Using OData with SQLAlchemy

Basic Usage

The easiest way to add OData filtering to a SQLAlchemy query is with the shorthand:

SQLAlchemy ORM

from odata_query.sqlalchemy import apply_odata_query

orm_query = select(MyModel)  # This is any form of Query or Selectable.
odata_query = "name eq 'test'"  # This will usually come from a query string parameter.

query = apply_odata_query(orm_query, odata_query)
results = session.execute(query).scalars().all()

SQLAlchemy Core

Attention

Basic support for SQLAlchemy Core is new since version 0.7.0. It currently does not support relationship traversal or any/all yet. Those operations will raise a NotImplementedException.

from odata_query.sqlalchemy import apply_odata_core

core_query = select(MyTable)  # This is any form of Query or Selectable.
odata_query = "name eq 'test'"  # This will usually come from a query string parameter.

query = apply_odata_query(core_query, odata_query)
results = session.execute(query).scalars().all()

Advanced Usage

If you need some more flexibility or advanced features, the implementation of the shorthand is a nice starting point: odata_query.sqlalchemy.shorthand

Let’s break it down real quick:

Parsing the OData Query

To get from a string representing an OData query to a usable representation, we need to tokenize and parse it as follows:

from odata_query.grammar import ODataParser, ODataLexer

lexer = ODataLexer()
parser = ODataParser()
ast = parser.parse(lexer.tokenize(my_odata_query))

This process is described in more detail in Parsing OData.

Optional: Modifying the AST

There are cases where you’ll want to modify the query before executing it. That’s what NodeTransformer’s are for!

One example might be that certain fields are exposed to end users under a different name than the one in the database. In this case, the odata_query.rewrite.AliasRewriter will come in handy. Just pass it a mapping of aliases to their full name and let it do its job:

from odata_query.rewrite import AliasRewriter

rewriter = AliasRewriter({
    "name": "author/name",
})
new_ast = rewriter.visit(ast)

Building a Query Filter

To get from an AST to something SQLAlchemy can use, you’ll need to use the odata_query.sqlalchemy.orm.AstToSqlAlchemyOrmVisitor (ORM mode) or the odata_query.sqlalchemy.core.AstToSqlAlchemyCoreVisitor (Core mode). It needs to know about the ‘root model’ or table of your query in order to see which fields exists and how objects are related.

SQLAlchemy ORM

from odata_query.sqlalchemy.orm import AstToSqlAlchemyOrmVisitor

visitor = AstToSqlAlchemyOrmVisitor(MyModel)
query_filter = visitor.visit(ast)

SQLAlchemy Core

from odata_query.sqlalchemy.core import AstToSqlAlchemyCoreVisitor

visitor = AstToSqlAlchemyCoreVisitor(MyTable)
query_filter = visitor.visit(ast)

Optional: Joins

Attention

Relationship traversal and automatic joins are not yet supported for SQLAlchemy Core mode.

If your query spans relationships, the AstToSqlAlchemyClauseVisitor will generate join statements. For the query to work, these will need to be applied explicitly:

for j in visitor.join_relationships:
    query = query.join(j)

Running the query

Finally, we’re ready to run the query:

query = query.where(query_filter)
results = s.execute(query).scalars().all()