#!/usr/bin/env bash
# Rebuild the per-dataset recon artefacts inside the api pod and copy them
# back to the host so they're visible next to the source YAMLs.
#
# Outputs per dataset under api/datasets/<name>/:
#   - extracted_schema.json   (auto, from Postgres introspection)
#   - recon.json              (merged: extracted + schema_docs.yaml + metrics.yaml)
#
# Usage:
#   bash ctrl/recon.sh                # all datasets
#   bash ctrl/recon.sh financial      # one dataset

set -euo pipefail
cd "$(dirname "$0")/.."

CTX="${KCTX:---context kind-nvi}"
NS="${KNS:--n nvi}"

POD=$(kubectl $CTX $NS get pod -l app=api -o jsonpath='{.items[0].metadata.name}')
if [ -z "$POD" ]; then
    echo "no api pod found; is \`make tilt-up\` running?" >&2
    exit 1
fi

# Run the build inside the pod (needs live Postgres access).
build_args=()
if [ "${1-}" != "" ]; then
    build_args+=(--dataset "$1")
fi
kubectl $CTX $NS exec "$POD" -- uv run python -m api.recon.build "${build_args[@]}"

# Pull each artefact back to the host.
for ds_path in api/datasets/*/; do
    ds=$(basename "$ds_path")
    case "$ds" in
        __*|.*) continue ;;
    esac
    if [ "${1-}" != "" ] && [ "$1" != "$ds" ]; then
        continue
    fi
    for f in extracted_schema.json recon.json; do
        if kubectl $CTX $NS cp "$POD:/app/api/datasets/$ds/$f" "api/datasets/$ds/$f" 2>/dev/null; then
            echo "  → api/datasets/$ds/$f"
        fi
    done
done