Go to file
2022-04-09 12:02:05 +02:00
.gitignore Use a new SQLite db every week 2017-08-28 18:03:36 +02:00
LICENSE Use a new SQLite db every week 2017-08-28 18:03:36 +02:00
README.md Update doc, drop config 2018-01-15 14:02:39 +01:00
velib.py Optimize SQL handling 2022-04-09 12:02:05 +02:00
visu.py Better documentation + fix a color inversion in visu 2017-08-31 14:52:54 +02:00


This code can be used to dump periodically all the available data from the Velib API, the bike sharing system in Paris.

It is basically a wrapper around pybikes to dump values periodically in a SQLite database.


  • Clone this repo.
  • Install pybikes.
  • Run python2 velib.py.

Note: For now, pybikes is only Python2 compatible.

Dumped data

Important: For the latest information about the dump available at https://pub.phyks.me/datasets/velib/, please refer to https://pub.phyks.me/datasets/velib/README.txt.

This script is used to dump the returned data from the Velib API every few minutes. Dumps are available at https://pub.phyks.me/datasets/velib/.

The script writes in a new SQLite file every week, put in a different folder by year, and labelled with the week number.

Each SQLite file has three tables:

  • A stations table, containing "permanent" information about each station (latitude, longitude, number of stands etc).
  • A stationsstats table which contains the available number of bikes and stands at each time, for each station. Not that these data are directly dumped from the API, hence updated field is coming from the API and is a timestamp in milliseconds.
  • A stationsevents table keeps tracks of modifications of fields in the stations table. For instance when a mobile station changes position, latitude and longitude are updated, or when a station gains new stands, this table keeps track of the changes.

You should have a look at the init_db function (or run .schema in the resulting SQLite database) to have more details about the structure of these tables, it should be rather self-explicit.

Note: There are currently no ways to explicitly list stations addition / removal. As the API response always contains the data for all the available stations, you can find when a station was created (removed) by looking at the first (last) time a line was added in stationsstats table for this station.


The visualization script generates sequences of PNG images from your database dump. You can then concatenate them in a x264 movie using ffmpeg (or avconv, should be the same command):

cat *.png | ffmpeg -f image2pipe -framerate 10 -i - output.mp4


Code is released under MIT license.