Upload Demand Forecast

Upload demand: the number of staff each role needs over a time range.

Each row has a role id, a startDate, an endDate and a demand, the number of staff needed throughout the range. Times are ISO-8601 with an explicit offset, on a quarter hour (:00, :15, :30 or :45). A row covers its start up to, but not including, its end. Rows may come in any order.

For each role in the request, everything stored between the role's earliest start and latest end is replaced by the rows sent, for every skill value, including stretches no row covers. Demand outside that span is unchanged.

A role broken down by skills accepts one extra column per skill. A cell holds one of the skill's values; an empty cell, or in JSON a missing key, null or "", means any value. Column names and values are matched exactly. GET /v1/forecasting/demand-config/{teamId} lists the columns and values each role accepts. Rows for the same role may overlap when their skill values differ, and overlapping demand adds up. Rows with the same role and skill values must not overlap.

The body is application/json, an array of rows with skill columns as extra keys, or text/csv with a header row: role, startDate, endDate, demand, then any skill columns, in any order.

Nothing is stored if any row is invalid; the response is a 400. A field in the wrong format, such as a time without an offset, a role that is not an id, or a missing or negative demand, is reported with the field and its row. A row that breaks an upload rule, such as an end before its start, a time off the quarter hour, a value a skill column does not allow, or an overlap, is reported as "Row 3: ...", counting from 1 and, in CSV, without the header. A role id that does not exist returns RoleNotFound. A column the role does not accept is ignored and listed in ignoredColumns.