The standalone library is intentionally independent of Paimon core. Paimon integration should be a thin adapter:
GlobalIndexerFactory that delegates to Java FullTextIndexWriter and FullTextIndexReader.Suggested index identifier:
fulltext
Suggested option namespace:
fulltext.tokenizer fulltext.text-fields fulltext.ngram.min-gram fulltext.ngram.max-gram fulltext.ngram.prefix-only fulltext.jieba.search-mode fulltext.jieba.ordinal-position fulltext.lower-case fulltext.max-token-length fulltext.ascii-folding fulltext.stem fulltext.language fulltext.remove-stop-words fulltext.stop-words fulltext.with-position
The standalone library accepts both unprefixed keys and fulltext. prefixed keys.
Query JSON supports match, multi_match, match_phrase, boolean, and boost-demotion queries. match accepts fuzziness, max_expansions, and prefix_length; use fuzziness: "auto" for automatic edit distance. If match.column is omitted, the native reader searches every text field configured in the index, so Paimon can derive native text fields from its extra-fields mechanism without adding a user-facing multi_match requirement.