diff --git a/docker/syncer/Dockerfile b/docker/syncer/Dockerfile new file mode 100644 index 0000000..16b7432 --- /dev/null +++ b/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 sync_helpers.py +RUN dnf install -y \ + dnf-plugins-core \ + createrepo_c \ + appstream \ + git \ + jq \ + openssh-clients \ + python3 \ + python3-pyyaml \ + cronie \ + gnupg2 \ + rpm-sign \ + findutils \ + util-linux-core \ + curl \ + wget \ + && dnf clean all + +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/entrypoint.sh /entrypoint.sh +RUN chmod +x /usr/local/bin/sync.sh /entrypoint.sh + +ENTRYPOINT ["/entrypoint.sh"] \ No newline at end of file diff --git a/docker/syncer/entrypoint.sh b/docker/syncer/entrypoint.sh new file mode 100644 index 0000000..40d9ff5 --- /dev/null +++ b/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 \ No newline at end of file diff --git a/docker/syncer/sync.sh b/docker/syncer/sync.sh new file mode 100644 index 0000000..0c666b2 --- /dev/null +++ b/docker/syncer/sync.sh @@ -0,0 +1,527 @@ +#!/usr/bin/env bash +# +# sync.sh - Mirror selected packages from third-party RPM repos into ONE +# combined local repo consumable by AlmaLinux clients (dnf / +# PackageKit / GNOME Software / Discover) - WITHOUT pulling in or +# exposing packages that AlmaLinux itself already ships +# (AppStream, BaseOS, Extras, CRB/"Extras for Enterprise"). +# +# ALL sources defined in the YAML end up in a SINGLE repo +# (REPO_BASE_DIR/pkgs/) with a SINGLE client-facing .repo file +# (CLIENT_REPO_DIR/mirror.repo) - clients only ever add one repo, no matter +# how many sources you configure. +# +# What it does, per entry in the YAML source list: +# 1. Registers the source repo temporarily and resolves the full +# dependency closure of the requested packages, using the source repo +# AND the local AlmaLinux repos (needed so resolution succeeds, e.g. +# glibc-style deps that only exist in AlmaLinux). +# 2. Throws away every RPM in that closure that AlmaLinux itself provides. +# Only RPMs that genuinely only exist in the third-party repo are kept. +# 3. Verifies each kept RPM against the source's own gpgkey (if given in +# the YAML), then STRIPS the original signature and RE-SIGNS it with +# the mirror's own GPG key. Clients therefore only ever need to trust +# one key (this mirror's), never the individual upstream repos' keys. +# 4. Moves the signed RPMs into the single shared repo directory +# (REPO_BASE_DIR/pkgs/). +# Once all sources have been processed, the script rebuilds repo metadata +# for that ONE directory with createrepo_c and generates AppStream "generic +# component" metadata for ONLY the packages explicitly listed under +# "packages:" across all sources (never for pulled-in dependencies), +# injecting it once into that same repo. Only those show up as installable +# items in GNOME Software/Discover. +# +# Set "metadata_only: true" on a source to skip steps 2-4 entirely for that +# source: no RPMs are downloaded/mirrored, only curated AppStream visibility +# is contributed to the combined repo. Use this when baseurl already points +# at a repo your clients have enabled directly (e.g. AlmaLinux's own +# AppStream repo, which on AlmaLinux 10 ships dotnet-* itself) - you just +# want to narrow down what shows up in Discover without duplicating RPM +# content clients can already get natively. +# +# NOTE ON NAME COLLISIONS: if two different sources define an RPM with the +# exact same NEVRA (name-version-release.arch), the later one silently wins +# in the shared directory (harmless - it's the same package). If two +# sources define the SAME PACKAGE NAME with genuinely different content, +# that's not supported here - keep package names distinct across sources. +# +# IMPORTANT LIMITATION: +# Discover/GNOME Software hide plain "generic" AppStream components +# entirely unless the user enables "technische Pakete anzeigen"/"Show +# technical packages" - so every package in packages.yaml needs either +# "binary: " (-> console-application, visible by default, +# findable via search) or "type: desktop-application" + "desktop_id:" +# (needs a real .desktop file, also shows up on the curated "Explore" +# front page and in categories - console-application does not). The +# AppStream generator (sync_helpers.py) warns loudly whenever a package +# would fall back to "generic" so this doesn't go unnoticed. This is +# Discover/GNOME Software/appstreamcli behaviour and cannot be fixed by +# repo metadata alone. +# +# SECURITY NOTE ON RE-SIGNING: +# The mirror's private signing key is generated once (unattended, without +# a passphrase - standard practice for automated repo signing) and kept +# under GPG_HOME. Only the exported PUBLIC key ends up in REPO_BASE_DIR +# where clients can fetch it. GPG_HOME itself must never be exposed via +# nginx/the webserver and should live on storage only the sync process can +# read. Anyone who can read GPG_HOME can sign packages as your mirror - +# treat it like any other private key material. +# +# A revocation certificate is generated alongside the key on first run +# (GPG_HOME/revocation-cert.asc) - since the key never expires, this +# certificate is the ONLY way to invalidate it later if it's ever +# compromised. Copy it to secure, offline storage right after first +# startup; see the log output / README for the exact command. +# +# Requires: dnf, dnf-plugins-core (for "dnf download"), createrepo_c, +# modifyrepo_c, rpm-sign (rpmsign/rpmkeys), gnupg2, python3, +# python3-pyyaml, git (if fetching the source list from git), curl. +# +# Run as root (needs to write a temporary .repo file to /etc/yum.repos.d). + +set -euo pipefail + +# --------------------------------------------------------------------------- +# Configuration (override via environment, e.g. in the systemd unit/cron job) +# --------------------------------------------------------------------------- + +# Where the YAML source list lives. Either point SOURCES_GIT_REPO at a git +# repo (SOURCES_YAML_FILE is the path *inside* that repo), or leave +# SOURCES_GIT_REPO empty and point SOURCES_YAML_FILE at a local file. +SOURCES_GIT_REPO="${SOURCES_GIT_REPO:-}" +SOURCES_GIT_BRANCH="${SOURCES_GIT_BRANCH:-main}" +SOURCES_YAML_FILE="${SOURCES_YAML_FILE:-sources.yaml}" + +# Local working/output paths. +WORK_DIR="${WORK_DIR:-/var/cache/rpm-mirror-sync}" +REPO_BASE_DIR="${REPO_BASE_DIR:-/srv/repo}" +CLIENT_REPO_DIR="${CLIENT_REPO_DIR:-${REPO_BASE_DIR}/client-repos}" +# The ONE combined repo directory all sources' packages end up in. +PKG_DIR="${REPO_BASE_DIR}/pkgs" + +# Base URL under which REPO_BASE_DIR is actually served to clients (web +# server / reverse proxy in front of REPO_BASE_DIR). Used only to generate +# ready-to-use .repo files for clients. +CLIENT_BASE_URL="${CLIENT_BASE_URL:-http://mirror.example.local/repo}" + +# repoids of AlmaLinux's own repos on THIS machine, as shown by +# `dnf repolist`. Adjust to match your system if they differ. +# ("Extras for Enterprise" is the "extras" repo in AlmaLinux 10 naming; crb +# is CodeReady Builder / "extras-common" on some setups - check your +# `dnf repolist` output and adjust