sources.yaml

# Quebec institutional repositories — all speak OAI-PMH 2.0.
#
# `metadata_prefix` defaults to `oai_dc`. We use richer prefixes when the
# source exposes a qualified discipline / department field that's lost in
# basic Dublin Core. See `docs/source_metadata_audit.md` for the full
# inventory of what each source carries.
#
# Mapping rationale
#   - DSpace (bishops, udem, usherbrooke, laval): `dim` preserves the
#     qualified `dc:subject@discipline` and `*:degree.discipline` fields
#   - Some EPrints expose `oai_etdms` with explicit `etdms:discipline`
#     (concordia, uqtr, rlibre, polymtl)
#   - Other EPrints expose `uketd_dc` with `uketdterms:department`
#     (uqam, uqac, uqar, uqo, uqat, inrs, ets)
#   - ÉTS exposes uketd_dc with `qualificationlevel` (doctoral/masters) —
#     critical because dc:type collapses everything to "Mémoire ou thèse"

- id: concordia
  name: Concordia University (Spectrum)
  short: Concordia
  platform: eprints
  base_url: https://spectrum.library.concordia.ca/cgi/oai2
  metadata_prefix: oai_etdms

- id: udem
  name: Université de Montréal (Papyrus)
  short: UdeM
  platform: dspace
  base_url: https://umontreal.scholaris.ca/server/oai/request
  metadata_prefix: dim

- id: usherbrooke
  name: Université de Sherbrooke (Savoirs UdeS)
  short: Sherbrooke
  platform: dspace
  base_url: https://usherbrooke.scholaris.ca/server/oai/request
  metadata_prefix: dim

- id: bishops
  name: Bishop's University (IRis)
  short: Bishop's
  platform: dspace
  base_url: https://ubishops.scholaris.ca/server/oai/request
  metadata_prefix: dim

- id: uqam
  name: UQAM (Archipel)
  short: UQAM
  platform: eprints
  base_url: https://archipel.uqam.ca/cgi/oai2
  metadata_prefix: uketd_dc

- id: uqtr
  name: UQTR (Cognitio)
  short: UQTR
  platform: eprints
  base_url: https://depot-e.uqtr.ca/cgi/oai2
  metadata_prefix: oai_etdms

- id: uqac
  name: UQAC (Constellation)
  short: UQAC
  platform: eprints
  base_url: https://constellation.uqac.ca/cgi/oai2
  metadata_prefix: uketd_dc

- id: uqar
  name: UQAR (Sémaphore)
  short: UQAR
  platform: eprints
  base_url: https://semaphore.uqar.ca/cgi/oai2
  metadata_prefix: uketd_dc

- id: uqo
  name: UQO (Dépôt institutionnel)
  short: UQO
  platform: eprints
  base_url: https://di.uqo.ca/cgi/oai2
  metadata_prefix: uketd_dc

- id: uqat
  name: UQAT (Depositum)
  short: UQAT
  platform: eprints
  base_url: https://depositum.uqat.ca/cgi/oai2
  metadata_prefix: uketd_dc

- id: inrs
  name: INRS (EspaceINRS)
  short: INRS
  platform: eprints
  base_url: https://espace.inrs.ca/cgi/oai2
  metadata_prefix: uketd_dc

- id: ets
  name: ÉTS (Espace ETS)
  short: ÉTS
  platform: eprints
  base_url: https://espace.etsmtl.ca/cgi/oai2
  metadata_prefix: uketd_dc

- id: rlibre
  name: TÉLUQ (R-Libre)
  short: TÉLUQ
  platform: eprints
  base_url: https://r-libre.teluq.ca/cgi/oai2
  metadata_prefix: oai_etdms

- id: polymtl
  name: Polytechnique Montréal (PolyPublie)
  short: Polytechnique
  platform: eprints
  base_url: https://publications.polymtl.ca/cgi/oai2
  metadata_prefix: oai_etdms
  # Polymtl's `set` filter (hex-encoded EPrints style) doesn't apply with
  # `oai_etdms`, but the etdms feed is already thesis-only — no set needed.

- id: laval
  name: Université Laval (Corpus UL)
  short: Laval
  platform: dspace
  base_url: https://corpus.ulaval.ca/oai/request
  metadata_prefix: dim
  set: col_20.500.11794_17461   # collection « Thèses et mémoires »

# TODO: Érudit OAI investigation results (roadmap 2.3)
#   - Endpoints tried:
#       https://oai.erudit.org/oai/request?verb=Identify              -> OK (Fedora-OAI provider)
#       https://www.erudit.org/oai/request?verb=Identify              -> 404 HTML
#       https://api.erudit.org/oai/request?verb=Identify              -> DNS no-resolve
#       https://api.erudit.org/v1/oai-pmh?verb=Identify               -> DNS no-resolve
#   - Findings:
#       * Identify works, oai_dc + nlm + oai_openaire prefixes are advertised.
#       * ListSets returns ~150 sets, ALL of them journals (`erudit.revue.*`)
#         or cultural magazines (`erudit.culture.*`). No thesis/memoire/dissertation
#         set exists.
#       * Sample records (000001ar … 000007ar) show dc:type=text and
#         dc:relation=<journal name; vol/no>, i.e. journal articles only.
#       * Érudit's thesis section (www.erudit.org/fr/theses/) federates theses
#         that already live in their home repositories (Papyrus, Savoirs UdeS,
#         Archipel, etc.) — those are already harvested directly by us.
#       * Conclusion: adding the OAI feed would harvest ~0 theses (or duplicate
#             what we already have via the home repos). Not added on purpose.
#   - Possible alternative: api.erudit.org/v1 (REST API) — would need a custom
#     harvester and per-thesis fetches; only worth doing if a future audit shows
#     theses Érudit hosts that are NOT in any of the 14 OAI sources we already
#     cover.