Castalia Quotes

Curated quotations by Castalia faculty authors, linked to texts in Bibliotech with EPUB locators (CFI) for Readest.

Site: https://quotes.castalia.institute/
Quote of the day: HTML · JSON

Supabase

Table faculty_quotes (Castalia Supabase project, shared with Bibliotech):

Column	Purpose
faculty_id	public.faculty.id (e.g. a.shakespeare)
book_id	public.books.id when the EPUB is in corpus
quote_text	The quotation
passage_label	Human citation (act, chapter, poem)
cfi	EPUB Canonical Fragment Identifier for Readest
epub_locator	JSON fallback before CFI is resolved
source / source_id	Bibliotech catalog keys (gutenberg, id)

View faculty_quotes_public adds faculty_name and readest_url.

Migration: supabase/migrations/20260517120000_create_faculty_quotes.sql (also copied to castalia.institute/supabase/migrations).

Seed famous quotes

cp ../castalia.institute/.env .env   # or export SUPABASE_URL + SUPABASE_SERVICE_ROLE_KEY
npm install
npm run seed:dry
npm run seed

Rows with matching books.source + books.source_id get book_id filled automatically.

Resolve CFI in Readest

1. Open the book via readest_url (or Bibliotech library ?book=<uuid>).
2. Navigate to the passage and copy the CFI from developer tools / note sync.
3. Update faculty_quotes.cfi so highlights deep-link correctly (&cfi= on the Readest URL is reserved for this).

GitHub Pages

Static site in docs/ with docs/CNAME = quotes.castalia.institute.

npm run build    # writes docs/quote-of-the-day.{json,html}

Enable Pages → GitHub Actions (workflow pages.yml). Daily cron rebuilds quote-of-the-day.

DNS (Cloudflare)

DNS-only CNAME quotes → castaliainstitute.github.io:

set -a && source ../castalia.institute/.env && set +a
[[ -z "${CLOUDFLARE_API_TOKEN:-}" && -n "${CLOUDFLARE_TOKEN:-}" ]] && export CLOUDFLARE_API_TOKEN="$CLOUDFLARE_TOKEN"
npm run dns

Then in GitHub: Settings → Pages → Custom domain → quotes.castalia.institute.

Topic search (RAG)

Each published quote has a search_text document (quote, passage, work title, tags) and a embedding (vector(1536), OpenAI text-embedding-3-small).

npm run embed              # embed via edge (OpenRouter → Gemini fallback)
npm run search -- "stoicism and resilience"
npm run search -- "libraries paradise" --json

Embeddings use Gemini gemini-embedding-001 at 1536 dimensions on the server (OpenRouter when credits are available). Re-embed after adding quotes: npm run embed or npm run embed:force.

SQL RPC: search_faculty_quotes(query_embedding, match_count, match_threshold, filter_faculty_id)

Edge function: POST /functions/v1/search-quotes with { "query": "…", "limit": 8, "faculty_id": "a.plato" }
Deploy from castalia.institute:

supabase functions deploy search-quotes --project-ref pilmscrodlitdrygabvo

Web UI: quotes.castalia.institute/search.html (requires search-config.js from npm run build:search-config).

When quotes change, re-run npm run embed (trigger clears embedding if search_text changes).

JSON shape (quote of the day)

{
  "date": "2026-05-17",
  "index": 3,
  "total": 12,
  "quote": {
    "faculty_id": "a.woolf",
    "quote_text": "...",
    "passage_label": "...",
    "book_title": "...",
    "readest_url": "https://bibliotech.castalia.institute/read/?book=..."
  }
}