Implement GC for ver4 dict contents.

Bug: 11073222
Change-Id: Ia4dff07c7f4576897e0d3fc4840aed0ccb535299

native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/bigram_dict_content.cpp

@@ -72,4 +72,66 @@ bool BigramDictContent::copyBigramList(const int bigramListPos, const int toPos)
     return true;
 }
 
+bool BigramDictContent::runGC(const TerminalPositionLookupTable::TerminalIdMap *const terminalIdMap,
+        const BigramDictContent *const originalBigramDictContent,
+        int *const outBigramEntryCount) {
+    for (TerminalPositionLookupTable::TerminalIdMap::const_iterator it = terminalIdMap->begin();
+            it != terminalIdMap->end(); ++it) {
+        const int originalBigramListPos =
+                originalBigramDictContent->getBigramListHeadPos(it->first);
+        if (originalBigramListPos == NOT_A_DICT_POS) {
+            // This terminal does not have a bigram list.
+            continue;
+        }
+        const int bigramListPos = getContentBuffer()->getTailPosition();
+        int bigramEntryCount = 0;
+        // Copy bigram list with GC from original content.
+        if (!runGCBigramList(originalBigramListPos, originalBigramDictContent, bigramListPos,
+                terminalIdMap, &bigramEntryCount)) {
+            return false;
+        }
+        if (bigramEntryCount == 0) {
+            // All bigram entries are useless. This terminal does not have a bigram list.
+            continue;
+        }
+        *outBigramEntryCount += bigramEntryCount;
+        // Set bigram list position to the lookup table.
+        if (!getUpdatableAddressLookupTable()->set(it->second, bigramListPos)) {
+            return false;
+        }
+    }
+    return true;
+}
+
+bool BigramDictContent::runGCBigramList(const int bigramListPos,
+        const BigramDictContent *const sourceBigramDictContent, const int toPos,
+        const TerminalPositionLookupTable::TerminalIdMap *const terminalIdMap,
+        int *const outEntrycount) {
+    bool hasNext = true;
+    int readingPos = bigramListPos;
+    int writingPos = toPos;
+    while (hasNext) {
+        int probability = NOT_A_PROBABILITY;
+        int targetTerminalId = Ver4DictConstants::NOT_A_TERMINAL_ID;
+        sourceBigramDictContent->getBigramEntryAndAdvancePosition(&probability, &hasNext,
+                &targetTerminalId, &readingPos);
+        if (targetTerminalId == Ver4DictConstants::NOT_A_TERMINAL_ID) {
+            continue;
+        }
+        TerminalPositionLookupTable::TerminalIdMap::const_iterator it =
+                terminalIdMap->find(targetTerminalId);
+        if (it == terminalIdMap->end()) {
+            AKLOGE("terminal Id %d is not in the terminal position map. map size: %zd",
+                    targetTerminalId, terminalIdMap->size());
+            return false;
+        }
+        if (!writeBigramEntryAndAdvancePosition(probability, hasNext, it->second,
+                &writingPos)) {
+            return false;
+        }
+        *outEntrycount += 1;
+    }
+    return true;
+}
+
 } // namespace latinime

native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/bigram_dict_content.h

@@ -19,6 +19,7 @@
 
 #include "defines.h"
 #include "suggest/policyimpl/dictionary/structure/v4/content/sparse_table_dict_content.h"
+#include "suggest/policyimpl/dictionary/structure/v4/content/terminal_position_lookup_table.h"
 #include "suggest/policyimpl/dictionary/structure/v4/ver4_dict_constants.h"
 
 namespace latinime {
@@ -65,6 +66,10 @@ class BigramDictContent : public SparseTableDictContent {
                 Ver4DictConstants::BIGRAM_FILE_EXTENSION);
     }
 
+    bool runGC(const TerminalPositionLookupTable::TerminalIdMap *const terminalIdMap,
+            const BigramDictContent *const originalBigramDictContent,
+            int *const outBigramEntryCount);
+
  private:
     DISALLOW_COPY_AND_ASSIGN(BigramDictContent);
 
