am b85bf4eb: Merge "Use word id to construct DicNode instead of isTerminal flag."

* commit 'b85bf4ebb80ad471b1f61ea92d971bbeffb77872':
  Use word id to construct DicNode instead of isTerminal flag.

native/jni/src/defines.h

@@ -299,6 +299,7 @@ static inline void prof_out(void) {
 #define NOT_AN_INDEX (-1)
 #define NOT_A_PROBABILITY (-1)
 #define NOT_A_DICT_POS (S_INT_MIN)
+#define NOT_A_WORD_ID (S_INT_MIN)
 #define NOT_A_TIMESTAMP (-1)
 #define NOT_A_LANGUAGE_WEIGHT (-1.0f)
 

native/jni/src/suggest/core/dicnode/dic_node.h

@@ -136,7 +136,7 @@ class DicNode {
     }
 
     void initAsChild(const DicNode *const dicNode, const int ptNodePos,
-            const int childrenPtNodeArrayPos, const int probability, const bool isTerminal,
+            const int childrenPtNodeArrayPos, const int probability, const int wordId,
             const bool hasChildren, const bool isBlacklistedOrNotAWord,
             const uint16_t mergedNodeCodePointCount, const int *const mergedNodeCodePoints) {
         uint16_t newDepth = static_cast<uint16_t>(dicNode->getNodeCodePointCount() + 1);
@@ -144,7 +144,7 @@ class DicNode {
         const uint16_t newLeavingDepth = static_cast<uint16_t>(
                 dicNode->mDicNodeProperties.getLeavingDepth() + mergedNodeCodePointCount);
         mDicNodeProperties.init(ptNodePos, childrenPtNodeArrayPos, mergedNodeCodePoints[0],
-                probability, isTerminal, hasChildren, isBlacklistedOrNotAWord, newDepth,
+                probability, wordId, hasChildren, isBlacklistedOrNotAWord, newDepth,
                 newLeavingDepth, dicNode->mDicNodeProperties.getPrevWordsTerminalPtNodePos());
         mDicNodeState.init(&dicNode->mDicNodeState, mergedNodeCodePointCount,
                 mergedNodeCodePoints);

native/jni/src/suggest/core/dicnode/dic_node_vector.h

@@ -59,13 +59,13 @@ class DicNodeVector {
     }
 
     void pushLeavingChild(const DicNode *const dicNode, const int ptNodePos,
-            const int childrenPtNodeArrayPos, const int probability, const bool isTerminal,
+            const int childrenPtNodeArrayPos, const int probability, const int wordId,
             const bool hasChildren, const bool isBlacklistedOrNotAWord,
             const uint16_t mergedNodeCodePointCount, const int *const mergedNodeCodePoints) {
         ASSERT(!mLock);
         mDicNodes.emplace_back();
         mDicNodes.back().initAsChild(dicNode, ptNodePos, childrenPtNodeArrayPos, probability,
-                isTerminal, hasChildren, isBlacklistedOrNotAWord, mergedNodeCodePointCount,
+                wordId, hasChildren, isBlacklistedOrNotAWord, mergedNodeCodePointCount,
                 mergedNodeCodePoints);
     }
 

native/jni/src/suggest/core/dicnode/internal/dic_node_properties.h

@@ -31,20 +31,20 @@ class DicNodeProperties {
     AK_FORCE_INLINE DicNodeProperties()
             : mPtNodePos(NOT_A_DICT_POS), mChildrenPtNodeArrayPos(NOT_A_DICT_POS),
               mProbability(NOT_A_PROBABILITY), mDicNodeCodePoint(NOT_A_CODE_POINT),
-              mIsTerminal(false), mHasChildrenPtNodes(false),
+              mWordId(NOT_A_WORD_ID), mHasChildrenPtNodes(false),
               mIsBlacklistedOrNotAWord(false), mDepth(0), mLeavingDepth(0) {}
 
