Release 3.3.0

Upgrading from 3.2.0

To upgrade to this release it is necessary to do the following configuration changes in the solrconfig.xml:

If present, replace the deprecated parameters pixolution.autoContext.fieldname.readFrom and pixolution.autoContext.fieldname.searchin of the queryParser configuration with their new names:

<queryParser name="pixolutionParser" class="de.pixolution.solr.search.PixolutionParserPlugin">
  [...]
  <str name="rank.ac.readfrom">keywords</str>
  <str name="rank.ac.searchin">keywords</str>
</queryParser>

New Features

Added HTTP API for descriptor generation

Besides the CLI and Java API we have added the new requesthandler de.pixolution.solr.handler.component.DescriptorHandler which is providing an HTTP endpoint for generating descriptors. Sending an image URL (file, http, https) as GET parameter to the endpoint will return the corresponding image descriptor. If Solr is running in cloud mode the requests will be randomly distributed to all shards automatically.

Runtime configuration of AutoContext fields

It is now possible to configure the Auto Context fields at runtime as query parameter. This enables faster testing of different fields as well as the possibility to switch between different language fields in production.

Additionally, we changed the parameter names to provide a shorter and more consistent naming scheme:

  • pixolution.autoContext.fieldname.readFrom -> rank.ac.readfrom
  • pixolution.autoContext.fieldname.searchIn -> rank.ac.searchin

Bugfixes

Image loading fails due to AWS URL signature checking

We removed the Content-Type: charset=UTF-8 header field from image loading GET requests. This could cause wrong signatures calculations in AWS cloud and leads to a 403 HTTP status code. Since this field is not necessary for GET requests, we could safely remove it.