rdeknijf/delpher-optimized

Delpher — Historical Dutch Newspapers, Magazines & Books

Search Delpher's full-text OCR'd collection of historical Dutch publications. Particularly useful for finding obituaries, death notices (familieberichten), birth/marriage announcements, and contextual mentions of family members.

No login required. Everything is freely accessible.

What to extract from the user's request

• search terms -- person name, place, event (use quotes for exact phrases)
• collection -- ddd (newspapers), dts (magazines), boeken (books)
• date range -- filter via period facets after searching
• article type -- familiebericht, advertentie, artikel, illustratie met onderschrift

Workflow

Primary method (JSON API)

Delpher exposes a JSON API at /nl/api/results that returns structured data directly. This is 10-20x faster than browser automation and parallelizable.

1. Search

Construct the search URL and fetch with curl:

curl -s 'https://www.delpher.nl/nl/api/results?query=QUERY&coll=COLL&page=PAGE&maxperpage=MAX&sortfield=SORT&order=ORDER'

Parameters:

| Parameter | Values | Default | Notes | |-----------|--------|---------|-------| | query | URL-encoded search string | required | Use %22 for quotes around exact phrases | | coll | ddd (newspapers), dts (magazines), boeken (books) | required | Must specify one collection per request | | page | 1, 2, 3... | 1 | Pagination | | maxperpage | 1-50 | 20 | Number of results per page | | sortfield | date, relevance | relevance | Sort order | | order | asc, desc | desc | Sort direction |

Facet filters (append to URL):

• Type: &facets%5Btype%5D%5B%5D=familiebericht
• Period: &facets%5Bperiode%5D%5B%5D=0%2F19e_eeuw%2F
• Period: &facets%5Bperiode%5D%5B%5D=0%2F20e_eeuw%2F
• Geography: &facets%5Bspatial%5D%5B%5D=Regionaal%2Flokaal

Example -- search for familieberichten mentioning "de Knijf" in newspapers, sorted by date:

curl -s 'https://www.delpher.nl/nl/api/results?query=%22de+Knijf%22&coll=ddd&page=1&maxperpage=50&facets%5Btype%5D%5B%5D=familiebericht&sortfield=date&order=asc'

2. Parse the JSON response

The response is a JSON object:

{
  "numberOfRecords": 253,
  "records": [...],
  "facets": [...]
}

Each record contains:

| Field | Description | Example | |-------|-------------|---------| | title | Article heading | "Familiebericht" | | type | Article type | "familiebericht", "artikel", "advertentie" | | date | Publication date | "1979/04/02 00:00:00" | | papertitle | Newspaper name | "Trouw" | | spatial | Geographic scope | "Landelijk", "Regionaal/lokaal" | | spatialCreation | Place of publication | "Meppel" | | publisher | Publisher | "Organisatie Trouw" | | source | Archive holding the original | "Koninklijke Bibliotheek" | | edition | Day/evening edition | "Dag" | | volume | Volume number | "37" | | issue | Issue number | "10811" | | metadataKey | Unique article identifier | "ABCDDD:010825991:mpeg21:a0169" | | identifier | Resolver URL | resolver.kb.nl link | | accessible | Whether content is viewable | "1" |

Facets show available filters and counts:

{
  "facets": [
    {"name": "periode", "values": ["0/19e_eeuw/ (144)", "0/20e_eeuw/ (596)"]},
    {"name": "type", "values": ["advertentie (123)", "artikel (366)", "familiebericht (253)"]},
    {"name": "spatial", "values": ["Landelijk (245)", "Regionaal/lokaal (459)"]}
  ]
}

3. Get full article OCR text

For any record, fetch the OCR text using the metadataKey:

curl -s 'https://www.delpher.nl/nl/api/resource?coll=ddd&identifier=METADATA_KEY&type=ocr'

This returns JSON with a text field containing the full OCR transcription.

Other available resource types:

| Type | Returns | |------|---------| | ocr | Plain text OCR of the article | | dc | Dublin Core metadata | | dcx | Extended Dublin Core metadata | | alto | ALTO XML with word coordinates | | pdf | PDF of the page | | thumbnail | Thumbnail image URL | | image | Full-size page image URL |

4. Construct article URLs for linking

https://www.delpher.nl/nl/kranten/view?coll=ddd&identifier=METADATA_KEY

For magazines: replace kranten with tijdschriften and ddd with dts. For books: replace kranten with boeken and ddd with boeken.

