Skip to main content

topojson - a powerful library to encode geographic data as topology in Python!🌍

Project description


PyPI version License build status

[Its not yet version 1.0, but that's merely because of missing documentation. With other words: you should be safe to use it!]

Topojson encodes geographic data structures into a shared topology. This repository describes the development of a Python implementation of the TopoJSON format.


The package can be used in multiple different ways, with the main purpose to create a TopoJSON topology:

import topojson

data = [
    {"type": "Polygon", "coordinates": [[[0, 0], [1, 0], [1, 1], [0, 1], [0, 0]]]},
    {"type": "Polygon", "coordinates": [[[1, 0], [2, 0], [2, 1], [1, 1], [1, 0]]]}

tj = topojson.Topology(data, prequantize=False, topology=True)

What results in the following TopoJSON object:

'{"type": "Topology", "objects": {"data": {"geometries": [{"type": "Polygon", "arcs": [[-2, 0]]}, {"type": "Polygon", "arcs": [[1, 2]]}], "type": "GeometryCollection"}}, "bbox": [0.0, 0.0, 2.0, 1.0], "arcs": [[[1.0, 0.0], [0.0, 0.0], [0.0, 1.0], [1.0, 1.0]], [[1.0, 0.0], [1.0, 1.0]], [[1.0, 1.0], [2.0, 1.0], [2.0, 0.0], [1.0, 0.0]]]}'

More or less all Python objects containing geographical data are supported through the __geo_interface__ attribute. This includes at least the Python packages geojson, shapely, geopandas, pyshp.

Moreover a dict of objects that provide a valid __geo_interface__, a list of objects that provide a valid __geo_interface__ and str objects with TopoJSON or GeoJSON geographic structures are supported too.

In the example above the output is parsed to a JSON string (.to_json()), but this is not the only thing we can do. Multiple functions are available to serialize the Topology object.

Functions Required Packages
topojson.Topology().to_json() Shapely, NumPy
topojson.Topology().to_dict() Shapely, NumPy
topojson.Topology().to_svg() Shapely, NumPy
topojson.Topology().to_alt() Shapely, NumPy, Altair*
topojson.Topology().to_gdf() Shapely, NumPy, GeoPandas*
topojson.Topology().to_widget() Shapely, NumPy, Altair*, Simplification*, ipywidgets* (+ labextension)

* optional dependencies

The TopoJSON format is merely designed to create smaller files than its GeoJSON counterpart. It is capable of doing so through a few options of which the following are currently available: compute topology, quantize the input and/or output, simplify the input and/or output.

The following parameters can be used to control these options for generating the Topology() object. Detailed information can be found in the docstring of the topojson.Topology() class.

  • topology
  • prequantize
  • topoquantize
  • presimplify
  • toposimplify
  • simplify_with
  • simplify_algorithm
  • winding_order

Where the toposimplify and topoquantize are supported by chaining as well. Meaning you could first compute the Topology (which can be cost-intensive) and afterwards apply the simplify and quantize settings on the computed Topology and visualize till pleased.

tj = topojson.Topology(data, prequantize=False, topology=True)

Or use the ipywidget approach described more below for an interactive approach.


Installation can be done by:

python3 -m pip install topojson

Topojson depends on the following packages:

  • numpy
  • shapely

Windows users: download the dependencies from OS X or Linux users: use pip as usual

Further, optional dependencies are:

  • altair (enlarge the experience by visualizing your TopoJSON output)
  • simplification (more and quicker simplification options)
  • geojson (parse string input with GeoJSON data)
  • geopandas (with fiona version >=1.8.6!, parse your TopoJSON output directly into a GeoDataFrame - converting it to GeoJSON)
  • ipywidgets + (lab)extension (make your life complete with the interactive experience)

Get in touch

For now, just use the Github issues. That can be:

  • usage questions
  • bug reports
  • feature suggestions
  • or anything related

Examples and tutorial notebooks

The followig examples present different input types parsed to different output types. The input types are not dependent on the used output type and vice versa, they are just possible examples.

