Dateien nach "docker/syncer" hochladen

docker/syncer/Dockerfile

@@ -0,0 +1,32 @@
+# Build-Kontext ist das Repo-Root (siehe docker-compose.yml: context: ..),
+# deshalb sind alle COPY-Pfade relativ zum Repo-Root angegeben.
+FROM almalinux:latest
+
+# dnf-plugins-core -> "dnf download" und "dnf repomanage"
+# createrepo_c     -> createrepo_c und modifyrepo_c
+# appstream        -> appstreamcli fuer die optionale Katalog-Validierung
+# python3          -> Interpreter fuer appstream-gen.py
+RUN dnf install -y \
+        dnf-plugins-core \
+        createrepo_c \
+        appstream \
+        python3 \
+        git \
+        jq \
+        openssh-clients \
+        wget \
+    && dnf clean all
+
+RUN wget -q https://github.com/mikefarah/yq/releases/latest/download/yq_linux_amd64 \
+        -O /usr/local/bin/yq \
+    && chmod +x /usr/local/bin/yq
+
+COPY docker/syncer/sync.sh       /usr/local/bin/sync.sh
+COPY docker/syncer/sync_helpers.py /usr/local/bin/sync_helpers.py
+COPY docker/syncer/appstream-gen.py /usr/local/bin/appstream-gen.py
+COPY docker/syncer/entrypoint.sh /entrypoint.sh
+RUN chmod +x /usr/local/bin/sync.sh /entrypoint.sh
+
+# Bewusst kein VOLUME und kein EXPOSE: die Volumes definiert docker-compose.yml,
+# und ausgeliefert wird ueber den separaten nginx-Container.
+ENTRYPOINT ["/entrypoint.sh"]

docker/syncer/appstream-gen.py

@@ -0,0 +1,328 @@
+#!/usr/bin/env python3
+"""
+Erzeugt einen AppStream-Katalog (Collection-XML) aus packages.yaml.
+
+Der Katalog referenziert die gespiegelten RPMs direkt ueber <pkgname>.
+Es werden keine Wrapper-Pakete gebaut. Das Ergebnis wird per
+`modifyrepo_c --mdtype=appstream` in die Repo-Metadaten eingehaengt,
+damit Discover / GNOME Software die Pakete findet.
+
+Aufruf:
+    appstream-gen.py packages.yaml -o appstream.xml.gz \
+        [--available-packages vorhanden.txt] [--origin firmen-repo]
+"""
+
+import argparse
+import gzip
+import re
+import sys
+import xml.etree.ElementTree as ET
+from xml.dom import minidom
+
+import yaml
+
+# AppStream-Spezifikationsversion des Katalogformats.
+CATALOG_VERSION = "0.14"
+
+DEFAULT_TYPE = "console-application"
+DEFAULT_ICON = "application-x-executable"
+DEFAULT_NOTE = (
+    "Bereitgestellt über das Unternehmens-Repo der IT-Abteilung. "
+    "Fragen und Paketwünsche bitte an den IT-Service-Desk."
+)
+
+
+def die(msg):
+    sys.stdout.flush()
+    print(f"[appstream-gen] FEHLER: {msg}", file=sys.stderr, flush=True)
+    sys.exit(1)
+
+
+def warn(msg):
+    sys.stdout.flush()
+    print(f"[appstream-gen] WARNUNG: {msg}", file=sys.stderr, flush=True)
+
+
+def info(msg):
+    print(f"[appstream-gen] {msg}", flush=True)
+
+
+def sanitize_id_part(value):
+    """Macht aus einem Paketnamen ein gueltiges AppStream-ID-Segment."""
+    part = re.sub(r"[^A-Za-z0-9]+", "-", value).strip("-")
+    if not part:
+        die(f"Paketname ergibt kein gueltiges ID-Segment: {value!r}")
+    if part[0].isdigit():
+        part = "pkg-" + part
+    return part
+
+
+def prettify_name(pkgname):
+    """Fallback-Anzeigename, wenn in der YAML keiner gesetzt ist.
+
+    dotnet-sdk-9.0 -> Dotnet Sdk 9.0
+    """
+    words = re.split(r"[-_]", pkgname)
+    return " ".join(w[:1].upper() + w[1:] if w else w for w in words)
+
+
+def as_list(value):
+    if value is None:
+        return []
+    if isinstance(value, (list, tuple)):
+        return list(value)
+    return [value]
+
+
+def normalize_entry(entry, source_name):
+    """Akzeptiert sowohl 'paketname' als auch ein Mapping mit Metadaten."""
+    if isinstance(entry, str):
+        return {"name": entry, "_source": source_name}
+    if isinstance(entry, dict):
+        if "name" not in entry:
+            die(f"Paketeintrag in Quelle {source_name!r} ohne 'name': {entry!r}")
+        data = dict(entry)
+        data["_source"] = source_name
+        return data
+    die(f"Unerwarteter Paketeintrag in Quelle {source_name!r}: {entry!r}")
+
+
+def pick(entry, defaults, key, fallback=None):
+    """Wert aus dem Paketeintrag, sonst aus defaults, sonst fallback."""
