DergiPark — National Journal Harvester (TÜBİTAK ULAKBİM)

DergiPark (https://dergipark.org.tr) is TÜBİTAK ULAKBİM's national journal-hosting platform — ~2,537 Turkish scholarly journals on OJS-based infrastructure. This skill harvests its metadata, abstracts, and full-text PDFs reproducibly, by going through the one stable machine surface: a platform-wide OAI-PMH (Open Archives Initiative Protocol for Metadata Harvesting) endpoint, plus open article pages. It never guesses bibliographic data — every field comes from the live endpoint, and it emits BibTeX/RIS locally from parsed meta tags.

The interactive search page is gated behind a human-verification challenge, so this skill does not scrape search over plain HTTP. OAI set-harvest and direct article-page fetches are the robust paths (see references/oai-cookbook.md).

When to Use This Skill

Use it when the request is about getting data out of DergiPark:

• harvest / fetch all articles of a Turkish (DergiPark) journal
• list a journal's archive or find its slug
• get BibTeX or RIS (an export format) for a DergiPark article
• download a DergiPark full-text PDF
• read a journal's aim-and-scope (öz/kapsam — abstract/scope) or its

self-declared index list

Canonical Turkish terms you may see: öz (abstract), kapsam (scope), dergi (journal), künye (citation/masthead). Preserve Turkish spelling with diacritics in all user-facing strings (e.g. Mülkiye Dergisi).

Does NOT Trigger — route adjacent requests to the right sibling

How It Works — Three Scripts

All run with uv run python from the skill directory. No API key. Each prefers requests if installed, else falls back to stdlib urllib. Full recipes and gotchas are in references/oai-cookbook.md; the verified endpoint table is in references/endpoints.md.

1. scripts/dergipark_oai.py — OAI harvest (primary path)

# Find a journal's slug (slug == OAI setSpec)
uv run python scripts/dergipark_oai.py list-journals --tsv | rg -i "mülkiye"

# Harvest a whole journal (follows resumptionToken paging automatically)
uv run python scripts/dergipark_oai.py harvest mulkiye --prefix oai_dc --out mulkiye.json

# One article's structured oai_dc record
uv run python scripts/dergipark_oai.py get 10 --prefix oai_dc

Verbs: Identify / ListSets / ListRecords / GetRecord. Prefixes: oai_dc (parsed into clean fields) and oai_mods / oai_marc / oai_etdms (returned as raw record XML for callers that want the richer schema).

2. scripts/article_meta.py — one article → BibTeX / RIS

uv run python scripts/article_meta.py /en/pub/mulkiye/article/10 --format bibtex
uv run python scripts/article_meta.py /en/pub/mulkiye/article/10 --format ris
uv run python scripts/article_meta.py /en/pub/mulkiye/article/10 --format json

Parses the page's citation_* + DC.* meta tags and formats BibTeX/RIS locally — it does not call any on-site export backend (that route is unverified). The JSON output includes a resolved pdf_url.

3. scripts/journal_info.py — aim-and-scope + self-declared indexes

uv run python scripts/journal_info.py mulkiye --lang en

Returns the aim-and-scope text and the /indexes page, the latter stamped with a DISCLAIMER and cross-check pointers to TR Dizin and DOAJ.

Three Rules That Prevent Wrong Answers

1. Publication date ≠ OAI datestamp. The OAI <datestamp> is the platform

re-index time, not the publication date. To slice by publication year, harvest then filter on `dc:date` locally — do not rely on from/until.

1. PDF id ≠ article id. Take the PDF link from the page's citation_pdf_url

meta tag (e.g. /en/download/article-file/9 for article /10); never build it from the article id.

1. Hosting ≠ indexing. Being on DergiPark says nothing about quality or TR

Dizin coverage. The /indexes page is self-declared and unverified by DergiPark. For an authoritative TR Dizin status verdict, route to alterlab-trdizin. See references/indexing.md.

Graceful Degradation

Network failures surface as a network_unavailable / oai_error JSON object on stderr (exit 1) — the scripts never fabricate a record or a verdict. If the endpoint is unreachable, say so and stop; do not invent metadata from memory.

References

• references/endpoints.md — every verified endpoint, prefix, and identifier

scheme (with what is explicitly UNVERIFIED), observed live 2026-06-06.

• references/oai-cookbook.md — copy-paste recipes: slug lookup, full harvest,

year filtering, PDF download, and the gated-search workaround via Playwright.

• references/indexing.md — the DergiPark-vs-TR Dizin distinction and how to

cross-check real indexing against TR Dizin and DOAJ.

Part of the AlterLab Academic Skills suite.