Skip to content

osm-rawdata

📖 Documentation

These is a module to work with OpenStreetMap data using postgres and a custom database schema. This code is derived from the HOT Export Tool, osm-fieldwork, and Underpass, and the Raw Data API, which is the new FastAPI backend for the HOT Export Tool.

Since multiple projects need to do data extracts from OpenStreetMap in a flexible way, this was designed to have a single body of code to maintain.

Installation

To install osm-rawdata, you can use pip. Here are two options:

  • Directly from the main branch: pip install git+https://github.com/hotosm/osm-rawdata.git

  • Latest on PyPi: pip install osm-rawdata

The Database Schema

This project is heavily dependant on postgres and postgis. This schema was optimized for data anaylsis more than display purposes. The traditional schema use for OSM shows how it has evolved over the years. Some tags are columns (usually empty), and others get put into an jsonb tag column where they have to be accessed directly. One big change in this datbase schema is all the tags are in a single column, reducing the data size considerably, while also being easier to query in a consistent manner. In the past a developer had to keep track of what was a column, and what was in the tags column, which was inefficient.

This schema has 4 tables, similar to the traditional ones. OSM data is imported using osm2pgsql but uses a lua script to create the custom schema. This module's usage is all read-only, as Underpass can keep the raw data updated every minute, and we just want to access that data.

Things get more interesting as this module supports both a local database and a remote one. They use different query languages. To simplify this, a configuration file is used, which then generates the proper query syntax.

The Config File

This reads in two different formats that describe the eventualy SQL query. The YAML format was originally used by Export Tool, but later abandoned for a JSON format. The YAML format was adopted by the osm-fieldwork project before this transistion happened, so uses an enhanced version to define the queries.

The JSON format is also supported, both parsing the config file and also generating that query from a YAML config file.