below.) +ALMA_REPO_IDS="${ALMA_REPO_IDS:-baseos appstream extras crb}" + +# Origin of the generated AppStream catalog. Must be unique - if it collides +# with a distribution's own origin (e.g. "almalinux"), the icon/metadata +# caches on the clients overwrite each other. Can also be set via +# "appstream: origin:" in the YAML; this variable wins. +APPSTREAM_ORIGIN="${APPSTREAM_ORIGIN:-rpm-mirror}" + +# The mirror's own signing identity. A key is generated once (on first run) +# under GPG_HOME and reused on every subsequent run - make sure GPG_HOME +# points at persistent storage (a volume), or you'll get a new key (and +# therefore a trust-breaking change for clients) on every restart. +GPG_HOME="${GPG_HOME:-${WORK_DIR}/gnupg}" +GPG_KEY_NAME="${GPG_KEY_NAME:-RPM Mirror}" +GPG_KEY_EMAIL="${GPG_KEY_EMAIL:-rpm-mirror@example.local}" +MIRROR_GPG_KEY_FILE="${REPO_BASE_DIR}/RPM-GPG-KEY-mirror" + +LOG_TAG="rpm-mirror-sync" + +# --------------------------------------------------------------------------- +# Helpers +# --------------------------------------------------------------------------- + +log() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] $*" | tee -a "${WORK_DIR}/sync.log" >&2; } +die() { log "ERROR: $*"; exit 1; } + +require_cmd() { + command -v "$1" >/dev/null 2>&1 || die "required command '$1' not found (install it first)" +} + +cleanup() { + # Remove any temp source .repo files we created, even on failure. + if [[ -n "${TEMP_REPO_FILES:-}" ]]; then + for f in "${TEMP_REPO_FILES[@]}"; do + [[ -n "$f" ]] && rm -f "$f" + done + fi +} + +# Remove one entry from TEMP_REPO_FILES. +# NOTE: "${ARRAY[@]/pattern}" does SUBSTRING REPLACEMENT, so the element is +# left behind as an empty string rather than removed - which then makes +# cleanup() call `rm -f ""`. Rebuild the array instead. +forget_repo_file() { + local drop="$1" keep=() f + for f in "${TEMP_REPO_FILES[@]}"; do + [[ "$f" == "$drop" ]] || keep+=( "$f" ) + done + TEMP_REPO_FILES=( "${keep[@]}" ) +} +trap cleanup EXIT + +SCRIPT_DIR="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" >/dev/null 2>&1 && pwd)" +HELPER="${SCRIPT_DIR}/sync_helpers.py" + +# --------------------------------------------------------------------------- +# Preconditions +# --------------------------------------------------------------------------- + +[[ $EUID -eq 0 ]] || die "must run as root (writes to /etc/yum.repos.d and ${REPO_BASE_DIR})" + +for c in dnf createrepo_c modifyrepo_c rpm rpmsign rpmkeys gpg python3 curl gzip; do + require_cmd "$c" +done +[[ -f "$HELPER" ]] || die "sync_helpers.py not found next to sync.sh (expected: $HELPER)" +python3 -c "import yaml" 2>/dev/null || die "python3-pyyaml not installed (dnf install -y python3-pyyaml)" + +mkdir -p "$WORK_DIR" "$REPO_BASE_DIR" "$CLIENT_REPO_DIR" "$PKG_DIR" + +# --------------------------------------------------------------------------- +# Mirror signing key: generate once, reuse forever, export the public part +# --------------------------------------------------------------------------- + +MIRROR_GPG_KEY_ID="" +REVOCATION_CERT_FILE="${GPG_HOME}/revocation-cert.asc" + +init_mirror_gpg_key() { + mkdir -p "$GPG_HOME" + chmod 700 "$GPG_HOME" + + if [[ -z "$(gpg --homedir "$GPG_HOME" --list-secret-keys --with-colons 2>/dev/null)" ]]; then + log "No mirror signing key found under ${GPG_HOME} - generating one (one-time)..." + gpg --homedir "$GPG_HOME" --batch --pinentry-mode loopback --passphrase '' \ + --quick-generate-key "${GPG_KEY_NAME} <${GPG_KEY_EMAIL}>" rsa4096 sign never \ + || die "failed to generate mirror GPG signing key" + + # GnuPG >= 2.1 automatically writes a revocation certificate under + # openpgp-revocs.d/.rev as part of key generation itself + # (visible in the gpg output above: "revocation certificate stored + # as ..."). We just need to copy it to a known, stable path - no + # need to (fragile-ly) drive gpg's interactive --gen-revoke menu + # ourselves, which behaves differently across gpg versions. + local auto_revoc + auto_revoc="$(find "${GPG_HOME}/openpgp-revocs.d" -maxdepth 1 -name '*.rev' 2>/dev/null | head -n1)" + [[ -n "$auto_revoc" && -s "$auto_revoc" ]] \ + || die "expected GnuPG to auto-generate a revocation certificate under ${GPG_HOME}/openpgp-revocs.d/ but none was found" + + cp "$auto_revoc" "$REVOCATION_CERT_FILE" + chmod 600 "$REVOCATION_CERT_FILE" + + log "!!! Revocation certificate written to: ${REVOCATION_CERT_FILE}" + log "!!! Copy it to secure, OFFLINE storage now, e.g.:" + log "!!! docker compose cp syncer:${REVOCATION_CERT_FILE} ./mirror-key-revocation-cert.asc" + log "!!! It is the only way to invalidate this key if it is ever compromised -" + log "!!! it never expires, so it needs the same protection as the private key." + fi + + MIRROR_GPG_KEY_ID="$(gpg --homedir "$GPG_HOME" --list-secret-keys --with-colons \ + | awk -F: '/^sec:/ {print $5; exit}')" + [[ -n "$MIRROR_GPG_KEY_ID" ]] || die "could not determine mirror GPG key id" + + # rpmsign/rpm --addsign etc. read the signing identity from macros. + cat > "${HOME:-/root}/.rpmmacros" < "$MIRROR_GPG_KEY_FILE" + + log "Mirror signing key ready: ${MIRROR_GPG_KEY_ID} (public key: ${MIRROR_GPG_KEY_FILE})" +} + +init_mirror_gpg_key + + + +# --------------------------------------------------------------------------- +# Fetch the YAML source list +# --------------------------------------------------------------------------- + +resolve_yaml_path() { + if [[ -n "$SOURCES_GIT_REPO" ]]; then + if [[ "$SOURCES_YAML_FILE" == /* ]]; then + die "SOURCES_GIT_REPO is set, but SOURCES_YAML_FILE ('${SOURCES_YAML_FILE}') looks like an absolute path. When pulling the source list from git, SOURCES_YAML_FILE must be a path RELATIVE to that repo's root (e.g. 'sources.yaml' or 'config/sources.yaml') - not the '/config/...' path used for the local bind-mount option. Fix SOURCES_YAML_FILE in .env." + fi + local clone_dir="${WORK_DIR}/sources-repo" + if [[ -d "${clone_dir}/.git" ]]; then + log "Updating source list git repo..." + git -C "$clone_dir" fetch --depth 1 origin "$SOURCES_GIT_BRANCH" + git -C "$clone_dir" reset --hard "origin/${SOURCES_GIT_BRANCH}" + else + log "Cloning source list git repo..." + rm -rf "$clone_dir" + git clone --depth 1 --branch "$SOURCES_GIT_BRANCH" "$SOURCES_GIT_REPO" "$clone_dir" + fi + echo "${clone_dir}/${SOURCES_YAML_FILE}" + else + echo "$SOURCES_YAML_FILE" + fi +} + +YAML_PATH="$(resolve_yaml_path)" +[[ -f "$YAML_PATH" ]] || die "source YAML not found: $YAML_PATH" + +SOURCES_JSON="$(python3 "$HELPER" parse-yaml "$YAML_PATH")" || die "failed to parse $YAML_PATH" +SOURCE_COUNT="$(python3 -c 'import json,sys; print(len(json.load(sys.stdin)))' <<<"$SOURCES_JSON")" +log "Loaded $SOURCE_COUNT source(s) from $YAML_PATH" + +# --------------------------------------------------------------------------- +# Per-source sync +# --------------------------------------------------------------------------- + +declare -a TEMP_REPO_FILES=() + +# Build --enablerepo flags for the AlmaLinux repos as a bash array once. +declare -a ALMA_ENABLE_FLAGS=() +for r in $ALMA_REPO_IDS; do + ALMA_ENABLE_FLAGS+=( "--enablerepo=${r}" ) +done + +sync_one_source() { + local