Stempel's Trie.reduce with Gener or Lift rewrites the cells of the trie it is given, so reducing PolishAnalyzer.getDefaultTable(), documented as unmodifiable, silently changes the stems of every later PolishAnalyzer and StempelPolishStemFilterFactory in the JVM

Finding LFV-696Area: analysis/stempel / egothor stemmer trie reductions (Trie.reduce with Lift and Gener; the shared PolishAnalyzer.getDefaultTable() table)Priority: lowImpact: silent wrong resultsNovelty: new

Finding page for LFV-696

AI-assisted finding, confirmed by automated checks, not yet triaged by a human Lucene committer. This site files nothing upstream.

What it shows

You call reduce(new Gener()) or reduce(new Lift(..)) on a stempel Trie, for example on PolishAnalyzer.getDefaultTable(), which is documented as 'an unmodifiable instance of the default stemmer table', and expect what Trie.reduce's javadoc says, 'the newly reduced Trie', with the input left as it was (Optimizer and Optimizer2 behave so). Instead Lift.optimize and Gener.optimize take the input's own row list and rewrite its cells in place (Lift moves commands up a level and clears references, Gener drops infrequent commands and references) before copying them into the result, so the input answers differently afterwards: after PolishAnalyzer.getDefaultTable().reduce(new Gener()) every new PolishAnalyzer stems 'książkami rzekami pracowali' to [książk, rzekami, pracowali] instead of [książek, rzekść, pracować], and a one-key trie {aa -> Da} reduced with Lift(false) starts to answer getLastOnPath('a') = Da instead of nothing. Copying the rows before the rewrite fixes it; upstream TestStemmer, which reduces one trie five times, then needs a fresh trie per reduction.

The test shows that Trie.reduce(new Lift(false)) on a one-key MultiTrie2 {aa -> Da} built by the caller rewrites the input trie in place (Lift.optimize works on orig.rows), although reduce is documented to return "the newly reduced Trie"; Gener has the same root cause.

Expected vs actual

Expected: the input still answers getLastOnPath("a") = "" after the reduction. Actual: "Da" (expected:<[]> but was:<[Da]>, with -ea and -da).

Files

Run it

# 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:$LUCENE/lucene/analysis/stempel/build/libs/lucene-analysis-stempel-$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.analysis.stempel.TestTrieReduceKeepsInput

Gradle 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.