This collection talks to the iVia installation at INFOMINE, Riverside, California. The software should work with any iVia installation. Cgi scripts are called to do searching of and retrieval from the database. All three are fairly generic CGI scripts; iVia's "theming" ability is used to get the data back in the format we want-- there is a gsdl3 theme. To create a new collection that talks to an iVia database, you need to create the two configuration files, collectionConfig.xml, and buildConfig.xml. The collectionConfig.xml needs to contain any display information for the collection, and any metadata that is available. Also it can contain format information for document display or search result lists if the default is not useful. The buildConfig.xml file needs to specify the IViaProxy serviceRack class. Inside this element should be the url of the iVia installation that you want to connect to. Here is an example: The following describes the iVia cgi scripts. For performing a search: ************************ /cgi-bin/canned_search Parameters: Required parameters: theme=gsdl3&query=xxx Useful parameters: no_of_records_per_page=10&start_page_no=1&fields='kw,su' Other parameters: http://infomine.ucr.edu/help/adders/index.php?section=11 Example: http://infomine.ucr.edu/cgi-bin/canned_search?query=tree& theme=gsdl3&no_of_records_per_page=10&start_page_no=1 Fields include Title (ti), Subject (su), Keyword (kw), Author (au), Description (de) and full text (fu). 'all' will search all fields. It's possible to do uch more advanced searches including phrase searching and bolean queries. The canned_search result page will look something like this: Resources: 58 Ids: 1015 22128 1730 6020 2981 1014 20147 2097 46569 4360 For getting an HTML representation of a record: *********************************************** /cgi-bin/view_record Parameters: Required parameters: theme=gsdl3&record_id=x Example: http://infomine.ucr.edu/cgi-bin/view_record?theme=gsdl3& record_id=2097 The result page will be an HTML fragment; specifically, it will be an HTML table. Again, the formatting is quite flexible. There's one current problem: a lot of the data is displayed as hypertext links to new infomine searches; This should be changed eventually. For now, the Greenstone services process the records, replacing infomine links with Greenstone links where possible. For getting metadata for a set of records: ****************************************** /cgi-bin/view_record_set Parameters: Required parameters: theme=gsdl3&record_id_list=id1,id2,id3 Optional parameter: field_list=f1,f2,f3 Example: http://infomine.ucr.edu/cgi-bin/view_record_set?theme=gsdl3& record_id_list=1015,2128,1730&field_list=url,title The result will look like this: Record: 1015 url: http://www.fruit.affrc.go.jp/index-e.html title: National Institute of Fruit Tree Science (Japan) Record: 2128 url: http://www.publish.csiro.au/journals/ajb/ title: Australian Journal of Botany Record: 1730 url: http://usda.mannlib.cornell.edu/reports/erssor/specialty/fts-bb/ title: Fruit and Tree Nuts : USDA Economics and Statistics System There are many different metadata fields that can be requested, including keywords, subjects, ivia_description, access, expert_created, foreign_source, foreign_id, and so on. Note that the metadata always fits on one line (whitespace is explicitly normalised to space) to make parsing easier. Generally, the metadata is in ISO-8859-15 format (i.e. ISO-8859-1 plus the Euro currency symbol). Metadata is usually plain text, though the description field may contain HTML entities and limited set of HTML elements (bold tags, italic tags, links, and a few others).