am ef63947f: Add boundary checking for PtNode Array reading. DO NOT MERGE

* commit 'ef63947f7d10bcbb3237661ad480b430d4208833':
  Add boundary checking for PtNode Array reading. DO NOT MERGE

3 files changed, 21 insertions, 1 deletions

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
index 456352c17..2fa3111d3 100644
--- 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
@@ -26,7 +26,8 @@ namespace latinime {
 void DynamicPatriciaTrieNodeReader::fetchPtNodeInfoFromBufferAndProcessMovedPtNode(
         const int ptNodePos, const int maxCodePointCount, int *const outCodePoints) {
     if (ptNodePos < 0 || ptNodePos >= mBuffer->getTailPosition()) {
-        AKLOGE("Fetching PtNode info form invalid dictionary position: %d, dictionary size: %d",
+        // Reading invalid position because of bug or broken dictionary.
+        AKLOGE("Fetching PtNode info from invalid dictionary position: %d, dictionary size: %d",
                 ptNodePos, mBuffer->getTailPosition());
         ASSERT(false);
         invalidatePtNodeInfo();
diff --git a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_reading_helper.cpp b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_reading_helper.cpp
index f4a2ef389..601ee663b 100644
--- a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_reading_helper.cpp
+++ b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_reading_helper.cpp
@@ -155,6 +155,15 @@ bool DynamicPatriciaTrieReadingHelper::traverseAllPtNodesInPtNodeArrayLevelPreor
 // Read node array size and process empty node arrays. Nodes and arrays are counted up in this
 // method to avoid an infinite loop.
 void DynamicPatriciaTrieReadingHelper::nextPtNodeArray() {
+    if (mReadingState.mPos < 0 || mReadingState.mPos >= mBuffer->getTailPosition()) {
+        // Reading invalid position because of a bug or a broken dictionary.
+        AKLOGE("Reading PtNode array info from invalid dictionary position: %d, dict size: %d",
+                mReadingState.mPos, mBuffer->getTailPosition());
+        ASSERT(false);
+        mIsError = true;
+        mReadingState.mPos = NOT_A_DICT_POS;
+        return;
+    }
     mReadingState.mPosOfLastPtNodeArrayHead = mReadingState.mPos;
     const bool usesAdditionalBuffer = mBuffer->isInAdditionalBuffer(mReadingState.mPos);
     const uint8_t *const dictBuf = mBuffer->getBuffer(usesAdditionalBuffer);
@@ -191,6 +200,15 @@ void DynamicPatriciaTrieReadingHelper::nextPtNodeArray() {
 
 // Follow the forward link and read the next node array if exists.
 void DynamicPatriciaTrieReadingHelper::followForwardLink() {
+    if (mReadingState.mPos < 0 || mReadingState.mPos >= mBuffer->getTailPosition()) {
+        // Reading invalid position because of bug or broken dictionary.
+        AKLOGE("Reading forward link from invalid dictionary position: %d, dict size: %d",
+                mReadingState.mPos, mBuffer->getTailPosition());
+        ASSERT(false);
+        mIsError = true;
+        mReadingState.mPos = NOT_A_DICT_POS;
+        return;
+    }
     const bool usesAdditionalBuffer = mBuffer->isInAdditionalBuffer(mReadingState.mPos);
     const uint8_t *const dictBuf = mBuffer->getBuffer(usesAdditionalBuffer);
     if (usesAdditionalBuffer) {
diff --git a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_reading_helper.h b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_reading_helper.h
index b033eee05..154590fbd 100644
--- a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_reading_helper.h
+++ b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_reading_helper.h
@@ -240,6 +240,7 @@ class DynamicPatriciaTrieReadingHelper {
     static const int MAX_NODE_ARRAY_COUNT_TO_AVOID_INFINITE_LOOP;
     static const size_t MAX_READING_STATE_STACK_SIZE;
 
+    // TODO: Introduce error code to track what caused the error.
     bool mIsError;
     ReadingState mReadingState;
     const BufferWithExtendableBuffer *const mBuffer;