name="$1" baseurl="$2" gpgkey="$3" packages_json="$4" metadata_only="${5:-false}" + + local repoid="mirror-src-${name}" + local repo_file="/etc/yum.repos.d/${repoid}.repo" + local tmp_dir="${WORK_DIR}/${name}/download" + local list_dir="${WORK_DIR}/${name}" + # This source's contribution to the combined AppStream metadata: one + # "namesummary" line per explicitly requested package. Collected + # here, merged across all sources, and turned into ONE appstream.xml + # after the whole source loop finishes. + local entries_tsv="${list_dir}/appstream-entries.tsv" + + log "=== Syncing source '${name}' (metadata_only=${metadata_only}) ===" + mkdir -p "$tmp_dir" "$list_dir" + : > "$entries_tsv" + + # readarray of requested top-level package names for this source. + # parse-yaml normalises every entry to an object, so a package written + # as a plain string and one written as a mapping with AppStream metadata + # both arrive here as {"name": ...}. + mapfile -t packages < <(python3 -c ' +import json,sys +for p in json.load(sys.stdin): + print(p["name"]) +' <<<"$packages_json") + [[ ${#packages[@]} -gt 0 ]] || die "source '${name}' has no packages" + + # --- 1. Register the source repo temporarily ------------------------- + { + echo "[${repoid}]" + echo "name=Mirror source: ${name}" + echo "baseurl=${baseurl}" + echo "enabled=1" + if [[ -n "$gpgkey" ]]; then + echo "gpgcheck=1" + echo "gpgkey=${gpgkey}" + else + echo "gpgcheck=0" + fi + } > "$repo_file" + TEMP_REPO_FILES+=( "$repo_file" ) + dnf clean expire-cache --disablerepo='*' --enablerepo="$repoid" >/dev/null 2>&1 || true + + # --- metadata_only sources: curate Discover/GNOME Software visibility + # WITHOUT mirroring any RPM content. Use this when baseurl already + # points at a repo your clients have enabled directly (e.g. AlmaLinux's + # own AppStream repo, which ships dotnet-* itself on AlmaLinux 10) - the + # actual package install is then resolved by dnf from that repo as + # normal; this source only contributes AppStream "generic component" + # data for the packages listed to the combined repo built after the + # source loop, so ONLY those show up as installable in Discover instead + # of AlmaLinux's entire (huge) AppStream catalog. + # PRECONDITION: clients must have the repo(s) that actually provide + # these packages enabled already (true for any stock AlmaLinux install - + # baseos/appstream/extras/crb are on by default). + if [[ "$metadata_only" == "true" ]]; then + for pkg in "${packages[@]}"; do + local summary + summary="$(dnf repoquery -y --disablerepo='*' --enablerepo="$repoid" \ + --qf '%{summary}' "$pkg" 2>/dev/null | head -n1)" + printf '%s\t%s\n' "$pkg" "$summary" >> "$entries_tsv" + done + + log "Source '${name}': metadata_only - queued AppStream visibility for ${#packages[@]} package(s), no RPMs mirrored" + + rm -f "$repo_file" + forget_repo_file "$repo_file" + return + fi + + # --- 2. List everything the source repo itself provides -------------- + # (used later to decide "does this RPM genuinely come from the source, + # or did AlmaLinux already have it".) + local source_nevra_list="${list_dir}/source-nevra.txt" + dnf repoquery -y --disablerepo='*' --enablerepo="$repoid" \ + --qf '%{name}-%{version}-%{release}.