Version 0.12.1

May 2012

This is the first maintenance release of the 0.12 series.

Overview

While bug fixes and minor feature were the focus for this release, some bigger changes were included because of their importance. The parsing of documents saw a complete rewrite being now class based and allows for more than one parser per mimetype with sequential fallback. This provides the best text extraction on deployments where users have control over the installation and basic extraction when deploying on the cloud or other environments where users don’t have the ability to install OS level binaries.

Changes

Fabric file (fabfile)

A Fabric file is included to help users not very familiar with Ubuntu, Python and Django install Mayan EDMS, or for system administrators looking to automate the install whether in local or remote systems. At the moment the fabfile will install Mayan EDMS in the same configurations listed in this documentation, that is: (Ubuntu/Debian/Fedora) + virtualenv + Apache + MySQL. Feel free to submit your configuration settings and files for different databases, webserver or Linux distribution. More configurations will be added to the fabfile as more are tested.

Documentation update

The installation instructions were updated to include the installation of the libpng-dev and libjpeg-dev libraries as well as the installation of the poppler-utils package. An additional step to help users test their new installation of Mayan EDMS was also added.

Translations

The Italian translation has been synchronized with the source files at Transifex and finished to %100 completion.

Usability improvements

The index instance view now feature the same multi document action buttons (Submit to OCR, delete, download, etc) as the mail and recent document views.

Better office document conversion

A new method of converting office documents has been implemented, this new method doesn’t require the use of the command line utility UNOCONV. If this new method proves to work better than previous solutions the use of UNOCONV may be deprecated in the future. The conversion method adds just one new configuration option: CONVERTER_LIBREOFFICE_PATH which defaults to /usr/bin/libreoffice.

Better PDF text parsing

Brian E. submitted a patch to use the Poppler package pdftotext utility to extract text from PDF files. This is now the default method Mayan EDMS will execute to try to extract text from a PDF and failing that will fallback to the previous method. This change add a new configuration option: OCR_PDFTOTEXT_PATH to specify the location of the pdftotext executable, it defaults to /usr/bin/pdftotext. Be sure to install the poppler-utils os package to take advantage of this new parser.

Changed defaults

The OCR queue is now active by default when first created during the syncdb phase and the OCR_AUTOMATIC_OCR option now defaults to True. These two changes are made to reduce the steps required for new users to start enjoying the benefits of automatic text extraction from uploaded documents without having to read the documentation and have a more functional default install.

Upgrading from a previous version

Start off by adding the new requirements:

$ pip install -r requirements/production.txt

Migrate existing database schema with:

$ ./manage.py migrate documents

Install the poppler-utils package:

  • Ubuntu, Debian:

    $ apt-get install -y poppler-utils
    
  • Fedora:

    $ yum install -y poppler-utils
    

The upgrade procedure is now complete.

Backward incompatible changes

  • None

Bugs fixed

GitHub issue #25 “Office document conversion error”

Removals

  • None