# Postgres + tsearch2, migrations, and unit testing

**URL:** <https://www.ruby-forum.com/t/postgres-tsearch2-migrations-and-unit-testing/84335>\
**Category:** Rails\
**Created:** [February 3, 2007, 8:31am UTC](https://www.ruby-forum.com/t/postgres-tsearch2-migrations-and-unit-testing/84335 "2007-02-03T08:31:07Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![phillyloco](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/p/258eb7/32.png) [@phillyloco](https://www.ruby-forum.com/u/phillyloco)\
**Post date:** [February 3, 2007, 8:31am UTC](https://www.ruby-forum.com/t/postgres-tsearch2-migrations-and-unit-testing/84335/1 "2007-02-03T08:31:07Z")

</div>

I’m sure people have encountered this before, but I’ve been  
unsuccessful at finding anything about it.

Basically, I got a rails app that uses postgres and tsearch2 for some  
full text indexing and I’d like to be performing unit tests. Now,  
there’s the whole deal of how to get the test database to load all the  
tsearch2 information properly.

I tried to come up with an elegant way to do it without overwriting  
any rails code or duplicating SQL or anything ugly, but didn’t manage  
to. The best I came up with is as follows:

1. In the migrations, perform all the tsearch2 related stuff for the  
table that requires full text indexing. This includes loading  
tsearch2.sql (and uninstall\_tsearch2.sql in the down part), creating  
the tsvector column, the index on it, and the trigger that updates the  
tsvector column.

2. config.active\_record.schema\_format = :sql in the environment since  
schemas would puke at the tsvector column.

3. Copy all the inserts from the tsearch2.sql file and add it to a  
separate file. Setting schema\_format to sql will only have the table  
structure copied to the test database, not the actual content of the  
pg\_\* tables which is needed to configure tsearch2. I put this in  
test/fixtures.

4. Overwrite the db:test:prepare task with the following:

desc ‘Prepare the test database and load the schema’  
task :prepare =\> :environment do  
if defined?(ActiveRecord::Base) &&  
!ActiveRecord::Base.configurations.blank?  
Rake::Task[{ :sql =\> “db:test:clone\_structure”, :ruby =\>  
“db:test:clone” }[ActiveRecord::Base.schema\_format]].invoke  
end  
ActiveRecord::Base.establish\_connection(ActiveRecord::Base.configurations[‘test’])  
ActiveRecord::Base.connection.execute File.read(  
File.join(RAILS\_ROOT, ‘test’, ‘fixtures’, ‘tsearch2.sql’) )  
end

Which is basically a copy of the original with the two extra lines at  
the end that populate the configuration tables.

And that all seems to work. My question would be this: is there a  
better way to do it? All this doesn’t seem to be too elegant.

–  
EPA Rating: 3000 Lines of Code / Gallon (of coffee)

---

<div class="post-metadata">

**Author:** ![phillyloco](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/p/258eb7/32.png) [@phillyloco](https://www.ruby-forum.com/u/phillyloco)\
**Post date:** [February 3, 2007, 11:17pm UTC](https://www.ruby-forum.com/t/postgres-tsearch2-migrations-and-unit-testing/84335/2 "2007-02-03T23:17:21Z")

</div>

Well, in case anybody cares (or anybody searches for this later), I  
managed to clean up the code a bit more by perparing the test database  
from migrations. I took a bit of code from the code snippets site and  
tweaked it to suit my needs a bit more (see my comment).

> **[BigBold - Informasi Tentang Bisnis dan Marketing](http://www.bigbold.com)**
>
> BigBold ialah situs yang memberikan segala macam Informasi Tentang Bisnis dan Marketing