aboutsummaryrefslogtreecommitdiffstats
path: root/native
diff options
context:
space:
mode:
Diffstat (limited to 'native')
-rw-r--r--native/jni/Android.mk2
-rw-r--r--native/jni/com_android_inputmethod_latin_BinaryDictionary.cpp2
-rw-r--r--native/jni/src/suggest/core/dictionary/bigram_dictionary.cpp58
-rw-r--r--native/jni/src/suggest/core/dictionary/bigram_dictionary.h7
-rw-r--r--native/jni/src/suggest/core/dictionary/dictionary.cpp7
-rw-r--r--native/jni/src/suggest/core/dictionary/dictionary.h4
-rw-r--r--native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.cpp74
-rw-r--r--native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.h134
-rw-r--r--native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_policy.cpp96
9 files changed, 319 insertions, 65 deletions
diff --git a/native/jni/Android.mk b/native/jni/Android.mk
index acd230ff2..e14cf5a71 100644
--- a/native/jni/Android.mk
+++ b/native/jni/Android.mk
@@ -71,7 +71,9 @@ LATIN_IME_CORE_SRC_FILES := \
suggest/core/policy/weighting.cpp \
suggest/core/session/dic_traverse_session.cpp \
$(addprefix suggest/policyimpl/dictionary/, \
+ dynamic_patricia_trie_node_reader.cpp \
dynamic_patricia_trie_policy.cpp \
+ dynamic_patricia_trie_reading_utils.cpp \
patricia_trie_policy.cpp \
patricia_trie_reading_utils.cpp) \
suggest/policyimpl/gesture/gesture_suggest_policy_factory.cpp \
diff --git a/native/jni/com_android_inputmethod_latin_BinaryDictionary.cpp b/native/jni/com_android_inputmethod_latin_BinaryDictionary.cpp
index 6e1b80ee0..8b46c2644 100644
--- a/native/jni/com_android_inputmethod_latin_BinaryDictionary.cpp
+++ b/native/jni/com_android_inputmethod_latin_BinaryDictionary.cpp
@@ -186,7 +186,7 @@ static int latinime_BinaryDictionary_getSuggestions(JNIEnv *env, jclass clazz, j
scores, spaceIndices, outputTypes);
} else {
count = dictionary->getBigrams(prevWordCodePoints, prevWordCodePointsLength,
- inputCodePoints, inputSize, outputCodePoints, scores, outputTypes);
+ outputCodePoints, scores, outputTypes);
}
// Copy back the output values
diff --git a/native/jni/src/suggest/core/dictionary/bigram_dictionary.cpp b/native/jni/src/suggest/core/dictionary/bigram_dictionary.cpp
index 532c769c6..3751ae500 100644
--- a/native/jni/src/suggest/core/dictionary/bigram_dictionary.cpp
+++ b/native/jni/src/suggest/core/dictionary/bigram_dictionary.cpp
@@ -87,21 +87,14 @@ void BigramDictionary::addWordBigram(int *word, int length, int probability, int
/* Parameters :
* prevWord: the word before, the one for which we need to look up bigrams.
* prevWordLength: its length.
- * inputCodePoints: what user typed, in the same format as for UnigramDictionary::getSuggestions.
- * inputSize: the size of the codes array.
- * bigramCodePoints: an array for output, at the same format as outwords for getSuggestions.
- * bigramProbability: an array to output frequencies.
+ * outBigramCodePoints: an array for output, at the same format as outwords for getSuggestions.
+ * outBigramProbability: an array to output frequencies.
* outputTypes: an array to output types.
* This method returns the number of bigrams this word has, for backward compatibility.
- * Note: this is not the number of bigrams output in the array, which is the number of
- * bigrams this word has WHOSE first letter also matches the letter the user typed.
- * TODO: this may not be a sensible thing to do. It makes sense when the bigrams are
- * used to match the first letter of the second word, but once the user has typed more
- * and the bigrams are used to boost unigram result scores, it makes little sense to
- * reduce their scope to the ones that match the first letter.