+    if key in entry and entry[key] is not None:
+        return entry[key]
+    if key in defaults and defaults[key] is not None:
+        return defaults[key]
+    return fallback
+
+
+def build_component(entry, defaults, id_prefix):
+    pkgname = entry["name"]
+
+    comp_type = pick(entry, defaults, "type", DEFAULT_TYPE)
+    component = ET.Element("component", {"type": comp_type})
+
+    comp_id = entry.get("id") or f"{id_prefix}.{sanitize_id_part(pkgname)}"
+    ET.SubElement(component, "id").text = comp_id
+    ET.SubElement(component, "pkgname").text = pkgname
+
+    display_name = entry.get("display_name") or entry.get("name_display")
+    if not display_name:
+        display_name = prettify_name(pkgname)
+    ET.SubElement(component, "name").text = display_name
+
+    note = pick(entry, defaults, "note", DEFAULT_NOTE)
+    summary = entry.get("summary") or note
+    ET.SubElement(component, "summary").text = summary
+
+    # Beschreibung: eigener Text plus Hinweis auf die Herkunft.
+    description = ET.SubElement(component, "description")
+    body = entry.get("description")
+    if body:
+        for paragraph in [p.strip() for p in str(body).split("\n\n") if p.strip()]:
+            ET.SubElement(description, "p").text = paragraph
+    if note and note != summary:
+        ET.SubElement(description, "p").text = note
+    elif not body and note == summary:
+        ET.SubElement(description, "p").text = note
+
+    icon = pick(entry, defaults, "icon", DEFAULT_ICON)
+    if icon:
+        ET.SubElement(component, "icon", {"type": "stock"}).text = icon
+
+    categories = as_list(pick(entry, defaults, "categories"))
+    if categories:
+        cat_el = ET.SubElement(component, "categories")
+        for category in categories:
+            ET.SubElement(cat_el, "category").text = str(category)
+
+    keywords = as_list(entry.get("keywords"))
+    if keywords:
+        kw_el = ET.SubElement(component, "keywords")
+        for keyword in keywords:
+            ET.SubElement(kw_el, "keyword").text = str(keyword)
+
+    binaries = as_list(entry.get("binary") or entry.get("binaries"))
+    if binaries:
+        provides = ET.SubElement(component, "provides")
+        for binary in binaries:
+            ET.SubElement(provides, "binary").text = str(binary)
+
+    homepage = entry.get("homepage") or defaults.get("homepage")
+    if homepage:
+        ET.SubElement(component, "url", {"type": "homepage"}).text = str(homepage)
+
+    help_url = entry.get("help_url") or defaults.get("help_url")
+    if help_url:
+        ET.SubElement(component, "url", {"type": "help"}).text = str(help_url)
+
+    developer = entry.get("developer") or defaults.get("developer")
+    if developer:
+        # Das id-Attribut ist seit AppStream 1.0 erwartet; ohne es meldet
+        # appstreamcli 'developer-id-missing'.
+        dev_id = entry.get("developer_id") or defaults.get("developer_id")
+        attrs = {"id": str(dev_id)} if dev_id else {}
+        dev_el = ET.SubElement(component, "developer", attrs)
+        ET.SubElement(dev_el, "name").text = str(developer)
+
+    project_license = entry.get("project_license") or defaults.get("project_license")
+    if project_license:
+        ET.SubElement(component, "project_license").text = str(project_license)
+
+    # desktop-application braucht einen Launchable-Eintrag, sonst ignoriert
+    # Discover die Komponente beim Starten der installierten App.
+    if comp_type == "desktop-application":
+        desktop_id = entry.get("desktop_id")
+        if desktop_id:
+            ET.SubElement(
+                component, "launchable", {"type": "desktop-id"}
+            ).text = str(desktop_id)
+        else:
+            warn(
+                f"{pkgname}: type=desktop-application ohne 'desktop_id'. "
+                "Discover kann die Anwendung dann nicht starten."
