Skip to content
forked from Kopano-dev/grapi

Kopano GRAPI provides a general REST web service for multiple groupware applications/standards. It aims to be largely compatible with Microsoft Graph.

License

Notifications You must be signed in to change notification settings

marlemion/grapi

 
 

Repository files navigation

Kopano GRAPI

Kopano GRAPI provides a general REST web service for multiple groupware applications/standards. It aims to be largely compatible with Microsoft Graph. See COMPAT.MD for compatibility details.

GRAPI is meant to be used directly by Kopano API (kapi) and requires Python 3. The endpoints exposed by GRAPI can map multiple data provider backends into the same API using backend implementations. See the Backends section below for details.

Technologies

  • Python
  • Multiprocessing
  • WSGI
  • MAPI
  • JSON
  • LDAP
  • CalDAV
  • IMAP

Running GRAPI

GRAPI uses a master runner to start its services in a way that they can be accessed in parallel using unix sockets. So to start GRAPI, one uses the MFR script (Master Fleet Runner).

./mfr.py

MFR behaviour can be controlled using commandline parameters. Use ./mfr.py --help to get the details.

Run with Docker

Kopano GRAPI supports Docker to easily be run inside a container. Aside from the obvious container reason, running with Docker also supports to run newer GRAPI versions independent of the global Kopano installation.

Run GRAPI with Docker Swarm

Setup the Docker container in swarm mode like this:

docker service create \
    --read-only \
    --mount type=tmpfs,destination=/tmp \
    --mount type=tmpfs,destination=/run \
	--env KOPANO_GRAPI_USER=$(id -u kapi) \
	--env KOPANO_GRAPI_GROUP=$(id -g kopano) \
    --mount type=bind,source=/etc/ssl/certs,target=/etc/ssl/certs,readonly \
    --mount type=bind,source=/run/kopano-grapi-docker,target=/run/kopano-grapi \
    --env KOPANO_GRAPI_KOPANO_SERVER_URI=https://email.kopano.com:237/kopano \
    --name=grapi \
    kopano/grapi \
    serve

Run GRAPI from Docker image

docker run --rm=true --name=grapi \
	--read-only \
	--tmpfs /tmp \
	--tmpfs /run \
	--env KOPANO_GRAPI_USER=$(id -u kapi) \
	--env KOPANO_GRAPI_GROUP=$(id -g kopano) \
	--volume /run/kopano-grapi-docker:/run/kopano-grapi \
	--volume /run/kopano:/run/kopano \
	kopano/grapi \
	serve

Running GRAPI with Docker exposes the normal command line parameters of GRAPI.

To ensure access, the KOPANO_GRAPI_USER and KOPANO_GRAPI_GROUP environment vars can be set to properly access Kopano Storage Server socket and create GRAPI sockets so that they can be accessed by Kapid (provided as volumes).

Development

GRAPI consists of separate WSGI applications which are run together by MFR in a scalable way to be picked up by Kopano API. Recommended way to develop is to run GRAPI together with Kopano API.

Custom (direct) development startup

During development, it is sometimes easier to use the individual GRAPI services directly. Start the WSGI applications directly for example using gunicorn:

gunicorn3 'grapi.api.v1:RestAPI()'
gunicorn3 'grapi:api.v1:NotifyAPI()'

Another option is to start the devrunner by invoking make. This will run it using the Kopano backend and listening by default on port 8000, with the notify port on 8001.

make start-devrunner

Arguments, such as running a specific backend, can be passed as following:

make start-devrunner ARGS='--backends ldap'

Run unit tests

The unit tests can be run like this:

make test

Integration tests for various backends are described in the backends section.

Backends

Kopano Groupware Storage server backend

The Kopano backend is currently the most tested and supported backend for mail, calendar and directory information. Other available backends are for example LDAP, CalDAV and IMAP, these should be treated as experimental at this moment. Multiple backend can be run simultaneous where for example mail is provided by the IMAP backend and calendar by the CalDAV backend.

Dependencies

  • kopano Python module
  • MAPI Python module

Tests

The kopano backend has integration tests, which require a running kopano-server without users with the DB backend to be used for testing. Tests can be run as following and incluenced by KOPANO_SOCKET, KOPANO_SSLKEY_FILE and KOPANO_SSLKEY_PASS environment variables.

make test-backend-kopano

Coverage can be generated and viewed as following:

make open-integration-cov

LDAP backend

The ldap backend requires two environment variables to be set:

  • LDAP_URI - the ldap uri
  • LDAP_BASEDN - the base dn

If the LDAP server needs authentication for listing users the following environment variables should be set:

  • LDAP_BINDDN
  • LDAP_BINDPW

Dependencies

  • python-ldap Python module

Caldav backend

The caldav backend requires three environment variables to be set:

  • CALDAV_SERVER - the caldav server url
  • GRAPI_USER - the caldav user
  • GRAPI_PASSWORD - the caldav password

Dependencies

  • vobject Python module
  • caldav Python module

IMAP Backend

The IMAP backend only supports IMAP over SSL/TLS and requires three environment variables to be set:

  • IMAP_SERVER - the IMAP server url
  • GRAPI_USER - the IMAP user
  • GRAPI_PASSWORD - the IMAP password

Metrics

GRAPI can expose prometheus metrics if prometheus_client is installed and metrics are enabled using the appropriate flag --with-metrics is set. The env variable prometheus_multiproc_dir must be set to a directory that the client library can use for metrics.

Profiling

GRAPI can generate pstats profiles using the Yappi Python module.

By setting the environment variable named PROFILE_DIR to a writeable directory, GRAPI collects profile information, and generates writes them as pstat files (named after the corresponding process) when GRAPI is stopped.

In addition, GRAPI can also generate profiles for individual requests. To change the profile mode, use the PROFILE_MODE environment variable to request.

To view such profile files, use pstats by running python3 -m pstats profile.prof or similar.

Profile files can be converted to a QCachegrind file with pyprof2calltree. Visualizing profiling can be done by installing pyprof2calltree and running pyprof2calltree -i profile.prof -k.

PROFILE_DIR=/path/to/profiledir/ make start-mfr

Translations

To add a new translation, create or update the pot file.

make -C i18n pot

Then initialize new po file with msginit for for example Dutch.

msginit -i i18n/messages.pot -o i18n/nl.po -l nl

License

See LICENSE.txt for licensing information of this project.

About

Kopano GRAPI provides a general REST web service for multiple groupware applications/standards. It aims to be largely compatible with Microsoft Graph.

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Python 96.6%
  • Shell 2.2%
  • Other 1.2%