*/
-int BigramDictionary::getPredictions(const int *prevWord, int prevWordLength, int *inputCodePoints,
- int inputSize, int *bigramCodePoints, int *bigramProbability, int *outputTypes) const {
+int BigramDictionary::getPredictions(const int *prevWord, const int prevWordLength,
+ int *const outBigramCodePoints, int *const outBigramProbability,
+ int *const outputTypes) const {
// TODO: remove unused arguments, and refrain from storing stuff in members of this class
// TODO: have "in" arguments before "out" ones, and make out args explicit in the name
@@ -126,21 +119,16 @@ int BigramDictionary::getPredictions(const int *prevWord, int prevWordLength, in
getCodePointsAndProbabilityAndReturnCodePointCount(
mBinaryDictionaryInfo, bigramsIt.getBigramPos(), MAX_WORD_LENGTH,
bigramBuffer, &unigramProbability);
-
- // inputSize == 0 means we are trying to find bigram predictions.
- if (inputSize < 1 || checkFirstCharacter(bigramBuffer, inputCodePoints)) {
- const int bigramProbabilityTemp = bigramsIt.getProbability();
- // Due to space constraints, the probability for bigrams is approximate - the lower the
- // unigram probability, the worse the precision. The theoritical maximum error in
- // resulting probability is 8 - although in the practice it's never bigger than 3 or 4
- // in very bad cases. This means that sometimes, we'll see some bigrams interverted
- // here, but it can't get too bad.
- const int probability = ProbabilityUtils::computeProbabilityForBigram(
- unigramProbability, bigramProbabilityTemp);
- addWordBigram(bigramBuffer, length, probability, bigramProbability, bigramCodePoints,
- outputTypes);
- ++bigramCount;
- }
+ // Due to space constraints, the probability for bigrams is approximate - the lower the
+ // unigram probability, the worse the precision. The theoritical maximum error in
+ // resulting probability is 8 - although in the practice it's never bigger than 3 or 4
+ // in very bad cases. This means that sometimes, we'll see some bigrams interverted
+ // here, but it can't get too bad.
+ const int probability = ProbabilityUtils::computeProbabilityForBigram(
+ unigramProbability, bigramsIt.getProbability());
+ addWordBigram(bigramBuffer, length, probability, outBigramProbability, outBigramCodePoints,
+ outputTypes);
+ ++bigramCount;
}
return min(bigramCount, MAX_RESULTS);
}
@@ -157,22 +145,6 @@ int BigramDictionary::getBigramListPositionForWord(const int *prevWord, const in
mBinaryDictionaryInfo, pos);
}
-bool BigramDictionary::checkFirstCharacter(int *word, int *inputCodePoints) const {
- // Checks whether this word starts with same character or neighboring characters of
- // what user typed.
-
- int maxAlt = MAX_ALTERNATIVES;
- const int firstBaseLowerCodePoint = CharUtils::toBaseLowerCase(*word);
- while (maxAlt > 0) {
- if (CharUtils::toBaseLowerCase(*inputCodePoints) == firstBaseLowerCodePoint) {
- return true;
- }
- inputCodePoints++;
- maxAlt--;
- }
- return false;
-}
-
bool BigramDictionary::isValidBigram(const int *word0, int length0, const int *word1,
int length1) const {
int pos = getBigramListPositionForWord(word0, length0, false /* forceLowerCaseSearch */);
diff --git a/native/jni/src/suggest/core/dictionary/bigram_dictionary.h b/native/jni/src/suggest/core/dictionary/bigram_dictionary.h
index 7706a2c22..438c34cac 100644
--- a/native/jni/src/suggest/core/dictionary/bigram_dictionary.h
+++ b/native/jni/src/suggest/core/dictionary/bigram_dictionary.h
@@ -27,8 +27,8 @@ class BigramDictionary {
public:
BigramDictionary(const BinaryDictionaryInfo *const binaryDictionaryInfo);
- int getPredictions(const int *word, int length, int *inputCodePoints, int inputSize,
- int *outWords, int *frequencies, int *outputTypes) const;
+ int getPredictions(const int *word, int length, int *outBigramCodePoints,
+ int *outBigramProbability, int *outputTypes) const;
bool isValidBigram(const int *word1, int length1, const int *word2, int length2) const;
~BigramDictionary();
@@ -37,13 +37,10 @@ class BigramDictionary {
void addWordBigram(int *word, int length, int probability, int *bigramProbability,
int *bigramCodePoints, int *outputTypes) const;
- bool checkFirstCharacter(int *word, int *inputCodePoints) const;
