collective.gsa 2.0rc2

GSA integration for external indexing and searching

Introduction

Package collective.gsa integrates Plone site with a Google Search Appliance (GSA). It provides an indexing processor with collective.indexing as well as a search capabilities.

collective.gsa runs on Plone 4.x.

Installation

Add collective.gsa to your buildout.cfg to both eggs and create a part for the separate feeder startup script:

[buildout]
parts = gsafeeder

eggs = collective.gsa

[gsafeeder]
recipe = zc.recipe.egg:scripts
eggs = ${instance:eggs}
arguments = 'gsa host:port', 'source', 'path_to_a_dir_with_generated_feeds'

After running buildout and restarting the server, you can install it via Quick Installer either ZMI or Plone Add/Remove Products. After installing the package the GSA settings and GSA maintenance configlets will appear in the Plone Control Panel. Follow the fields’ description to set it up.

To start or stop the processing daemon run:

./bin/gsafeeder [start|stop|restart]

Global reindex

In the GSA maintenance configlet there is a tool to globally reindex the whole site. If the site is a large one, memory related issues may appear. Thus the reindex allows you to run it piece by piece by batching the objects.

If it is more suitable to run rather more small batches then there is an example script global_reindex.py in the example folder which runs the batch reindexes repeatedly.

Indexing

A indexing subscriber is created and set upon IObjectEditedEvent, IObjectInitializedEvent, IActionSucceededEvent, IObjectRemovedEvent and create relevant xml feeds which are saved on a filesystem where the gsafeeder daemon should process them.

The package contains content providers for objects implementing IATDocument, IATFile and IATContentType.

• For document CTs (Page, News Items etc.) the default page is rendered.

• For file CTs the primary file field is sent.

• For other archetype based CTs the title and description.

To create support for other types just create your own content provider implementing interface IContentProvider and register it via zcml. For details look at the content_provider module and gsa’s configure.zcml

The package supports dual indexing if you have two sites - e.g. secure for edit access and public for anonymous access. The object’s identifier in GSA is its url which is obtained using object’s absolute_url method. Thus all the indexing has to be done from the url you want it to be indexed for ( e.i. not from localhost). In the GSA’s control panel you can set a dual base url for anonymous site. Then the url is constructed using the dual url plus absolute_url_path method.

When reindexing object, the feed id added to a persistent queue and is removed when successfully sent to GSA hence if GSA is unreachable the feed will be send when another object is reindexed.

Fact that GSA received the feed does not mean that it is going to be indexed ( e.i. the url is not in the Matched URLs settings ) If your objects are not indexed, please, check the GSA’s Crawl and Index settings.

Searching

This package replaces the search template and livesearch script to use GSA as a search engine. This is done by adding a gsasearch=on into the search request to avoid using GSA search for internal searches ( such as navigation, folder contents etc. )

The plone’s advanced search is at the default search_form template and does not use GSA at all, because GSA does not handle indexes as zope’s ZCatalog does. However you can use the GSA’s advanced search which url you can set at the local GSA control panel.

Uninstall

To remove collective.gsa just uninstall it via QuickInstaller and remove it from buildout.

Current Status

The basic implementation is nearly finished and we aim to write the neccessary tests for it.

Credit

This code was inspired by collective.solr package and it was kindly sponsored by University of Leicester.

Changelog

2.0rc2 - 2012-04-27

• updated history

• removed purge feature as this may be done on the filesystem

• added timeout to control panel

2.0rc1 - 2012-04-27

• added compatibility for Plone 4.x

• removed dependency on collective.indexing

• moved feed sending into a separate process

• replaced schemaadapterbased configlet by registry

1.0.9 - 2011-01-25

• do not use ?searchterm= if url is not in the portal

1.0.8 - 2011-01-21

• fixed keyword error while not doing gsa search

1.0.7 - 2010-11-11

• use default view for all archetypes based objects

• added client’s frontend to settings

• added ConflictError bypass to all-catching exceptions

• fix login issues with GSA 6.4.0

• when GSA does not accept auth cookie, return public search and display info viewlet to the user

1.0.6 - 2010-06-21

• fix issue when adding searchterm to results url plone did not expect the results to have additional parameters - hardcoded ‘?’ which was then duplicate

• changed some info logging into debug level

1.0.5 - 2009-08-11

• added filter on bad characters - fixes GSA not finishing indexing

• add uninstall method

1.0.4 - 2009-07-13

• added render method to overriden searchbox viewlet - fixes compatibility with 3.1.2

• remove mechanize from required packages ( zope2 has its own with different version)

1.0.3 - 2009-06-24

• Included global_reindex script to run the reindex (by Steven Hayles)

• ‘Start over’ button at gsa-maintenance view only resets the ‘already reindexed objects’ number

1.0.2 - 2009-06-08

• Removed rank from LiveSearch if zero

• Filtering creators only to the existing in the current plone instance

• When reindexing files, commiting after one

• Added straight option for GSA indexer to skip persistent queue ( for global reindex )

1.0.1 - 2009-05-29

• Added better Memory error handling

1.0 - Initial release

• Initial release