Search::Fulltext - Fulltext search module


    use Search::Fulltext;
    my @docs = (
        'I like beer the best',
        'Wine makes people saticefied',  # does not include beer
        'Beer makes people happy',
    my $fts = Search::Fulltext->new({
        docs => \@docs,
    my $results = $fts->search('beer');
    is_deeply($results, [0, 2]);         # 1st & 3rd doc include 'beer'
    my $results = $fts->search('beer AND happy');
    is_deeply($results, [2]);            # 3rd doc includes both 'beer' & 'happy'


Search::Fulltext is a fulltext search module. It can be used in a few steps.

Search::Fulltext has pluggable tokenizer feature, which possibly provides fulltext search for any language. Currently, English and Japanese fulltext search are officially supported, although any other languages which have spaces for separating words could be also used. See CUSTOM TOKENIZERS section to learn how to search non-English languages.

SQLite's FTS4 is used as an indexer. Various queries supported by FTS4 (AND, OR, NEAR, ...) are fully provided. See "QUERIES" section for details.



Creates fulltext index for documents.

@param docs [required]

Reference to array whose contents are document to be searched.

@param index_file [optional]

File path to write fulltext index. By default, on-memory index is used.

@param tokenizer [optional]

Tokenizer name to use. simple (default) and porter must be supported. icu and unicode61 could be used if your SQLite libarary used via DBD::SQLite module support them. See for more details on FTS4 tokenizers.

Japanese tokenizer perl 'Search::Fulltext::Tokenizer::MeCab::tokenizer' is also available after you install Search::Fulltext::Tokenizer::MeCab module.

See CUSTOM TOKENIZERS section for developing other tokenizers.


Search terms in documents by query language.


Array of indexes of docs passed through Search::Fulltext->new in which query is matched.

@param query

Query to search from documents. See "QUERIES" section for types of queries.


The simplest query would be a term.

    my $results = $fts->search('beer');

Other queries below and combination of them can be also used.

    my $results = $fts->search('beer AND happy');
    my $results = $fts->search('saticefied OR happy');
    my $results = $fts->search('people NOT beer');
    my $results = $fts->search('make*');
    my $results = $fts->search('"makes people"');
    my $results = $fts->search('beer NEAR happy');
    my $results = $fts->search('beer NEAR/1 happy');

See for an explanation of each type of query.

NOTE: Some custom tokenizers might not support full of these queries above. Check the document of each tokenizer before using complex queries.


Custom tokenizers can be implemented by pure perl thanks to "Perl_tokenizers" in DBD::SQLite. Search::Fulltext::Tokenizer::MeCab is an example of custom tokenizers.

See "Perl_tokenizers" in DBD::SQLite and Search::Fulltext::Tokenizer::MeCab module to learn how to develop custom tokenizers.


Bug reports and pull requests are welcome at !


Version 1.03


Sho Nakatani <>, a.k.a. @laysakura