Feature 2339 enhance data comparison workflow

docker/ontop-reload.sh

@@ -0,0 +1,72 @@
+#!/usr/bin/env bash
+# SPDX-FileCopyrightText: 2026 Jonas Huber <https://github.com/jh-RLI> © Reiner Lemoine Institut
+#
+# SPDX-License-Identifier: AGPL-3.0-or-later
+
+# Restart the dev ontop container and verify that what it serves matches
+# docker/serviceConfigs/ontop/mapping.obda.
+#
+# ontop reads the OBDA mapping ONLY at startup: after a branch switch or a
+# mapping edit, a running container silently keeps serving the old mapping
+# (newly mapped tables return 0 rows). Run this after any mapping change.
+#
+# Usage:
+#   ./docker/ontop-reload.sh            restart, wait for the endpoint, verify
+#   ./docker/ontop-reload.sh --check    verify only (no restart)
+#
+# Environment overrides: ONTOP_CONTAINER (default: ontop),
+# ONTOP_ENDPOINT (default: http://localhost:8080/sparql), ONTOP_TIMEOUT (120s).
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+CONTAINER="${ONTOP_CONTAINER:-ontop}"
+ENDPOINT="${ONTOP_ENDPOINT:-http://localhost:8080/sparql}"
+MAPPING="$SCRIPT_DIR/serviceConfigs/ontop/mapping.obda"
+# table-name predicate emitted by every row anchor (see oekg/registry)
+TABLE_PRED="https://openenergyplatform.org/ontology/oeo/OEO_00000504"
+TIMEOUT="${ONTOP_TIMEOUT:-120}"
+
+if [[ "${1:-}" != "--check" ]]; then
+  echo "Restarting '$CONTAINER'..."
+  docker restart "$CONTAINER" >/dev/null
+fi
+
+printf 'Waiting for %s ' "$ENDPOINT"
+deadline=$((SECONDS + TIMEOUT))
+until curl -sf -o /dev/null --data-urlencode 'query=ASK {}' "$ENDPOINT"; do
+  if ((SECONDS >= deadline)); then
+    printf '\nERROR: endpoint not answering after %ss\n' "$TIMEOUT" >&2
+    exit 1
+  fi
+  printf '.'
+  sleep 3
+done
+printf ' up.\n'
+
+# Tables the mapping file on disk maps (FROM "schema"."table" in the source SQL).
+mapfile -t expected < <(grep -oE 'FROM[[:space:]]+"[^"]+"\."[^"]+"' "$MAPPING" |
+  sed -E 's/.*\."([^"]+)"$/\1/' | sort -u)
+if ((${#expected[@]} == 0)); then
+  echo "ERROR: no mapped tables found in $MAPPING" >&2
+  exit 1
+fi
+
+# Tables the running endpoint actually serves, with row counts.
+query="SELECT ?t (COUNT(?s) AS ?c) WHERE { ?s <$TABLE_PRED> ?t } GROUP BY ?t ORDER BY ?t"
+csv="$(curl -sf -H 'Accept: text/csv' --data-urlencode "query=$query" "$ENDPOINT")"
+
+status=0
+echo "Mapped tables (mapping.obda on disk vs SPARQL endpoint):"
+for t in "${expected[@]}"; do
+  count="$(printf '%s\n' "$csv" | tr -d '\r' | awk -F, -v t="$t" '$1 == t {print $2}')"
+  if [[ -n "${count:-}" && "$count" != "0" ]]; then
+    printf '  OK    %-45s %s rows\n' "$t" "$count"
+  else
+    printf '  STALE %-45s not served (0 rows)\n' "$t"
+    status=1
+  fi
+done
+if ((status != 0)); then
+  echo "Verification FAILED: the container serves an older mapping (or the table is empty in Postgres). Re-run without --check to restart." >&2
+fi
+exit $status

docker/serviceConfigs/ontop/README.md

@@ -5,3 +5,30 @@ Download the database JDBC driver for ontop:
 - <https://jdbc.postgresql.org/>
 
 Add the file postgresql.jar to this directory.
+
+## Reloading the mapping (dev)
+
+ontop reads `mapping.obda` **only at startup**. If the file changes while the
+container is running — a mapping edit, or a `git checkout` that rewrites it —
+the container silently keeps serving the old mapping: newly mapped tables return
+0 rows in SPARQL and the comparison frontend shows no data.
+
+After **any** change to `mapping.obda` (including branch switches), run:
+
+```sh
+./docker/ontop-reload.sh
+```
+
+It restarts the `ontop` container, waits for the SPARQL endpoint, and then
+verifies that every table mapped in `mapping.obda` on disk is actually served
+(row count > 0 via the table-name predicate `oeo:OEO_00000504`).
+
+To only check whether the running container matches the file on disk, without
+restarting:
+
+```sh
+./docker/ontop-reload.sh --check
+```
+
+A `STALE` line means the container loaded an older mapping (restart it) — or the
+table exists in the mapping but is empty in Postgres.