Skip to content

Latest commit

 

History

History
81 lines (59 loc) · 3.65 KB

README.md

File metadata and controls

81 lines (59 loc) · 3.65 KB
Raw

AWS Cloudsearch Plugin

Cloudsearch Plugin interfaces with the Amazon Cloudsearch service to provide search capabilities for any FarCry application. Cloudsearch is an API based implementation of the SOLR search engine, and the plugin works similarly to the FarCry Solr Pro Plugin.

NOTE: This plugin is compatible with FarCry 7.x and over.

Cloudsearch plugin is available under LGPL and compatible with the open source and commercial licenses of FarCry Core

Base features include:

  • config for general settings
  • a content type with records for each indexed type, and settings to configure property weighting
  • an event handler that triggers index updates on save and delete
  • a library in the application scope that encapsulates searches and index updates

Setup

  • Set up an account with Amazon Cloudsearch
  • Add this plugin to the project's plugin directory ./plugins/cloudsearch
  • Register the plugin in the projects constructor; ./www/farcryConstructor.cfm
  • Restart your application
  • Open the "Cloudsearch" config and fill in the details of your service account
  • Update the Dockerfile to install the relevant AWS jars. The entire SDK can be installed using: 
    RUN wget -q https://daemon-provisioning-resources.s3.ap-southeast-2.amazonaws.com/aws/aws-java-sdk-1.12.233.zip -O /tmp/aws-sdk.zip && \
  unzip -j /tmp/aws-sdk.zip -d /usr/local/tomcat/lib/ && \
  rm /tmp/aws-sdk.zip

AWS Cloudsearch

Notes about CloudSearch architecture:

  • index configs (i.e. fields, field types, weighting) are pre-defined in CloudSearch
  • after changing index configs, a re-index must be requested
  • changing index configs can potentially cause FailedToValidate errors on document fields (usually if the field type has changed)during the next index, and indexing will fail until the offending documents are updated

Cloudsearch Features Implemented

  1. CloudSearch initialization, discovery, and basic troubleshooting tools, i.e. are we connected, how many documents are there
  2. Basic configuration for type indexes, suitable for dmHTML for example
  3. Manual index updates; webtop overview tab that shows what would be sent to CloudSearch / is in CloudSearch
  4. Search tool in webtop for troubleshooting
  5. Handling of FailedToValidate, i.e. re-uploading problem documents
  6. Automatic index updates on save / delete
  7. Bulk update function for
    • new indexes
    • changed index configs, i.e. adding properties to index
  8. Faceted searches

Cloudsearch Features Not Yet Implemented

We didn't get time for some features, but would like to have them included at some point:

  • a default search form and results interface (this needs to be implemented in the project)
  • document boosting / search
  • search term highlighting
  • autocomplete / suggestions
  • file document content (such as PDF, DOC, etc)

Cloudsearch - Reindex Content Type

In Farcry webtop, when in 'Content Type Indexes', there is an option to 'Index all records' for each Content Type. If a new record for that content is added during the indexing, it sets 'build to date' to now, and only records after that timestamp will be index, which means older records will not be indexed.

To index all records for a given type, run this script: /webtop/index.cfm?typename=csContentType&view=webtopPageModal&bodyView=webtopBodyUploadTypeEverything&CONTENTTYPE=dspArticle

Cloudsearch - Export / Import

/webtop/index.cfm?typename=csContentType&view=webtopPageModal&bodyView=webtopBodyIndexExport /webtop/index.cfm?typename=csContentType&view=webtopPageModal&bodyView=webtopBodyIndexImport