@@ -72,6 +77,11 @@ class BigramDictContent : public SparseTableDictContent {
         return (probability & Ver4DictConstants::BIGRAM_PROBABILITY_MASK)
                 | (hasNext ? Ver4DictConstants::BIGRAM_HAS_NEXT_MASK : 0);
     }
+
+    bool runGCBigramList(const int bigramListPos,
+            const BigramDictContent *const sourceBigramDictContent, const int toPos,
+            const TerminalPositionLookupTable::TerminalIdMap *const terminalIdMap,
+            int *const outEntryCount);
 };
 } // namespace latinime
 #endif /* LATINIME_BIGRAM_DICT_CONTENT_H */

native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/probability_dict_content.h

@@ -19,6 +19,7 @@
 
 #include "defines.h"
 #include "suggest/policyimpl/dictionary/structure/v4/content/single_dict_content.h"
+#include "suggest/policyimpl/dictionary/structure/v4/content/terminal_position_lookup_table.h"
 #include "suggest/policyimpl/dictionary/structure/v4/ver4_dict_constants.h"
 #include "suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_reading_utils.h"
 #include "suggest/policyimpl/dictionary/utils/buffer_with_extendable_buffer.h"
@@ -28,13 +29,14 @@ namespace latinime {
 class ProbabilityDictContent : public SingleDictContent {
  public:
     ProbabilityDictContent(const char *const dictDirPath, const bool isUpdatable)
-            : SingleDictContent(dictDirPath, Ver4DictConstants::FREQ_FILE_EXTENSION,
+            : SingleDictContent(dictDirPath, Ver4DictConstants::FREQ_FILE_EXTENSION, isUpdatable),
-                    isUpdatable) {}
+              mSize(getBuffer()->getTailPosition() / (Ver4DictConstants::PROBABILITY_SIZE
+                      + Ver4DictConstants::FLAGS_IN_PROBABILITY_FILE_SIZE)) {}
 
-    ProbabilityDictContent() {}
+    ProbabilityDictContent() : mSize(0) {}
 
     int getProbability(const int terminalId) const {
-        if (terminalId < 0 || terminalId >= getSize()) {
+        if (terminalId < 0 || terminalId >= mSize) {
             return NOT_A_PROBABILITY;
         }
         return Ver4PatriciaTrieReadingUtils::getProbability(getBuffer(), terminalId);
@@ -44,7 +46,7 @@ class ProbabilityDictContent : public SingleDictContent {
         if (terminalId < 0) {
             return false;
         }
-        if (terminalId >= getSize()) {
+        if (terminalId >= mSize) {
             // Write new entry.
             int writingPos = getBuffer()->getTailPosition();
             while (writingPos <= getEntryPos(terminalId)) {
@@ -58,6 +60,7 @@ class ProbabilityDictContent : public SingleDictContent {
                         Ver4DictConstants::PROBABILITY_SIZE, &writingPos)) {
                     return false;
                 }
+                mSize++;
             }
         }
         const int probabilityWritingPos = getEntryPos(terminalId)
@@ -67,8 +70,33 @@ class ProbabilityDictContent : public SingleDictContent {
     }
 
     bool flushToFile(const char *const dictDirPath) const {
+        if (getEntryPos(mSize) < getBuffer()->getTailPosition()) {
+            ProbabilityDictContent probabilityDictContentToWrite;
+            for (int i = 0; i < mSize; ++i) {
+                if (!probabilityDictContentToWrite.setProbability(i, getProbability(i))) {
+                    return false;
+                }
+            }
+            return probabilityDictContentToWrite.flush(dictDirPath,
+                    Ver4DictConstants::FREQ_FILE_EXTENSION);
+        } else {
             return flush(dictDirPath, Ver4DictConstants::FREQ_FILE_EXTENSION);
         }
+    }
+
+    bool runGC(const TerminalPositionLookupTable::TerminalIdMap *const terminalIdMap,
+            const ProbabilityDictContent *const originalProbabilityDictContent) {
+        mSize = 0;
+        for (TerminalPositionLookupTable::TerminalIdMap::const_iterator it = terminalIdMap->begin();
+                it != terminalIdMap->end(); ++it) {
+            if (!setProbability(it->second,
+                    originalProbabilityDictContent->getProbability(it->first))) {
+                return false;
+            }
+            mSize++;
+        }
+        return true;
+    }
 
