#!/bin/bash
# Full REAL-data EHD pipeline. Every input is built from the actual 1.5 deg MARS
# download (txtm/srf_daily/srf_24h/plev), so nothing is mixed with the proxy.
# Steps 1-3 are moderate login-node preprocessing; step 4 submits the multi-node
# SLURM array + reduce. Proxy outputs are left untouched (separate dirs).
set -e
cd /home/tanpv/Progs/claude/HW_EC_10
PY=/home/tanpv/.conda/envs/ehd/bin/python

COARSE=outputs/real/tx_ec_1.5_real.nc
DSDIR=outputs/downscale_real
TXDS=$DSDIR/pred/pred_bcsd.npy
PRED=outputs/real/ec15_predictors.nc

echo "==== [1/4] real coarse Tx from txtm (native 1.5 deg) $(date '+%F %T') ===="
if [ ! -f "$COARSE" ]; then
  $PY -u src/build_ec_tx.py --root data/nc/txtm \
      --pattern "txtm_cf_{year}-{mmdd}.nc" --out "$COARSE" --jobs 10
else echo "  exists, skip"; fi

echo "==== [2/4] real BCSD downscaling $(date '+%F %T') ===="
if [ ! -f "$TXDS" ]; then
  $PY -u src/downscale.py --method bcsd --coarse "$COARSE" \
      --out "$DSDIR/baselines.json"
else echo "  exists, skip"; fi

echo "==== [3/4] real 46-var predictor cube $(date '+%F %T') ===="
if [ ! -f "$PRED" ]; then
  $PY -u src/build_ec15_real.py --jobs 10 --out "$PRED"
else echo "  exists, skip"; fi

echo "==== [4/4] submit multi-node EHD on REAL inputs $(date '+%F %T') ===="
bash src/launch_ehd_full.sh \
    --pred "$PRED" --txds "$TXDS" \
    --out outputs/downscale/ehd_full_real.json \
    --part-dir outputs/downscale/fold_parts_real
echo "==== pipeline dispatched $(date '+%F %T') ===="