The scripts in this repository attempt to match real-time bus position information against timetabled journeys and to present the results as a table from which trips and journeys can be viewed on a map.

This document provides instructions for installing them on the TFC platform. See DESCRIPTION.md for a more general description and instructions for a stand-alone installation.

1. Clone this repository in the home directory of the tfc_prod user

2. cd timetable_matching

3. mkdir -p /mnt/sdb1/tfc/cam_tt_matching/json

4. mkdir -p /mnt/sdb1/tfc/tnds/sections

5. ln -s /mnt/sdb1/tfc/cam_tt_matching/ /media/tfc/cam_tt_matching

6. ln -s /mnt/sdb1/tfc/tnds/ /media/tfc/tnds

7. (as root) cp nginx/tt_matching.conf /etc/nginx/includes2/

8. (as root) service nginx restart

9. python3 -m venv venv

10. pip install -r requirements.txt

11. cp setup_environment.skel setup_environment. Edit setup_environment to set at least API_TOKEN, TNDS_USERNAME and TNDS_PASSWORD (or copy from one of the other servers). The default values of the other variables should be fine.

The script refresh_timetable.sh will fetch a copy of the requred timetable files into /media/tfc/tnds. It runs no more than once every 2 hours and only fetched timetable data when it changes.

The script process_day.sh (which itself runs refresh_timetable.sh) takes zero or more 'YYYY-MM-DD' command-line parameters and generates merged data files for the corresponding days. If run with no command-line parameters it generates merged data files for "yesterday". It shouldn't be run for a day that doesn't have complete trip data (like 'today').

On tfc-app4, tfc_prod's crontab file has run process_day.sh daily at 02:45 since late September 2018:

# Extract yesterday's timetable and journeys
45 2 * * * cd /home/tfc_prod/timetable_matching/ && ./process_day.sh >/var/log/tfc_prod/process_day.err 2>&1 && echo $(date --iso-8601=seconds) > /var/log/tfc_prod/process_day.timestamp

The web interface for interrogating the analysed data is at http://<hostname>/backdoor/tt_matching/index.html. It needs to be run with a query parameter identifying the day to analyse, e.g.

http://<hostname>/backdoor/tt_matching/index.html?2018-10-01