OSMAlchemy 0.1.5

OpenStreetMap to SQLAlchemy bridge

OSMAlchemy is a bridge between SQLAlchemy and the OpenStreetMap API.

Goals

OSMAlchemy’s goal is to provide completely transparent integration of the real-world OpenStreetMap data within projects using SQLAlchemy. It provides two things:

1. Model declaratives resembling the structure of the main OpenStreetMap database, with some limitations, usable wherever SQLAlchemy is used, and

2. Transparent proxying and data-fetching from OpenStreetMap data.

The idea is that the model can be queried using SQLAlchemy, and OSMAlchemy will either satisfy the query from the database directly or fetch data from OpenStreetMap. That way, projects already using SQLAlchemy do not need another database framework to use OpenStreetMap data, and the necessity to keep a local copy of planet.osm is relaxed.

If, for example, a node with a certain id is queried, OSMAlchemy will…

• …try to get the node from the database/ORM directly, then…

• …if it is available, check its caching age, and…

  • …if it is too old, refresh it from OSM, or…

  • …else, fetch it from OSM, and…

• …finally create a real, ORM-mapped database object.

That’s the rough idea, and it counts for all kinds of OSM elements and queries.

OSMAlchemy uses Overpass to satisfy complex queries.

Non-goals

OSMAlchemy does not aim to replace large-scale OSM data frameworks like PostGIS, Osmosis or whatever. In fact, in terms of performance and otherwise, it cannot keep up with them.

If you are running a huge project that handles massive amounts of map data, has millions of requests or users, then OSMAlchemy is not for you (YMMV).

OSMAlchemy fills a niche for projects that have limited resources and cannot handle a full copy of planet.osm and an own API backend and expect to handle limited amounts of map data.

It might, however, be cool to use OSMAlchemy as ORM proxy with an own API backend. Who knows?

It might, as well, turn out that OSMAlchemy is an incredibly silly idea under all circumstances.

Usage

Here are a few tiny examples of how to basically use OSMAlchemy:

Installation

OSMAlchemy can be installed just like any other standard Python package by one of…

# pip3 install OSMAlchemy
# python3 setup.py install

…or what ever kind of distribution and install system you prefer.

Using plain SQLAlchemy

Make sure to get at least an engine from SQLAlchemy. Even better, get a declarative base and a scoped session:

from sqlalchemy import create_engine
from sqlalchemy.ext.declarative import declarative_base
from sqlalchemy.orm import sessionmaker, scoped_session

engine = create_engine("sqlite:////tmp/foo.db")
base = declarative_base(bind=engine)
session = scoped_session(sessionmaker(bind=engine))

You can then initialise OSMAlchemy like so:

osmalchemy = OSMAlchemy((engine, base, session), overpass=True)

And probably install the databases:

base.metadata.create_all()

Using Flask-SQLAlchemy and Flask-Restless

Imagine you have an SQLAlchemy object from Flask-SQLAlchemy bound to your Flask application. called db, and a Flask-Restless API manager as manager:

from osmalchemy import OSMAlchemy
osm = OSMAlchemy(db, overpass=True)
db.create_all()
osm.create_api(manager)

You should now magically be able to query OSM via the REST API. Keep in mind that, with no filter provided, OSMAlchemy refuses to do automatic updates from Overpass. However, providing a query in the default JSON query way in Flask-Restless will give you live data and cache it in the database.

Limitations

Only some basic SQL queries are supported by the online update code. This is because compiling SQLAlchemy’s queries to OverpassQL is very complex. If you are very good at algorithms and building compilers, feel free to help us out!

The following kinds of queries are fully supported:

# A node with a specific id
session.query(osmalchemy.node).filter_by(id=12345).one()

# All nodes within a bounding box
session.query(osmalchemy.node).filter(
    and_(latitude>51.0, latitude<51.1, longitude>7.0, longitude<7.1)
).all()

# All nodes having a specific tag
session.query(osmalchemy.node).filter(
    osmalchemy.node.tags.any(key="name", value="Schwarzrheindorf Kirche")
).all()

You can go mad combining the two with and_() and or_(). You can also query for tags of ways and relations and for ways and relations by id.

Not supported (yet) are queries for ways or relations by coordinates. You also cannot query for nodes related to a way or anything related to a relation - having a way or a relation, accessing it will, however, magically pull and update the nodes and members and add them to the database:

# Get all nodes that are members of a (unique) named way
session.query(osmalchemy.way).filter(
    osmalchemy.way.tags.any(key="name", value="My Unique Way")
).one().nodes

This should, in reality, cover most use cases. If you encounter a use case that is not supported, please open an issue asking whether it can be supported (if you have an idea how it can be, please add it or even implement it and open a pull request).

Projects using OSMAlchemy

OSMAlchemy was designed for use in the Veripeditus Augmented Reality framework.

Development and standards

Albeit taking the above into account, OSMAlchemy is developed with quality and good support in mind. That means code shall be well-tested and well-documented.

OSMAlchemy is tested against the following SQLAlchemy backends:

• SQLite

• PostgreSQL

• MySQL

However, we recommend PostgreSQL. MySQL acts strangely with some data and is incredibly slow, and SQLite just doesn’t scale too well (however, it is incredibly fast, in comparison).

Authors and credits

Authors:

Dominik George, Eike Tim Jesinghaus

Credits:

Special thanks to Mike Bayer from SQLAlchemy for his help with some SQLAlchemy bugs and pitfalls, and also some heads-up.

Contact:

E-mail to [e-mail]

License

OSMAlchemy is licensed under the MIT license. Alternatively, you are free to use OSMAlchemy under Simplified BSD, The MirOS Licence, GPL-2+, LGPL-2.1+, AGPL-3+ or the same terms as Python itself.