PUT /datasets/{id}/data — Upload Data to a Dataset

Upload a CSV or gzip-compressed CSV file to add, update, or replace dataset rows. Supports a column manifest, conflict resolution flags, and async processing.

Use this endpoint to load rows into a dataset by uploading a CSV file. The upload is processed asynchronously: the request returns immediately with a process record, and the rows become available in the dataset once the background job completes.

This page is the API reference. For the concepts behind the request — what a manifest is, what gets validated, and how the conflict-resolution flags interact — see the Uploading Data section.

Authentication

Authenticate using HTTP Basic Auth. Pass your API key as the username and leave the password empty. You must have at least Write permission on the target dataset.

Request

Method: PUT
URL: https://api.alphacast.io/datasets/{dataset_id}/data
Content-Type: multipart/form-data

Path parameters

dataset_id · integer · required

The numeric ID of the dataset to upload data into.

Form fields

data · file · required

The CSV file to upload. Accepts plain .csv files or gzip-compressed .csv.gz files. The first row must be a header row containing column names.

manifest · string

A JSON string declaring each column’s data type and entity flag. Required on the dataset’s first upload (or set the manifest beforehand from the web UI). On subsequent uploads it is optional — pass it only when you want to replace the dataset’s stored manifest. See the Manifest reference for the full schema.

Query parameters

deleteMissingFromDB · boolean · default: "false"

When true, rows present in the existing dataset but absent from the upload are deleted. Use this to perform a full replacement. Forced to false if the dataset’s manifest is locked. See Upload modes.

onConflictUpdateDB · boolean · default: "false"

When true, rows whose entity key matches an existing row are updated with the new value. When false, the existing value is kept and the upload’s value is discarded for those rows.

acceptNewColumns · boolean · default: "false"

When true, columns in the CSV that are not in the dataset’s existing schema are added. When false, unknown columns cause the upload to fail with Unknown column(s): .... Forced to false if the manifest is locked.

Response

Returns 201 Created with a process object describing the background upload job.

id · string

The ID of the upload process. Use this to poll the dataset processes endpoint and inspect status, stats, and any error message.

status · string

Initial status — typically Requested. Transitions to Processing → Processed (or Error) as the background worker runs. See the upload lifecycle.

Example

Upload a CSV file with a manifest:

cURL

curl -u YOUR_API_KEY: \
  -X PUT https://api.alphacast.io/datasets/123/data \
  -F "data=@gdp_data.csv" \
  -F 'manifest=[{"sourceName":"Date","dataType":"Date","isEntity":true},{"sourceName":"Country","dataType":"String","isEntity":true},{"sourceName":"GDP","dataType":"Decimal","isEntity":false}]'

Upload a gzip-compressed CSV and replace any existing row not present in the file:

curl -u YOUR_API_KEY: \
  -X PUT "https://api.alphacast.io/datasets/123/data?deleteMissingFromDB=true&onConflictUpdateDB=true" \
  -F "data=@gdp_data.csv.gz"

Upload a plain CSV and allow new columns to extend the schema:

curl -u YOUR_API_KEY: \
  -X PUT "https://api.alphacast.io/datasets/123/data?acceptNewColumns=true" \
  -F "data=@gdp_data.csv"

Error responses

The API surfaces validation errors via the upload process record (status Error, statusDescription set to the message). The HTTP response itself uses standard codes:

Status	Cause
400 Bad Request	The manifest is malformed, or required form fields are missing.
401 Unauthorized	Missing or invalid API key.
403 Forbidden	You do not have Write permission on this dataset.
404 Not Found	No dataset with the given ID exists or is accessible to your account.

For row-level validation failures (Data has not Date column, duplicate rows, out-of-range Short Integer, etc.) the HTTP call returns 201 with a process record, and the failure surfaces when the process transitions to Error. See Validation rules for the complete list.

When deleteMissingFromDB is true, any row in the existing dataset whose key does not appear in the uploaded file is permanently deleted. Verify your file contains every row you wish to keep before setting this flag.

Uploads are processed asynchronously. The API returns immediately with a process object, and the data becomes available in the dataset once the background job completes. Identical re-uploads (same content + same flags as the previous successful run) are detected via a content hash and skipped — see Idempotency.