+            )
+
+    return comp_id, component
+
+
+def load_available(path):
+    if not path:
+        return None
+    try:
+        with open(path, "r", encoding="utf-8") as handle:
+            names = {line.strip() for line in handle if line.strip()}
+    except OSError as exc:
+        die(f"Konnte Paketliste {path!r} nicht lesen: {exc}")
+    info(f"{len(names)} tatsaechlich vorhandene Pakete eingelesen")
+    return names
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Erzeugt einen AppStream-Katalog aus packages.yaml"
+    )
+    parser.add_argument("packages_yaml", help="Pfad zur packages.yaml")
+    parser.add_argument(
+        "-o", "--output", default="appstream.xml.gz",
+        help="Zieldatei (.xml.gz oder .xml), Standard: appstream.xml.gz",
+    )
+    parser.add_argument(
+        "--origin",
+        help="Origin des Katalogs. Ueberschreibt 'appstream.origin' aus der YAML.",
+    )
+    parser.add_argument(
+        "--available-packages",
+        help="Datei mit den tatsaechlich gespiegelten Paketnamen (einer pro Zeile). "
+             "Nicht vorhandene Pakete werden dann uebersprungen.",
+    )
+    parser.add_argument(
+        "--source",
+        help="Nur Pakete dieser einen Quelle beruecksichtigen. Fuer Setups "
+             "mit einem Repo-Verzeichnis pro Quelle.",
+    )
+    parser.add_argument(
+        "--strict", action="store_true",
+        help="Mit Fehler abbrechen, wenn ein gelistetes Paket nicht vorhanden ist.",
+    )
+    args = parser.parse_args()
+
+    try:
+        with open(args.packages_yaml, "r", encoding="utf-8") as handle:
+            config = yaml.safe_load(handle) or {}
+    except OSError as exc:
+        die(f"Konnte {args.packages_yaml!r} nicht lesen: {exc}")
+    except yaml.YAMLError as exc:
+        die(f"YAML-Fehler in {args.packages_yaml!r}: {exc}")
+
+    appstream_cfg = config.get("appstream") or {}
+    defaults = appstream_cfg.get("defaults") or {}
+    origin = args.origin or appstream_cfg.get("origin") or "firmen-repo"
+    id_prefix = appstream_cfg.get("id_prefix") or "de.firma.repo"
+
+    # Falls kein developer_id gesetzt ist, die ersten beiden Segmente des
+    # id_prefix verwenden: de.firma.repo -> de.firma
+    if "developer_id" not in defaults or not defaults.get("developer_id"):
+        segments = id_prefix.split(".")
+        if len(segments) >= 2:
+            defaults = dict(defaults)
+            defaults["developer_id"] = ".".join(segments[:2])
+
+    available = load_available(args.available_packages)
+
+    sources = config.get("sources") or []
+    if not sources:
+        die("Keine 'sources' in der packages.yaml gefunden.")
+
+    if args.source:
+        sources = [s for s in sources if s.get("name") == args.source]
+        if not sources:
+            die(f"Keine Quelle mit dem Namen {args.source!r} in der YAML gefunden.")
+        info(f"Beschraenke auf Quelle {args.source!r}")
+
+    root = ET.Element("components", {"version": CATALOG_VERSION, "origin": origin})
+
+    seen_ids = {}
+    seen_pkgs = set()
+    skipped = []
+    count = 0
+
+    for source in sources:
+        source_name = source.get("name", "<unbenannt>")
+        for raw_entry in source.get("packages") or []:
+            entry = normalize_entry(raw_entry, source_name)
+            pkgname = entry["name"]
+
+            if entry.get("appstream") is False or entry.get("hidden") is True:
+                info(f"{pkgname}: per Konfiguration von AppStream ausgenommen")
+                continue
+
+            if available is not None and pkgname not in available:
+                skipped.append(pkgname)
+                continue
+
+            if pkgname in seen_pkgs:
+                warn(f"{pkgname}: mehrfach gelistet, spaetere Eintraege ignoriert")
+                continue
+            seen_pkgs.add(pkgname)
+
+            comp_id, component = build_component(entry, defaults, id_prefix)
+            if comp_id in seen_ids:
+                die(
+                    f"Doppelte AppStream-ID {comp_id!r} "
+                    f"({seen_ids[comp_id]} und {pkgname}). "
+                    "Setze 'id:' fuer eines der beiden Pakete."
+                )
+            seen_ids[comp_id] = pkgname
+            root.append(component)
+            count += 1
+
+    if skipped:
+        message = (
+            f"{len(skipped)} gelistete Pakete sind nicht im Repo vorhanden "
+            f"und wurden uebersprungen: {', '.join(sorted(skipped))}"
+        )
+        if args.strict:
+            die(message)
+        warn(message)
+
+    raw = ET.tostring(root, encoding="utf-8")
+    pretty = minidom.parseString(raw).toprettyxml(indent="  ", encoding="UTF-8")
+
+    try:
+        if args.output.endswith(".gz"):
+            with gzip.open(args.output, "wb") as handle:
+                handle.write(pretty)
+        else:
+            with open(args.output, "wb") as handle:
+                handle.write(pretty)
+    except OSError as exc:
+        die(f"Konnte {args.output!r} nicht schreiben: {exc}")
+
+    info(f"{count} Komponente(n) nach {args.output} geschrieben (origin={origin})")
+    if count == 0:
+        warn("Katalog ist leer. Discover wird nichts anzeigen.")
+
+
+if __name__ == "__main__":
+    main()

docker/syncer/entrypoint.sh

@@ -0,0 +1,43 @@
+#!/usr/bin/env bash
+# Container entrypoint: run sync.sh once, or in a loop every
+# SYNC_INTERVAL_SECONDS. There's no systemd/cron inside the container, so a
+# simple sleep-loop is the pragmatic choice.
