Rows that could not be processed

4 min read

A job can easily finish even though some rows gave no answer — for example an address that can't be recognised, or a BBR id that doesn't exist. The job doesn't stop for that; the row is still in the result with its idx, and the reason is recorded.

If it is instead the whole job that stopped or was rejected, see When a job fails — error codes.

Errors and warnings

Error — the row gave no answer. Counts in the job's number of errors (fejl in the API).
Warning — the row gave an answer, but it is uncertain or empty. Doesn't count as an error.

Rows with errors cost nothing. Rows with a warning did get a result and count as one lookup each in the quota.

Where do I see them?

In the dashboard: the Errors and warnings tab on the job's detail page shows the rows with number, input and reason. Download error rows (CSV) downloads them.
Via the API: GET /v1/jobs/{id} has the field raekke_fejl with the count per reason (pr_aarsag) and warnings separately. GET /v1/jobs/{id}/fejl downloads the rows themselves as CSV (or NDJSON with ?format=ndjson) — see the columns in Follow a job and download the result.

Fix and run again: The error CSV from GET /v1/jobs/{id}/fejl has the input columns with exactly the names the job reads (lat, lon, radius, limit, betegnelse, id, postnr). Fix the values directly in the file and upload it as a new job — the extra columns (raekke, aarsag, forklaring …) are ignored. Use raekke to put the new answers back in the right place in your own data.

The reasons

`ikke_tolket`

Could not be parsed · error · cleansing

The text can't be read as an address. Cleansing must be able to find a street name, a house number and a four-digit postal code in the text — typos in the street name and town are fine, but a missing postal code is not.

Input	Problem
`Rådhuspladsen 1`	No postal code
`Vestergade 2, Aarhus C`	No postal code — town name only
`Rådhuspladsen, 1550 København V`	No house number
`Lageret bag hallen`	Not an address

How to fix it: Write the row as street name, house number and postal code, e.g. Rådhuspladsen 1, 1550 København V. Include floor and door if you use the unit addresses variant (Vesterbrogade 12, 3. th, 1620 København V).

`ikke_fundet`

Not found · error · DAGI

There are no addresses at the coordinates — for example a point at sea or in the middle of an area without addresses.

How to fix it: Check the coordinates — that they are precise enough (preferably 4 decimals, e.g. 55.6761 and 12.5683) and hit the place you mean. If you have the postal code, you can use that instead.

`ukendt_id`

Unknown BBR id · error · BBR

The id doesn't exist in BBR for the selected variant. The row is in the result with fundet = false.

Why: Most often it is a building id in a job with the units variant (or the other way round), or an id from a register other than BBR.

How to fix it: Check that the ids are id_lokalid from BBR, and that the variant matches: building ids → buildings, unit ids → units.

`ukendt_postnr`

Unknown postal code · error · DAGI

The postal code doesn't exist, or there are no addresses in it — e.g. 9999.

How to fix it: Correct it to a Danish four-digit postal code. A missing leading zero is added automatically in CSV (800 → 0800).

`intet_match`

No match · warning · cleansing and reverse

The row gave no reliable answer:

Cleansing: category C — the address couldn't be matched reliably. If there is a guess anyway, it is in bedste_bud.
Reverse: no address within the radius (antal_match = 0).

How to fix it: Cleansing: check the spelling of the street name, and that the postal code matches the address. Reverse: increase radius (up to 5000 metres) for points far from buildings.

`flere_mulige`

Uncertain match · warning · cleansing

The address was found, but not unambiguously — category B. Typically there is a typo, or the floor and door don't quite match. The best guess is in the result and in bedste_bud.

How to fix it: Check the best guess. If it is right, you can use it. Otherwise make the address more precise — fix the spelling, and check the floor and door.

`anden_fejl`

Other error · error · cleansing and DAGI

The row returned an error that doesn't fit the other categories. The error text is in the fejl column of the result CSV (NDJSON: the error field).

How to fix it: Read the error text and fix the row. If it affects many rows, or you don't understand the text, contact support at support@danskadresseapi.dk with the job id and a few of the row numbers.

Was this helpful?