Input Type: list

The list should contain items that supports the __geo_interface__

import topojson

list_geoms = [
    {"type": "Polygon", "coordinates": [[[0, 0], [1, 0], [1, 1], [0, 1], [0, 0]]]},
    {"type": "Polygon", "coordinates": [[[1, 0], [2, 0], [2, 1], [1, 1], [1, 0]]]}

apply Topology and present the output as dict

tj = topojson.Topology(data, prequantize=False, topology=True)
{'type': 'Topology',
  [[1.0, 0.0], [1.0, 1.0]],
  [[1.0, 1.0], [2.0, 1.0], [2.0, 0.0], [1.0, 0.0]]],
 'objects': {'data': {'geometries': [{'type': 'Polygon', 'arcs': [[-2, 0]]},
    {'type': 'Polygon', 'arcs': [[1, 2]]}],
   'type': 'GeometryCollection'}},
 'options': TopoOptions(
   {'prequantize': False,
  'presimplify': False,
  'simplify_with': 'shapely',
  'topology': True,
  'topoquantize': False,
  'toposimplify': 0.0001,
  'winding_order': 'CW_CCW'}
 'bbox': (0.0, 0.0, 2.0, 1.0),
 'arcs': [[[1.0, 0.0], [0.0, 0.0], [0.0, 1.0], [1.0, 1.0]],
  [[1.0, 0.0], [1.0, 1.0]],
  [[1.0, 1.0], [2.0, 1.0], [2.0, 0.0], [1.0, 0.0]]]}

Input Type: dict

The dictionary should be structured like {key1: obj1, key2: obj2}.

import topojson

dictionary = {
    0: {
        "type": "Polygon",
        "coordinates": [[[0, 0], [1, 0], [1, 1], [0, 1], [0, 0]]],
    1: {
        "type": "Polygon",
        "coordinates": [[[1, 0], [2, 0], [2, 1], [1, 1], [1, 0]]],

apply Topology and present the output as scalable vector graphic

tj = topojson.Topology(dictionary, prequantize=False, topology=True)

Input Type: GeoDataFrame from package geopandas (if installed)

import geopandas
import topojson
from shapely import geometry
%matplotlib inline

gdf = geopandas.GeoDataFrame({
    "name": ["abc", "def"],
    "geometry": [
        geometry.Polygon([[0, 0], [1, 0], [1, 1], [0, 1], [0, 0]]),
        geometry.Polygon([[1, 0], [2, 0], [2, 1], [1, 1], [1, 0]])
name geometry
0 abc POLYGON ((0 0, 1 0, 1 1, 0 1, 0 0))
1 def POLYGON ((1 0, 2 0, 2 1, 1 1, 1 0))

apply Topology and present output as altair chart (if installed)

tj = topojson.Topology(gdf, prequantize=False, topology=True)

Input Type: FeatureCollection from package geojson (if installed)

from geojson import Feature, Polygon, FeatureCollection

feature_1 = Feature(
    geometry=Polygon([[[0, 0], [1, 0], [1, 1], [0, 1], [0, 0]]]),
feature_2 = Feature(
    geometry=Polygon([[[1, 0], [2, 0], [2, 1], [1, 1], [1, 0]]]),
feature_collection = FeatureCollection([feature_1, feature_2])

apply Topology and present output as geodataframe (if geopandas is installed)

tj = topojson.Topology(feature_collection, prequantize=False, topology=True)
geometry id name
0 POLYGON ((1 1, 1 0, 0 0, 0 1, 1 1)) None abc
1 POLYGON ((1 0, 1 1, 2 1, 2 0, 1 0)) None def

Currently parsing TopoJSON as string input requires geopandas (fiona version >=1.8.6) and parsing GeoJSON as string requires the package geojson.

The package simplification can be used if you want to adopt the Visvalingam-Whyatt algorithm for simplifying or for having a speedup on the Douglas-Peucker algorithm (compared to the shapely-integrated version).

The .to_widget() function depends on ipywidgets and can be a bit tricky to get it installed and properly working. But if you do, something like the following will show up:


To install, use the ipywidgets website for installation.
Initially I ran very often in errors like the following after I thought I'd install everything correctly:

[IPKernelApp] WARNING | No such comm: xxxyyyzzz123etc.