int getBigramListPositionForWord(const int *prevWord, const int prevWordLength,
const bool forceLowerCaseSearch) const;
const BinaryDictionaryInfo *const mBinaryDictionaryInfo;
- // TODO: Re-implement proximity correction for bigram correction
- static const int MAX_ALTERNATIVES = 1;
};
} // namespace latinime
#endif // LATINIME_BIGRAM_DICTIONARY_H
diff --git a/native/jni/src/suggest/core/dictionary/dictionary.cpp b/native/jni/src/suggest/core/dictionary/dictionary.cpp
index 4f5d29f6a..4a9e38fe8 100644
--- a/native/jni/src/suggest/core/dictionary/dictionary.cpp
+++ b/native/jni/src/suggest/core/dictionary/dictionary.cpp
@@ -76,11 +76,10 @@ int Dictionary::getSuggestions(ProximityInfo *proximityInfo, DicTraverseSession
}
}
-int Dictionary::getBigrams(const int *word, int length, int *inputCodePoints, int inputSize,
- int *outWords, int *frequencies, int *outputTypes) const {
+int Dictionary::getBigrams(const int *word, int length, int *outWords, int *frequencies,
+ int *outputTypes) const {
if (length <= 0) return 0;
- return mBigramDictionary->getPredictions(word, length, inputCodePoints, inputSize, outWords,
- frequencies, outputTypes);
+ return mBigramDictionary->getPredictions(word, length, outWords, frequencies, outputTypes);
}
int Dictionary::getProbability(const int *word, int length) const {
diff --git a/native/jni/src/suggest/core/dictionary/dictionary.h b/native/jni/src/suggest/core/dictionary/dictionary.h
index 1bf24a85b..9f1e0729d 100644
--- a/native/jni/src/suggest/core/dictionary/dictionary.h
+++ b/native/jni/src/suggest/core/dictionary/dictionary.h
@@ -62,8 +62,8 @@ class Dictionary {
const SuggestOptions *const suggestOptions, int *outWords, int *frequencies,
int *spaceIndices, int *outputTypes) const;
- int getBigrams(const int *word, int length, int *inputCodePoints, int inputSize, int *outWords,
- int *frequencies, int *outputTypes) const;
+ int getBigrams(const int *word, int length, int *outWords, int *frequencies,
+ int *outputTypes) const;
int getProbability(const int *word, int length) const;
diff --git a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.cpp b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.cpp
new file mode 100644
index 000000000..20cda91a3
--- /dev/null
+++ b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.cpp
@@ -0,0 +1,74 @@
+/*
+ * Copyright (C) 2013, The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.h"
+
+#include "suggest/core/dictionary/binary_dictionary_info.h"
+#include "suggest/core/dictionary/binary_dictionary_terminal_attributes_reading_utils.h"
+#include "suggest/policyimpl/dictionary/dynamic_patricia_trie_reading_utils.h"
+
+namespace latinime {
+
+void DynamicPatriciaTrieNodeReader::fetchNodeInfoFromBufferAndProcessMovedNode(const int nodePos,
+ const int maxCodePointCount, int *const outCodePoints) {
+ const uint8_t *const dictRoot = mBinaryDictionaryInfo->getDictRoot();
+ int pos = nodePos;
+ mFlags = PatriciaTrieReadingUtils::getFlagsAndAdvancePosition(dictRoot, &pos);
+ mParentPos = DynamicPatriciaTrieReadingUtils::getParentPosAndAdvancePosition(dictRoot, &pos);
+ if (outCodePoints != 0) {
+ mCodePointCount = PatriciaTrieReadingUtils::getCharsAndAdvancePosition(
+ dictRoot, mFlags, maxCodePointCount, outCodePoints, &pos);
+ } else {
+ mCodePointCount = PatriciaTrieReadingUtils::skipCharacters(
+ dictRoot, mFlags, MAX_WORD_LENGTH, &pos);
+ }
+ if (isTerminal()) {
+ mProbability = PatriciaTrieReadingUtils::readProbabilityAndAdvancePosition(dictRoot, &pos);
+ } else {
+ mProbability = NOT_A_PROBABILITY;
+ }
+ if (hasChildren()) {
+ mChildrenPos = DynamicPatriciaTrieReadingUtils::readChildrenPositionAndAdvancePosition(
+ dictRoot, mFlags, &pos);
+ } else {
+ mChildrenPos = NOT_A_DICT_POS;
+ }
+ if (PatriciaTrieReadingUtils::hasShortcutTargets(mFlags)) {
+ mShortcutPos = pos;
+ BinaryDictionaryTerminalAttributesReadingUtils::skipShortcuts(mBinaryDictionaryInfo, &pos);
+ } else {
+ mShortcutPos = NOT_A_DICT_POS;
+ }
+ if (PatriciaTrieReadingUtils::hasBigrams(mFlags)) {
+ mBigramPos = pos;
+ BinaryDictionaryTerminalAttributesReadingUtils::skipExistingBigrams(
+ mBinaryDictionaryInfo, &pos);
+ } else {
+ mBigramPos = NOT_A_DICT_POS;
+ }
+ // Update siblingPos if needed.
