We just released new versions of the experimental highlighter that highlights the 150 million searches we handle every day at the Wikimedia Foundation. The new versions are: 1.3.0 for Elasticsearch 1.3, 1.4.2 for Elasticsearch 1.4.X, and 1.5.0 for Elasticsearch 1.5. This is not an elastic.co supported plugin.
What's new:
- Fixed and error caused by single term phrase queries. They come up pretty rarely but it wasn't cool that we were throwing an ArrayIndexOutOfBoundsException on them.
- Fix the merge rules for overlapping matches so the widest match wins. It used to take the skinniest match.
- Properly handle function_score queries. They were being ignored before, now their contents are considered for terms to highlight.
- Handle common_terms queries in a more intuitive way. Before this release all terms in the common_terms query were highlighted. After it only the uncommon terms are highlighted. The old behavior is available using a setting.
- Add max_expanded_terms option. Some multi-term queries are handled without expansion by the highlighter but for the ones that aren't we were hard coding the number of expanded terms to 100. We've made that an option and changed made the default 1024 which is what the fast vector highlighter uses so its probably safe.
- Elasticsearch 1.5 support.