# clojure-sql

A DSL for [Clojure][1] to compose queries using the constructs of
[Relational Algebra][2]. The RA constructs are then compiled into an
SQL query to be run on a DBMS. `clojure-sql` doesn't connect to a
database itself, but can be configured to perform an action when a
query is to be executed.

`clojure-sql` provides some utility functions beyond strict RA to
allow for data to be inserted, updated and deleted.

`clojure-sql` provides no mechanism to create database schemas.

[1]: http://clojure.org/
[2]: http://en.wikipedia.org/wiki/Relational_Algebra

## Database support

`clojure-sql` aims to have an extensible compiler. Compilation of
queries is performed by multimethods which dispatch on an arbitrary
(and otherwise unused) `db` parameter. This allows the compilation of
queries to be special-cased per database. By default `clojure-sql`
will produce SQL for PostgreSQL (work on other databases may be
undertaken in future).

## Usage

In `clojure-sql` expressions it is generally assumed that:

* Clojure keywords represent *column names* (exception: the `table`
  function)
* Clojure symbols represent *SQL functions* and *SQL operators*
* Clojure strings represent *literal strings*
* Clojure numbers represent *literal numbers*

Thus: `(and (= :name "Barry") (= (length :username) 10))` compiles
into an SQL expression equivalent to `("name" = 'Barry' AND
"length"("username") = 10)`.

A few more involved query examples:

    (require '[clojure-sql.core :as s])

    (-> (s/table :users) ;; tables are automatically given internal aliases
        (s/project [:id :username]))
    ;; => ["SELECT \"users3205\".\"id\" AS \"id\", \"users3205\".\"username\" AS \"username\" FROM \"users\" AS \"users3205\""]

    (-> (s/table :users)
        (s/project [:id :username])
        (s/rename {:id :uid})
        (s/join (-> (s/table :people)
                    (s/project {:id :pid, :fname :first})
                    (s/select '(= :first "Henry")))
                :on '(= :uid :pid))
        (s/project [:username]))
    ;; => ["SELECT \"users3430\".\"username\" AS \"username\" FROM (\"users\" AS \"users3430\" INNER JOIN \"people\" AS \"people3432\" ON (\"users3430\".\"id\" = \"people3432\".\"id\")) WHERE (\"people3432\".\"fname\" = ?)" "Henry"]

### clojure.java.jdbc

[`clojure.java.jdbc`][3] support is provided, but must be included
explicitly. `clojure-sql` does not depend on `clojure.java.jdbc`.

    (require '[clojure-sql.jdbc :as jdbc])

    (jdbc/use-jdbc! "postgres://user:pass@localhost:5432/db")
    ;; => nil

    (deref (-> (s/table :users)
               (s/project [:id :username])))
    ;; => [{:id 5, :username "username"}]

Results are returned from queries in an eager vector.

[3]: https://github.com/clojure/java.jdbc

### Other query executors

If `clojure.java.jdbc` support is not what you're looking for (or if
you're looking for more control of how `clojure.java.jdbc` is
executed) then you can register a query executor with
`clojure-sql.core/set-query-executor!`. A query executor is a function
which is used to run a query when it is executed.

When a query is executed the executor is passed two things: the type
of query being executed, and the compiled query (as vector of
`[query-string & query-args]`). The query executor's return value will
be the return value of the associated query function call (`deref`,
`insert!`, `update!` or `delete!`).

## Generated queries

In general, `clojure-sql` will try to avoid creating subqueries. For
any of the RA primitives (`select`, `join`, `project`, `rename`) no
subqueries will be introduced. The use of non-RA operations (`group`,
`sort`, `take`/`drop`) may introduce a subquery.

The particular operations which will create a subquery are:

* `join`ing a query which has been `group`ed
* `sort`ing a query which has been `take`n or `drop`ped from
* `group`ing a query which has been previously `group`ed

## License

Copyright © 2013 Carlo Zancanaro

Distributed under the Eclipse Public License, the same as Clojure.