+ if (mSiblingPos == NOT_A_VALID_WORD_POS) {
+ // Sibling position is the tail position of current node.
+ mSiblingPos = pos;
+ }
+ // Read destination node if the read node is a moved node.
+ if (DynamicPatriciaTrieReadingUtils::isMoved(mFlags)) {
+ // The destination position is stored at the same place as the parent position.
+ fetchNodeInfoFromBufferAndProcessMovedNode(mParentPos, maxCodePointCount, outCodePoints);
+ }
+}
+
+}
diff --git a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.h b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.h
new file mode 100644
index 000000000..b668aab78
--- /dev/null
+++ b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.h
@@ -0,0 +1,134 @@
+/*
+ * Copyright (C) 2013, The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef LATINIME_DYNAMIC_PATRICIA_TRIE_NODE_READER_H
+#define LATINIME_DYNAMIC_PATRICIA_TRIE_NODE_READER_H
+
+#include "defines.h"
+#include "suggest/policyimpl/dictionary/dynamic_patricia_trie_reading_utils.h"
+#include "suggest/policyimpl/dictionary/patricia_trie_reading_utils.h"
+
+namespace latinime {
+
+class BinaryDictionaryInfo;
+
+/*
+ * This class is used for helping to read nodes of dynamic patricia trie. This class handles moved
+ * node and reads node attributes.
+ */
+class DynamicPatriciaTrieNodeReader {
+ public:
+ explicit DynamicPatriciaTrieNodeReader(const BinaryDictionaryInfo *const binaryDictionaryInfo)
+ : mBinaryDictionaryInfo(binaryDictionaryInfo), mNodePos(NOT_A_VALID_WORD_POS),
+ mFlags(0), mParentPos(NOT_A_DICT_POS), mCodePointCount(0),
+ mProbability(NOT_A_PROBABILITY), mChildrenPos(NOT_A_DICT_POS),
+ mShortcutPos(NOT_A_DICT_POS), mBigramPos(NOT_A_DICT_POS),
+ mSiblingPos(NOT_A_VALID_WORD_POS) {}
+
+ ~DynamicPatriciaTrieNodeReader() {}
+
+ // Reads node information from dictionary buffer and updates members with the information.
+ AK_FORCE_INLINE void fetchNodeInfoFromBuffer(const int nodePos) {
+ fetchNodeInfoFromBufferAndGetNodeCodePoints(mNodePos , 0 /* maxCodePointCount */,
+ 0 /* outCodePoints */);
+ }
+
+ AK_FORCE_INLINE void fetchNodeInfoFromBufferAndGetNodeCodePoints(const int nodePos,
+ const int maxCodePointCount, int *const outCodePoints) {
+ mNodePos = nodePos;
+ mSiblingPos = NOT_A_VALID_WORD_POS;
+ fetchNodeInfoFromBufferAndProcessMovedNode(mNodePos, maxCodePointCount, outCodePoints);
+ }
+
+ AK_FORCE_INLINE int getNodePos() const {
+ return mNodePos;
+ }
+
+ // Flags
+ AK_FORCE_INLINE bool isDeleted() const {
+ return DynamicPatriciaTrieReadingUtils::isDeleted(mFlags);
+ }
+
+ AK_FORCE_INLINE bool hasChildren() const {
+ return PatriciaTrieReadingUtils::hasChildrenInFlags(mFlags);
+ }
+
+ AK_FORCE_INLINE bool isTerminal() const {
+ return PatriciaTrieReadingUtils::isTerminal(mFlags);
+ }
+
+ AK_FORCE_INLINE bool isBlacklisted() const {
+ return PatriciaTrieReadingUtils::isBlacklisted(mFlags);
+ }
+
+ AK_FORCE_INLINE bool isNotAWord() const {
+ return PatriciaTrieReadingUtils::isNotAWord(mFlags);
+ }
+
+ // Parent node position
+ AK_FORCE_INLINE int getParentPos() const {
+ return mParentPos;
+ }
+
+ // Number of code points
+ AK_FORCE_INLINE uint8_t getCodePointCount() const {
+ return mCodePointCount;
+ }
+
+ // Probability
+ AK_FORCE_INLINE int getProbability() const {
+ return mProbability;
+ }
+
+ // Children node group position
+ AK_FORCE_INLINE int getChildrenPos() const {
+ return mChildrenPos;
+ }
+
+ // Shortcutlist position
+ AK_FORCE_INLINE int getShortcutPos() const {
+ return mShortcutPos;
+ }
+