To solve this error I found out that I'd first had to pip uninstall JupyterLab, then install the lab extension of ipywidgets and then install JupyterLab again. Then when starting JupyterLab for the first time it asks to rebuild to include the ipywidgets lab extension. Click Yes or OK and wait till JupyterLab refresh, afterwards these errors did not appear for me anymore (both Windows and macOS). If you got all installed I suggest starting from In [5] in the following notebook to test if all works.

Futher, the many tests as part of this package also can be used as example material.


Version 1.0rc6:

  • fix linemerging of non-duplicate arcs #50
  • include __geo_interface__ attributed as input #53
  • include travis testing on GitHub (thanks @Casyfill!)
  • migrate from unittests to pytest (thanks @Casyfill!)

Version 1.0rc5:

  • change TopoOptions in to_dict to be serializable #46
  • changed all int to np.int64, since it is platform specific #49, #45

Version 1.0rc4:

  • no linestring key in topojson
  • serialize str of TopoJSON or GeoJSON data
  • add vw as algoritm type and update widget

Version 1.0rc3:

  • changed class object to inherit sequence
  • removed the topojson.topology function
  • introducted the topojson.Topology class
  • speedups and bug fixes, see PR#15-#36
  • introduced multiple options see #8

Version 1.0rc2:

  • apply linemerge on non-duplicate arcs
  • fix computing topology without shared boundaries (#1, #3)
  • use geopandas and geojson solely for tests, but recognize them as type (#2, #4)
  • use simplification as option to simplify linestrings
  • include option to snap vertices to grid
  • removed rdtree as dependency, use SRTtree from shapely instead

Version 1.0rc1:

  • initial release


There are many tests writen to make sure all type of corner-cases are covered. To make sure all tests will pass, you must have version >=0.5.0 of geopandas in combination with fiona version >=1.8.6. Shapely version 1.7a2 is recommended (because of, but all tests pass from version >=1.6.3.

Development Notes

Development of this packages started by reading:

The reason for development of this package was the willingness:

  • To adopt shapely (GEOS) and numpy for the core-functionalities in deriving the Topology.
  • To provide integration with other geographical packages within the Python ecosystem (eg. geopandas and altair).
  • Also the possibility of including the many tests available in the JavaScript implementation was hoped-for.

To create a certain synergy between the JavaScript and Python implementation the same naming conventions was adopted for the processing steps (extract, join, cut, dedup, hashmap). Even though the actual code differs significant.

Some subtile differences are existing between the JavaScript implementation and the current Python implementation for deriving the Topology. Some of these deviations are briefly mentioned here:

  1. The extraction class stores all the different geometrical objects as Shapely LineStrings in linestrings and keeps a record of these linestrings available under the key bookkeeping_geoms. In the JavaScript implementation there is a differentiation of the geometries between lines, rings and a seperate object containing all coordinates. Since the current approach adopts shapely for much of the heavy lifting this extraction is working against us (in the cut-process).

  2. In the join class only the geometries that have shared paths are considered to have junctions. This means that the intersection of two crossing lines at a single coordinate is not considered as a junction. This also means that the two ends of a LineString are not automatically considered as being a junction. So if a segment starts or finish on another segment, with that coordinate being the only coordinate in common, it is not considered as a junction.

  3. In the computation of a shared path, a junction can be created on an existing coordinate in one of the geometries. Where in the JavaScript implementation this only can be considered when both geometries contain the coordinate.

  4. In the process of cutting lines; the rings are rotated in the JavaScript implementation to make sure they start at a junction. This reduces the number of cuts. This rotation is done before cutting. In the current Python implementation this is done differently. First the linestrings are cut using the junction coordinates and afterwards there is tried to apply a linemerge on the non-duplicate arcs of a geometry containing at least one shared arc.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for topojson, version 1.0rc7
Filename, size File type Python version Upload date Hashes
Filename, size topojson-1.0rc7-py2.py3-none-any.whl (109.0 kB) File type Wheel Python version py2.py3 Upload date Hashes View hashes
Filename, size topojson-1.0rc7.tar.gz (4.7 MB) File type Source Python version None Upload date Hashes View hashes

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page