#!/usr/bin/env bash
# Refresh the test database from production. Runs daily via cron on pms1.
#
# pelagia_test is a throwaway mirror of prod (pelagia) so the autofix Claude can
# run integration tests / a dev server against realistic data WITHOUT touching
# production. The test DB is owned by pelagia_user (created once as superuser);
# this refresh runs purely as pelagia_user using the prod connection string.

set -uo pipefail

ENV_FILE="${1:-/home/shad0w/pms/App/.env}"
PROD_DB="pelagia"
TEST_DB="pelagia_test"

log() { echo "$(date '+%F %T') $*"; }

PROD_URL=$(grep -E '^DATABASE_URL' "$ENV_FILE" | sed -E 's/^DATABASE_URL=//; s/^"//; s/"$//')
[ -n "$PROD_URL" ] || { log "ERROR: no DATABASE_URL in $ENV_FILE"; exit 1; }

# Derive the test URL by swapping ONLY the database-name path segment (anchored on
# @host/ so the 'pelagia' inside the username is never touched).
TEST_URL=$(printf '%s' "$PROD_URL" | sed -E "s#(@[^/]+/)$PROD_DB([?]|\$)#\1$TEST_DB\2#")

if [ "$TEST_URL" = "$PROD_URL" ]; then
    log "ERROR: failed to derive test URL (db name not found in connection string)"; exit 1
fi

log "Refreshing $TEST_DB from $PROD_DB ..."
# --clean --if-exists drops+recreates each object in place (first run on an empty
# DB is a no-op for the DROPs). --no-owner/--no-privileges keep it portable.
errfile=$(mktemp)
pg_dump --clean --if-exists --no-owner --no-privileges "$PROD_URL" \
    | psql "$TEST_URL" >/dev/null 2>"$errfile"

prod_tables=$(psql -tAc "SELECT count(*) FROM information_schema.tables WHERE table_schema='public';" "$PROD_URL")
test_tables=$(psql -tAc "SELECT count(*) FROM information_schema.tables WHERE table_schema='public';" "$TEST_URL")

if [ "$test_tables" = "$prod_tables" ] && [ "$test_tables" -gt 0 ]; then
    log "Done. $TEST_DB has $test_tables public tables (prod has $prod_tables)."
    rm -f "$errfile"
else
    log "WARNING: table counts differ (test=$test_tables prod=$prod_tables). Recent errors:"
    tail -8 "$errfile"
    rm -f "$errfile"
    exit 1
fi