Skip to contents

Export the raw trips table to the cross-country API bucket

Source: R/api.R
export_api_raw.Rd

Projects the weighted (pre-validation) long catch table onto the 22-column schema peskas-api-{dev,prod} publishes for every country, and uploads it to conf$api$trips$raw$cloud_path.

Usage

export_api_raw(log_threshold = logger::DEBUG)

Arguments

log_threshold

The (standard Apache logj4) log level used as a threshold for the logging infrastructure. See logger::log_levels for more details.

Value

no outputs. This function is used for its side effects.

Details

The schema was read off the live peskas-api-prod objects (2026-08-10) and is the same 22 columns, in the same order, for Kenya, Mozambique and Zanzibar, raw and validated alike:

survey_id, trip_id, landing_date, gaul_1_code, gaul_1_name, gaul_2_code,
gaul_2_name, landing_site, n_fishers, trip_duration_hrs, gear, vessel_type,
catch_habitat, catch_outcome, n_catch, catch_taxon, scientific_name,
length_cm, catch_kg, catch_price, tot_catch_kg, tot_catch_price

Three things to know about the mapping:

  • Grain. The API is one row per (trip, catch); Timor's long tables are one row per (submission, catch, length bin). The bins are collapsed: catch_kg is their sum and length_cm their mean weighted by the number of individuals counted in each bin. Empty bins are kept in the input and contribute nothing to either, so no catch record is lost.

  • trip_id is TRIP_<submission_id>, as in all three other countries — a trip identifier, not the PDS tracker_trip_id. coasts::summarize_data() does group_by(trip_id) |> slice(1), so it must be unique per landing, and only half of Timor's landings match a tracked trip.

  • catch_price is NA: Timor records revenue per landing, not per catch. The landing's revenue is tot_catch_price, which is what Mozambique — also without per-catch prices — publishes.