Dovecot FTS Tika Support Experimental

Dovecot Pro can use Apache Tika to index text in various mail attachments to improve the user's search experience.

Additionally, Tika supports Optical Character Recognition (OCR) which enables certain types of image data (e.g. scanned PDFs, JPEG) to be indexed as well.

Configuration

Apache Tika Software

Docker

Dovecot Pro provides a pre-built Apache Tika OCI container that contains the necessary indexing service. This container includes the necessary components to perform OCR.

These containers are located in the Open-Xchange Container Registry.

Access Instructions

1. Authenticate to registry.open-xchange.com

   docker login registry.open-xchange.com

2. Pull the image.

   docker pull registry.open-xchange.com/dovecot-pro/apache-tika:latest

3. Example command to run the container:

   docker run -d -p 9998:9998 registry.open-xchange.com/dovecot-pro/apache-tika:latest

See Custom Config for instructions on how to configure the Tika container.

Dovecot Configuration

Add fts_decoder_tika_url to the Dovecot configuration:

fts_decoder_driver = tika
fts_decoder_tika_url = http://example.com:9998/tika/

Scaling

Tika acts as a microservice, and each request from Dovecot is independent. Therefore, Tika can be scaled horizontally by adding more nodes.

The customer will need to provide the service to distribute requests between the various Tika nodes. Dovecot should then be configured to point to this distribution endpoint.

Security

By default, the Tika container exposes its service via an unsecured HTTP port.

The customer will need to provide TLS protection if the service needs to be secured. fts_decoder_tika_url supports https URLs.

Tika License

The Tika container is provided under the terms of the Apache License v2.0.