NOTE && TODO: This has been tested on Rails 2.0.2, 2.1.0 and 2.1.1 only. It contains some activesupport'isms, and... they should definitely be stripped out and this plugin made into a proper gem.

Description

This is an extremely light weight interface to postgresql 8.3's built in text searching. There exists an acts_as_tsearch, but this is its smaller and built specifically for 8.3, more pointed, brother. It lacks many features, but is useful in most text searching scenerios.

This plugin assumes a lot. The lot is: - Default language used for full text search is english. Use :catalog => 'pg_catalog.french' to change that. - GIN index by default. You can pass in :index => 'GIST' if you'd like to change that. - Default tsvector is a coellesced composite of the fields you pass in. - OPTIONAL: if you want to use suggestion feature ("do you mean x ?") pg_trgm has to be activated like this: psql database_name < /path_to_your_pg_contrib_folder/pg_trgm.sql (see http://www.postgresql.org/docs/8.3/static/pgtrgm.html for more details about pg_trgm)

Usage

• ensure your table has a tsvector ts_index column ... in create table block: t.tsvector :ts_index

• ensure your model declares text_searchable properly ... tsearchable :fields => [:name, :description]

• create tsvector index and trigger using rake task ... rake tsearchable:vectors:create VECTOR=name_of_the_model ... rake tsearchable:triggers:create VECTOR=name_of_the_model

• YourModel.find_by_text_search("moose OR mousse") ... returns result set. supports most googlish ands and ors and double quoting ... supports WillPaginate pagination, and the will_paginate view helper

• YourModel.count_by_text_search("moose OR mousse") ... returns the number of matches

OPTIONAL: to use suggestion feature (pg_trgm MUST be activated)

• ensure your table has a column with gist_trgm_ops or gin_trgm_ops index type ... TODO: create rake task to do that

• ensure your model declares suggestion fields ... tsearchable :suggests => [ :title ]

• YourModel.find_by_trgm("keyword") ... returns result set.

TODO

most importantly... document, and turn it into a gem

also:

• weighting, highlighting,
• custom parsers/templates/rankers,
• multiple vectors,
• multitable support