The good ship Bibliosight was due into port at the end of November with the rest of the jiscri fleet, however, as I reported at the time, she found herself in a spot of heavy weather and, after experimenting throughout the project with a more general, unrestricted API, we activated our subscription to Web Services Light only to discover that is a different enough product that it would need another reasonable chunk of time to learn and implement. I’m pleased to report, however, that Mike has been at the helm night and day, battling manfully through the storm, and has managed to bring us back on course!
After some initial problems dealing with an authentication step and setting up a query in such a way that it actually returned an appropriate XML response, it appears that the structure of the XML returned from WS Lite is actually somewhat better organised than from the general API, and more customisable meaning that for our XML transformation step we can simply create our own XML file in the format that we want such that we can transform without having to worry about the oddities that we were seeing with the general API. Mike initially thought that we could do without the XSLT altogether (i.e. have code to output in the formats we need) but that would reduce the flexibility of the process.
A sample record is reproduced below:
<?xml version=”1.0″ encoding=”UTF-8″?>
<searchResponse>
<!– Number of records in the database/editions selected –>
<numberOfItemsSearched>1000</numberOfItemsSearched>
<!– Number of records that match the query parameters –>
<numberOfItemsFound>1</numberOfItemsFound>
<!– Number of records in the result set –>
<numberOfItemsListed>1</numberOfItemsListed>
<!– Date this file was created (generally would be used to date the query execution time) –>
<dateCreated>2009-12-09T15:30:00Z</dateCreated>
<items>
<item>
<!– Seems to be always present –>
<title>Record title</title>
<!– Seems to be always present –>
<authors count=”3″>
<author>Bloggs, J</author>
<author>Smith, J</author>
<author>Sheppard, N</author>
</authors>
<source>
<!– Not always present –>
<bookSeriesTitle>Book series title</bookSeriesTitle>
<!– Seems to be always present –>
<title>Source title</title>
<!– Not always present –>
<volume>10</volume>
<!– Not always present –>
<issue>1</issue>
<!– Not always present –>
<pages>116-126</pages>
<!– Not always present –>
<published>
<!– Not always present –>
<date>JAN</date>
<!– Seems to be always present –>
<year>2008</year>
</published>
</source>
<!– Not always present –>
<keywords count=”2″>
<keyword>keyword 1</keyword>
<keyword>keyword 2</keyword>
</keywords>
<!– Seems to be always present –>
<ut>000252821700009</ut>
</item>
</items>
<!– This section echoes the query parameters used to generate the results –>
<searchRequest>
<queryParameters>
<databaseId>WOS</databaseId>
<!– These are the only editions we seem to be entitled to –>
<editions count=”4″>
<edition collection=”WOS”>SCI</edition>
<edition collection=”WOS”>SSCI</edition>
<edition collection=”WOS”>AHCI</edition>
<edition collection=”WOS”>ISTP</edition>
</editions>
<!– Symbolic time span can’t be used in conjunction with time span –>
<symbolicTimeSpan>1week</symbolicTimeSpan>
<!– This is a DATABASE time span, not a publication time span –>
<timeSpan>
<begin>2008-01-01</begin>
<end>2008-12-31</end>
</timeSpan>
<!– Language is always ‘en” –>
<userQuery language=”en”>AD=(leeds met* univ*)</userQuery>
</queryParameters>
<retrieveParameters>
<!– Currently this is the only available sort field –>
<fields count=”1″>
<field>
<name>Date</name>
<sort>A</sort>
</field>
</fields>
<!– Max returned records (1 – 100) –>
<count>100</count>
<!– Record offset –>
<firstRecord>1</firstRecord>
</retrieveParameters>
</searchRequest>
</searchResponse>
And here is a diagram of how we expect to map the XML onto LOM XML for ingest to intraLibrary (click on the image for full size):
So far so good, now all we need is a UI:
The UI is not yet coupled to the API but the basic components are now pretty much all in place; Mike has aimed to ensure that the client is as flexible as possible – it will allow users to limit a query by a specified date range including recent updates and can also accommodate additional Database IDs should it be possible to plug-in additional databases in the future, for example.
Hopefully we will get the boat floating early in the New Year when we will finally be able to do some user testing as well as disseminating the code under an appropriate licence (probably GNU GENERAL PUBLIC LICENSE Version 3 – http://www.gnu.org/copyleft/gpl.html)
Merry Christmas!