     ~DicNodeProperties() {}
 
     // Should be called only once per DicNode is initialized.
     void init(const int pos, const int childrenPos, const int nodeCodePoint, const int probability,
-            const bool isTerminal, const bool hasChildren, const bool isBlacklistedOrNotAWord,
+            const int wordId, const bool hasChildren, const bool isBlacklistedOrNotAWord,
             const uint16_t depth, const uint16_t leavingDepth, const int *const prevWordsNodePos) {
         mPtNodePos = pos;
         mChildrenPtNodeArrayPos = childrenPos;
         mDicNodeCodePoint = nodeCodePoint;
         mProbability = probability;
-        mIsTerminal = isTerminal;
+        mWordId = wordId;
         mHasChildrenPtNodes = hasChildren;
         mIsBlacklistedOrNotAWord = isBlacklistedOrNotAWord;
         mDepth = depth;
@@ -58,7 +58,7 @@ class DicNodeProperties {
         mChildrenPtNodeArrayPos = rootPtNodeArrayPos;
         mDicNodeCodePoint = NOT_A_CODE_POINT;
         mProbability = NOT_A_PROBABILITY;
-        mIsTerminal = false;
+        mWordId = NOT_A_WORD_ID;
         mHasChildrenPtNodes = true;
         mIsBlacklistedOrNotAWord = false;
         mDepth = 0;
@@ -71,7 +71,7 @@ class DicNodeProperties {
         mChildrenPtNodeArrayPos = dicNodeProp->mChildrenPtNodeArrayPos;
         mDicNodeCodePoint = dicNodeProp->mDicNodeCodePoint;
         mProbability = dicNodeProp->mProbability;
-        mIsTerminal = dicNodeProp->mIsTerminal;
+        mWordId = dicNodeProp->mWordId;
         mHasChildrenPtNodes = dicNodeProp->mHasChildrenPtNodes;
         mIsBlacklistedOrNotAWord = dicNodeProp->mIsBlacklistedOrNotAWord;
         mDepth = dicNodeProp->mDepth;
@@ -86,7 +86,7 @@ class DicNodeProperties {
         mChildrenPtNodeArrayPos = dicNodeProp->mChildrenPtNodeArrayPos;
         mDicNodeCodePoint = codePoint; // Overwrite the node char of a passing child
         mProbability = dicNodeProp->mProbability;
-        mIsTerminal = dicNodeProp->mIsTerminal;
+        mWordId = dicNodeProp->mWordId;
         mHasChildrenPtNodes = dicNodeProp->mHasChildrenPtNodes;
         mIsBlacklistedOrNotAWord = dicNodeProp->mIsBlacklistedOrNotAWord;
         mDepth = dicNodeProp->mDepth + 1; // Increment the depth of a passing child
@@ -121,7 +121,7 @@ class DicNodeProperties {
     }
 
     bool isTerminal() const {
-        return mIsTerminal;
+        return mWordId != NOT_A_WORD_ID;
     }
 
     bool hasChildren() const {
@@ -144,7 +144,7 @@ class DicNodeProperties {
     int mChildrenPtNodeArrayPos;
     int mProbability;
     int mDicNodeCodePoint;
-    bool mIsTerminal;
+    int mWordId;
     bool mHasChildrenPtNodes;
     bool mIsBlacklistedOrNotAWord;
     uint16_t mDepth;

native/jni/src/suggest/core/policy/dictionary_structure_with_buffer_policy.h

@@ -36,6 +36,7 @@ class UnigramProperty;
  * This class abstracts the structure of dictionaries.
  * Implement this policy to support additional dictionaries.
  */
+// TODO: Use word id instead of terminal PtNode position.
 class DictionaryStructureWithBufferPolicy {
  public:
     typedef std::unique_ptr<DictionaryStructureWithBufferPolicy> StructurePolicyPtr;