5. Pagination strategy

• First request: check numberOfRecords to know total results
• For genealogy, usually filter on familiebericht first (reduces result set)
• Use maxperpage=50 to get more results per request
• Iterate pages: page=1, page=2, etc.
• For large result sets, combine search terms to narrow down

SRU endpoint (alternative)

The KB also exposes an SRU 1.2 endpoint for more advanced queries:

curl -s 'https://jsru.kb.nl/sru/sru?version=1.2&operation=searchRetrieve&x-collection=DDD_artikel&query=%22de+Knijf%22&recordSchema=ddd&maximumRecords=20'

This returns XML with the same metadata. Use when you need:

• CQL query syntax for complex boolean searches
• Specific index searches (title, creator, date ranges)
• Integration with library search tools

The JSON API is simpler and preferred for most genealogy use cases.

Fallback (browser automation)

If the API is unavailable, returns errors, or the response format changes, fall back to Playwright browser automation:

Navigate and search

browser_navigate -> https://www.delpher.nl/

Fill the search box (labeled "Zoeken in alle tekstcollecties") with the query. Use double quotes around names for exact phrase matching, e.g. "de Knijf" Bennekom.

Click the "Zoeken" button. Wait for results -- page shows "Bezig met laden" while loading.

Read the overview

Results show counts per collection type:

• krantenartikelen (newspaper articles)
• tijdschriften (magazines)
• boeken (books)
• externe krantenpagina's (external newspaper pages)
• radiobulletins (radio bulletins)

Click the relevant collection to see individual results.

Browse results

Each result shows:

• Article title/heading
• Text snippet with search terms highlighted
• Krantentitel (newspaper name)
• Datum (date)

Filter results

Left sidebar has filters:

• Periode -- century or custom date range
• Soort bericht -- Advertentie, Artikel, Familiebericht (most useful for genealogy)
• Verspreidingsgebied -- Landelijk (national), Regionaal/lokaal
• Krantentitel -- specific newspaper

For genealogy, filter on "Familiebericht" to find death notices and family announcements.

View full article

Click an article to see the full text and a scan of the original newspaper page.

Search tips

• Exact name matching: %22Jan+de+Knijf%22 (URL-encoded quotes)
• Name + place: %22de+Knijf%22+Bennekom finds the name near the place
• Obituaries: add &facets%5Btype%5D%5B%5D=familiebericht to filter
• Spelling variants: try both old and modern spellings. OCR errors are common, especially with f/s, ij/y, and broken ligatures
• Date sorting: use &sortfield=date&order=asc for chronological order
• Parallel searches: run multiple curl commands simultaneously for different name variants or collections (impossible with browser automation)

What you'll typically find

| Type | Genealogical value | |------|-------------------| | Familieberichten | Death notices with age, place, surviving relatives -- confirms dates and family relationships | | Geboorte/Huwelijk | Birth and marriage announcements -- social context, witnesses | | Troepenschepen | Military transport lists -- name, rank, home address | | Examenuitslagen | School exam results -- places people in time and location | | Rechtswezen | Legal proceedings -- property, disputes, context | | Advertenties | Job ads, business listings -- occupations and addresses |

Output format

## Delpher Result -- [article type]

**Source:** [newspaper name], [date]
**Type:** [Familiebericht / Artikel / Advertentie]
**Text:** [OCR text excerpt with key details]

**Genealogical relevance:** [what this tells us -- confirms a date, reveals a relationship, etc.]
**Confidence:** Tier B -- contemporary newspaper record
**Link:** https://www.delpher.nl/nl/kranten/view?coll=ddd&identifier=[metadataKey]

Confidence tier

Newspaper mentions are Tier B evidence -- they're contemporary published records, though OCR can introduce errors in names/dates. The original scan is always available to verify the OCR text. Family notices (familieberichten) are particularly reliable as they were submitted by the family themselves.

Quick reference

# Search newspapers for familieberichten
curl -s 'https://www.delpher.nl/nl/api/results?query=%22SURNAME%22&coll=ddd&facets%5Btype%5D%5B%5D=familiebericht&maxperpage=50&sortfield=date&order=asc'

# Search magazines
curl -s 'https://www.delpher.nl/nl/api/results?query=%22SURNAME%22&coll=dts&maxperpage=50'

# Search books
curl -s 'https://www.delpher.nl/nl/api/results?query=%22SURNAME%22&coll=boeken&maxperpage=50'

# Get OCR text for an article
curl -s 'https://www.delpher.nl/nl/api/resource?coll=ddd&identifier=METADATA_KEY&type=ocr'

# Get metadata for an article
curl -s 'https://www.delpher.nl/nl/api/resource?coll=ddd&identifier=METADATA_KEY&type=dcx'