  private:
     DISALLOW_COPY_AND_ASSIGN(ProbabilityDictContent);
@@ -78,10 +106,7 @@ class ProbabilityDictContent : public SingleDictContent {
                 + Ver4DictConstants::PROBABILITY_SIZE);
     }
 
-    int getSize() const {
+    int mSize;
-        return getBuffer()->getTailPosition() / (Ver4DictConstants::PROBABILITY_SIZE
-                + Ver4DictConstants::FLAGS_IN_PROBABILITY_FILE_SIZE);
-    }
 };
 } // namespace latinime
 #endif /* LATINIME_PROBABILITY_DICT_CONTENT_H */

native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/shortcut_dict_content.cpp

@@ -48,6 +48,30 @@ bool ShortcutDictContent::flushToFile(const char *const dictDirPath) const {
             Ver4DictConstants::SHORTCUT_FILE_EXTENSION);
 }
 
+bool ShortcutDictContent::runGC(
+        const TerminalPositionLookupTable::TerminalIdMap *const terminalIdMap,
+        const ShortcutDictContent *const originalShortcutDictContent) {
+   for (TerminalPositionLookupTable::TerminalIdMap::const_iterator it = terminalIdMap->begin();
+           it != terminalIdMap->end(); ++it) {
+       const int originalShortcutListPos =
+               originalShortcutDictContent->getShortcutListHeadPos(it->first);
+       if (originalShortcutListPos == NOT_A_DICT_POS) {
+           continue;
+       }
+       const int shortcutListPos = getContentBuffer()->getTailPosition();
+       // Copy shortcut list with GC from original content.
+       if (!copyShortcutList(originalShortcutListPos, originalShortcutDictContent,
+               shortcutListPos)) {
+           return false;
+       }
+       // Set shortcut list position to the lookup table.
+       if (!getUpdatableAddressLookupTable()->set(it->second, shortcutListPos)) {
+           return false;
+       }
+   }
+   return true;
+}
+
 bool ShortcutDictContent::copyShortcutList(const int shortcutListPos,
         const ShortcutDictContent *const sourceShortcutDictContent, const int toPos) {
     bool hasNext = true;

native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/shortcut_dict_content.h

@@ -47,6 +47,9 @@ class ShortcutDictContent : public SparseTableDictContent {
 
    bool flushToFile(const char *const dictDirPath) const;
 
+   bool runGC(const TerminalPositionLookupTable::TerminalIdMap *const terminalIdMap,
+           const ShortcutDictContent *const originalShortcutDictContent);
+
  private:
     DISALLOW_COPY_AND_ASSIGN(ShortcutDictContent);
 

native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/sparse_table_dict_content.cpp

@@ -18,22 +18,6 @@
 
 namespace latinime {
 
-bool SparseTableDictContent::copyContent(
-        const SparseTableDictContent *const sparseTableDictContent) {
-    if (!mExpandableLookupTableBuffer.copy(
-            &sparseTableDictContent->mExpandableLookupTableBuffer)) {
-        return false;
-    }
-    if (!mExpandableAddressTableBuffer.copy(
-            &sparseTableDictContent->mExpandableAddressTableBuffer)) {
-        return false;
-    }
-    if (!mExpandableContentBuffer.copy(&sparseTableDictContent->mExpandableContentBuffer)) {
-        return false;
-    }
-    return true;
-}
-
 bool SparseTableDictContent::flush(const char *const dictDirPath,
         const char *const lookupTableFileName, const char *const addressTableFileName,
         const char *const contentFileName) const {

native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/sparse_table_dict_content.h

@@ -75,8 +75,6 @@ class SparseTableDictContent : public DictContent {
                 || mExpandableContentBuffer.isNearSizeLimit();
     }
 
