kgpz_web/techstack.md

Tech Stack for KGPZ

Dynamic

• Server Rendered
• No DB, just use Git for versioning
  • Need to not reclone everything on change, but pull
  • Need to have a URL to register a webhook (in process)
  • Need to have a fallback -> first clone & and all pulls to disk? -> we use this to be able to manually put the repo in the base dir
  • Settings: Github Webhook URL, Maybe second path for dynamically cloned data
• Either: This is dynamic, implement metadata search as dynamically generated
  • Or: This is static, implement metadata search with lunr.js or sth similar.

Static

• Implement SSG, by enumeration of endpoint parameters
• Implement basic search for metadata (persons, places, works, etc) in lunr.js and have it work client side -> Maybe we need to implement metadata search client-side with priority since it can be combined with the dynamic approach. -> {{- if .enumerate -}} could work implementing different functions on fitering
• No full text search available in this model

Common

• HTMX + Go, with progressive enhancement
  • Settings: Git URL, folder path for repo
• Integration with Geonames:
  • Need to pull in and read geonames data for places: alternative names, country, lat/long, wikipedia, maybe even dbpedia
  • Cache geonames data in a local folder
  • Settings: Geonames URL, folder path for geonames data
• First Version: use simple image folder for images -> Maybe push everything to S3 storage
• We could implement aggressive caching, but im not sure if it is necessary

On Startup

• Read in settings
• Read local git directory and parse as described below, creating caches (use geonames and gnd folders for caching to not overload the services)

On Parse

• Read in files and save in local folder
• Get commit date & hash from git if possible
• Read in structs: persons, works, places, stuecke, beitraege
  • Persons: Get metadata from GND (background job) to enrich person structs, where not availble, update, if possible -> Use local folder as cache, and dont update in development
    • Endpoints: Person overview(s), person search
  • Works: do nothing
    • Endpoints: Works, all Beiträge pages
  • Places: get metadata from geonames, if available, dont update (just do it once) -> Use local folder as cache, and dont update
    • Endpoints: places overview
  • Stücke: validate dates, get weekdays, get human readable months
    • Endpoints: Stücke overview (Index), Date search
  • Categories: read in category descriptions
    • Endpoints: Categories overview
  • Beiträge: devise categories from beiträge, if possible
    • Endpoints: all Beträge pages

On Webhook Update

• Pull in data
• Restart

File Inaccurancies:

• No canonical names for Actors, can use GND for most of them
  • Done, we can sort by akteur@id
• URL values YYYY/ST and YYYY/ST/P and YYYY/ST/Beilage/P? Are they unique?
  • Yes, they are unique
• Which names of places should be used? Example: Mitau, Jelgava, Jelgava (Mitau), Mitau (Jelgava)
  • Historical names
• Titel der Werke: not saved yet
• Kurzzitat: KGPZ St. 2, 14 Januar 1774 ?
  • Besser: KGPZ 2/1774, 14. Januar
  • Done
• Navigation: Galerie / Durchklicken??
• Weniger Kästen / Volle Seitenbreite ausnutzen
  • Horizontale Jahresnavigation durch vertikale ersetzen, so wie in der Einzelansicht
• Was ist "Link auf seite teilen"?
  • Permalink

Überarbeiten

• Sekundärüberlieferung in Stück/Beitrag
• Fehlende Daten in Stück/Beitrag
• Kurzinformation Überlieferung
• Link je Jahrgang gleich, in der Primärüberlieferung
• Anzahl der Beiträge in der Primärüberlieferung
• Logo
• Sortiername ohne GND eintrag
  • Done

(([^(\d{4}–)(])(\d{4}–)(\d{4})([^)]*)) ($1$2$3$4)

([\s(])((?:16|17|18|19|20)\d{2}–)*((?:16|17|18|19|20)\d{2}) $1$2$3