Introduction

It's surprisingly easy to get this up and running. I'll document this at a high level to begin with, then follow up with more details later.

Screencast Install

Watch fac-back-opac being installed on windows xp from start to finish. Download, unzip, and open fbowindowsinstall.html.

Requirements

• Subversion client
• Java JDK 1.5 or higher - Note: You need a JDK or SDK to run Solr, not just a JRE.
• Solr - tested successfully with Solr versions 1.1 and 1.2
• Python - tested successfully with Python versions 2.4 and (if you're willing to give up I18N support) 2.3
• Django - use the latest development version from Subversion, as Unicode support is required

Details

1. Using Subversion, check out the latest version of FacBackOPAC from the repository. This will create a directory called 'fac-back-opac/' in your current working directory, with two subdirectories ('catalog/' and 'indexer/').
2. Get a dump of your MARC records. Currently, the records must be in MARC21 format, although as part of the indexing process converts them to MARC21XML format it would be wonderful to add an option to directly index MARC21XML records if you have them. It will make your life easier if you include the holdings (in the 999 field) and your accession number is in the 001 field. This is the default with Unicorn.
3. Install Solr:
1. Unzip the zipped nightly build, or extract the tarball: tar xzf solr-YYYY-MM-DD.tar.gz
2. Copy the following files from the indexer/ directory into the solr/conf/ directory:
• solrconfig.xml
• schema.xml
4. Start Solr; we're assuming you're going to be indexing a lot of records, so we use the -Xmx256M option to increase the memory dedicated to Solr. If you have lots of RAM, you can increase this number and increase the SOLR_INDEX_BATCH_SIZE variable in indexer/config/indexer.properties to speed up your indexing:

cd solr
java -Xmx256M -jar start.jar

5. Set your indexing preferences in indexer/config/indexer.properties to match your site's needs. Notice that MAX_RECORDS_TO_ADD is set to a very low number; this is to encourage rapid testing of all the pieces before you try to index all 1 million of your records (or what have you).
6. Index your MARC records. Use the batchIndexer.bat file if you're running on Windows.

cd indexer
sh batchIndexer.sh <MARC_file>

7. Assuming that the indexing was successful, configure catalog/catalog/settings.py with values corresponding to your institution.
8. Start the Django Web application. We'll use the built-in Web server for test purposes. You can optionally specify an IP address that the Web server should listen on (0.0.0.0 means listen to all requests) and a port number that the Web server will run on. The built-in Web server defaults to 127.0.0.1:8000, so it will only listen to requests from a browser running on the same machine.

cd catalog
python manage.py syncdb
python manage.py runserver [IP address[:port-number]]

9. Test the search interface by connecting to the catalog at http://127.0.0.1:8000/catalog/ (trailing slash is required - adjust to reflect your hostname and port number, depending on the options you passed to the runserver command in the previous step). Type in a search term and you should see a list of facets on the left, with a column of Amazon book covers in the next column, the titles matching the OR ed search terms ordered by relevance (by default). Depending on the setting of LOCAL_ITEM_VIEW in catalog/catalog/config.py, the highlighted title will either take you to the "real" ILS catalog detailed record for that item, or to a detailed view entirely within fac-back-opac. You will also see call numbers for each item so fac-back-opac can serve as a backup catalogue when the main ILS dies.