eduroam-db

This repository contains tools to convert institution.xml eduroam database files to new version 2 in JSON format.

It also containts web interface to edit converted JSON files. The web interface is very simple. There is just one simple page with form to edit all the available information. The backend is really simple. There are configuration files for backend that abstract from our data storage, so its simpler if anyone else wants to use this project.

conversion tools

dependencies for conversion tools

Before you begin with the conversion, make sure all the needed dependencies are available:

• python3
• python3-lxml
• python3-dateutil
• python3-tz
• python3-click
• python3-pydispatch
• jq

preparations for conversion

Before running the conversion take a look at config. Edit all the necessary values to suit your needs. You should edit at least ROid, default_lang and local_timezone. Also there is hardcoded default value for number of wired socket for eduroam set to 1!

Also all xml documents from connected institutions must be gathered.

converting institution.xml to institution.json v2

Input filename is used as field instid, ie if input filename is inst1.xml instid is set to inst1. In case there is some NRO's external evidence and you would like to match the instid somehow with your evidence, this is what you can use to do it.

To convert xml files run script:

./converter.py input_directory output_directory

This takes all the .xml files from input_directory and tries to convert them. If the conversion is successfull, all the JSON files are created in the output_directory.

There are various options that can affect the conversion. To see all the options run:

./converter.py --help

building the institution.json

To build the whole institution.json file, use inst_json.sh. This script uses the output of converter.py. It takes the directory with all the created JSON files and concatenates them into desired structure.

Use the script as:

./inst_json.sh converter_output_directory destination_filename

This information changes based on new institutions connecting so a cronjob building the institution.json should be setup if this is used.

validation & debugging

Validator is available here. When the validator reports any errors it could be useful to print out the documents containing errors. This can be simply done with jq. For example if the error is [property] => institutions.institution[2].info_URL[0].data, the problematic document can be printed with jq '.institutions.institution[2].info_URL[0].data' insitution.json.

web interface

The web application is built in node.js.

frontend

The frontend is just one simple page with form to enable user to edit data.

backend

The backend is composed from several API endpoints. The main one is /api/:inst_id for both GET a POST request types.

The GET variant first checks the user permissions. Then it checks the mapping of given realm to institution. If the JSON data file exists, it reads the data and sends them to frontend. If the JSON data file does not exist it performs an LDAP query to get all the data available. After querying LDAP is finished, all the data are sent to frontend.

The POST variant first checks the user permissions. Then it checks the mapping of given realm to institution. POST input is then checked if it is a valid JSON. Finally the input data are validated against eduroam db v2 schema. If all checks are OK, then the data are saved.

configuration

The main idea to use configuration for this was to separate create additional layer that is not dependent on any specific data source.

Configuration is done via JSON format. There are two main configuration files. The first configuration file specifies mapping of realms to administrators. The second configuration file speficies mapping of realms to institution ids as used in conversion process. See below for specific details.

admin to realm mapping

It is necessary for the web application to know which administrator has permissions to edit specific realms. This is done via configration.

The configuration file is placed in web/coverage/config/realm_to_admin.js. The file is not part of the repository for security reasons.

Sample configuration may look like:

module.exports =
{
    "user1@university1.tld": [
        "realm1",
        "realm2",
        "realm3",
        "realm4",
        "realm5"
    ],
    "user1@university2.tld": [
        "realm1",
        "realm3"
    ],
    "user2@university2.tld": [
        "realm6"
    ],
}

The module.exports is just a node.js way of exposing file contents to be able to use one file in another.

realm to institution mapping

It is necessary for the web application to know which realm corresponds to specific instid as used in converter. This is done via configration.

The configuration file is placed in web/coverage/config/realm_to_inst.js. The file is not part of the repository for security reasons.

Sample configuration may look like:

module.exports =
{
    "realm1.cz": "instid1",
    "realm2.cz": "instid2",
    "realm3.cz": "instid3",
    "realm4.cz": "instid4",
}

super admin mode

Application includes a super admin mode as an options for NRO adminitors to edit institutions' coverage information. An administator needs to be manually configured to become super admin. The configuration file is placed in web/coverage/config/admins.js.

Sample configuration may look like:

module.exports =
{
  "admins" : [ "admin1@nro1.tld", "admin2@nro2.tld" ]
}

dependencies

The web application is built in node.js. It needs node.js to run.

automated data update

Since some organizations connected to eduroam might have their eduroam coverage information in some kind of their own database we wanted to provide some way to synchronize their data with this application.

There is an authentication exception for eiditing api in the webserver configuration so anyone can access it even without providing any credentials. If anyone wants to synchronize the data automatically, they need to contact us to provide them an access token which server as an alternative authentication method.

Example:

curl -H "Authorization:my_secret_token" -H 'Content-Type: application/json;charset=UTF-8' -H 'Accept: application/json, text/plain, */*' --data-binary $'{"my":"data"}'

eduroam.cz setup

This seection briefly describes how the export and related stuff is done in eduroam.cz.

The export is done once a day by cronjob. The main script is export.sh. This script uses inst_json.sh to create the institution.json file from all the connected institution's stored coverage data.

Map creation for eduroam.cz website is also part of the export process. Source data for maps are created by map_data.sh.