#!/bin/bash
# AFIR historical XLSX importer wrapper.
#
# Downloads a yearly AFIR FEADR/FEGA XLSX, normalizes to pipe-TSV, ships to
# satra, COPYs into fonduri.staging_afir, then INSERTs into fonduri.afir_plati
# with source_year tagging.
#
# Idempotent: rows with the matching source_year are deleted before insert
# (XLSX dumps are stateless reflections of AFIR DB at publication time).
#
# Usage:
#   ./import-afir-historical.sh URL YEAR FUND [LIMIT]
#     URL:   AFIR XLSX direct download URL
#     YEAR:  4-digit source year, e.g. 2023
#     FUND:  'feadr' or 'fega' (informational; schema is identical)
#     LIMIT: optional integer — only insert first N rows (smoke test)
#
# Example:
#   ./import-afir-historical.sh \
#     'https://www.afir.ro/media/35cm3jdr/listaplati_2023_feadr_actualizata.xlsx' \
#     2023 feadr
#
# Smoke test (1000 rows):
#   ./import-afir-historical.sh '<url>' 2023 feadr 1000

set -euo pipefail

URL="${1:?URL required}"
YEAR="${2:?YEAR required}"
FUND="${3:?FUND required (feadr|fega)}"
LIMIT="${4:-}"

if ! [[ "$YEAR" =~ ^20[0-9]{2}$ ]]; then
  echo "[afir-historical] ERROR: YEAR must be 4-digit (got: $YEAR)" >&2
  exit 2
fi
if [[ "$FUND" != "feadr" && "$FUND" != "fega" ]]; then
  echo "[afir-historical] ERROR: FUND must be 'feadr' or 'fega' (got: $FUND)" >&2
  exit 2
fi

WORK_LOCAL="/tmp/afir-historical-$$"
WORK_REMOTE="/tmp/afir-historical-$YEAR-$FUND"
trap "rm -rf $WORK_LOCAL" EXIT
mkdir -p "$WORK_LOCAL"

XLSX_LOCAL="$WORK_LOCAL/listaplati_${YEAR}_${FUND}.xlsx"
TSV_LOCAL="$WORK_LOCAL/listaplati_${YEAR}_${FUND}.tsv"

echo "[afir-historical] === ${YEAR} ${FUND} ==="

# 1. Download (resume-friendly, large file safe). Run on satra to skip the
# upload-back-to-server hop — the XLSX is 30 MB.
echo "[afir-historical] downloading on satra..."
ssh satra "mkdir -p $WORK_REMOTE && curl -sLkf --max-time 600 -o $WORK_REMOTE/listaplati.xlsx '$URL' && ls -lh $WORK_REMOTE/listaplati.xlsx"

# 2. Normalize to pipe-delimited TSV using existing python3-openpyxl on satra.
SCRIPT_DIR="$(cd "$(dirname "$0")/.." && pwd)/scripts"
echo "[afir-historical] uploading normalizer..."
scp -q "$SCRIPT_DIR/import-afir-historical.py" satra:$WORK_REMOTE/normalize.py

echo "[afir-historical] normalizing XLSX → TSV (this takes ~2-5 min for 500K rows)..."
ssh satra "python3 $WORK_REMOTE/normalize.py $WORK_REMOTE/listaplati.xlsx $WORK_REMOTE/data.tsv 2>&1 | tail -20"

# 3. Optional smoke-test truncation
TSV_REMOTE="$WORK_REMOTE/data.tsv"
if [ -n "$LIMIT" ]; then
  echo "[afir-historical] LIMIT=$LIMIT — truncating TSV for smoke test..."
  ssh satra "head -n $LIMIT $WORK_REMOTE/data.tsv > $WORK_REMOTE/data.smoke.tsv && wc -l $WORK_REMOTE/data.smoke.tsv"
  TSV_REMOTE="$WORK_REMOTE/data.smoke.tsv"
fi

# 4. Stage + INSERT on Postgres via /tmp/baseline.sh (Infisical-aware psql wrapper).
echo "[afir-historical] staging + insert..."
ssh satra "/tmp/baseline.sh <<SQL
\\set ON_ERROR_STOP on

TRUNCATE TABLE fonduri.staging_afir;

\\copy fonduri.staging_afir (beneficiar_name, last_name, mama_cui, localitate, cod_masura, obiectiv, data_start, data_end, fega_op, fega_total, feadr_op, feadr_total, op_amount, cofinantare, ue_total) FROM '$TSV_REMOTE' WITH (FORMAT text, DELIMITER '|', NULL '')

SELECT 'staging_loaded' AS step, COUNT(*) AS rows FROM fonduri.staging_afir;

-- Idempotent: drop existing rows for (year, fund) before reinsert.
-- We use cod_masura prefix as a fund discriminator: FEGA codes start with
-- a single letter or specific scheme (DPB, ANTPDD, etc); FEADR is 'M ' prefix
-- or numeric. For safety in the LIMIT smoke test we DON'T delete; only
-- delete on a full run (LIMIT empty).
SQL"

if [ -z "$LIMIT" ]; then
  echo "[afir-historical] full run — deleting prior rows for source_year=$YEAR..."
  ssh satra "/tmp/baseline.sh -c \"DELETE FROM fonduri.afir_plati WHERE source_year = $YEAR;\""
fi

ssh satra "/tmp/baseline.sh <<SQL
\\set ON_ERROR_STOP on

INSERT INTO fonduri.afir_plati (
  source_year, beneficiar_name, last_name, mama_cui, localitate,
  cod_masura, obiectiv, data_start, data_end,
  fega_op, fega_total, feadr_op, feadr_total,
  op_amount, cofinantare, ue_total
)
SELECT
  $YEAR,
  beneficiar_name, NULLIF(last_name, ''), NULLIF(mama_cui, ''), NULLIF(localitate, ''),
  NULLIF(cod_masura, ''), NULLIF(obiectiv, ''), NULLIF(data_start, ''), NULLIF(data_end, ''),
  NULLIF(fega_op, '')::numeric,
  NULLIF(fega_total, '')::numeric,
  NULLIF(feadr_op, '')::numeric,
  NULLIF(feadr_total, '')::numeric,
  NULLIF(op_amount, '')::numeric,
  NULLIF(cofinantare, '')::numeric,
  NULLIF(ue_total, '')::numeric
FROM fonduri.staging_afir;

SELECT '$YEAR-$FUND' AS run,
       COUNT(*)                                  AS rows_inserted,
       COUNT(DISTINCT beneficiar_name)           AS distinct_beneficiars,
       SUM(CASE WHEN feadr_total > 0 THEN 1 END) AS with_feadr,
       SUM(CASE WHEN fega_total  > 0 THEN 1 END) AS with_fega,
       SUM(ue_total)::bigint                     AS sum_ue_eur
FROM fonduri.afir_plati WHERE source_year = $YEAR;
SQL"

if [ -z "$LIMIT" ]; then
  echo "[afir-historical] cleaning up remote workdir..."
  ssh satra "rm -rf $WORK_REMOTE"
fi

echo "[afir-historical] === done ($YEAR $FUND) ==="