TSTLookup.store/load writes each tree node's split char with DataOutput.writeString, so the two surrogate halves of an emoji or CJK Extension B character come back as U+FFFD: after load(), get() and lookups whose prefix runs through such a character silently find nothing, and some neighbouring BMP keys are lost too (every release since 4.7.0)
Finding LFV-699Area: suggest / ternary search tree suggester persistence (TSTLookup.store and load, writeRecursively and readRecursively)Priority: lowImpact: silent wrong resultsNovelty: new
AI-assisted finding, confirmed by automated checks, not yet triaged by a human Lucene committer. This site files nothing upstream.
What it shows
You build a TSTLookup from ordinary text that contains an emoji or a CJK Extension B character (for example the key 𠮷野家 with weight 5), store() it, load() it back, and expect the reloaded copy to answer like the original, as Lookup.load promises ('load it from a previously saved copy'). Instead lookup('𠮷', true, 5) returns [] instead of [𠮷野家/5] and get('𠮷野家') returns null after the reload, because TSTLookup keeps one tree node per UTF-16 code unit, writeRecursively writes each node's split char with DataOutput.writeString of a one-char String, and a lone surrogate has no UTF-8 form, so it is written and reloaded as U+FFFD and the tree's char comparisons then take the wrong branches. ASCII keys and getCount() survive, while some BMP keys next to a surrogate node are lost too (in the finder's test lookup('Ａ') returns [] instead of [Ａb/4]).
The test shows that TSTLookup.store/load loses a key that contains a supplementary character (U+20BB7): each surrogate split char is written with writeString and comes back as U+FFFD.
Expected vs actual
Expected get(key) on the reloaded lookup to return the weight 5; actual null (expected:<5> but was:<null>), with -ea and -da.
Files
Run it
- Tested on Lucene main e1faeba08bde (2026-11-02): still reproduces.
- Tested on Lucene main 645d4fe14a20 (2026-10-28): reproduced.
- Recorded outcome of run 1 on Lucene main 645d4fe14a20: 1 test(s) run, 1 failed.
- How to read the outcome: a failing test is the finding reproducing (the failure message states expected against actual).
- Seeds: a rerun with the same seed replays the same random choices. Assertions off (-da -dsa -Dtests.asserts=false) is the production-like setting.
# Lucene checkout at the tested commit, built with: ./gradlew --no-daemon assemble (JDK 25)
LUCENE=/path/to/lucene; V=12.0.0-SNAPSHOT
CP=$LUCENE/lucene/core/build/libs/lucene-core-$V.jar
CP=$CP:$LUCENE/lucene/codecs/build/libs/lucene-codecs-$V.jar
CP=$CP:$LUCENE/lucene/test-framework/build/libs/lucene-test-framework-$V.jar
CP=$CP:$LUCENE/lucene/analysis/common/build/libs/lucene-analysis-common-$V.jar
CP=$CP:$LUCENE/lucene/suggest/build/libs/lucene-suggest-$V.jar
CP=$CP:randomizedtesting-runner-2.9.2.jar:junit-4.13.2.jar:hamcrest-3.0.jar # test-framework dependencies (Maven Central)
CP=$CP:junit-platform-console-standalone-6.1.3.jar # JUnit Platform console launcher (Maven Central)
javac -d classes -cp "$CP" $(find src -name '*.java')
# run 1: assertions on, seed DEADBEEF
java -ea -esa --add-modules jdk.incubator.vector --enable-native-access=ALL-UNNAMED -Dtests.seed=DEADBEEF \
-cp "classes:$CP" org.junit.platform.console.ConsoleLauncher \
execute --select-class org.apache.lucene.search.suggest.tst.TestTSTLookupStoreLoadSupplementaryGradle alternative: copy the test into the matching module's src/test/ tree of the checkout and run ./gradlew -p lucene/<module> test --tests <class> -Ptests.seed=<seed> (add -Ptests.asserts=false for a run without assertions).
Feedback
Open an issue in lucene-ai-bug-hunt. Please do not open issues or pull requests in apache/lucene about this finding unless a Lucene committer agrees it should be filed.
Snapshot of 2026-11-04. AI-assisted; not yet triaged by a human Lucene committer. Feedback: open an issue in lucene-ai-bug-hunt.