+set -euo pipefail
+
+: "${SYNC_INTERVAL_SECONDS:=3600}"
+: "${RUN_ONCE:=false}"
+
+log() { echo "[entrypoint] $(date '+%Y-%m-%d %H:%M:%S') $*"; }
+
+run_once() {
+    /usr/local/bin/sync.sh
+}
+
+if [[ "$RUN_ONCE" == "true" ]]; then
+    log "RUN_ONCE=true - running sync.sh once and exiting"
+    run_once
+    exit $?
+fi
+
+log "Starting sync loop (interval: ${SYNC_INTERVAL_SECONDS}s)"
+
+# Graceful shutdown on docker stop / compose down
+term_handler() {
+    log "Received SIGTERM, shutting down"
+    exit 0
+}
+trap term_handler SIGTERM SIGINT
+
+while true; do
+    # NOTE: don't inline `run_once` into `if ! run_once; then ...` - inside
+    # that then-branch, $? would reflect the (always-0) negated test itself,
+    # not run_once's real exit code. Capture it explicitly instead.
+    rc=0
+    run_once || rc=$?
+    if [[ $rc -ne 0 ]]; then
+        log "sync.sh failed (exit ${rc}) - will retry after the next interval"
+    fi
+    log "Sleeping ${SYNC_INTERVAL_SECONDS}s until next sync..."
+    sleep "${SYNC_INTERVAL_SECONDS}" &
+    wait $!
+done

docker/syncer/sync.sh

@@ -0,0 +1,191 @@
+#!/bin/bash
+# Synct die Paket-Whitelist aus Git (oder aus einer lokal gemounteten Datei)
+# und laedt die selektierten Pakete aus den jeweils angegebenen Fremd-Repos.
+# Erzeugt anschliessend RPM- und - sofern der Generator vorhanden ist -
+# AppStream-Metadaten.
+set -euo pipefail
+
+# createrepo_c uebernimmt die umask des Aufrufers. Ohne "umask 022" kann
+# repodata/ mit 0700 landen; der nginx-Worker (User nginx) kommt dann nicht
+# hinein und liefert 403 - bzw. 404, falls die nginx-Config try_files nutzt.
+umask 022
+
+# --- Pfade ----------------------------------------------------------------
+# Muessen zu den Volumes in docker-compose.yml passen:
+#   repo-data  -> /srv/repo                   (im nginx-Container: /usr/share/nginx/html/repo)
+#   sync-cache -> /var/cache/rpm-mirror-sync
+# Daraus ergibt sich die oeffentliche URL: <host>/repo/pkgs/repodata/repomd.xml
+WORK_DIR="${WORK_DIR:-/var/cache/rpm-mirror-sync}"
+REPO_ROOT="${REPO_ROOT:-/srv/repo}"
+DEST="$REPO_ROOT/pkgs"
+
+CONFIG_DIR="$WORK_DIR/config"
+LOCAL_PACKAGES_FILE=/config/packages.yaml
+TMP_REPO_DIR=/etc/yum.repos.d
+LOCKFILE="$WORK_DIR/sync.lock"
+APPSTREAM_GEN="${APPSTREAM_GEN:-/usr/local/bin/appstream-gen.py}"
+
+mkdir -p "$WORK_DIR" "$DEST"
+
+exec 200>"$LOCKFILE"
+if ! flock -n 200; then
+  echo "[sync] Ein anderer Sync-Lauf ist bereits aktiv, breche ab."
+  exit 0
+fi
+
+echo "[sync] $(date -Iseconds) Starte Sync-Lauf"
+
+# --- 1. Whitelist besorgen ------------------------------------------------
+# Eine gemountete /config/packages.yaml hat Vorrang. Nur wenn sie fehlt,
+# wird SOURCES_GIT_REPO gebraucht.
+if [ -f "$LOCAL_PACKAGES_FILE" ]; then
+  PACKAGES_FILE="$LOCAL_PACKAGES_FILE"
+  echo "[sync] Nutze lokal gemountete $PACKAGES_FILE"
+else
+  : "${SOURCES_GIT_REPO:?weder SOURCES_GIT_REPO gesetzt noch /config/packages.yaml gemountet}"
+  CONFIG_REPO_BRANCH="${CONFIG_REPO_BRANCH:-main}"
+  PACKAGES_FILE="$CONFIG_DIR/packages.yaml"
+
+  if [ -n "${GIT_SSH_KEY_PATH:-}" ] && [ -f "$GIT_SSH_KEY_PATH" ]; then
+    export GIT_SSH_COMMAND="ssh -i ${GIT_SSH_KEY_PATH} -o StrictHostKeyChecking=accept-new"
+  fi
+
+  if [ -d "$CONFIG_DIR/.git" ]; then
+    git -C "$CONFIG_DIR" fetch --depth=1 origin "$CONFIG_REPO_BRANCH"
+    git -C "$CONFIG_DIR" reset --hard "origin/$CONFIG_REPO_BRANCH"
+  else
+    # Reste eines abgebrochenen Clones wegraeumen, sonst scheitert git clone.