%{arch}' -a \ + > "$source_nevra_list" \ + || die "failed to query source repo '${name}' (check baseurl/gpgkey)" + [[ -s "$source_nevra_list" ]] || die "source repo '${name}' returned no packages - check baseurl" + + # --- 3. Resolve full dependency closure for the requested packages --- + # Enable the source repo AND the local AlmaLinux repos so dependency + # resolution succeeds even for deps AlmaLinux normally provides - + # we filter those back out in step 4. + rm -rf "$tmp_dir"; mkdir -p "$tmp_dir" + log "Resolving + downloading dependency closure for: ${packages[*]}" + dnf download -y --resolve --alldeps \ + --destdir="$tmp_dir" \ + --disablerepo='*' \ + --enablerepo="$repoid" \ + "${ALMA_ENABLE_FLAGS[@]}" \ + "${packages[@]}" \ + || die "dnf download failed for source '${name}'" + + # --- 4. Drop everything AlmaLinux already provides, verify + re-sign -- + # the rest with the mirror's own key ----------------------------------- + local verify_dbpath="${list_dir}/verify-rpmdb" + if [[ -n "$gpgkey" ]]; then + local source_gpgkey_file="${list_dir}/source-gpgkey.asc" + curl -fsSL "$gpgkey" -o "$source_gpgkey_file" \ + || die "could not fetch gpgkey for source '${name}' from ${gpgkey}" + rm -rf "$verify_dbpath"; mkdir -p "$verify_dbpath" + rpm --dbpath "$verify_dbpath" --initdb + rpm --dbpath "$verify_dbpath" --import "$source_gpgkey_file" \ + || die "could not import gpgkey for source '${name}'" + else + log " WARNING: source '${name}' has no gpgkey configured - skipping upstream signature verification" + fi + + local kept=0 dropped=0 kept_requested=0 kept_dependency=0 + for f in "$tmp_dir"/*.rpm; do + [[ -e "$f" ]] || continue + local base; base="$(basename "$f" .rpm)" + + if ! grep -qxF "$base" "$source_nevra_list"; then + log " dropping (provided by AlmaLinux): $base" + rm -f "$f" + dropped=$((dropped + 1)) + continue + fi + + if [[ -n "$gpgkey" ]]; then + rpmkeys --dbpath "$verify_dbpath" --checksig "$f" >/dev/null 2>&1 \ + || die "signature verification FAILED for '${base}' from source '${name}' - refusing to mirror it (tampered download or wrong gpgkey configured)" + fi + + # Strip the upstream signature and sign with the mirror's own key, + # so clients only ever need to trust MIRROR_GPG_KEY_FILE. + rpmsign --resign "$f" >/dev/null \ + || die "failed to re-sign '${base}' with the mirror key" + + # Label each kept package as explicitly "requested" (listed under + # packages: in the YAML) or a pulled-in "dependency" - packages the + # requested ones need but AlmaLinux doesn't provide, so they have to + # be mirrored too or `dnf install ` breaks on clients. + # This is purely informational (log output); both kinds are kept. + local pkg_name tag + pkg_name="$(rpm -qp --qf '%{NAME}' "$f" 2>/dev/null)" + tag="dependency" + for req in "${packages[@]}"; do + if [[ "$pkg_name" == "$req" ]]; then + tag="requested" + break + fi + done + if [[ "$tag" == "requested" ]]; then + kept_requested=$((kept_requested + 1)) + local summary + summary="$(rpm -qp --qf '%{SUMMARY}' "$f" 2>/dev/null)" + printf '%s\t%s\n' "$pkg_name" "$summary" >> "$entries_tsv" + else + kept_dependency=$((kept_dependency + 1)) + fi + log " keeping (${tag}): $base" + + mv "$f" "${PKG_DIR}/" + kept=$((kept + 1)) + done + log "Source '${name}': kept ${kept} package(s) - ${kept_requested} explicitly requested + ${kept_dependency} pulled-in third-party dependencies (re-signed with mirror key), dropped ${dropped} AlmaLinux-provided package(s)" + [[ $kept -gt 0 ]] || die "nothing kept for source '${name}' - check baseurl/package names" + + # --- cleanup temp repo file for this source --------------------------- + rm -f "$repo_file" + forget_repo_file "$repo_file" +} + +# Iterate over sources (JSON array -> one JSON object per line via jq-less python) +while IFS= read -r src_json; do + name="$(python3 -c 'import json,sys;print(json.loads(sys.argv[1])["name"])' "$src_json")" + baseurl="$(python3 -c 'import json,sys;print(json.loads(sys.argv[1])["baseurl"])' "$src_json")" + gpgkey="$(python3 -c 'import json,sys;print(json.loads(sys.argv[1]).get("gpgkey",""))' "$src_json")" + pkgs_json="$(python3 -c 'import json,sys;print(json.dumps(json.loads(sys.argv[1])["packages"]))' "$src_json")" + metadata_only="$(python3 -c 'import json,sys;print(str(bool(json.loads(sys.argv[1]).get("metadata_only", False))).lower())' "$src_json")" + sync_one_source "$name" "$baseurl" "$gpgkey" "$pkgs_json" "$metadata_only" +done < <(python3 -c ' +import json,sys +for s in json.load(sys.stdin): + print(json.dumps(s)) +' <<<"$SOURCES_JSON") + +# --------------------------------------------------------------------------- +# Build the ONE combined repo from everything all sources contributed +# --------------------------------------------------------------------------- + +createrepo_c --update "$PKG_DIR" >/dev/null + +# Build ONE AppStream catalog for the single shared repo. The helper reads +# every source's appstream-entries.tsv (written during its own processing +# above) to know WHICH packages made it into the mirror, and packages.yaml +# to know HOW to present them (display name, description, categories, +# keywords). YAML metadata wins; the upstream RPM summary from the TSV is +# the fallback. +APPSTREAM_XML="${WORK_DIR}/appstream.xml" +python3 "$HELPER" gen-appstream \ + --origin "$APPSTREAM_ORIGIN" \ + --yaml "$YAML_PATH" \ + --entries-dir "$WORK_DIR" \ + > "$APPSTREAM_XML" \ + || die "failed to generate AppStream catalog" + +# Purely informational: appstreamcli's checks target the style expectations +# for public app stores (content ratings, screenshots, ...). They do not +# affect whether Discover shows the packages. +if command -v appstreamcli >/dev/null 2>&1; then + appstreamcli validate --no-net "$APPSTREAM_XML" >/dev/null 2>&1 \ + || log "note: AppStream catalog has style hints (harmless) - run 'appstreamcli validate ${APPSTREAM_XML}' to see them" +fi + +gzip -fk "$APPSTREAM_XML" +modifyrepo_c --mdtype=appstream "${APPSTREAM_XML}.gz" "${PKG_DIR}/repodata" >/dev/null + +# Detached-sign repomd.xml itself with the mirror key, so clients can verify +# the METADATA (which packages/versions exist, their checksums) hasn't been +# tampered with - not just each individual RPM. Must happen last, after +# createrepo_c/modifyrepo_c are done touching repomd.xml (they update its +# checksums), otherwise the signature would be over stale content. +gpg --homedir "$GPG_HOME" --batch --pinentry-mode loopback --passphrase '' \ + --detach-sign --armor \ + --output "${PKG_DIR}/repodata/repomd.xml.asc" \ + "${PKG_DIR}/repodata/repomd.xml" \ + || die "failed to sign repomd.xml with the mirror key" + +# --- The ONE client-facing .repo file - this is all clients ever need ----- +cat > "${CLIENT_REPO_DIR}/mirror.repo" < + 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: + Discover/GNOME Software hide plain 'generic' AppStream components by + default - they only appear once the user enables "technische Pakete + anzeigen"/"Show technical packages". This module therefore only lets a + package fall back to 'generic' (and warns loudly when it does) when + packages.yaml sets neither 'binary:' nor an explicit 'type:'. Every + package should set 'binary: ' (becomes a console-application, + visible by default, requires a real CLI entry point after install) or + 'type: desktop-application' + 'desktop_id:' (requires the package to + ship a .desktop file) - only these two types are guaranteed visible + without the user opening that hidden setting. A