A Rust library for reading the OpenStreetMap PBF file format (*.osm.pbf). It strives to offer the best performance using parallelization and lazy-decoding with a simple interface while also exposing iterators for items of every level in a PBF file.

Add this to your Cargo.toml:

[dependencies]
osmpbf = "0.2"

Here's a simple example that counts all the ways in a file:

use osmpbf::{ElementReader, Element};

let reader = ElementReader::from_path("tests/test.osm.pbf")?;
let mut ways = 0_u64;

// Increment the counter by one for each way.
reader.for_each(|element| {
    if let Element::Way(_) = element {
        ways += 1;
    }
})?;

println!("Number of ways: {ways}");

In this second example, we also count the ways but make use of all cores by decoding the file in parallel:

use osmpbf::{ElementReader, Element};

let reader = ElementReader::from_path("tests/test.osm.pbf")?;

// Count the ways
let ways = reader.par_map_reduce(
    |element| {
        match element {
            Element::Way(_) => 1,
            _ => 0,
        }
    },
    || 0_u64,      // Zero is the identity value for addition
    |a, b| a + b   // Sum the partial results
)?;

println!("Number of ways: {ways}");

• rust-zlib (default) -- use the pure Rust zlib implementationminiz_oxide
• zlib -- use the widely available zlib library
• zlib-ng -- use the zlib-ng library for better performance.

To effectively use the more lower-level features of this library it is useful to have an overview of the structure of a PBF file. For a more detailed format description see here or take a look at the .proto files in this repository.

The PBF format as a hierarchy (square brackets [] denote arrays):

Blob[]
├── HeaderBlock
└── PrimitiveBlock
    └── PrimitiveGroup[]
    	├── Node[]
    	├── DenseNodes
    	├── Way[]
        └── Relation[]

At the highest level a PBF file consists of a sequence of blobs. Each Blob can be decoded into either a HeaderBlock or a PrimitiveBlock.

Iterating over blobs is very fast, but decoding might involve a more expensive decompression step. So especially for larger files it is advisable to parallelize at the blob level as each blob can be decompressed independently. (See the reader module in this library for parallel methods)

Usually the first Blob of a file decodes to a HeaderBlock which holds global information for all following PrimitiveBlocks, such as a list of required parser features.

A PrimitiveBlock contains an array of PrimitiveGroups. Each PrimitiveGroup only contains one element type: Node, Way, Relation or DenseNodes. A DenseNodes item is an alternative and space-saving representation of a Node array. So, do not forget to check for DenseNodes when aggregating all nodes in a file.

Elements reference each other using integer IDs. Corresponding elements could be stored in any blob, so finding them can involve iterating over the whole file. Some files declare an optional feature "Sort.Type_then_ID" in the HeaderBlock to indicate that elements are stored sorted by their type and then ID. This can be used to dramatically reduce the search space.

This project is licensed under either of

• Apache License, Version 2.0, (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
• MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)

at your option.