FixedShingleFilter takes the gap before each token of a graph route from that token's stream position increment, so tokens stacked right after a hole (a removed stop word followed by a preserved original, a stemmed variant or a synonym) get shingles that skip the filler ('at café' for 'at the café') while the correct ones ('at _ café') are missing, and a token after a multi-position token gets a filler it should not have ('login _' instead of 'login page'); released since 8.0.0
Finding LFV-781Area: analysis-common / shingle (FixedShingleFilter, FixedShingleFilterFactory) over core analysis GraphTokenFilter (route walk)Priority: mediumImpact: 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 run 'coffee at the café' through WhitespaceTokenizer, StopFilter{the}, ASCIIFoldingFilter(preserveOriginal=true) and FixedShingleFilter(2), and expect the filler that stands for the removed 'the' in every shingle that crosses it, as the folded 'cafe' gets it ('at _'), because the class javadoc says the filter 'handles position increments > 1 by inserting filler tokens' and 'will output stacked shingles' for stacked tokens. Instead the original 'café', stacked behind 'cafe' with position increment 0, is joined straight to 'at': you get [coffee at, at _, at café], at size 3 'coffee at café' appears and 'at _ café' is missing, and on a canned stream a(+1) b(+2) c(+0) swapping only the stream order of the two stacked tokens changes the output from [a _, a c] to [a _, a b]. The same misreading gives 'login _' instead of 'login page' after a two-position token 'login' over 'log' plus a removed 'in', and nothing throws, with assertions on or off.
FixedShingleFilter(2) on a(+1) b(+2) c(+0) joins a and c across the hole, because the stacked c keeps its stream increment 0 instead of the route gap.
Expected vs actual
Expected no shingle "a c" (c is two positions after a), actual [a _, a c].
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: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.shingle.TestFixedShingleFilterStackedAfterHoleGradle 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.