Revision history for Perl extension Search::Indexer.
1.01 01.07.2021
- bug fix : CORE::fc did not compile with perl versions < 5.016
1.0 29.06.2021
- new algorithm (Okapi bm25) for relevancy scores, like ElasticSearch or SQLite FTS5
- default tokenizer now relies on unicode \p{Word}
- use BerkeleyDB "partials" for more efficient indexing
- general revision of the source code and documentation
- upgrade tests to use Test2::V0
INDEXES ARE NOT COMPATIBLE WITH PREVIOUS VERSION, you need to reindex your docs.
0.78 11.02.2016
- suppress spurious warnings in Indexer.pm
0.77 04.05.2013
- fix warning in dump() & identify stopwords
0.76 26.06.2009
- suppress spurious warning in addToScore
0.75 24.09.2007
- bug fix: remove() did not work on indexes without word positions
0.74 11.03.2007
- bug fix in checking "positions" option
0.73 11.03.2007
- added option "positions => 0" for smaller (yet less precise) indexes
0.72
- fixed regex bug (missing "quotemeta")
0.71 25.05.2005
- Internal structure change : use Btree instead of Hash for
ixw and ixp. This takes more space but allows us to walk
through words in the dictionary, or through words in a
given doc. INDEXES ARE NOT COMPATIBLE WITH PREVIOUS
VERSION, you need to reindex your docs.
- add" method : check that $docId is a fresh number
- "remove" method : no longer necessary to supply text buffer;
just give the docId to remove
- new method "words" and "wordIds"
- modified "translateQuery" for compatibility with
Search::QueryParser v0.91
0.70 Feb 6th. First public release.