c) {
- return c - 'A' + 'a';
- }
-
- static AK_FORCE_INLINE bool isAscii(int c) {
- return isascii(c) != 0;
- }
-
static AK_FORCE_INLINE int toLowerCase(const int c) {
if (isAsciiUpper(c)) {
return toAsciiLower(c);
@@ -48,7 +42,7 @@ class CharUtils {
if (isAscii(c)) {
return c;
}
- return static_cast<int>(latin_tolower(static_cast<unsigned short>(c)));
+ return latin_tolower(c);
}
static AK_FORCE_INLINE int toBaseLowerCase(const int c) {
@@ -59,7 +53,6 @@ class CharUtils {
// TODO: Do not hardcode here
return codePoint == KEYCODE_SINGLE_QUOTE || codePoint == KEYCODE_HYPHEN_MINUS;
}
-
static AK_FORCE_INLINE int getCodePointCount(const int arraySize, const int *const codePoints) {
int size = 0;
for (; size < arraySize; ++size) {
@@ -91,9 +84,6 @@ class CharUtils {
return codePoint >= MIN_UNICODE_CODE_POINT && codePoint <= MAX_UNICODE_CODE_POINT;
}
- static unsigned short latin_tolower(const unsigned short c);
- static const std::vector<int> EMPTY_STRING;
-
// Returns updated code point count. Returns 0 when the code points cannot be marked as a
// Beginning-of-Sentence.
static AK_FORCE_INLINE int attachBeginningOfSentenceMarker(int *const codePoints,
@@ -111,6 +101,17 @@ class CharUtils {
return codePointCount + 1;
}
+ // Returns updated code point count.
+ static AK_FORCE_INLINE int removeBeginningOfSentenceMarker(int *const codePoints,
+ const int codePointCount) {
+ if (codePointCount <= 0 || codePoints[0] != CODE_POINT_BEGINNING_OF_SENTENCE) {
+ return codePointCount;
+ }
+ const int newCodePointCount = codePointCount - 1;
+ memmove(codePoints, codePoints + 1, sizeof(int) * newCodePointCount);
+ return newCodePointCount;
+ }
+
private:
DISALLOW_IMPLICIT_CONSTRUCTORS(CharUtils);
@@ -125,6 +126,16 @@ class CharUtils {
*/
static const int BASE_CHARS_SIZE = 0x0500;
static const unsigned short BASE_CHARS[BASE_CHARS_SIZE];
+
+ static AK_FORCE_INLINE bool isAscii(int c) {
+ return isascii(c) != 0;
+ }
+
+ static AK_FORCE_INLINE int toAsciiLower(int c) {
+ return c - 'A' + 'a';
+ }
+
+ static int latin_tolower(const int c);
};
} // namespace latinime
#endif // LATINIME_CHAR_UTILS_H
diff --git a/native/jni/src/utils/int_array_view.h b/native/jni/src/utils/int_array_view.h
index c1ddc9812..408373176 100644
--- a/native/jni/src/utils/int_array_view.h
+++ b/native/jni/src/utils/int_array_view.h
@@ -17,8 +17,10 @@
#ifndef LATINIME_INT_ARRAY_VIEW_H
#define LATINIME_INT_ARRAY_VIEW_H
+#include <algorithm>
+#include <array>
#include <cstdint>
-#include <cstdlib>
+#include <cstring>
#include <vector>
#include "defines.h"
@@ -56,14 +58,14 @@ class IntArrayView {
explicit IntArrayView(const std::vector<int> &vector)
: mPtr(vector.data()), mSize(vector.size()) {}
- template <int N>
- AK_FORCE_INLINE static IntArrayView fromFixedSizeArray(const int (&array)[N]) {
- return IntArrayView(array, N);
+ template <size_t N>
+ AK_FORCE_INLINE static IntArrayView fromArray(const std::array<int, N> &array) {
+ return IntArrayView(array.data(), array.size());
}
- // Returns a view that points one int object. Does not take ownership of the given object.
- AK_FORCE_INLINE static IntArrayView fromObject(const int *const object) {
- return IntArrayView(object, 1);
+ // Returns a view that points one int object.
+ AK_FORCE_INLINE static IntArrayView singleElementView(const int *const ptr) {
+ return IntArrayView(ptr, 1);
}
AK_FORCE_INLINE int operator[](const size_t index) const {
@@ -91,6 +93,46 @@ class IntArrayView {
return mPtr + mSize;
}
+ AK_FORCE_INLINE bool contains(const int value) const {
+ return std::find(begin(), end(), value) != end();
+ }
+
+ // Returns the view whose size is smaller than or equal to the given count.
