Skip to content

paulmach/osm

Repository files navigation

osm CI Go Report Card Go Reference

This package is a general purpose library for reading, writing and working with OpenStreetMap data in Go (golang). It has the ability to:

Made available by the package are the following types:

  • Node
  • Way
  • Relation
  • Changeset
  • Note
  • User

And the following “container” types:

List of sub-package utilities

  • annotate - adds lon/lat, version, changeset and orientation data to way and relation members
  • osmapi - supports all the v0.6 read/data endpoints
  • osmgeojson - OSM to GeoJSON conversion compatible with osmtogeojson
  • osmpbf - stream processing of *.osm.pbf files
  • osmxml - stream processing of *.osm xml files
  • replication - fetch replication state and change files

Concepts

This package refers to the core OSM data types as Objects. The Node, Way, Relation, Changeset, Note and User types implement the osm.Object interface and can be referenced using the osm.ObjectID type. As a result it is possible to have a slice of []osm.Object that contains nodes, changesets and users.

Individual versions of the core OSM Map Data types are referred to as Elements and the set of versions for a give Node, Way or Relation is referred to as a Feature. For example, an osm.ElementID could refer to "Node with id 10 and version 3" and the osm.FeatureID would refer to "all versions of node with id 10." Put another way, features represent a road and how it's changed over time and an element is a specific version of that feature.

A number of helper methods are provided for dealing with features and elements. The idea is to make it easy to work with a Way and its member nodes, for example.

Scanning large data files

For small data it is possible to use the encoding/xml package in the Go standard library to marshal/unmarshal the data. This is typically done using the osm.OSM or osm.Change "container" structs.

For large data the package defines the Scanner interface implemented in both the osmxml and osmpbf sub-packages.

type osm.Scanner interface {
	Scan() bool
	Object() osm.Object
	Err() error
	Close() error
}

This interface is designed to mimic the bufio.Scanner interface found in the Go standard library.

Example usage:

f, err := os.Open("./delaware-latest.osm.pbf")
if err != nil {
	panic(err)
}
defer f.Close()

scanner := osmpbf.New(context.Background(), f, 3)
defer scanner.Close()

for scanner.Scan() {
	o := scanner.Object()
	// do something
}

scanErr := scanner.Err()
if scanErr != nil {
	panic(scanErr)
}

Note: Scanners are not safe for parallel use. One should feed the objects into a channel and have workers read from that.

Working with JSON

This library supports reading and writing OSM JSON. This format is returned by the Overpass API and can be optionally returned by the OSM API.

If performance is important, this library supports third party "encoding/json" replacements such as github.com/json-iterator/go.

They can be enabled with something like this:

import (
  jsoniter "github.com/json-iterator/go"
  "github.com/paulmach/osm"
)

var c = jsoniter.Config{
  EscapeHTML:              true,
  SortMapKeys:             false,
  MarshalFloatWith6Digits: true,
}.Froze()

osm.CustomJSONMarshaler = c
osm.CustomJSONUnmarshaler = c

The above change can have dramatic performance implications, see the benchmarks below on a large OSM Change object.

benchmark                            old ns/op     new ns/op     delta
BenchmarkChange_MarshalJSON-12       604496        461349        -23.68%
BenchmarkChange_UnmarshalJSON-12     1633834       1051630       -35.63%

benchmark                            old allocs    new allocs    delta
BenchmarkChange_MarshalJSON-12       1277          1081          -15.35%
BenchmarkChange_UnmarshalJSON-12     5133          8580          +67.15%

benchmark                            old bytes     new bytes     delta
BenchmarkChange_MarshalJSON-12       180583        162727        -9.89%
BenchmarkChange_UnmarshalJSON-12     287707        317723        +10.43%

CGO and zlib

OSM PBF data comes in blocks, each block is zlib compressed. Decompressing this data takes about 33% of the total read time. DataDog/czlib is used to speed this process. See osmpbf/README.md for more details.

As a result, a C compiler is necessary to install this module. On macOS this may require installing pkg-config using something like brew install pkg-config

CGO can be disabled at build time using the CGO_ENABLED ENV variable. For example, CGO_ENABLED=0 go build. The code will fallback to the stdlib implementation of zlib.