|Phyks (Lucas Verney) aaa9b19cd3 Rework SQLite database handling to make the script more efficient||7 months ago|
|.gitignore||1 year ago|
|LICENSE||1 year ago|
|README.md||1 year ago|
|velib.py||7 months ago|
|visu.py||1 year ago|
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.
Note: For now,
pybikes is only Python2 compatible.
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:
stationstable, containing “permanent” information about each station (latitude, longitude, number of stands etc).
stationsstatstable 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
updatedfield is coming from the API and is a timestamp in milliseconds.
stationseventstable keeps tracks of modifications of fields in the
stationstable. For instance when a mobile station changes position,
longitudeare 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
avconv, should be the same command):
cat *.png | ffmpeg -f image2pipe -framerate 10 -i - output.mp4
Code is released under MIT license.