+ // Bigrams position
+ AK_FORCE_INLINE int getBigramsPos() const {
+ return mBigramPos;
+ }
+
+ // Sibling node position
+ AK_FORCE_INLINE int getSiblingNodePos() const {
+ return mSiblingPos;
+ }
+
+ private:
+ DISALLOW_COPY_AND_ASSIGN(DynamicPatriciaTrieNodeReader);
+
+ const BinaryDictionaryInfo *const mBinaryDictionaryInfo;
+ int mNodePos;
+ DynamicPatriciaTrieReadingUtils::NodeFlags mFlags;
+ int mParentPos;
+ uint8_t mCodePointCount;
+ int mProbability;
+ int mChildrenPos;
+ int mShortcutPos;
+ int mBigramPos;
+ int mSiblingPos;
+
+ void fetchNodeInfoFromBufferAndProcessMovedNode(const int nodePos, const int maxCodePointCount,
+ int *const outCodePoints);
+};
+} // namespace latinime
+#endif /* LATINIME_DYNAMIC_PATRICIA_TRIE_NODE_READER_H */
diff --git a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_policy.cpp b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_policy.cpp
index c7314ecf1..9a180e6f7 100644
--- a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_policy.cpp
+++ b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_policy.cpp
@@ -14,12 +14,15 @@
* limitations under the License.
*/
-#include "suggest/policyimpl/dictionary/patricia_trie_policy.h"
+#include "suggest/policyimpl/dictionary/dynamic_patricia_trie_policy.h"
#include "defines.h"
#include "suggest/core/dicnode/dic_node.h"
#include "suggest/core/dicnode/dic_node_vector.h"
#include "suggest/core/dictionary/binary_dictionary_info.h"
+#include "suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.h"
+#include "suggest/policyimpl/dictionary/dynamic_patricia_trie_reading_utils.h"
+#include "suggest/policyimpl/dictionary/patricia_trie_reading_utils.h"
namespace latinime {
@@ -28,15 +31,76 @@ const DynamicPatriciaTriePolicy DynamicPatriciaTriePolicy::sInstance;
void DynamicPatriciaTriePolicy::createAndGetAllChildNodes(const DicNode *const dicNode,
const BinaryDictionaryInfo *const binaryDictionaryInfo,
const NodeFilter *const nodeFilter, DicNodeVector *const childDicNodes) const {
- // TODO: Implement.
+ if (!dicNode->hasChildren()) {
+ return;
+ }
+ DynamicPatriciaTrieNodeReader nodeReader(binaryDictionaryInfo);
+ int mergedNodeCodePoints[MAX_WORD_LENGTH];
+ int nextPos = dicNode->getChildrenPos();
+ do {
+ const int childCount = PatriciaTrieReadingUtils::getGroupCountAndAdvancePosition(
+ binaryDictionaryInfo->getDictRoot(), &nextPos);
+ for (int i = 0; i < childCount; i++) {
+ nodeReader.fetchNodeInfoFromBufferAndGetNodeCodePoints(nextPos, MAX_WORD_LENGTH,
+ mergedNodeCodePoints);
+ if (!nodeReader.isDeleted() && !nodeFilter->isFilteredOut(mergedNodeCodePoints[0])) {
+ // Push child note when the node is not deleted and not filtered out.
+ childDicNodes->pushLeavingChild(dicNode, nodeReader.getNodePos(),
+ nodeReader.getChildrenPos(), nodeReader.getProbability(),
+ nodeReader.isTerminal(), nodeReader.hasChildren(),
+ nodeReader.isBlacklisted() || nodeReader.isNotAWord(),
+ nodeReader.getCodePointCount(), mergedNodeCodePoints);
+ }
+ nextPos = nodeReader.getSiblingNodePos();
+ }
+ nextPos = DynamicPatriciaTrieReadingUtils::getForwardLinkPosition(
+ binaryDictionaryInfo->getDictRoot(), nextPos);
+ } while(DynamicPatriciaTrieReadingUtils::isValidForwardLinkPosition(nextPos));
}
int DynamicPatriciaTriePolicy::getCodePointsAndProbabilityAndReturnCodePointCount(
const BinaryDictionaryInfo *const binaryDictionaryInfo,
const int nodePos, const int maxCodePointCount, int *const outCodePoints,
int *const outUnigramProbability) const {
- // TODO: Implement.
