Defining SQL Queries
====================

Query Names
-----------

Name definitions are how aiosql determines the name of the methods that SQL
code blocks are accessible by.
A query name is defined by a SQL comment of the form "-- name: ".
As a readability convenience, dash characters (``-``) in the name are turned
into underlines (``_``).

.. code:: sql

    -- name: get-all-blogs
    select * from blogs;

This query will be available in aiosql under the python method name ``.get_all_blogs(conn)``

Query Comments
--------------

*./sql/blogs.sql*

.. code:: sql

    -- name: get-all-blogs
    -- Fetch all fields for every blog in the database.
    select * from blogs;

Any other SQL comments you make between the name definition and your code will
be used a the python documentation string for the generated method.
You can use ``help()`` in the Python REPL to view these comments while using python.

.. code:: pycon

    Python 3.8.3 (default, Jul 29 2020, 11:15:42)
    [GCC 10.1.0] on linux
    Type "help", "copyright", "credits" or "license" for more information.
    >>> import aiosql
    >>> queries = aiosql.from_path("sql", "sqlite3")
    >>> help(queries.get_all_blogs)
    Help on method get_all_blogs in module aiosql.queries:

    get_all_blogs(conn, *args, **kwargs) method of aiosql.queries.Queries instance
        Fetch all fields for every blog in the database.


Named Parameters
----------------

Named parameters ``:param`` are accepted by all supported drivers and taken
from Python named parameters passed to the query.
In addition, simple attributes can be referenced with the ``.``-syntax.

.. code:: sql

    -- name: with-params
    select length(:name), :x.real + :x.imaj;

Then the generated function expects two named parameters:

.. code:: python

    res = queries.with_params(name="Calvin", x=(1+1j))
    # => (6, 2.0)

Operators
---------

This section describes the usage of various query operator symbols that you can
annotate query names with in order to direct how aiosql will execute and return
results.

No Operator (Default)
~~~~~~~~~~~~~~~~~~~~~

In the above `Query Names <#query-names>`__ section the ``get-all-blogs``
name is written without any trailing operators.

.. code:: sql

    -- name: get-all-blogs

The lack of an operator is actually the most basic operator used by default for
your queries. This tells aiosql to execute the query and to return all the results.
In the case of ``get-all-blogs`` that means a ``select`` statement will be executed
and a list of rows will be returned. When writing your application you will often
need to perform other operations besides ``select``, like ``insert``, ``delete``,
and perhaps bulk operations. The operators detailed in the other sections of this
doc let you declare in your SQL code how that query should be executed by a python
database driver.

``^`` Select One
~~~~~~~~~~~~~~~~

The ``^`` operator executes a query and returns the first row of a result set.
When there are no rows in the result set it returns ``None``.
This is useful when you know there should be one, and exactly one result from your query.

As an example, if you have a unique constraint on the ``username`` field in your
``users`` table which makes it impossible for two users to share the same username,
you could use ``^`` to direct aiosql to select a single user rather than a list of
rows of length 1.

.. code:: sql

    -- name: get_user_by_username^
    select userid,
           username,
           name
      from users
     where username = :username;

When used from Python this query will either return ``None`` or the singular selected row.

.. code:: python

    queries.get_user_by_username(conn, username="willvaughn")
    # => (1, "willvaughn", "William Vaughn") or None

``$`` Select Value
~~~~~~~~~~~~~~~~~~

The ``$`` operator will execute the query, and only return the first value of the first row
of a result set. If there are no rows in the result set it returns ``None``.
This is implemented by returing the first element of the tuple returned by ``cur.fetchone()``
from the underlying driver. This is mostly useful for queries returning IDs, COUNTs or
other aggregates.

.. code:: sql

    -- name: get-count$
    select count(*) from users

When used from Python:

.. code:: python

    queries.get_count(conn)
    # => 3

``!`` Insert/Update/Delete
~~~~~~~~~~~~~~~~~~~~~~~~~~

The ``!`` operator executes SQL without returning any results.
It is meant for statements that use ``insert``, ``update``, and ``delete`` to make
modifications to database rows without a necessary return value.

.. code:: sql

    -- name: publish_blog!
    insert into blogs(userid, title, content) values (:userid, :title, :content);

    -- name: remove_blog!
    -- Remove a blog from the database
    delete from blogs where blogid = :blogid;

The methods generated are:

.. code:: text

    publish_blog(conn, userid: int, title: str, content: str) -> int:
    remove_blog(conn, blogid: int) -> int:

Each can be called to alter the database, and returns the number of affected rows
if available.

``<!`` Insert/Update/Delete Returning
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

When performing a modification of rows, or adding new rows, sometimes it is
necessary to return values using the ``returning`` clause where available.

When using SQLite this operator will return the id of the inserted row using
```cur.lastrowid`` <https://docs.python.org/3/library/sqlite3.html#sqlite3.Cursor.lastrowid>`__.

As recent version of SQLite do support the ``returning`` clause, forget about
this, use the clause explicitely and treat the whole command as a standard
select with the *empty* operator (relation), or ``^`` (tuple), or ``$``
(scalar).

.. code:: sql

    -- name: publish_blog<!
    insert into blogs(userid, title, content) values (:userid, :title, :content);

Executing this query in python will return the ``blogid`` of the inserted row.

.. code:: python

    queries = aiosql.from_path("blogs.sql", "sqlite3")
    # ... connection code ...
    blogid = queries.publish_blog(conn, userid=1, title="Hi", content="blah blah.")

``*!`` Insert/Update/Delete Many
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The ``*!`` operator directs aiosql to execute a SQL statement over all items of a given sequence.
Under the hood this calls the ``executemany`` method of many database drivers.
See `sqlite3 Cursor.executemany <https://docs.python.org/3/library/sqlite3.html#sqlite3.Cursor.executemany>`__
for an example.

In aiosql we can use this for a bulk publish method that operates over a list of blog entries.

.. code:: sql

    -- name: bulk_publish*!
    -- Insert many blogs at once
    insert into blogs (userid, title, content, published)
    values (:userid, :title, :content, :published);

.. code:: python

    queries = aiosql.from_path("blogs.sql", "psycopg2")
    # ... connection code ...
    blogs = [
        {"userid": 1, "title": "First Blog", "content": "...", published: datetime(2018, 1, 1)},
        {"userid": 1, "title": "Next Blog", "content": "...", published: datetime(2018, 1, 2)},
        {"userid": 2, "title": "Hey, Hey!", "content": "...", published: datetime(2018, 7, 28)},
    ]
    queries.bulk_publish(conn, blogs)

The methods returns the number of affected rows, if available.

``#`` Execute Scripts
~~~~~~~~~~~~~~~~~~~~~

Using this operarator will execute sql statements as a script.
You can't do variable substitution with the ``#`` operator.
An example usecase is using data definition statements like create table in order to setup a database.

.. code:: sql

    -- name: create_schema#
    create table users (
        userid integer not null primary key,
        username text not null,
        firstname integer not null,
        lastname text not null
    );

    create table blogs (
        blogid integer not null primary key,
        userid integer not null,
        title text not null,
        content text not null,
        published date not null default CURRENT_DATE,
        foreign key(userid) references users(userid)
    );

.. code:: python

    queries = aiosql.from_path("create_schema.sql", "sqlite3")
    # ... connection code ...
    queries.create_schema(conn)