Monday was a holiday this week and I spent most of the four working days on the Books and Borrowing project, although unfortunately my access to the Stirling University VPN stopped working on Wednesday and access wasn’t restored until late on Thursday afternoon.  As I am unable to access the project’s server and database without VPN access this limited what I could do, although I did manage to work on some code ‘blind’ (i.e. without uploading it to the server and testing it out) for much of Wednesday.

For the project this week I generated a list of Haddington book holdings that don’t have any borrowings so the team could test some things.  I also added in author gender as a field and wrote a script to import author gender from a spreadsheet, together with tweaks made to all of the author name fields.  I still need to fully integrate author gender into the CMS, API and front-end, which I will focus on next week.

I spent the rest of my time on the project this week developing a new visualisation for the library facts and figures page.  It is a line chart for plotting the number of borrowing records for book holdings over time and it features an autocomplete box where you can enter the name of a book holding.  As you type, any matching titles appear, with the total number of borrowings in brackets after the title.  If the title is longer than 50 characters it is cropped and ‘…’ is added.  Once you select a book title a line chart is generated with the book’s borrowings plotted.  You can repeat this process to add as many books as you want, and you can also remove a book by pressing on the ‘delete’ icon in the list of selected books above the chart.  You can also press on the book’s title here to perform a search to view all of the associated borrowing records.

Hopefully this chart will be of some use, but its usefulness will really depend on the library and the range and number of borrowings.  So for example the image below for Wigtown shows a comparison of the borrowings for four popular journals, which I think could be pretty useful.  A similar comparison at Chambers is less useful, though, due to the limited date range of borrowing records.

I also spent some time working for the DSL this week.  This included investigating an issue with one of the DSL’s website that I’m not involved with (https://www.macwordle.co.uk/) which was no longer generating any Google Analytics stats.  It turned out that the site had been updated recently and the GA code had not been carried over.  I also fixed another couple of occurrences of the ‘Forbidden’ error that had cropped up due to a change in the way Apache handles space characters.

The rest of my time was spent on the new Solr instance that we’d set up for the project last week.   The DSL team had been testing out the new instance, which I had connected to our test version of the DSL site, and had spotted some inconsistencies.  With the new Solr instance, when a search is limited to a particular source dictionary it fails to return any snippets (the sections of the search results with the term highlighted) while removing the ‘source’ part of the query works fine.  I was unable to find anything online about why this is happening, but by moving the ‘source’ part from the ‘q’ variable to ‘fq’ the highlighting and snippets are returned successfully.

There was also an issue with the length of the snippets being returned, and some snippets being amalgamated and featuring multiple terms rather than being treated separately.  It would appear that the new version of Solr uses a new highlighting method called ‘unified’ and this does not seem to be paying attention to the ‘fragsize’ variable that should set the desired size of the snippet.  I’d set this to 100 but some of the returned snippets were thousands of characters long.  In addition, it amalgamates snippets when the highlighted term is found multiple times in close proximity.  I have now figured out how to revert to the ‘original’ highlighting method and this seems to have got things working in the same way as the old Solr instance (and also addressed the ‘source’ issue mentioned above).  With this identified and fixed the results for the new Solr instance displayed identically to the old Solr instance and on Friday I made the switch to make the live DSL site use the new Solr instance.

Also this week I made some further updates to the STAR website, fixing a number of typos and changing the wording of a few sections.  I also added in a ‘spoiler’ effect to the phonetic transcriptions as found here: https://www.seeingspeech.ac.uk/speechstar/child-speech-error-database/ which blanks out the transcriptions until pressed on.  This is to facilitate teaching, enabling students to write their own transcriptions and then check theirs against the definitive version.  I also finally added citation information for all videos on the live STAR site now.  Video popups should all now contain citation information and links to the videos.  You can also now share URLs to specific videos, for example https://seeingspeech.ac.uk/speechstar/disordered-child-speech-sentences-database/#location=1 and this also works for multiple videos where this option is given, for example https://seeingspeech.ac.uk/speechstar/speech-database/#location=617|549.

Finally this week I made a number of changes to Ophira Gamliel’s new project website, which is now ready to launch.