alfresco-indexer

What is it?

Alfresco Indexer is an API that allows to index content stored in Alfresco, when you want, how you want, selecting the content you're interested to.

Compatibility Matrix

Alfresco Indexer version	(shipped with) ManifoldCF version	(tested wth) Alfresco edition/version
0.7.x	1.8.0 to 2.2.0-RC0	Community 5.0.[a,b,c,d], Enterprise 4.2.x
0.8.x	trunk (master) - WIP	Community 5.0.d, Enterprise 5.0.x

Community 5.1.[a,b,c]-EA is work in progress (add issue link) There may be other permutations that work but haven't been tested.

Run Tests

git clone git@github.com:maoo/alfresco-indexer.git
mvn clean install -DskipTests
cd alfresco-indexer-webscripts-war
mvn clean integration-test

To know how to build the master and test it against ManifoldCF, follow these instructions

Project Structure

• Alfresco Indexer Webscripts - A server-side component (an AMP that needs to be installed in Alfresco) that exposes a set of Webscripts on Alfresco Repository
• Alfresco Indexer Client - A Java API that wraps HTTP invocations to Alfresco Indexer Webscripts and publishes a simple client interface to interact with Alfresco contents; hereby the most important methods you get access to:

/**
* Fetches nodes from Alfresco which has changed since the provided timestamp.
*
* @param lastAclChangesetId
*         the id of the last ACL changeset already being indexed; it can be considered a "startFrom" param
* @param lastTransactionId
*         the id of the last transaction already being indexed; it can be considered a "startFrom" param
* @return an {@link AlfrescoResponse}
*/
AlfrescoResponse fetchNodes(long lastTransactionId, long lastAclChangesetId, AlfrescoFilters filters) throws
AlfrescoDownException;

/**
* Fetches Node Info from Alfresco for a given node.
* @param nodeUuid the UUID for the node
* @return an {@link AlfrescoResponse}
* @throws AlfrescoDownException
*/
AlfrescoResponse fetchNode(String nodeUuid) throws AlfrescoDownException;

/**
* Fetches metadata from Alfresco for a given node.
* @param nodeUuid
*        the UUID for the node
* @return a map with metadata created from a json object
*/
Map<String, Object> fetchMetadata(String nodeUuid) throws AlfrescoDownException;

Differences with Alfresco-Solr integration

The software architecture of Alfresco Indexer is the same delivered by Alfresco-Solr integration:

• A collection of webscripts (accessible via /alfresco/api/solr/* endpoints) that allow to track transactions and acl change events on Alfresco side
• A Java client that interacts with webscripts and updates Apache Solr indexes

Nevertheless, the following differences can be noted:

• Alfresco Indexer Webscripts are delivered by an AMP, they're not part of the core Alfresco code, as opposed to Alfresco Solr Integration
• Alfresco Indexer is an unsupported, community, experimental effort; Alfresco Solr integration is stable and supported by Alfresco
• Alfresco Indexer is agnostic to the Search Engine to adopt, as opposed to Alfresco-Solr integration; Alfresco ManifoldCF Connector is a great example on how to use Alfresco with other Search Engines (i.e. Elasticsearch)
• Alfresco-Solr integration maintains 2 isolated index structures for transactions and changesets; Alfresco-Indexer maintains 1 index structure with one index entry per Alfresco node, containing a list of readable authorities (readablaAuthorities); as a result:
  1. Alfresco-Solr integration is slower at query time, since document index entries must be cross-referenced with ACL index entries to understand which documents are accessible from the current user
  2. Alfresco Indexer triggers a reindexing of all nodes whose ACL change; a change to /app:Company_Home would trigger a full re-indexing; on the other hand, it doesn't need complex query logic to implement authorisation query parsers for the Search Engine of your choice
• Alfresco Indexer does not provide any Search Engine query parser, as opposed to Alfresco Solr integration, that delivers CMISQL, FTS and Lucene Query query predicates to implement advanced query capabilities; this makes Alfresco Indexer not suitable for any integration with Alfresco clients that rely on these search capabilities, such as Alfresco Share

To summarise, advantages of using Alfresco Indexer:

• Simplified Search Index structure, it improves integration of Alfresco indexing with existing Search engines and index data structures
• The authorization checks are implemented by query parsers by adding security constraints to a given query; there is no post-processing or data-joining activity involved during a query execution

Disadvantages of using Alfresco Indexer:

• If an ACL changes on a node, also all other nodes that inherit from it will be re-indexed, including node properties and content
• Alfresco query parsers cannot be used with this solution, therefore Alfresco Share won't work out of the box

Configuration

Alfresco Indexer Webscripts can be configured to tweak the indexing process; in alfresco-global.properties you can override the following default parameters.

Url Prefixes

indexer.properties.url.prefix = http://localhost:8080/alfresco/service/node/details
indexer.document.url.prefix = http://localhost:8080/alfresco/service/slingshot/node
indexer.content.url.prefix = http://localhost:8080/alfresco/service
indexer.share.url.prefix = http://localhost:8888/share
indexer.preview.url.prefix = http://localhost:8080/alfresco/service
indexer.thumbnail.url.prefix = http://localhost:8080/alfresco/service

Node Changes paging parameters

indexer.changes.nodesperacl=10
indexer.changes.nodespertxn=10

Node Changes allowed Node Types (whitelist)

indexer.changes.allowedTypes={http://www.alfresco.org/model/content/1.0}content,{http://www.alfresco.org/model/content/1.0}folder

Other examples of allowed types:

{http://www.alfresco.org/model/forum/1.0}topic
{http://www.alfresco.org/model/forum/1.0}post
{http://www.alfresco.org/model/content/1.0}person
{http://www.alfresco.org/model/content/1.0}link
{http://www.alfresco.org/model/calendar}calendar
{http://www.alfresco.org/model/calendar}calendarEvent
{http://www.alfresco.org/model/datalist/1.0}dataList
{http://www.alfresco.org/model/datalist/1.0}dataListItem (includes all sub-types, such as dl:task, dl:event and dl:issue)
{http://www.alfresco.org/model/blogintegration/1.0}blogDetails
{http://www.alfresco.org/model/blogintegration/1.0}blogPost

Binaries

Alfresco Indexer binaries can be found in Maven Central; you can use Alfresco Indexer using Apache Maven, simply adding the following dependency in your pom.xml file:

  <dependency>
      <groupId>com.github.maoo.indexer</groupId>
      <artifactId>alfresco-indexer-client</artifactId>
      <version>0.8.0</version>
  </dependency>

Release

Before releasing, make sure you can upload artifacts to Maven Central:

mvn deploy -Pgpg

If everything goes fine, make sure you're up-to-date with git master and run the release command:

git status
netstat -anl | grep 8080 #make sure local port 8080 is free
mvn clean -Ppurge
mvn release:prepare release:perform

Follow sonatype docs for setting up your environment.

Credits

This project was have been developed by

• Alfresco Consultant Maurizio Pillitu
• Findwise (Martin Nycander and Andreas Salomonsson)
• Zaizi (Rafa Haro and Ivan Arroyo)

License

Please see the file LICENSE.md for the copyright licensing conditions attached to this codebase