.PBF

Introduction

The PBF format is the compact binary form of OpenStreetMap data, the free map of the world built by volunteers. The whole planet and regional extracts are published as .osm.pbf files, which are much smaller and faster to read than the same data in OpenStreetMap’s XML format (.osm). PBF stands for Protocolbuffer Binary Format: the data is encoded with Protocol Buffers.

The file holds OpenStreetMap’s three kinds of element: nodes, points with a latitude and longitude; ways, ordered lists of nodes that make roads, rivers and building outlines; and relations, groups of other elements. Each element carries tags, such as highway=residential.

Structure

The file is a sequence of blocks with no overall header. Each block is:

• a 4-byte big-endian length
• a BlobHeader message of that length, giving the block type, OSMHeader or OSMData, and the size of the blob
• a Blob message holding the block’s data, raw or compressed, usually with zlib

The first block is an OSMHeader with the bounding box and the features a reader must support. Each OSMData block is a primitive block of up to 8,000 elements. Its strings are kept in one table and referred to by number, and coordinates and IDs are stored as differences from the previous one, often in the packed DenseNodes form, which keeps the file small.

Links