geraldlarner.com: a public archive
About 6,700 old word-processor files, turned into a free public archive that anyone can search. It holds 5,388 programme notes on 445 composers.
Who it is for
Gerald Larner (1936 to 2018) was a music critic. Over five decades he wrote notes for concert programmes. The archive makes them free to performers, students and listeners.
What I started with
About 6,700 old word-processor files, in folders by composer. They were in several old formats, some of which today's software cannot open.
What I built
A set of steps that reads every file, whatever its format, and pulls out the writing. It works out which composer and which piece of music each note is about. It sets aside the files that are not notes at all, such as indexes and logs.
Where the steps could not be sure, I decided by hand and recorded the decision. The whole archive can be rebuilt from the files and come out the same.
A report lists every entry that still needs a human eye. It now lists none.
Then the website: a page for each composer and each piece, and a search across every word.
The site is small and quick. The pages are made in advance, so the server has very little to do when someone asks for one.
What it does
Writing that sat in old files can now be read and searched by anyone: 5,388 notes on 3,522 pieces of music by 445 composers.
Where it is now
It has been public since May 2026.
What I did here
-
Hosting and care
Website hosting and care from the person who built it. Software kept up to date, security kept tight, and one person to call when something needs fixing.
-
Moving off WordPress
Moving your website off WordPress, with your content and web addresses kept. A smaller site your team can still edit, built and looked after by me.
For IT people
A public, searchable archive, built from a pile of old files.
- About 6,700 legacy word-processor files: WordPerfect, RTF (including old Nisus RTF) and Word.
- A Python pipeline converts them with headless LibreOffice, with its own parser for the files LibreOffice cannot read, then classifies each as a work, an essay or source material.
- Manual corrections are keyed by file path, so a rebuild is repeatable.
- 5,388 notes on 3,522 works by 445 composers.
- A thin Express server over one SQLite file of pre-rendered pages, with full-text search and a JSON API.
Tell me what you need
A few lines about the project is enough. I read every message myself. I meet people face to face within about 30 miles of Portsmouth, and work with organisations anywhere in the UK.
Or email hello@passionfruit.design, or call or WhatsApp 07385 509278.