+    rm -rf "$CONFIG_DIR"
+    git clone --depth=1 --branch "$CONFIG_REPO_BRANCH" "$SOURCES_GIT_REPO" "$CONFIG_DIR"
+  fi
+fi
+
+if [ ! -f "$PACKAGES_FILE" ]; then
+  echo "[sync] FEHLER: $PACKAGES_FILE nicht gefunden."
+  exit 1
+fi
+
+# --- 2. Pro Quelle: temporaeres Repo aktivieren, Pakete gezielt ziehen -----
+source_count=$(yq -o=json '.sources | length' "$PACKAGES_FILE")
+download_failed=0
+
+for i in $(seq 0 $((source_count - 1))); do
+  src=$(yq -o=json ".sources[$i]" "$PACKAGES_FILE")
+  name=$(echo "$src" | jq -r '.name')
+  baseurl=$(echo "$src" | jq -r '.baseurl')
+  gpgkey=$(echo "$src" | jq -r '.gpgkey // empty')
+
+  # Standard: false. Nur bei echten Fremd-Repos aktivieren, deren
+  # Abhaengigkeiten die Clients nicht ueber ihre Basis-Repos bekommen.
+  resolve_deps=$(echo "$src" | jq -r '.resolve_deps // false')
+
+  # Paketeintraege koennen Strings ODER Mappings mit Metadaten sein.
+  mapfile -t packages < <(
+    echo "$src" | jq -r '.packages[] | if type == "string" then . else .name end'
+  )
+
+  if [ "${#packages[@]}" -eq 0 ]; then
+    echo "[sync] Quelle '$name': keine Pakete gelistet, ueberspringe."
+    continue
+  fi
+
+  echo "[sync] Quelle '$name': ${#packages[@]} Paket(e) -> $baseurl"
+
+  repofile="$TMP_REPO_DIR/tmp-${name}.repo"
+  {
+    echo "[tmp-${name}]"
+    echo "name=${name}"
+    echo "baseurl=${baseurl}"
+    echo "enabled=0"
+    if [ -n "$gpgkey" ]; then
+      echo "gpgcheck=1"
+      echo "gpgkey=${gpgkey}"
+    else
+      echo "gpgcheck=0"
+    fi
+  } > "$repofile"
+
+  # Die Basis-Repos des Containers bleiben aktiviert, damit dnf
+  # Abhaengigkeiten wie glibc oder libstdc++ ueberhaupt aufloesen kann.
+  # Ohne sie scheitert bereits die Aufloesung, nicht erst der Download.
+  dl_args=(--destdir="$DEST" --enablerepo="tmp-${name}")
+
+  if [ "$resolve_deps" = "true" ]; then
+    echo "[sync] Quelle '$name': Abhaengigkeiten werden mitgespiegelt"
+    # Bewusst OHNE --alldeps: bereits im Container installierte Pakete
+    # (glibc & Co.) haben die Clients ohnehin und muessen nicht ins Repo.
+    dl_args+=(--resolve)
+  fi
+
+  if ! dnf download "${dl_args[@]}" "${packages[@]}"; then
+    echo "[sync] WARNUNG: Download fuer Quelle '$name' fehlgeschlagen."
+    download_failed=1
+  fi
+
+  rm -f "$repofile"
+done
+
+# --- 3. Alte Paketversionen aufraeumen ------------------------------------
+# repomanage braucht bereits vorhandene Metadaten. Beim allerersten Lauf
+# gibt es noch keine, deshalb der Existenztest.
+if [ -f "$DEST/repodata/repomd.xml" ]; then
+  echo "[sync] Raeume veraltete Paketversionen auf"
+  if ! dnf repomanage --old "$DEST" | xargs -r rm -f; then
+    echo "[sync] WARNUNG: repomanage fehlgeschlagen, ueberspringe Aufraeumen."
+  fi
+else
+  echo "[sync] Erster Lauf, kein Aufraeumen noetig."
+fi
+
+# --- 4. RPM-Metadaten neu erzeugen ----------------------------------------
+echo "[sync] Aktualisiere Repo-Metadaten"
+createrepo_c --update "$DEST"
+
+# --- 5. AppStream-Katalog erzeugen und einhaengen --------------------------
+# Ohne diesen Schritt sind die Pakete zwar per dnf installierbar, tauchen
+# aber weder in Discover noch in GNOME Software auf. Der Generator ist
+# optional: fehlt er, laeuft der Sync trotzdem sauber durch.
+if [ -x "$APPSTREAM_GEN" ]; then
+  echo "[sync] Erzeuge AppStream-Katalog"
+
+  ASWORK=$(mktemp -d)
+  trap 'rm -rf "$ASWORK"' EXIT
+
+  # Welche Pakete liegen tatsaechlich im Repo? Verhindert, dass Discover
+  # Eintraege anzeigt, deren RPM gar nicht da ist.
+  if compgen -G "$DEST"/*.rpm > /dev/null; then
+    rpm -qp --qf '%{NAME}\n' "$DEST"/*.rpm 2>/dev/null \
+      | sort -u > "$ASWORK/vorhanden.txt"
+  else
+    echo "[sync] WARNUNG: Keine RPMs in $DEST gefunden."
