#!/bin/sh
# Extrait les données d'une seule instance depuis un dump global.
#
# Ne touche JAMAIS la base de production : le dump est restauré dans une base
# jetable, réduit à l'instance demandée, puis re-dumpé. La cible d'un
# pg_restore est toujours une base neuve.
#
# usage: extract-instance.sh <dump-global> <instanceId>
set -eu

FILE=${1:?usage: extract-instance.sh <dump-global> <instanceId>}
INSTANCE=${2:?usage: extract-instance.sh <dump-global> <instanceId>}
CONTAINER=${CONTAINER:-myim_postgres}
DB_USER=${POSTGRES_USER:-mym}
OUT=${OUT:-./backups}
SCRATCH="extract_$(date -u +%s)"

psql() { docker exec "$CONTAINER" psql -qtAX -v ON_ERROR_STOP=1 --username="$DB_USER" --dbname="$1" -c "$2"; }

docker exec "$CONTAINER" createdb --username="$DB_USER" "$SCRATCH"
trap 'docker exec "$CONTAINER" dropdb --username="$DB_USER" --if-exists "$SCRATCH" >/dev/null 2>&1' EXIT

docker exec -i "$CONTAINER" pg_restore \
  --username="$DB_USER" --dbname="$SCRATCH" --no-owner --no-privileges < "$FILE"

if [ "$(psql "$SCRATCH" "select count(*) from \"Instances\" where \"Id\" = '$INSTANCE'")" != "1" ]; then
  echo "Instance '$INSTANCE' absente de ce dump. Instances disponibles :" >&2
  psql "$SCRATCH" 'select "Id" || ' ' || "Name" from "Instances"' >&2 || true
  exit 1
fi

# "is distinct from" et non "<>" : une ligne dont InstanceId est NULL n appartient
# a aucune instance, et "<> x" ne la supprime pas — elle fuiterait dans l export
# de chaque client. Repere en verifiant que la somme des extractions egale le global.
#
# Suppression des enfants avant les parents. Les 7 premières tables n'ont pas
# d'InstanceId : elles se rattachent par leur parent, et c'est là que se cachent
# les oublis silencieux — un quiz sans ses questions, un agenda sans ses events.
psql "$SCRATCH" "
begin;
delete from \"QuizQuestions\" q where
     (q.\"SectionQuizId\" is not null and q.\"SectionQuizId\" not in (select \"Id\" from \"Sections\" where \"InstanceId\" = '$INSTANCE'))
  or (q.\"GuidedStepId\" is not null and q.\"GuidedStepId\" not in (select s.\"Id\" from \"GuidedSteps\" s join \"GuidedPaths\" p on p.\"Id\" = s.\"GuidedPathId\" where p.\"InstanceId\" = '$INSTANCE'))
  or (q.\"SectionQuizId\" is null and q.\"GuidedStepId\" is null);
delete from \"MapAnnotations\" a where
     (a.\"SectionEventId\" is not null and a.\"SectionEventId\" not in (select \"Id\" from \"Sections\" where \"InstanceId\" = '$INSTANCE'))
  or (a.\"ProgrammeBlockId\" is not null and a.\"ProgrammeBlockId\" not in (select b.\"Id\" from \"ProgrammeBlocks\" b join \"Sections\" s on s.\"Id\" = b.\"SectionEventId\" where s.\"InstanceId\" = '$INSTANCE'));
delete from \"GuidedSteps\" s where s.\"GuidedPathId\" not in (select \"Id\" from \"GuidedPaths\" where \"InstanceId\" = '$INSTANCE');
delete from \"EventAgendas\" e where coalesce(e.\"SectionAgendaId\", e.\"SectionEventId\") not in (select \"Id\" from \"Sections\" where \"InstanceId\" = '$INSTANCE');
delete from \"GeoPoints\" g where coalesce(g.\"SectionMapId\", g.\"SectionEventId\") not in (select \"Id\" from \"Sections\" where \"InstanceId\" = '$INSTANCE');
delete from \"ProgrammeBlocks\" b where b.\"SectionEventId\" not in (select \"Id\" from \"Sections\" where \"InstanceId\" = '$INSTANCE');
delete from \"AppConfigurationLinks\" l where
     l.\"ConfigurationId\" not in (select \"Id\" from \"Configurations\" where \"InstanceId\" = '$INSTANCE')
  or (l.\"ApplicationInstanceId\" is not null and l.\"ApplicationInstanceId\" not in (select \"Id\" from \"ApplicationInstances\" where \"InstanceId\" = '$INSTANCE'))
  or (l.\"DeviceId\" is not null and l.\"DeviceId\" not in (select \"Id\" from \"Devices\" where \"InstanceId\" = '$INSTANCE'));

delete from \"ApplicationInstances\" where \"InstanceId\" is distinct from '$INSTANCE';
delete from \"GuidedPaths\" where \"InstanceId\" is distinct from '$INSTANCE';
delete from \"Sections\" where \"InstanceId\" is distinct from '$INSTANCE';
delete from \"Devices\" where \"InstanceId\" is distinct from '$INSTANCE';
delete from \"Configurations\" where \"InstanceId\" is distinct from '$INSTANCE';
delete from \"Resources\" where \"InstanceId\" is distinct from '$INSTANCE';
delete from \"Users\" where \"InstanceId\" is distinct from '$INSTANCE';
delete from \"ApiKeys\" where \"InstanceId\" is distinct from '$INSTANCE';
delete from \"AuditLogs\" where \"InstanceId\" is distinct from '$INSTANCE';
delete from \"ContentEmbeddings\" where \"InstanceId\" is distinct from '$INSTANCE';
delete from \"PushNotifications\" where \"InstanceId\" is distinct from '$INSTANCE';
delete from \"QuestionThemeMonthlies\" where \"InstanceId\" is distinct from '$INSTANCE';
delete from \"VisitEvents\" where \"InstanceId\" is distinct from '$INSTANCE';
delete from \"VisitorQuestions\" where \"InstanceId\" is distinct from '$INSTANCE';
delete from \"Instances\" where \"Id\" <> '$INSTANCE';
commit;"

mkdir -p "$OUT"
DEST="$OUT/instance_${INSTANCE}_$(date -u +%Y%m%dT%H%M%SZ).dump"
docker exec "$CONTAINER" pg_dump --username="$DB_USER" --dbname="$SCRATCH" \
  --format=custom --compress=6 --no-owner --no-privileges > "$DEST"

echo "--- contenu extrait ---"
psql "$SCRATCH" "
select table_name || ' ' || (xpath('/row/c/text()',
         query_to_xml('select count(*) as c from public.\"' || table_name || '\"', false, true, '')))[1]::text::int
from information_schema.tables
where table_schema = 'public' and table_type = 'BASE TABLE'
order by table_name" | awk '{ if ($2 > 0) printf "  %-24s %s\n", $1, $2 }'

echo "$DEST ($(du -h "$DEST" | cut -f1))"