Ok I have two approached possible for this ticket:

• https://github.com/fedora-infra/anitya/compare/exact_match
  We remove doing pattern* search by default. This will work and do exact match by default but this also means we are changing the default behavior.
• https://github.com/fedora-infra/anitya/compare/rework_search
  If the pattern starts with a !, then do an exact match search, keeping the current default behavior.

This might prove to be an interesting third option: http://www.postgresql.org/docs/8.3/static/textsearch-intro.html#TEXTSEARCH-MATCHING

Well, isn't the question here about being able to do exact match search rather than text search?

Yes. But textsearch might be a more flexible method for specifying searches (and it might return results in a nicer order).

Ever played with textsearch? In sqlalchemy?

There seems to be some support: http://docs.sqlalchemy.org/en/latest/dialects/postgresql.html#full-text-search

So I guess there are two issues here:

• improve the search by using textsearch (table.c.column.match(pattern) in sqlalchemy)
• allow perfect match search

I just did some testing of using match vs ilike when doing the search:

- Packages.package_name.match('python')
- Project.name.match('python')
  -> 451 projects found

- Packages.package_name.ilike('python%')
- Project.name.ilike('python%')
  -> 454 projects found

- Packages.package_name.ilike('python')
- Project.name.ilike('python')
  -> 454 projects found

What is the first result? is it 'python'? :)

In both case it's not (after removing the order_by constraint)

Bummer. I was hoping textsearch would provide the "best" match first like xapian does.