#!/bin/bash # RegAS scraper — runs scrape-regas.ts in a node:22-alpine container. # Mirrors the enrich-anaf.sh pattern: Infisical Machine Identity → env-file # → docker run --env-file (NEVER -e $VAR), file deleted post-launch. # # Idempotent (uses ON CONFLICT (id) DO UPDATE). Safe to run from cron. set -euo pipefail PAGE_SIZE="${PAGE_SIZE:-5000}" START_PAGE="${START_PAGE:-0}" MAX_PAGES="${MAX_PAGES:-0}" LOG=/var/log/vreaudigital-regas.log log() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] $1" | tee -a "$LOG"; } log "=== RegAS scrape started (page-size=$PAGE_SIZE start-page=$START_PAGE max-pages=$MAX_PAGES) ===" if docker ps --filter name=vreaudigital-regas --format '{{.Names}}' | grep -q '^vreaudigital-regas$'; then log "WARN: vreaudigital-regas already running, skipping this tick" exit 0 fi docker rm -f vreaudigital-regas 2>/dev/null || true # ── Fetch DATABASE_URL via Infisical Machine Identity ── source /opt/vreaudigital/.infisical-mi TOKEN=$(infisical login --method=universal-auth \ --domain="$INFISICAL_API_URL" \ --client-id="$INFISICAL_CLIENT_ID" \ --client-secret="$INFISICAL_CLIENT_SECRET" \ --silent --plain) umask 077 ENVF=$(mktemp /tmp/.vreaudigital-env.XXXXXX) DBURL=$(infisical secrets get DATABASE_URL \ --domain="$INFISICAL_API_URL" \ --projectId="$INFISICAL_PROJECT_ID" \ --env="$INFISICAL_ENV" --path="$INFISICAL_PATH" \ --token="$TOKEN" --plain --silent) echo "DATABASE_URL=$DBURL" > "$ENVF" # RegAS uses an intermediate CA cert chain that node's bundle doesn't trust. # Cert is valid (verified OOB), bypass for this scraper only. echo "NODE_TLS_REJECT_UNAUTHORIZED=0" >> "$ENVF" unset DBURL TOKEN # ── Launch detached docker container ── cd /opt/vreaudigital/services/seap-scraper if [ ! -d node_modules/tsx ]; then log "Installing seap-scraper deps..." docker run --rm -v "$(pwd):/work" -w /work --user "$(id -u):$(id -g)" \ node:22-alpine npm install --omit=optional 2>&1 | tee -a "$LOG" >/dev/null fi CID=$(docker run -d \ --name vreaudigital-regas \ --network host \ --env-file "$ENVF" \ -v "$(pwd):/work" \ -w /work \ --user "$(id -u):$(id -g)" \ --restart no \ node:22-alpine \ npx tsx src/scrape-regas.ts \ --page-size="$PAGE_SIZE" \ --start-page="$START_PAGE" \ --max-pages="$MAX_PAGES") log "container started: $CID" sleep 3 rm -f "$ENVF" log "envfile cleaned" docker wait vreaudigital-regas >/dev/null EXIT_CODE=$(docker inspect -f '{{.State.ExitCode}}' vreaudigital-regas 2>/dev/null || echo "?") docker logs vreaudigital-regas 2>&1 | tail -10 | tee -a "$LOG" log "=== RegAS scrape done (exit=$EXIT_CODE) ===" exit "$EXIT_CODE"