For more information visit the API docs here.
Quill.d is a data access library for the D programming language that sits on top of DDBC. After getting set up you'll be able to write plain SQL in a file or a string, and run it. Quill.d embraces SQL as a language and does not attempt to abstract the database away. As it turns out, SQL is a pretty good language in which to query a database.
Here are a few high level examples:
Fetch some records
auto models = database.list!(Model, "list.sql")();
where list.sql
contains
select * from models;
Fetch a single record
auto model = database.single!(Model, "select.sql")(Variant(4));
where select.sql
contains
select * from models where id = ?;
Insert a record
database.execute!(Model, "insert.sql")(Variant("name"));
where insert.sql
contains
insert into models(name) values(?);
Add Quill.d to dub.json
{
...
"dependencies": {
"quill-d": "~>0.1.4"
}
}
Specify a database configuration to use.
Add the PostgreSQL configuration to dub.json
{
...
"subConfigurations": {
"quill-d": "PostgreSQL"
}
}
Install PostgreSQL Client
If you don't already have it, install the PostgreSQL client
sudo apt-get install postgresql-client
On Linux you may get an error cannot find -lpq
. The linker is having trouble finding the client library. To fix this, you can add a symlink like this:
ln -s /usr/lib/libpq.so.5 /usr/lib/libpq.so
Create a new PostgreSQL client:
import quill;
auto database = new Database("127.0.0.1", to!(ushort)(54320), "testdb", "admin", "password", true);
Add the MySQL configuration to dub.json
{
...
"subConfigurations": {
"quill-d": "MySQL"
}
}
Create a new MySQL client:
import quill;
auto database = new Database("127.0.0.1", to!(ushort)(33060), "testdb", "admin", "password");
Add the SQLite configuration to dub.json
{
...
"subConfigurations": {
"quill-d": "SQLite"
}
}
Install SQLite3
If you don't already have it, install SQLite3
sudo apt-get install sqlite3 libsqlite3-dev
Create a new SQLite client:
import quill;
auto database = new Database("/path/to/db.sqlite3");
Quill.d uses string imports to run SQL statements in files embedded in the compiled binary. The paths must be added to dub.json
to allow for those files to be imported as strings.
{
...
"stringImportPaths": ["queries"]
}
SQL queries can now be imported and run relative to the queries
directory like this:
database.execute!("statement.sql")();
The test suite is a collection of integration tests that actually runs SQL in all of the supported databases. Other than SQLite, you'll have to have a database to connect to. If you do not have a database, you can use Database Quickstart to spin up a server for each supported database. The connection details are in the test here. Once there is a database to connect to, the test suite can be run in all of the supported databases from the command line like this:
dub test
There are a bunch of overloads that can handle various kinds of queries. They are divided up by the expected return type.
Returns | Method Name |
---|---|
none | execute |
many | list |
one | single |
For each return type there can be no parameters, model based parameters, or Variant
based parameters.
This will execute a SQL statement in queries/statement.sql
that returns nothing and takes no parameters:
database.execute!("statement.sql")();
Model based parameters can be used by making a class that has fields that map to the column names in the result and the parameter names in the query. In D, bind
is used to specify the name of the parameter or the column name in a result set. In SQL, ?(parameter_name)
can be used to match the name of the field or the value of bind
.
Given a class like this:
class Model
{
int id;
@(bind("name")) string name;
}
auto model = new Model();
model.name = "value";
It can map the fields of the class into a query like this:
database.execute!("statement.sql")(model);
where statement.sql
looks like this:
insert into models(name) values(?(name));
It can also map the column names from the result of a query like this:
auto models = database.list!("statement.sql")();
where statement.sql
looks like this:
select id, name from models;
Variant
based parameters are ordered parameters that can be of any type.
Given a class like this:
class Model
{
int id;
@(bind("name")) string name;
}
A Variant
or array of Variant
will map to each parameter
auto model = database.single!("statement.sql")(Variant(4));
where statement.sql
looks like this:
select * from models where id = ?;
Properties that are not public
or include the @omit
attribute will be ignored. Given the following class, both name
and ignored
will be omitted.
class Model
{
int id;
protected string name;
@omit string ignored;
}
There are a ton more examples in the test and doc comments of database.d.
See license