After spending much of my time over the past three weeks adding genre to the Books and Borrowing project I turned my attention to other projects for most of this week.  One of my main tasks was to go through the feedback from the Dictionaries of the Scots Language people regarding the new date and quotation searches I’d developed back in September.  There was quite a lot to go through, fixing bugs and updating the functionality and layout of the new features.  This included fixing a bug with the full text Boolean search, which was querying the headword field rather than the full text and changing the way quotation search ranking works.  Previously quotation search results were ranked by the percentage of matching quotes, and if this was the same then the entry with the largest number of quotes would appear higher.  Unfortunately this meant that entries with only one quote ended up ranked higher than entries with large numbers of quotes, not all of which contained the term.  I updated this so that the algorithm now counts the number of matching quotes and ranks primarily on this, only using the percentage of matching quotes when two entries have the same number of matching quotes.  So now a quotation search for ‘dreich’ ranks what are hopefully the most important entries first.

I also updated the display of dates in quotations to make them bold and updated the CSV download option to limit the number of fields that get returned.  I also noticed that when a quotation search exceeded the maximum number of allowed results (e.g. ‘heid’) it was returning no results due to a bug in the code, which I fixed.  I also fixed a bug that was stopping wildcards in quick searches from working as intended and fixed an issue with the question mark wildcard in the advanced headword search.

I then made updates to the layout of the advanced search page, including adding placeholder ‘YYYY’ text to the year boxes, adding a warning about the date range when dates provided are beyond the scope of the dictionaries and overhauling the search help layout down the right of the search form.  The help text scroll down/up was always a bit clunky so I’ve replaced it with what I think is a neater version.  You can see this, and the year warning in the following screenshot:

I also tweaked the layout of the search results page, including updating the way the information about what was search for is displayed, moving some text to a tooltip, moving the ‘hide snippets’ option to the top menu bar and ensuring the warning that is displayed when too many results are returned appears directly above the results.  You can see all of this in the following screenshot:

I then moved onto updates to the sparklines.  The team decided they wanted the gap length between attestations to be increased from 25 to 50 years.  This would mean individual narrow lines would then be grouped into thicker blocks.  They also wanted the SND sparkline to extend to 2005, whereas previously it was cut off at 2000 (with any attestations after this point given the year 2000 in the visualisation).  These updates required me to make changes to the scripts that generate the Solr data and to then regenerate the data and import it into Solr.  This took some time to develop and process, and currently the results are only running on my laptop as it’s likely the team will want further changes made to the data.  The following screenshot shows a sparkline when the gap length was set to 25 years:

And the following screenshot shows the same sparkline with the gap length set to 50 years:

I also updated the dates that are displayed in an entry beside the sparkline to include the full dates of attestation as found in the sparkline tooltip rather than just displaying the first and last dates of attestation.

I completed going through the feedback and making updates on Wednesday and now I need to want and see whether further updates are required before we go live with the new date and quotation search facilities.

I spent the rest of the week working on various projects.  I made a small tweak to remove an erroneous category from the Old English Thesaurus and dealt with a few data issues for the Books and Borrowing project too, including generating spreadsheets of data for checking (e.g. list of all of the distinct borrower titles) and then making updates to the online database after these spreadsheets had been checked.  I also fixed a bug with the genre search, which was joining multiple genre selections with Boolean AND when it should have been joining them with Boolean OR.

I also returned to working for the Anglo-Norman Dictionary.  This included updating the XSLT so that legiturs in variant lists displayed properly (see ‘la noitement (l. l’anoitement))’ here: https://anglo-norman.net/entry/anoitement).  Whilst sorting this out I noticed that some entries would appear to have multiple ‘active’ records in the database – a situation that should not have happened.  After spotting this I did some frantic investigation to understand what was going on.  Thankfully it turned out that the issue has only affected 23 entries, with all but two of them having two active records.  I’m not sure what happened with ‘bland’ to result in 36 active records, and ‘anoitement’ with 9, but I figured out a way to resolve the issue and ensure it doesn’t happen again in future.  I updated the script that publishes holding area entries to ensure any existing ‘active’ records are removed when the new record is published.  Previously the script was only dealing with one ‘active’ entry (as that is all there should have been), which I think may have been how the issue cropped up.  In future the duplicate issue will rectify itself whenever one of the records with duplicate active records is edited – at the point of publication all existing ‘active’ records will be moved to the ‘history’ table.

Also for the AND this week I updated the DTD to ensure that superscript text is allowed in commentaries.  I also removed the embedded Twitter feed from the homepage as it looks like this facility has been permanently removed by Twitter / X.  I’ve also tweaked the logo on narrow screens so it doesn’t display so large, which should make the site better to use on mobile phones and I fixed an issue with the entry proofreader which was referencing an older version of jQuery that no longer existed.  I also fixed the dictionary’s ‘browse up’ facility, which had broken.

I also found some time to return to working on the new map interface for the Iona place-names project and have now added in the full record details.  When you press on a marker to open the popup there is now a ‘View full record’ button.  Pressing on this opens an overlay the same as the ‘Advanced search’ that contains all of the information about the record, in the same way as the record page on the other place-name resources.  This is divided into a tab for general information and another for historical forms as you can see from the following screenshot:

Finally this week I kept project teams updated on another server move that took place overnight on Thursday.  This resulted in downtime for all affected websites, but all was working again the next morning.  I needed to go through all of the websites to ensure they were working as intended after the move, and thankfully all was well.