+    : > "$ASWORK/vorhanden.txt"
+  fi
+
+  python3 "$APPSTREAM_GEN" "$PACKAGES_FILE" \
+    --output "$ASWORK/appstream.xml.gz" \
+    --available-packages "$ASWORK/vorhanden.txt"
+
+  # Optionale Validierung, falls appstreamcli im Image vorhanden ist.
+  if command -v appstreamcli > /dev/null 2>&1; then
+    if ! appstreamcli validate --no-net "$ASWORK/appstream.xml.gz"; then
+      echo "[sync] WARNUNG: AppStream-Katalog hat Validierungshinweise."
+    fi
+  fi
+
+  # modifyrepo_c MUSS nach createrepo_c laufen, da createrepo_c die
+  # repomd.xml komplett neu schreibt.
+  modifyrepo_c --mdtype=appstream "$ASWORK/appstream.xml.gz" "$DEST/repodata/"
+else
+  echo "[sync] HINWEIS: $APPSTREAM_GEN nicht vorhanden - AppStream wird uebersprungen."
+fi
+
+# --- 6. Leserechte fuer den Webserver sicherstellen ------------------------
+# a+rX = Leserecht auf alles, Betretungsrecht nur auf Verzeichnisse.
+chmod -R a+rX "$REPO_ROOT"
+
+if [ "$download_failed" -ne 0 ]; then
+  echo "[sync] $(date -Iseconds) Sync-Lauf mit Fehlern abgeschlossen"
+  exit 1
+fi
+
+echo "[sync] $(date -Iseconds) Sync-Lauf abgeschlossen"

docker/syncer/sync_helpers.py

@@ -0,0 +1,367 @@
+#!/usr/bin/env python3
+"""
+sync_helpers.py - small helper used by sync.sh
+
+Reliable YAML parsing and XML escaping in pure bash/sed is a losing game, so
+these two jobs live here:
+
+  parse-yaml <file>
+      Reads the "sources:" list from the YAML file, NORMALISES it and prints
+      it as JSON on stdout, so sync.sh can loop over it safely.
+
+      Normalisation: every entry under "packages:" becomes an object with at
+      least a "name" key, whether it was written as a plain string or as a
+      mapping with extra AppStream metadata. sync.sh can therefore always do
+      p["name"] without caring which form the YAML used.
+
+  gen-appstream --origin O --yaml F --entries-dir D
+      Builds the AppStream catalog for the combined repo.
+
+      Metadata precedence, highest first:
+        1. what packages.yaml says about the package
+        2. the upstream RPM summary collected by sync.sh into the
+           appstream-entries.tsv files under --entries-dir
+        3. a generated fallback
+
+      Only packages that actually made it into the mirror (i.e. that appear
+      in one of the TSV files) get a component - never pulled-in
+      dependencies, and never packages whose download failed.
+
+  NOTE ON DISCOVER/GNOME SOFTWARE:
+      Components without an icon/screenshots/.desktop file make a package
+      installable and findable via SEARCH, but they will not appear on the
+      curated "Explore" front page or when browsing categories. That is a
+      frontend behaviour for non-desktop-app components and cannot be fixed
+      at the repo-metadata level. Only a real desktop-application component
+      (type: desktop-application + desktop_id: in packages.yaml, which
+      requires the package to ship a .desktop file) shows up there.
+"""
+import argparse
+import glob
+import json
+import os
+import re
+import sys
+from xml.sax.saxutils import escape, quoteattr
+
+CATALOG_VERSION = "0.14"
+DEFAULT_NOTE = (
+    "Bereitgestellt über das Unternehmens-Repo der IT-Abteilung. "
+    "Fragen und Paketwünsche bitte an den IT-Service-Desk."
+)
+
+
+def die(msg):
+    sys.stdout.flush()
+    sys.stderr.write(f"ERROR: {msg}\n")
+    sys.exit(1)
+
+
+def warn(msg):
+    sys.stdout.flush()
+    sys.stderr.write(f"WARNING: {msg}\n")
+
+
+def load_yaml(path):
+    try:
+        import yaml
+    except ImportError:
+        die(
+            "python3-pyyaml (PyYAML) is not installed.\n"
+            "Install it with: dnf install -y python3-pyyaml"
+        )
+    try:
+        with open(path, "r", encoding="utf-8") as fh:
+            return yaml.safe_load(fh) or {}
+    except OSError as exc:
+        die(f"could not read {path!r}: {exc}")
+    except Exception as exc:  # yaml.YAMLError and friends
+        die(f"could not parse {path!r}: {exc}")
+
+
+def normalise_package(entry, source_name, index):
+    """Accept both 'pkgname' and {name: pkgname, display_name: ...}."""
