estonian-newspaper-archive

Estonian newspaper archive (DEA / DIGAR)

Base: https://dea.digar.ee/ (Veridian). Be polite: ~1 s between requests, honest User-Agent.

Endpoints

# 1. Search (25 hits/page; r=26 for page 2; quoted phrases work; year range optional)
curl "https://dea.digar.ee/?a=q&hs=1&r=1&results=1&txq=<url_encoded_query>&txf=txIN&ssnip=txt&dafyq=<start_year>&datyq=<end_year>"
# hit doc_ids are in: href="/?a=d&d=<doc_id>&srpos=..."

# 2. OCR text of one document (strip tags to read)
curl "https://dea.digar.ee/?a=d&d=<doc_id>&f=XML"

# 3. Clipping image of the document (~100 KB JPEG; best for reading/verifying)
curl "https://dea.digar.ee/?a=is&oid=<doc_id>&type=blockimage&area=1&width=1400"

# 4. Whole issue PDF (10–30 MB); issue_id = doc_id prefix before the first dot
curl -L "https://dea.digar.ee/?a=is&oid=<issue_id>&type=staticpdf"

Human-readable page for citing: https://dea.digar.ee/?a=d&d=<doc_id>

doc_id carries metadata — use it before fetching

virumaateataja20150526.2.6.1 = title + date (YYYYMMDD) + node path. So the search hit list alone gives you dated, sourced results. Nodes .2.N… are articles; .1.N are whole-page nodes — their block images are useless header fragments, use the DEA page or issue PDF for those.

Traps

• Restricted documents: XML containing piiratud is searchable, but its text and images may be unavailable outside authorized networks. Detect and report the restriction; do not claim to have retrieved blocked content.
• OCR structure: flattened XML may include navigation document IDs. Remove tokens matching [a-z]+\d{8}[\d.]* before parsing extracted data. OCR can scramble page structure, so verify important quotations, names, dates, and numbers against the clipping image or issue PDF.
• Block areas: area 1 is often just the headline; try areas 1–3 (non-image or tiny response = no more areas).