-    bool copyContent(const SparseTableDictContent *const sparseTableDictContent);
-
  protected:
     SparseTable *getUpdatableAddressLookupTable() {
         return &mAddressLookupTable;

native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_node_writer.cpp

@@ -126,6 +126,12 @@ bool Ver4PatriciaTrieNodeWriter::updateChildrenPosition(
             newChildrenPosition, &childrenPosFieldPos);
 }
 
+bool Ver4PatriciaTrieNodeWriter::updateTerminalId(const PtNodeParams *const toBeUpdatedPtNodeParams,
+        const int newTerminalId) {
+    return mTrieBuffer->writeUint(newTerminalId, Ver4DictConstants::TERMINAL_ID_FIELD_SIZE,
+            toBeUpdatedPtNodeParams->getTerminalIdFieldPos());
+}
+
 bool Ver4PatriciaTrieNodeWriter::writePtNodeAndAdvancePosition(
         const PtNodeParams *const ptNodeParams, int *const ptNodeWritingPos) {
     const int nodePos = *ptNodeWritingPos;

native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_node_writer.h

@@ -57,6 +57,9 @@ class Ver4PatriciaTrieNodeWriter : public PtNodeWriter {
     virtual bool updateChildrenPosition(const PtNodeParams *const toBeUpdatedPtNodeParams,
             const int newChildrenPosition);
 
+    bool updateTerminalId(const PtNodeParams *const toBeUpdatedPtNodeParams,
+            const int newTerminalId);
+
     virtual bool writePtNodeAndAdvancePosition(const PtNodeParams *const ptNodeParams,
             int *const ptNodeWritingPos);
 

native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_writing_helper.cpp

@@ -20,7 +20,6 @@
 
 #include "suggest/policyimpl/dictionary/bigram/ver4_bigram_list_policy.h"
 #include "suggest/policyimpl/dictionary/header/header_policy.h"
-#include "suggest/policyimpl/dictionary/structure/v3/dynamic_patricia_trie_gc_event_listeners.h"
 #include "suggest/policyimpl/dictionary/shortcut/ver4_shortcut_list_policy.h"
 #include "suggest/policyimpl/dictionary/structure/v4/ver4_dict_buffers.h"
 #include "suggest/policyimpl/dictionary/structure/v4/ver4_dict_constants.h"
@@ -141,15 +140,6 @@ bool Ver4PatriciaTrieWritingHelper::runGC(const int rootPtNodeArrayPos,
     Ver4PatriciaTrieNodeWriter newPtNodeWriter(buffersToWrite->getWritableTrieBuffer(),
             buffersToWrite, &newPtNodeReader, &newBigramPolicy, &newShortcutPolicy);
 
-    if(!buffersToWrite->getUpdatableBigramDictContent()->copyContent(
-            mBuffers->getBigramDictContent())) {
-        return false;
-    }
-    if(!buffersToWrite->getUpdatableShortcutDictContent()->copyContent(
-            mBuffers->getShortcutDictContent())) {
-        return false;
-    }
-
     DynamicPatriciaTrieReadingHelper newDictReadingHelper(buffersToWrite->getTrieBuffer(),
             &newPtNodeReader);
     newDictReadingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
@@ -160,11 +150,50 @@ bool Ver4PatriciaTrieWritingHelper::runGC(const int rootPtNodeArrayPos,
         return false;
     }
 