+    if isinstance(entry, str):
+        return {"name": entry}
+    if isinstance(entry, dict):
+        if not entry.get("name"):
+            die(
+                f"package #{index} in source {source_name!r} has no 'name' key: "
+                f"{entry!r}"
+            )
+        return dict(entry)
+    die(
+        f"package #{index} in source {source_name!r} must be a string or a "
+        f"mapping, got {type(entry).__name__}"
+    )
+
+
+def normalise_sources(data):
+    sources = data.get("sources") or []
+    if not isinstance(sources, list):
+        die("'sources' in YAML is not a list")
+
+    for i, src in enumerate(sources):
+        if not isinstance(src, dict):
+            die(f"source #{i} is not a mapping")
+        for field in ("name", "baseurl", "packages"):
+            if field not in src:
+                die(f"source #{i} is missing required field '{field}'")
+        if not isinstance(src["packages"], list) or not src["packages"]:
+            die(
+                f"source '{src.get('name')}' has an empty/invalid packages list"
+            )
+        src["packages"] = [
+            normalise_package(p, src.get("name"), j)
+            for j, p in enumerate(src["packages"])
+        ]
+    return sources
+
+
+def cmd_parse_yaml(path):
+    data = load_yaml(path)
+    json.dump(normalise_sources(data), sys.stdout)
+
+
+# ---------------------------------------------------------------------------
+# AppStream catalog generation
+# ---------------------------------------------------------------------------
+
+def sanitize_id_part(value):
+    part = re.sub(r"[^A-Za-z0-9]+", "-", value).strip("-")
+    if not part:
+        die(f"package name yields no usable AppStream id segment: {value!r}")
+    if part[0].isdigit():
+        part = "pkg-" + part
+    return part
+
+
+def as_list(value):
+    if value is None:
+        return []
+    if isinstance(value, (list, tuple)):
+        return list(value)
+    return [value]
+
+
+def pick(meta, defaults, key, fallback=None):
+    if meta.get(key) is not None:
+        return meta[key]
+    if defaults.get(key) is not None:
+        return defaults[key]
+    return fallback
+
+
+def read_entries(entries_dir):
+    """Merge every source's appstream-entries.tsv into {name: summary}."""
+    entries = {}
+    pattern = os.path.join(entries_dir, "*", "appstream-entries.tsv")
+    for path in sorted(glob.glob(pattern)):
+        try:
+            with open(path, encoding="utf-8") as fh:
+                for line in fh:
+                    line = line.rstrip("\n")
+                    if not line:
+                        continue
+                    parts = line.split("\t", 1)
+                    name = parts[0].strip()
+                    summary = parts[1].strip() if len(parts) > 1 else ""
+                    if not name:
+                        continue
+                    if name in entries and entries[name] and summary \
+                            and entries[name] != summary:
+                        warn(
+                            f"package '{name}' was contributed by more than one "
+                            "source with differing summaries - keeping the first"
+                        )
+                        continue
+                    entries.setdefault(name, summary)
+        except OSError:
+            continue
+    return entries
+
+
+def collect_metadata(yaml_path):
+    """Return (defaults, {pkgname: metadata}, id_prefix, yaml_origin)."""
+    data = load_yaml(yaml_path)
+    appstream_cfg = data.get("appstream") or {}
+    defaults = dict(appstream_cfg.get("defaults") or {})
+    id_prefix = appstream_cfg.get("id_prefix") or "de.firma.repo"
+    yaml_origin = appstream_cfg.get("origin")
+
+    if not defaults.get("developer_id"):
+        segments = id_prefix.split(".")
+        if len(segments) >= 2:
+            defaults["developer_id"] = ".".join(segments[:2])
+
+    by_name = {}
+    for src in normalise_sources(data):
+        for pkg in src["packages"]:
+            name = pkg["name"]
+            if name in by_name:
+                warn(f"package '{name}' listed more than once - keeping the first")
+                continue
+            by_name[name] = pkg
+    return defaults, by_name, id_prefix, yaml_origin
+
+
+def emit(out, depth, text):
+    out.append("  " * depth + text)
+
+
+def build_component(pkgname, meta, upstream_summary, defaults, id_prefix, out):
+    if meta.get("appstream") is False or meta.get("hidden") is True:
+        return False
+
+    binaries = as_list(meta.get("binary") or meta.get("binaries"))
+
+    # A console-application component without <provides><binary> trips
+    # appstreamcli's 'console-app-no-binary' check, so only claim that type
+    # when a binary is actually declared. Runtimes and libraries stay
+    # 'generic', which Discover still finds via search.