native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_policy.cpp

@@ -76,8 +76,9 @@ void Ver4PatriciaTriePolicy::createAndGetAllChildDicNodes(const DicNode *const d
             // Skip PtNodes that represent non-word information.
             continue;
         }
+        const int wordId = isTerminal ? ptNodeParams.getHeadPos() : NOT_A_WORD_ID;
         childDicNodes->pushLeavingChild(dicNode, ptNodeParams.getHeadPos(),
-                ptNodeParams.getChildrenPos(), ptNodeParams.getProbability(), isTerminal,
+                ptNodeParams.getChildrenPos(), ptNodeParams.getProbability(), wordId,
                 ptNodeParams.hasChildren(),
                 ptNodeParams.isBlacklisted()
                         || ptNodeParams.isNotAWord() /* isBlacklistedOrNotAWord */,

native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_policy.h

@@ -55,6 +55,7 @@ class DicNodeVector;
 namespace backward {
 namespace v402 {
 
+// Word id = Position of a PtNode that represents the word.
 class Ver4PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy {
  public:
     Ver4PatriciaTriePolicy(Ver4DictBuffers::Ver4DictBuffersPtr buffers)

native/jni/src/suggest/policyimpl/dictionary/structure/v2/patricia_trie_policy.cpp

@@ -367,8 +367,8 @@ int PatriciaTriePolicy::createAndGetLeavingChildNode(const DicNode *const dicNod
             &probability, &childrenPos, &shortcutPos, &bigramPos, &siblingPos);
     // Skip PtNodes don't start with Unicode code point because they represent non-word information.
     if (CharUtils::isInUnicodeSpace(mergedNodeCodePoints[0])) {
-        childDicNodes->pushLeavingChild(dicNode, ptNodePos, childrenPos, probability,
+        const int wordId = PatriciaTrieReadingUtils::isTerminal(flags) ? ptNodePos : NOT_A_WORD_ID;
-                PatriciaTrieReadingUtils::isTerminal(flags),
+        childDicNodes->pushLeavingChild(dicNode, ptNodePos, childrenPos, probability, wordId,
                 PatriciaTrieReadingUtils::hasChildrenInFlags(flags),
                 PatriciaTrieReadingUtils::isBlacklisted(flags)
                         || PatriciaTrieReadingUtils::isNotAWord(flags),

native/jni/src/suggest/policyimpl/dictionary/structure/v2/patricia_trie_policy.h

@@ -36,6 +36,7 @@ namespace latinime {
 class DicNode;
 class DicNodeVector;
 
+// Word id = Position of a PtNode that represents the word.
 class PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy {
  public:
     PatriciaTriePolicy(MmappedBuffer::MmappedBufferPtr mmappedBuffer)

native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_policy.cpp

@@ -66,8 +66,9 @@ void Ver4PatriciaTriePolicy::createAndGetAllChildDicNodes(const DicNode *const d
             // Skip PtNodes that represent non-word information.
             continue;
         }
+        const int wordId = isTerminal ? ptNodeParams.getTerminalId() : NOT_A_WORD_ID;
         childDicNodes->pushLeavingChild(dicNode, ptNodeParams.getHeadPos(),
-                ptNodeParams.getChildrenPos(), ptNodeParams.getProbability(), isTerminal,
+                ptNodeParams.getChildrenPos(), ptNodeParams.getProbability(), wordId,
                 ptNodeParams.hasChildren(),
                 ptNodeParams.isBlacklisted()
                         || ptNodeParams.isNotAWord() /* isBlacklistedOrNotAWord */,

native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_policy.h

@@ -36,6 +36,7 @@ namespace latinime {
 class DicNode;
 class DicNodeVector;
 
+// Word id = Artificial id that is stored in the PtNode looked up by the word.
 class Ver4PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy {
  public:
     Ver4PatriciaTriePolicy(Ver4DictBuffers::Ver4DictBuffersPtr buffers)