Auto tagging for Plone
This module allows automatic keyword tagging of multimedia content with a strong focus on images.
Inqbus.tagging processes metadata of the following keyword sources: Filename/Title, EXIF, IPTC, XMP. Any Plone Object-Type can be processed.
Keywords from Filename
The metadata is filtered by configurable filters. For each metadata source there is a filter which can be switched on/off independently.
The filename/title may be filtered by a regular expression you are free to craft. The regular expression enables you to filter, strip, split into words, what you like. You can decide if you like to have new keywords to be extracted from the title or if you want use only keywords from the title that match existing ones, or both strategies with different regular expressions in parallel.
Keywords from EXIF, IPTC and XMP
EXIF, IPTC and XMP are filtered firstly by a positive list of tags for each category.
For every tag in a filter can be defined a regular expression and a format string to cut and format the way you like. Lets assume you have metadata tags with the structure “Asimov, Isaac; Bradley, Alex” you can transform them into the Plone tags “Isaac Asimov” and “Alex Bradley” easily.
Since there are lots of possible tags available inqbus.tagging comes with tag import views for each metadata category. Each tag import view allows for opening an arbitrary file in Plone to inspect its metadata and to select and transfer metadata tag names into the tag configuration.
Improved manual tagging of images
Inqbus.tagging supports manual tagging of images by providing a preview image column in the folder_contents-View. Also inqbus.tagging brings lossless EXIF image auto rotation back to Plone.
Retagging of already uploaded content
If you have changed your autotagging config you can use the “retag” Button in the folder_contents-View to rerun the auto tagging on certain objects.
This code is work in progress crafted in after hours. It may not be save for production sites, but runs fine in our setup. We welcome anyone to improve the code.
- Plone 5
- (only for auto rotation) jpegtran-cffi (libjpeg8, cython, cffi)
Add inqbus.tagging to your buildout eggs.:
eggs=\ ... inqbus.tagging
To use inqbus.tagging with jpegtran-cffi add inqbus.tagging[jpegtran] to your buildout eggs.:
eggs=\ ... inqbus.tagging[jpegtran]
jpegtran-cffi is used for the rotation of the images. If you do not use it images will be rotated using Pillow. This leads to metadata loss especially xmp-data.
For removing all stored configurations run the profile profile-inqbus.tagging:purge.
Then go to Configuration -> Extensions. Select uninstall inqbus.tagging.
Go to Configuration -> inqbus.tagging Configuration:
Enable Auto-Tagging for other Contenttypes
Register a subscriber like:
<subscriber for="plone.app.contenttypes.content.Image zope.lifecycleevent.IObjectCreatedEvent" handler="inqbus.tagging.subscriber.title_based.title_to_tag" />
for every contenttype you want to tag.
To make working with Images more easier a preview-column was added to the folder_contents-View and can be selected like all other columns.
To migrate existing objects you can use the Retag-Action in the folder_contents-View.
Processing XMP is done by parsing the RDF:XML-Data structure directly utilizing LXML. Usually XMP is parsed by tools based on Adobe’s XMP-Toolkit which fiddles a lot with the tag values to make them appear right. There seems to be no Python XMP-Lib out there which does not requires an image file name to process a file. Even in the underlying C++ Code there appears no good entry point for processing image data from a ZODB-Blob. So we decided to parse the XMP “by hand” and do without the Adobe corrections. So please be not disappointed if the XMP tags you have applied to your image by Photoshop 10 years ago may come out a bit askew.
We take part in a small foto group in germany. Our site http://fotogruppe-altenstadt.de presents 20.000+ high quality images tagged by 2000+ keywords under CC license. Using Plone for a long time we like to give back our knowledge to the community.
- Added MANIFEST.in to make package installable
1.0.1 and 1.0.2
- Some smaller changes ind Readme and setup.py to improve the documentation
1.0 - Initial release
- Autotagging for Images based on meta information and title
- Autorotation for Images based on Exif
- Possibilities to configure Autotagging
- Managing Keywords by a modified KeywordManager
- Retaging and Previews for Folder-Contents-View