+    comp_type = pick(meta, defaults, "type")
+    if not comp_type:
+        comp_type = "console-application" if binaries else "generic"
+
+    comp_id = meta.get("id") or f"{id_prefix}.{sanitize_id_part(pkgname)}"
+
+    display_name = meta.get("display_name") or pkgname
+
+    note = pick(meta, defaults, "note", DEFAULT_NOTE)
+    summary = meta.get("summary") or upstream_summary or note
+
+    emit(out, 1, f"<component type={quoteattr(comp_type)}>")
+    emit(out, 2, f"<id>{escape(comp_id)}</id>")
+    emit(out, 2, f"<pkgname>{escape(pkgname)}</pkgname>")
+    emit(out, 2, f"<name>{escape(display_name)}</name>")
+    emit(out, 2, f"<summary>{escape(summary)}</summary>")
+
+    emit(out, 2, "<description>")
+    body = meta.get("description")
+    wrote_paragraph = False
+    if body:
+        for para in [p.strip() for p in str(body).split("\n\n") if p.strip()]:
+            emit(out, 3, f"<p>{escape(para)}</p>")
+            wrote_paragraph = True
+    if note and (not wrote_paragraph or note != summary):
+        emit(out, 3, f"<p>{escape(note)}</p>")
+    elif not wrote_paragraph:
+        emit(out, 3, f"<p>{escape(summary)}</p>")
+    emit(out, 2, "</description>")
+
+    icon = pick(meta, defaults, "icon", "application-x-executable")
+    if icon:
+        emit(out, 2, f'<icon type="stock">{escape(str(icon))}</icon>')
+
+    categories = as_list(pick(meta, defaults, "categories"))
+    if categories:
+        emit(out, 2, "<categories>")
+        for cat in categories:
+            emit(out, 3, f"<category>{escape(str(cat))}</category>")
+        emit(out, 2, "</categories>")
+
+    keywords = as_list(meta.get("keywords"))
+    if keywords:
+        emit(out, 2, "<keywords>")
+        for kw in keywords:
+            emit(out, 3, f"<keyword>{escape(str(kw))}</keyword>")
+        emit(out, 2, "</keywords>")
+
+    if binaries:
+        emit(out, 2, "<provides>")
+        for binary in binaries:
+            emit(out, 3, f"<binary>{escape(str(binary))}</binary>")
+        emit(out, 2, "</provides>")
+
+    homepage = meta.get("homepage") or defaults.get("homepage")
+    if homepage:
+        emit(out, 2, f'<url type="homepage">{escape(str(homepage))}</url>')
+    help_url = meta.get("help_url") or defaults.get("help_url")
+    if help_url:
+        emit(out, 2, f'<url type="help">{escape(str(help_url))}</url>')
+
+    developer = meta.get("developer") or defaults.get("developer")
+    if developer:
+        dev_id = meta.get("developer_id") or defaults.get("developer_id")
+        attr = f" id={quoteattr(str(dev_id))}" if dev_id else ""
+        emit(out, 2, f"<developer{attr}>")
+        emit(out, 3, f"<name>{escape(str(developer))}</name>")
+        emit(out, 2, "</developer>")
+
+    project_license = meta.get("project_license") or defaults.get("project_license")
+    if project_license:
+        emit(out, 2, f"<project_license>{escape(str(project_license))}</project_license>")
+
+    # A desktop-application component that Discover cannot launch is worse
+    # than a generic one, so insist on the desktop file id.
+    if comp_type == "desktop-application":
+        desktop_id = meta.get("desktop_id")
+        if desktop_id:
+            emit(
+                out, 2,
+                f'<launchable type="desktop-id">{escape(str(desktop_id))}</launchable>',
+            )
+        else:
+            warn(
+                f"{pkgname}: type=desktop-application without 'desktop_id' - "
+                "Discover will not be able to launch it after installation"
+            )
+
+    emit(out, 1, "</component>")
+    return True
+
+
+def cmd_gen_appstream(origin, yaml_path, entries_dir):
+    defaults, by_name, id_prefix, yaml_origin = collect_metadata(yaml_path)
+    entries = read_entries(entries_dir)
+
+    if not entries:
+        warn(
+            f"no appstream-entries.tsv found under {entries_dir} - the catalog "
+            "will be empty and nothing will show up in Discover"
+        )
+
+    effective_origin = origin or yaml_origin or "rpm-mirror"
+
+    out = []
+    out.append('<?xml version="1.0" encoding="UTF-8"?>')
+    out.append(
+        f'<components version="{CATALOG_VERSION}" origin={quoteattr(effective_origin)}>'
+    )
+
+    written = 0
+    for pkgname in sorted(entries):
+        meta = by_name.get(pkgname, {})
+        if build_component(pkgname, meta, entries[pkgname], defaults,
+                           id_prefix, out):
+            written += 1
+
+    out.append("</components>")
+    sys.stdout.write("\n".join(out) + "\n")
+    sys.stderr.write(
+        f"gen-appstream: wrote {written} component(s), origin={effective_origin}\n"
+    )
+
+
+def main():
+    ap = argparse.ArgumentParser()
+    sub = ap.add_subparsers(dest="cmd", required=True)
+
+    p1 = sub.add_parser("parse-yaml")
+    p1.add_argument("path")
+
+    p2 = sub.add_parser("gen-appstream")
+    p2.add_argument("--origin", default="")
+    p2.add_argument("--yaml", required=True)
+    p2.add_argument("--entries-dir", required=True)
+
+    args = ap.parse_args()
+
+    if args.cmd == "parse-yaml":
+        cmd_parse_yaml(args.path)
+    elif args.cmd == "gen-appstream":
+        cmd_gen_appstream(args.origin, args.yaml, args.entries_dir)
+
+
+if __name__ == "__main__":
+    main()