#!/usr/bin/env bash
# CancerIndex — ordered first ingestion (idempotent; safe to re-run).
#
# Runs connectors in the recommended order (terminology → genes → evidence/genomics/variants →
# trials → literature → epidemiology) with per-connector time budgets, then rebuilds counters and
# rankings. Connectors are restartable (cursor) and idempotent (payload hash), so re-running this
# script only fetches what is missing. A connector that is not registered yet, paused, awaiting
# credentials or under license review is skipped with a message — never treated as an error.
#
# Usage: cd ~/apps/cancerindex && bash deploy/first-run.sh [--only id,id] [--skip id,id] [--dry-run]
set -uo pipefail

cd "$(dirname "$0")/.." || exit 1
export DATABASE_URL="${DATABASE_URL:-postgres://localhost:5432/cancerindex}"
export CI_DATA_DIR="${CI_DATA_DIR:-$PWD/data}"
export CI_SERVICE=first-run
LOG_DIR="${LOG_DIR:-$PWD/logs}"
mkdir -p "$LOG_DIR" data/raw data/cache

# connector id : time budget (minutes) : max attempts within this script
ORDER=(
  "ncit-evs:45:2"
  "oncotree:10:2"
  "hgnc:15:2"
  "civic:30:2"
  "gdc:30:2"
  "clinvar:45:2"
  "clinicaltrials:45:3"
  "pubmed:45:3"
  "cdc-wonder:30:2"
)

ONLY=""; SKIP=""; DRY=""
while [ $# -gt 0 ]; do
  case "$1" in
    --only) ONLY="$2"; shift 2 ;;
    --skip) SKIP="$2"; shift 2 ;;
    --dry-run) DRY="--mode dry_run --max-records 100"; shift ;;
    *) echo "unknown option $1"; exit 2 ;;
  esac
done

ts() { date -u +%Y-%m-%dT%H:%M:%SZ; }
log() { echo "[$(ts)] $*" | tee -a "$LOG_DIR/first-run.log"; }
in_list() { [ -z "$2" ] && return 1; echo ",$2," | grep -q ",$1,"; }

log "== CancerIndex first run (db=$DATABASE_URL, data=$CI_DATA_DIR) =="
pnpm cix sources:sync >>"$LOG_DIR/first-run.log" 2>&1 || log "warning: sources:sync failed (continuing)"
REGISTERED="$(pnpm --silent cix connectors 2>/dev/null | awk '{print $1}' | tr '\n' ',')"

last_status() {
  psql "$DATABASE_URL" -Atc "SELECT status FROM ingest_runs WHERE connector_id = '$1' ORDER BY started_at DESC LIMIT 1" 2>/dev/null
}
cursor_health() {
  psql "$DATABASE_URL" -Atc "SELECT coalesce(health,'unknown') || '|' || coalesce(paused::text,'false') FROM connector_cursors WHERE connector_id = '$1'" 2>/dev/null
}

for entry in "${ORDER[@]}"; do
  IFS=: read -r id minutes attempts <<<"$entry"
  if [ -n "$ONLY" ] && ! in_list "$id" "$ONLY"; then continue; fi
  if in_list "$id" "$SKIP"; then log "-- $id: skipped (--skip)"; continue; fi
  if ! echo ",$REGISTERED" | grep -q ",$id,"; then log "-- $id: not registered in packages/connectors/src/registry.ts — skipped"; continue; fi
  hp="$(cursor_health "$id")"
  case "$hp" in
    awaiting_credentials*) log "-- $id: awaiting credentials — skipped"; continue ;;
    *"|true") log "-- $id: paused — skipped"; continue ;;
  esac
  if [ -z "$DRY" ] && [ "$(last_status "$id")" = "succeeded" ] && [ "${FORCE:-}" != "1" ]; then
    log "-- $id: last run already succeeded — running incremental refresh (FORCE=1 to force full)"
  fi
  n=0
  while [ "$n" -lt "$attempts" ]; do
    n=$((n + 1))
    log "-> $id attempt $n/$attempts (budget ${minutes} min)"
    # shellcheck disable=SC2086
    if pnpm cix run "$id" --max-minutes "$minutes" $DRY >>"$LOG_DIR/first-run.$id.log" 2>&1; then
      st="$(last_status "$id")"
      log "   $id: $st"
      # partial = time budget hit with cursor saved → run again while attempts remain
      [ "$st" = "partial" ] && [ "$n" -lt "$attempts" ] && continue
      break
    else
      st="$(last_status "$id")"
      log "   $id: FAILED (status=${st:-none}); see $LOG_DIR/first-run.$id.log"
      case "$st" in aborted) break ;; esac   # license/credentials gate: do not retry
    fi
  done
done

if [ -z "$DRY" ]; then
  log "-> counters"
  pnpm cix counters >>"$LOG_DIR/first-run.log" 2>&1 && log "   counters ok" || log "   counters FAILED"
  log "-> rank"
  pnpm cix rank >>"$LOG_DIR/first-run.log" 2>&1 && log "   rank ok" || log "   rank FAILED"
fi

log "== summary =="
pnpm --silent cix connectors 2>/dev/null | tee -a "$LOG_DIR/first-run.log"
pnpm --silent cix stats 2>/dev/null | tee -a "$LOG_DIR/first-run.log"
log "== done =="