KAPSI Journal Archive — Proof of Concept

Proof of concept · July 2026

From demo to production

This local pilot combines four live Journal PDFs from the SharePoint corpus with scan-queued placeholder covers for the broader archive structure. The point it proves: we can build the archive ourselves, on infrastructure we already run, behind the login members already have.

1 · Where the scans live

Each publication is scanned (or exported) to PDF, then a one-time pipeline renders every page to web-ready images (WebP, two sizes: reading + thumbnail) stored in Cloudflare R2. R2 has zero egress fees — members reading thousands of pages costs nothing extra.

Rough scale: 500 issues × ~50 pages × ~150 KB ≈ 4 GB of page images (~$0.06/month) plus ~30 GB of archival PDFs (~$0.45/month).

2 · How it's organized

One metadata table in the Supabase database KAPSI Connect already uses. Collections, volumes, seasons, and tags are data — adding a new publication type never requires new code.

create table publications (
  id          uuid primary key default gen_random_uuid(),
  collection  text not null,   -- 'journal' | 'convention' | 'directory' | 'governance'
  title       text not null,
  volume      int,
  issue_no    int,
  season      text,            -- Winter / Spring / Summer / Fall
  pub_date    date not null,
  page_count  int not null,
  pdf_key     text not null,   -- r2: pdfs/journal/1948-winter.pdf
  pages_prefix text not null,  -- r2: pages/journal/1948-winter/
  tags        text[] default '{}',
  ocr_indexed boolean default false
);

3 · Who can see it

The archive mounts as a page inside KAPSI Connect (kapsi1911.com). Members use their existing login; access rules ride on the RBAC roles already built. Final IGS/NGS eligibility and any NGS payment treatment remain governance decisions. Page images are served through signed URLs so links can't be shared outside the portal.

4 · The reading experience

The flipbook you just used is plain CSS/JS — no licensed viewer. In production each page face is simply the scanned image. Phase two adds OCR (searching inside articles, not just titles) using Postgres full-text search.

Build vs. buy

	FlippingBook / issuu class SaaS	In-house (this POC)
Annual cost	~$3–10K/yr, per-publication tiers	< $20/yr infrastructure
Member login	Separate accounts or public links	Existing KAPSI Connect auth
Branding & control	Vendor templates, vendor lock-in	Fully ours, matches the portal
Search across 500 issues	Varies by tier	OCR + full-text, roadmap phase 2

Suggested pilot

Expand the real-PDF test batch to 10 issues spanning several decades.
Run the PDF → page-image pipeline into R2; load metadata into Supabase.
Mount this UI in KAPSI Connect behind member login.
Review with leadership, then schedule bulk digitization of the remaining ~490.