A model your application can use.

Available now: PostgreSQL and SQLite. Coming soon: MySQL and Microsoft SQL Server (MSSQL).

Start with a recipe

The recipe declares the logical structure. Asqueel resolves it into table and column objects. Building those objects does not open a database connection or create tables.

from asqueel import AsqueelDb
from invoicing import Invoicing

db = AsqueelDb(Invoicing)
invoice = db.table('invc.invoice')

Invoicing is the model used throughout the course. It contains customers, products, invoices, invoice rows and notes. Explore the recipe →

Describe a relationship once

A relation belongs to the model. Queries can follow that relation with an @ path. An alias column can give a frequently used path a name.

tbl.aliasColumn(
    'customer_name',
    relation_path='@customer_id.account_name',
)

Now every query on this table can refer to $customer_name. The compiler resolves the relation and generates the join. Follow the relations →

Keep calculations close to the data

A formula column describes a calculated value using SQL. It is queryable by name without adding a stored column.

tbl.formulaColumn(
    'line_total',
    sql_formula='$quantity * $unit_price',
    dtype='N',
)

The course also covers correlated subqueries and aggregations over related rows. These examples use the development API described in the compatibility notes.

Inspect a query before executing it

SQL remains the expression language. Asqueel resolves model references; values are passed separately.

query = invoice.query(
    columns='$inv_number, $customer_name, $total',
    where='$total >= :minimum',
    minimum=100,
    order_by='$inv_number',
)
print(query.sqltext)  # Compile without opening a connection.
rows = query.fetch()  # Execute against the database.

Fetched rows are mutable NamedList objects with access by name or position. Use dict(row) when you need a dictionary. Choose a fetch terminal →

Use the same model for business logic

Table objects expose writes and can carry application behavior. The course follows what happens before and after an insert, update or delete: field triggers, table hooks and relation actions.

Transactions are explicit. db.commit() saves a unit of work. db.rollback() discards it. Understand the write cycle →

Choose the right boundary

The model can carry external metadata through grammar extensions. That does not imply a UI renderer or an authorization system. See the full course →