console-application + still only shows up via SEARCH, not on the curated "Explore" front page + or when browsing categories - that page is reserved for + desktop-application components with an icon/screenshots. That is a + frontend behaviour and cannot be fixed at the repo-metadata level. +""" +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 trips + # appstreamcli's 'console-app-no-binary' check, so only claim that type + # when a binary is actually declared. + comp_type = pick(meta, defaults, "type") + if not comp_type: + if binaries: + comp_type = "console-application" + else: + comp_type = "generic" + # Discover/GNOME Software hide 'generic' components by default - + # they only show up once the user enables "technische Pakete + # anzeigen"/"Show technical packages". Every package listed here + # should therefore set either 'binary: ' (the package + # becomes a console-application, visible by default) or + # 'type: desktop-application' + 'desktop_id:' for a real GUI + # app. Falling through to 'generic' almost always means the + # package silently won't be visible for end users. + warn( + f"{pkgname}: no 'binary' and no 'type' set - this package " + "becomes a 'generic' AppStream component, which Discover/" + "GNOME Software hide unless the user has \"technische " + "Pakete anzeigen\"/\"Show technical packages\" enabled. Set " + "'binary: ' (console-application) or " + "'type: desktop-application' with 'desktop_id:' in " + "packages.yaml so it's visible by default." + ) + + 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"") + emit(out, 2, f"{escape(comp_id)}") + emit(out, 2, f"{escape(pkgname)}") + emit(out, 2, f"{escape(display_name)}") + emit(out, 2, f"{escape(summary)}") + + emit(out, 2, "") + 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"

{escape(para)}

") + wrote_paragraph = True + if note and (not wrote_paragraph or note != summary): + emit(out, 3, f"

{escape(note)}

") + elif not wrote_paragraph: + emit(out, 3, f"

{escape(summary)}

") + emit(out, 2, "") + + icon = pick(meta, defaults, "icon", "application-x-executable") + if icon: + emit(out, 2, f'{escape(str(icon))}') + + categories = as_list(pick(meta, defaults, "categories")) + if categories: + emit(out, 2, "") + for cat in categories: + emit(out, 3, f"{escape(str(cat))}") + emit(out, 2, "") + + keywords = as_list(meta.get("keywords")) + if keywords: + emit(out, 2, "") + for kw in keywords: + emit(out, 3, f"{escape(str(kw))}") + emit(out, 2, "") + + if binaries: + emit(out, 2, "") + for binary in binaries: + emit(out, 3, f"{escape(str(binary))}") + emit(out, 2, "") + + homepage = meta.get("homepage") or defaults.get("homepage") + if homepage: + emit(out, 2, f'{escape(str(homepage))}') + help_url = meta.get("help_url") or defaults.get("help_url") + if help_url: + emit(out, 2, f'{escape(str(help_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"") + emit(out, 3, f"{escape(str(developer))}") + emit(out, 2, "") + + project_license = meta.get("project_license") or defaults.get("project_license") + if project_license: + emit(out, 2, f"{escape(str(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'{escape(str(desktop_id))}', + ) + else: + warn( + f"{pkgname}: type=desktop-application without 'desktop_id' - " + "Discover will not be able to launch it after installation" + ) + + emit(out, 1, "") + 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('') + out.append( + f'' + ) + + 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("") + 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() \ No newline at end of file