+ AK_FORCE_INLINE const IntArrayView limit(const size_t maxSize) const {
+ return IntArrayView(mPtr, std::min(maxSize, mSize));
+ }
+
+ AK_FORCE_INLINE const IntArrayView skip(const size_t n) const {
+ if (mSize <= n) {
+ return IntArrayView();
+ }
+ return IntArrayView(mPtr + n, mSize - n);
+ }
+
+ template <size_t N>
+ void copyToArray(std::array<int, N> *const buffer, const size_t offset) const {
+ ASSERT(mSize + offset <= N);
+ memmove(buffer->data() + offset, mPtr, sizeof(int) * mSize);
+ }
+
+ AK_FORCE_INLINE int firstOrDefault(const int defaultValue) const {
+ if (empty()) {
+ return defaultValue;
+ }
+ return mPtr[0];
+ }
+
+ AK_FORCE_INLINE int lastOrDefault(const int defaultValue) const {
+ if (empty()) {
+ return defaultValue;
+ }
+ return mPtr[mSize - 1];
+ }
+
+ AK_FORCE_INLINE std::vector<int> toVector() const {
+ return std::vector<int>(begin(), end());
+ }
+
private:
DISALLOW_ASSIGNMENT_OPERATOR(IntArrayView);
@@ -100,6 +142,9 @@ class IntArrayView {
using WordIdArrayView = IntArrayView;
using PtNodePosArrayView = IntArrayView;
+using CodePointArrayView = IntArrayView;
+template <size_t size>
+using WordIdArray = std::array<int, size>;
} // namespace latinime
#endif // LATINIME_MEMORY_VIEW_H
diff --git a/native/jni/src/utils/jni_data_utils.h b/native/jni/src/utils/jni_data_utils.h
index cb82d3c3b..a259e1cd0 100644
--- a/native/jni/src/utils/jni_data_utils.h
+++ b/native/jni/src/utils/jni_data_utils.h
@@ -21,7 +21,7 @@
#include "defines.h"
#include "jni.h"
-#include "suggest/core/session/prev_words_info.h"
+#include "suggest/core/session/ngram_context.h"
#include "suggest/core/policy/dictionary_header_structure_policy.h"
#include "suggest/policyimpl/dictionary/header/header_read_write_utils.h"
#include "utils/char_utils.h"
@@ -50,6 +50,7 @@ class JniDataUtils {
const jsize keyUtf8Length = env->GetStringUTFLength(keyString);
char keyChars[keyUtf8Length + 1];
env->GetStringUTFRegion(keyString, 0, env->GetStringLength(keyString), keyChars);
+ env->DeleteLocalRef(keyString);
keyChars[keyUtf8Length] = '\0';
DictionaryHeaderStructurePolicy::AttributeMap::key_type key;
HeaderReadWriteUtils::insertCharactersIntoVector(keyChars, &key);
@@ -59,6 +60,7 @@ class JniDataUtils {
const jsize valueUtf8Length = env->GetStringUTFLength(valueString);
char valueChars[valueUtf8Length + 1];
env->GetStringUTFRegion(valueString, 0, env->GetStringLength(valueString), valueChars);
+ env->DeleteLocalRef(valueString);
valueChars[valueUtf8Length] = '\0';
DictionaryHeaderStructurePolicy::AttributeMap::mapped_type value;
HeaderReadWriteUtils::insertCharactersIntoVector(valueChars, &value);
@@ -96,18 +98,14 @@ class JniDataUtils {
}
}
- static PrevWordsInfo constructPrevWordsInfo(JNIEnv *env, jobjectArray prevWordCodePointArrays,
- jbooleanArray isBeginningOfSentenceArray) {
+ static NgramContext constructNgramContext(JNIEnv *env, jobjectArray prevWordCodePointArrays,
+ jbooleanArray isBeginningOfSentenceArray, const size_t prevWordCount) {
int prevWordCodePoints[MAX_PREV_WORD_COUNT_FOR_N_GRAM][MAX_WORD_LENGTH];
int prevWordCodePointCount[MAX_PREV_WORD_COUNT_FOR_N_GRAM];
bool isBeginningOfSentence[MAX_PREV_WORD_COUNT_FOR_N_GRAM];
- jsize prevWordsCount = env->GetArrayLength(prevWordCodePointArrays);
- for (size_t i = 0; i < NELEMS(prevWordCodePoints); ++i) {
+ for (size_t i = 0; i < prevWordCount; ++i) {
prevWordCodePointCount[i] = 0;
isBeginningOfSentence[i] = false;
- if (prevWordsCount <= static_cast<int>(i)) {
- continue;
- }
jintArray prevWord = (jintArray)env->GetObjectArrayElement(prevWordCodePointArrays, i);
if (!prevWord) {
continue;
@@ -117,14 +115,15 @@ class JniDataUtils {
continue;
}
env->GetIntArrayRegion(prevWord, 0, prevWordLength, prevWordCodePoints[i]);
+ env->DeleteLocalRef(prevWord);
prevWordCodePointCount[i] = prevWordLength;
jboolean isBeginningOfSentenceBoolean = JNI_FALSE;
env->GetBooleanArrayRegion(isBeginningOfSentenceArray, i, 1 /* len */,
&isBeginningOfSentenceBoolean);
isBeginningOfSentence[i] = isBeginningOfSentenceBoolean == JNI_TRUE;
}
- return PrevWordsInfo(prevWordCodePoints, prevWordCodePointCount, isBeginningOfSentence,
- MAX_PREV_WORD_COUNT_FOR_N_GRAM);
+ return NgramContext(prevWordCodePoints, prevWordCodePointCount, isBeginningOfSentence,
+ prevWordCount);
}
static void putBooleanToArray(JNIEnv *env, jbooleanArray array, const int index,
generated by cgit v1.2.3-83-g751a (git 2.47.1) at 2025-05-08 00:25:47 +0000