-    // TODO: GC for dict contents.
+    // Re-assign terminal IDs for valid terminal PtNodes.
-
+    TerminalPositionLookupTable::TerminalIdMap terminalIdMap;
+    if(!buffersToWrite->getUpdatableTerminalPositionLookupTable()->runGCTerminalIds(
+            &terminalIdMap)) {
+        return false;
+    }
+    TraversePolicyToUpdateAllTerminalIds traversePolicyToUpdateAllTerminalIds(&newPtNodeWriter,
+            &terminalIdMap);
+    if (!newDictReadingHelper.traverseAllPtNodesInPostorderDepthFirstManner(
+            &traversePolicyToUpdateAllTerminalIds)) {
+        return false;
+    }
+    // Run GC for probability dict content.
+    if (!buffersToWrite->getUpdatableProbabilityDictContent()->runGC(&terminalIdMap,
+            mBuffers->getProbabilityDictContent())) {
+        return false;
+    }
+    // Run GC for bigram dict content.
+    if(!buffersToWrite->getUpdatableBigramDictContent()->runGC(&terminalIdMap,
+            mBuffers->getBigramDictContent(), outBigramCount)) {
+        return false;
+    }
+    // Run GC for shortcut dict content.
+    if(!buffersToWrite->getUpdatableShortcutDictContent()->runGC(&terminalIdMap,
+            mBuffers->getShortcutDictContent())) {
+        return false;
+    }
     *outUnigramCount = traversePolicyToUpdateAllPositionFields.getUnigramCount();
-    *outBigramCount = traversePolicyToUpdateAllPositionFields.getBigramCount();
     return true;
 }
 
+bool Ver4PatriciaTrieWritingHelper::TraversePolicyToUpdateAllTerminalIds::onVisitingPtNode(
+        const PtNodeParams *const ptNodeParams) {
+    if (!ptNodeParams->isTerminal()) {
+        return true;
+    }
+    TerminalPositionLookupTable::TerminalIdMap::const_iterator it =
+            mTerminalIdMap->find(ptNodeParams->getTerminalId());
+    if (it == mTerminalIdMap->end()) {
+        AKLOGE("terminal Id %d is not in the terminal position map. map size: %zd",
+                ptNodeParams->getTerminalId(), mTerminalIdMap->size());
+        return false;
+    }
+    return mPtNodeWriter->updateTerminalId(ptNodeParams, it->second);
+}
+
 } // namespace latinime

native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_writing_helper.h

@@ -18,11 +18,14 @@
 #define LATINIME_VER4_PATRICIA_TRIE_WRITING_HELPER_H
 
 #include "defines.h"
+#include "suggest/policyimpl/dictionary/structure/v3/dynamic_patricia_trie_gc_event_listeners.h"
+#include "suggest/policyimpl/dictionary/structure/v4/content/terminal_position_lookup_table.h"
 
 namespace latinime {
 
 class HeaderPolicy;
 class Ver4DictBuffers;
+class Ver4PatriciaTrieNodeWriter;
 
 class Ver4PatriciaTrieWritingHelper {
  public:
@@ -39,6 +42,28 @@ class Ver4PatriciaTrieWritingHelper {
  private:
     DISALLOW_IMPLICIT_CONSTRUCTORS(Ver4PatriciaTrieWritingHelper);
 
+    class TraversePolicyToUpdateAllTerminalIds
+            : public DynamicPatriciaTrieReadingHelper::TraversingEventListener {
+     public:
+          TraversePolicyToUpdateAllTerminalIds(Ver4PatriciaTrieNodeWriter *const ptNodeWriter,
+                const TerminalPositionLookupTable::TerminalIdMap *const terminalIdMap)
+                : mPtNodeWriter(ptNodeWriter), mTerminalIdMap(terminalIdMap) {};
+
+        bool onAscend() { return true; }
+
+        bool onDescend(const int ptNodeArrayPos) { return true; }
+
+        bool onReadingPtNodeArrayTail() { return true; }
+
+        bool onVisitingPtNode(const PtNodeParams *const ptNodeParams);
+
+     private:
+        DISALLOW_IMPLICIT_CONSTRUCTORS(TraversePolicyToUpdateAllTerminalIds);
+
+        Ver4PatriciaTrieNodeWriter *const mPtNodeWriter;
+        const TerminalPositionLookupTable::TerminalIdMap *const mTerminalIdMap;
+    };
+
     bool runGC(const int rootPtNodeArrayPos, const HeaderPolicy *const headerPolicy,
             Ver4DictBuffers *const buffersToWrite, int *const outUnigramCount,
             int *const outBigramCount, const bool needsToDecay);