Get Job Result Rows

Paginate, filter, and sort the rows of an inference-job result file server-side. Sibling of GET /jobs/{job_id}/results (which returns the signed download URL + summary aggregates); this endpoint lets the console render an interactive row table without downloading the full parquet.

GET /jobs/{job_id}/results/rows

Get Job Result Rows

cURL

curl --request GET \
  --url https://api.woodwide.ai/jobs/{job_id}/results/rows \
  --header 'Authorization: Bearer <token>'

200

{
  "rows": [
    {}
  ],
  "offset": 123,
  "limit": 123,
  "total": 123,
  "total_filtered": 123,
  "notice": "<string>"
}

Authorizations

Authorization · string · header · required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Path Parameters

job_id · string · required

Query Parameters

offset · integer · default: 0

Zero-based offset of the first row.

Required range: x >= 0

limit · integer · default: 50

Page size (capped at 500 by the server).

Required range: 1 <= x <= 500

search · string | null

Case-insensitive substring match across every column (mirrors the console's old client-side global filter).

Maximum string length: 200

sort_by · string | null

Column name to sort by. Unknown columns are silently ignored.

Maximum string length: 200

sort_dir · string · default: asc

Sort direction. Ignored when sort_by is not set.

Pattern: ^(asc|desc)$

filter_column · string | null

Exact-value filter on a single column (combined with search). Required together with filter_value. Unknown columns are silently ignored. Used by the segment drill-down to fetch one cluster's rows without abusing the substring search param.

Maximum string length: 200

filter_value · string | null

Value to match against filter_column (string-coerced on both sides). Ignored when filter_column is not set.

Maximum string length: 200

columns · string[] | null

Subset of columns to return (repeatable: ?columns=a&columns=b). Omit to return every column. Unknown names are silently ignored. Lets the console fetch only the columns it renders so a wide combined result file doesn't ship every input column on every page. search still matches across all columns; the projection only narrows the columns returned.

Response

200

application/json

Successful Response

One page of inference-result rows for the console output viewer.

Paginated server-side from the underlying parquet via polars; lets the console drop its in-browser DuckDB-WASM windowed reader for this path. Sibling endpoint of GET /jobs/{id}/results (which returns the signed download URL + summary aggregates).

rows · Rows · object[] · required

One page of inference-result rows as plain JSON objects.

offset · integer · required

Zero-based offset of the first row in this page.

limit · integer · required

Number of rows requested (may be capped server-side).

total · integer · required

Total unfiltered rows in the underlying result file. Independent of any search parameter, so the UI can show of N against the real dataset size.

total_filtered · integer · required

Rows matching the requested search filter (or equal to total when no search applied).

notice · string | null

Optional human-readable note about degraded results — e.g. set when the original input is no longer available and only model predictions (without input columns) could be returned.