#!/bin/bash
# Discovers the latest ONRC bulk dataset on data.gov.ro, downloads any newer
# CSVs, and runs import-onrc.sh — but only if the dataset is fresher than
# what's already on disk. Idempotent: re-running on the same day is a no-op.
#
# Dataset on data.gov.ro is published ~monthly with slug pattern
# `firme-DD-MM-YYYY`. Resource UUIDs change each release, so we can't
# hardcode URLs — query CKAN to discover the current ones.

set -euo pipefail

DATA_DIR=/opt/vreaudigital/data/onrc
LOG=/var/log/vreaudigital-onrc-import.log
STAMP_FILE="$DATA_DIR/.dataset-name"
SCRIPT_DIR="$(dirname "$(readlink -f "$0")")"

log() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] $1" | tee -a "$LOG"; }

mkdir -p "$DATA_DIR"

log "=== ONRC fresh-check started ==="

# Query CKAN for the most recently modified `firme-...` dataset.
LATEST_NAME=$(curl -fsS --max-time 30 \
  "https://data.gov.ro/api/3/action/package_search?q=firme&sort=metadata_modified+desc&rows=10" \
  | jq -r '[.result.results[] | select(.name | test("^firme-[0-9]{2}-[0-9]{2}-[0-9]{4}$"))][0].name // empty')

if [ -z "$LATEST_NAME" ]; then
  log "ERROR: could not find a firme-DD-MM-YYYY dataset on data.gov.ro"
  exit 1
fi
log "Latest dataset on data.gov.ro: $LATEST_NAME"

# Skip if we've already imported this snapshot.
if [ -f "$STAMP_FILE" ] && [ "$(cat "$STAMP_FILE")" = "$LATEST_NAME" ]; then
  log "Already imported $LATEST_NAME — nothing to do."
  exit 0
fi

# Fetch resource URLs for the dataset. We need 4 of them (the rest are unused).
log "Fetching resource URLs for $LATEST_NAME..."
RESOURCES_JSON=$(curl -fsS --max-time 30 \
  "https://data.gov.ro/api/3/action/package_show?id=$LATEST_NAME")

declare -A NEEDED=(
  [od_firme.csv]=""
  [od_caen_autorizat.csv]=""
  [od_stare_firma.csv]=""
  [od_reprezentanti_legali.csv]=""
)

while IFS=$'\t' read -r url; do
  fname=$(basename "$url" | tr 'A-Z' 'a-z')
  if [ -n "${NEEDED[$fname]+x}" ]; then
    NEEDED[$fname]="$url"
  fi
done < <(echo "$RESOURCES_JSON" | jq -r '.result.resources[] | "\(.url)"')

for f in "${!NEEDED[@]}"; do
  if [ -z "${NEEDED[$f]}" ]; then
    log "ERROR: resource $f not found in dataset $LATEST_NAME"
    exit 1
  fi
done

# Download each CSV (curl -z compares against existing file's mtime).
for f in od_firme.csv od_caen_autorizat.csv od_stare_firma.csv od_reprezentanti_legali.csv; do
  url="${NEEDED[$f]}"
  log "Downloading $f..."
  curl -fL --max-time 600 -o "$DATA_DIR/$f.tmp" "$url" 2>&1 | tail -3 | tee -a "$LOG"
  mv -f "$DATA_DIR/$f.tmp" "$DATA_DIR/$f"
done

log "Running import-onrc.sh..."
"$SCRIPT_DIR/import-onrc.sh"

# ONRC import inserts new firms without lat/lng. Run the full geocoding
# fallback chain (geonames_postal → uat_centroid → photon → judet_centroid)
# so /harta + UI map clustering have coordinates for every fresh-import row.
log "Running geocode-firms.sh fallback chain..."
"$SCRIPT_DIR/geocode-firms.sh" || log "WARN: geocode-firms.sh exited non-zero; continuing"

# Record the snapshot we just successfully imported.
echo "$LATEST_NAME" > "$STAMP_FILE"
log "=== ONRC fresh-import done (snapshot=$LATEST_NAME) ==="