- return 0;
+ // This method traverses parent nodes from the terminal by following parent pointers; thus,
+ // node code points are stored in the buffer in the reverse order.
+ int reverseCodePoints[maxCodePointCount];
+ int mergedNodeCodePoints[maxCodePointCount];
+ int codePointCount = 0;
+
+ DynamicPatriciaTrieNodeReader nodeReader(binaryDictionaryInfo);
+ // First, read terminal node and get its probability.
+ nodeReader.fetchNodeInfoFromBufferAndGetNodeCodePoints(nodePos, maxCodePointCount,
+ mergedNodeCodePoints);
+ // Store terminal node probability.
+ *outUnigramProbability = nodeReader.getProbability();
+ // Store terminal node code points to buffer in the reverse order.
+ for (int i = nodeReader.getCodePointCount() - 1; i >= 0; --i) {
+ reverseCodePoints[codePointCount++] = mergedNodeCodePoints[i];
+ }
+ // Then, follow parent pos toward the root node.
+ while (nodeReader.getParentPos() != getRootPosition()) {
+ // codePointCount must be incremented at least once in each iteration to ensure preventing
+ // infinite loop.
+ if (nodeReader.isDeleted() || codePointCount > maxCodePointCount
+ || nodeReader.getCodePointCount() <= 0) {
+ // The nodePos is not a valid terminal node position in the dictionary.
+ *outUnigramProbability = NOT_A_PROBABILITY;
+ return 0;
+ }
+ // Read parent node.
+ nodeReader.fetchNodeInfoFromBufferAndGetNodeCodePoints(nodeReader.getParentPos(),
+ maxCodePointCount, mergedNodeCodePoints);
+ // Store node code points to buffer in the reverse order.
+ for (int i = nodeReader.getCodePointCount() - 1; i >= 0; --i) {
+ reverseCodePoints[codePointCount++] = mergedNodeCodePoints[i];
+ }
+ }
+ // Reverse the stored code points to output them.
+ for (int i = 0; i < codePointCount; ++i) {
+ outCodePoints[i] = reverseCodePoints[codePointCount - i - 1];
+ }
+ return codePointCount;
}
int DynamicPatriciaTriePolicy::getTerminalNodePositionOfWord(
@@ -48,22 +112,34 @@ int DynamicPatriciaTriePolicy::getTerminalNodePositionOfWord(
int DynamicPatriciaTriePolicy::getUnigramProbability(
const BinaryDictionaryInfo *const binaryDictionaryInfo, const int nodePos) const {
- // TODO: Implement.
- return NOT_A_PROBABILITY;
+ DynamicPatriciaTrieNodeReader nodeReader(binaryDictionaryInfo);
+ nodeReader.fetchNodeInfoFromBuffer(nodePos);
+ if (nodeReader.isDeleted() || nodeReader.isBlacklisted() || nodeReader.isNotAWord()) {
+ return NOT_A_PROBABILITY;
+ }
+ return nodeReader.getProbability();
}
int DynamicPatriciaTriePolicy::getShortcutPositionOfNode(
const BinaryDictionaryInfo *const binaryDictionaryInfo,
const int nodePos) const {
- // TODO: Implement.
- return NOT_A_DICT_POS;
+ DynamicPatriciaTrieNodeReader nodeReader(binaryDictionaryInfo);
+ nodeReader.fetchNodeInfoFromBuffer(nodePos);
+ if (nodeReader.isDeleted()) {
+ return NOT_A_DICT_POS;
+ }
+ return nodeReader.getShortcutPos();
}
int DynamicPatriciaTriePolicy::getBigramsPositionOfNode(
const BinaryDictionaryInfo *const binaryDictionaryInfo,
const int nodePos) const {
- // TODO: Implement.
- return NOT_A_DICT_POS;
+ DynamicPatriciaTrieNodeReader nodeReader(binaryDictionaryInfo);
+ nodeReader.fetchNodeInfoFromBuffer(nodePos);
+ if (nodeReader.isDeleted()) {
+ return NOT_A_DICT_POS;
+ }
+ return nodeReader.getBigramsPos();
}
} // namespace latinime