ioggstream/italia-backend

App della Cittadinanza Digitale (backend)

Digital citizenship web and mobile backend

This repository contains the code of the backend used by the web and mobile applications of the Digital citizenship project.

Table of content

• What is this?
• How to run the application
  • Dependencies
  • Installation steps
  • Containers description
  • Environment variables
  • Logs
  • SPID user management
• How to contribute
  • Dependencies
  • Starting steps
  • Generate the API client
• Troubleshooting

---

What is this?

This is the backend that supports the italia-app mobile application.

This project is part of the Italian Digital Citizenship initiative, see the main repository for further information.

Authentication process

The italia-app application will authenticate to the backend in two steps:

1. an initial user initiated SPID authentication process (SAML2 based) that identifies the user and, on success, triggers the creation of a new authentication session (associated to a session token)
2. subsequent requests to the backend will be authenticated via a bearer session token

User authentication

When a client (the mobile app or a browser) wants to login with the backend it will call the /login endpoint with the IDP entityID as parameter in the query string. The backend will then builds an authorization URL and performs a redirect to the chosen IDP. The authentication process will continue to the IDP website. If the authentication process ends with success the IDP will redirect the client to an HTML page with a form that will auto-post itself to the /assertionConsumerService endpoint with a SAMLResponse as an hidden field. The backend will parse and validate the SAMLResponse to extract all the user attributes (fiscal code, first name, last name, email), then it will generates an unique alphanumeric string as token and saves an User object to the SessionStorage service using the token as key. Finally the backend will redirect the client to the value of the environment variable CLIENT_REDIRECTION_URL with the token in the query string. The client must saves the token and use it in all API request.

The code that manage this flow are in the spid-passport package (more info here), and in the src/strategies/spidStrategy.js and src/controllers/authenticationController.js files.

Token authentication

All API requests sent by the client to the backend must have an Authorization: Bearer header with the value of the token obtained from the SPID authentication process. The token is used to retrieve the User object from the SessionStorage service.

The code that manage this flow are in the src/strategies/bearerTokenStrategy.js file.

How to run the application

Dependencies

• Docker and Docker Compose

To fully simulate the SPID authentication process we use the images provided by spid-testenv-backoffice and spid-testenv-identityserver projects.

A Linux/macOS environment is required at the moment.

Installation steps

1. clone the project in a folder called italia-backend
2. go to the project's folder
3. run scripts/build-tools.sh to build the tools Docker image
4. run scripts/yarn.sh to install backend dependencies
5. run scripts/generate-proxy-api-models.sh to generate the models defined in api_proxy.yaml and api_notifications.yaml
6. run scripts/generate-api-client.sh to generate the Autorest API Client
7. run scripts/build.sh to compile the Typescript files
8. run docker-compose up -d to start the containers
9. edit your /etc/hosts file by adding:

localhost    spid-testenv-identityserver
localhost    italia-backend

7. wait a couple of minutes to let the IDP start (or monitor the process with $ tail -f logs/idp/wso2carbon.log)
8. run scripts/import-spid-data.sh to configure the local IDP
9. copy app/.env.example to app/.env and fill the variables with your values
10. point your browser to https://italia-backend

If you are using Docker with a Docker Machine replace localhost with the IP of the Docker Machine (More details here).

Container description

• backend: the backend Node application that serves the web and mobile applications
• spid-testenv-identityserver: the test IDP server
• spid-testenv-backoffice: simple configuration interface to manage the test IDP server

Nginx is reachable at https://italia-backend:80
IDP is reachable at https://spid-testenv-identityserver:9443 (user: admin, password: admin)
IDP simple backoffice is reachable at https://spid-testenv-identityserver:8080

Environment variables

Those are all Environment variables needed by the application:

Variable name	Description	type
API_KEY	The key used to authenticate to the API backend	string
API_URL	The API backend URL	string
CLIENT_REDIRECTION_URL	The path where the user will be redirected after a successful SPID login	string
CLIENT_ERROR_REDIRECTION_URL	The path where the user will be redirected when en error occurs during SPID login	string
PORT	The HTTP port the Express server is listening to	int
REDIS_URL	The URL of a Redis instance	string
TOKEN_DURATION_IN_SECONDS	The number of seconds a session token is considered valid	int
SAML_CALLBACK_URL	The absolute URL of the assertion consumer service endpoint	string
SAML_ISSUER	The issuer id for this Service Provider	string
SAML_ATTRIBUTE_CONSUMING_SERVICE_INDEX	The index in the attribute consumer list	int
PRE_SHARED_KEY	The key shared with the API backend to authenticate the webhook notifications	string
ALLOW_NOTIFY_IP_SOURCE_RANGE	The range in CIDR form of allowed IPs for the webhook notifications	string
AZURE_NH_HUB_NAME	The hub name configured in the Azure Notification HUB	string
AZURE_NH_ENDPOINT	The endpoint URL configured in the Azure Notification HUB	string

Logs

Application logs are saved into the logs folder.

SPID user management

The setup procedure adds some test users to the test IDP server, the full list could be retrieved in spid-batch-import/spid-users.json. To add more users connect to https://spid-testenv-identityserver:8080 and navigate to: service provider > Servizi registrati and click on Utenti.

---

How to contribute

Dependencies

• nodenv
• YARN
• Docker (optional)

A Linux/macOS environment is required at the moment.

Starting steps

• use nodenv to run the correct version of Nodejs as specified in app/.node-version
• run Jest tests directly or with scripts/test.sh

In general follow the Node Best Practices.

Generate the API client

The API client is generated with the AutoRest tool, in case of API change you need to regenerate the client code:

• run the command yarn generate:api-client

Generate SAML (SPID) certs

The backend implements a SAML Service Provider - for authenticating the clients it needs a certificate that you can generate with the following command (you need to have openssl available in your path):

$ yarn generate:test-certs

Architecture decision records

In a world of evolutionary architecture, it's important to record certain design decisions for the benefit of future team members as well as for external oversight. Architecture Decision Records is a technique for capturing important architectural decisions along with their context and consequences. We store these details in source control, along with code, as then they can provide a record that remains in sync with the code itself.

We use ADRs to track architectural decisions of this initiative.

This repository is configured for Nat Pryce's adr-tools.

Here's the decisions we taken so far:

ADR	Title	PR (discussion)
1	Record architecture decisions
2	Backend runs on Docker on local environments
3	Use OpenAPI to defined the API specs
4	Use a dependency injection container

---

Troubleshooting

I installed on my mac but seems that https://italia-backend:80 is not working (ping italia-backend return a host error)

Check out /etc/hosts Remember that in some cases you need to use your docker-machine ip (get it from >docker-machine ip) instead of localhost.

I followed all the steps but when i go to https://italia-backend it shows me the same as https://italia-backend:8080

This problem seems to be dependent on how Docker for Mac (doesn't) manage well the /etc/hosts file. If you install Docker Toolbox it works fine (and can coexist) (Read more at https://medium.com/@itseranga/set-hosts-in-docker-for-mac-2029276fd448)

When i run the scripts/import-spid-data.sh file, after the first entries the script display a lot of errors like # users imported: -- Error [object Object]

Have you waited the IDP to start successfully? Wait a minute and retry.

italia-backend's Issues

Integrate circleci-local for testing

Reproduce

testing goes thru a long series of steps, documented in README.md
The same steps are contained in .circleci/config.yml

I expect

To reuse locally the .circleci/config.yml via circleci-local https://circleci.com/docs/2.0/local-cli/

circleci-local is a script with a docker image capable of running locally your .circleci/config.yml

Simplify the installation

I expect

The installation steps:

• to be easily reproducible
• to be copy-paste able

The README.md not to be a build descriptor ;)

Instead

I need to go thru the README.md.
Changing in the installation procedure require a change to the README.md.

Note

A single-command installation simplifies contributing.

Integrate fixers for linter rules

Reproduce

• modify the source (eg. rename an import)
• run tslint --fix && prettify
• build

I expect

• tslint && prettify to fix everything
• a fixer configuration to be provided

instead

• there are some rules without fixers.