usm/scripts/dcba-e2e.sh

#!/bin/bash
#
# DCBA end-to-end verification — no network, file:// repositories only.
#
# Builds USM, then drives two fully isolated USM roots (producer A and
# consumer B) through the whole DCBA story: a tiny fixture package is
# packaged, published in a file:// source repository and installed; the
# resulting build cache is turned into a .dcba artifact published in a
# file:// DCBA repository; a second root installs from scratch and must
# receive the pre-built artifact as its build.tar.xz build cache.
#
# Assertions (each printed PASS/FAIL, exit non-zero on any failure):
#   a. <state>/packages/<name>-<version>/build.tar.xz exists right after
#      the download phase, and no local build ran (the build script's
#      one-time fingerprint is absent from the artifact)
#   b. origin-info records the DCBA repository name in "dcba"
#   c. the package installs successfully
#   d. usm clean builds removes build.tar.xz and build.tar.xz.digest
#   e. a second install after removing the cache re-downloads the
#      artifact (the "Retrieved pre-built artifact" line reappears)
#   f. a tampered artifact in the repository (content no longer matching
#      the signed listing checksum) is rejected with a warning and the
#      install still succeeds via a source build
#
# Environment quirks this script works around (keep in mind when
# running it elsewhere):
#   * The freshly built usm binary links libusm-0.1.so by SONAME, and the
#     host carries an older /opt/usm/lib/libusm-0.1.so without the DCBA
#     symbols — every invocation needs LD_LIBRARY_PATH=<build>/lib or it
#     silently talks to the stale library.
#   * meson only finds the Invercargill libraries through
#     PKG_CONFIG_PATH=/opt/usm/lib/pkgconfig.
#   * USM never creates the state root itself: each root's
#     <state>/{lists,packages,installed} must exist before the first
#     repository refresh/install.
#   * usm.config must exist per root with is_managed=true and a
#     "managed"."state_path" pointing at that root's state directory;
#     USM_CONFIGDIR/USM_DESTDIR (see Paths.override_from_environment)
#     keep the roots apart.
#
# Usage: scripts/dcba-e2e.sh [build-dir]
#   build-dir defaults to a fresh temp directory; pass an existing meson
#   build directory to skip the (slow) compile.

set -u
set -o pipefail

SCRIPT_DIR=$(cd -- "$(dirname -- "$0")" && pwd)
SOURCE_DIR=$(cd -- "$SCRIPT_DIR/.." && pwd)

WORK=$(mktemp -d /tmp/usm-dcba-e2e.XXXXXX)
trap 'rm -rf "$WORK"' EXIT

PASS_COUNT=0
FAIL_COUNT=0

report() {
    if [ "$1" = 0 ]; then
        printf 'PASS  %s\n' "$2"
        PASS_COUNT=$((PASS_COUNT + 1))
    else
        printf 'FAIL  %s\n' "$2"
        FAIL_COUNT=$((FAIL_COUNT + 1))
    fi
}

# ---------------------------------------------------------------------------
# 0. Build
# ---------------------------------------------------------------------------

BUILD_DIR=${1:-}

if [ -z "$BUILD_DIR" ]; then
    BUILD_DIR="$WORK/usm-build"
    printf '== Building USM into %s\n' "$BUILD_DIR"
    meson setup "$BUILD_DIR" "$SOURCE_DIR/src" >/dev/null 2>&1 \
        || { echo "meson setup failed"; exit 1; }
    (cd "$BUILD_DIR" && PKG_CONFIG_PATH=/opt/usm/lib/pkgconfig ninja >/dev/null 2>&1) \
        || { echo "ninja failed"; exit 1; }
fi

USM_BIN="$BUILD_DIR/cli/usm"
[ -x "$USM_BIN" ] || { echo "no usm binary at $USM_BIN"; exit 1; }

# The stale-host-library workaround described in the header
export LD_LIBRARY_PATH="$BUILD_DIR/lib${LD_LIBRARY_PATH:+:$LD_LIBRARY_PATH}"

# ---------------------------------------------------------------------------
# 1. Two isolated roots
# ---------------------------------------------------------------------------

write_config() {
    mkdir -p "$1"
    cat > "$1/usm.config" <<EOF
{
  "is_managed": true,
  "managed": { "state_path": "$2" },
  "download_timeout": 300,
  "paths": {
    "prefix": "/usr", "bin": "bin", "sbin": "sbin", "lib": "lib64", "canonlib": "lib",
    "libexec": "libexec", "include": "include", "data": "share", "info": "share/info",
    "man": "share/man", "locale": "share/locale", "localState": "var", "sharedState": "com",
    "sysConfig": "etc", "tags": "share/usm-tags"
  }
}
EOF
}

make_root() {
    # <root>/config (USM_CONFIGDIR) and <root>/state, with the
    # subdirectories USM expects to already exist
    rm -rf "$1" "$2"
    mkdir -p "$1" "$2/lists" "$2/packages" "$2/installed"
    write_config "$1" "$2"
}

ROOT_A_CONFIG="$WORK/rootA/config"
ROOT_A_STATE="$WORK/rootA/state"
ROOT_B_CONFIG="$WORK/rootB/config"
ROOT_B_STATE="$WORK/rootB/state"

make_root "$ROOT_A_CONFIG" "$ROOT_A_STATE"
make_root "$ROOT_B_CONFIG" "$ROOT_B_STATE"

# usm_a/usm_b/usm_pkg run the built binary against one isolated root
usm_a() {
    (cd "$WORK" \
        && USM_CONFIGDIR="$ROOT_A_CONFIG" USM_DESTDIR="$WORK/rootA/dest" "$USM_BIN" "$@")
}
usm_b() {
    # Root B builds (only in the tamper fallback) carry B's fingerprint
    (cd "$WORK" \
        && DCBA_E2E_FINGERPRINT="$FINGERPRINT_B" \
           USM_CONFIGDIR="$ROOT_B_CONFIG" USM_DESTDIR="$WORK/rootB/dest" "$USM_BIN" "$@")
}

# ---------------------------------------------------------------------------
# 2. Root A: package a fixture, publish it, install it
# ---------------------------------------------------------------------------

PKG="$WORK/pkg"
mkdir -p "$PKG/usm-scripts"

# Two fingerprints, one per root: the build script embeds the one its
# environment passes through DCBA_E2E_FINGERPRINT, so root A's artifact
# and root B's local build are distinguishable — root B reusing the
# downloaded artifact must contain A's fingerprint and not B's
# (assertion a), while the tamper fallback build (assertion f) must
# contain B's.
FINGERPRINT_A="dcba-e2e-a-$(date +%s)-$RANDOM"
FINGERPRINT_B="dcba-e2e-b-$(date +%s)-$RANDOM"

cat > "$PKG/MANIFEST.usm" <<EOF
{
  "name": "dcba-e2e-demo",
  "version": "1.0.0",
  "summary": "tiny dcba end-to-end fixture",
  "licences": [{ "name": "MIT", "text": "LICENCE", "category": "libre" }],
  "flags": [],
  "provides": { "res:dcba-e2e-demo/hello.txt": "build:hello.txt" },
  "depends": { "runtime": [], "build": [], "manage": [] },
  "execs": { "build": "usm-scripts/build.sh" }
}
EOF

printf 'MIT\n' > "$PKG/LICENCE"
# The fingerprint rides in through the environment so the very same
# .usmc (whose bytes must not differ between the roots) builds with a
# per-root marker. The script is incremental like any real build system:
# an up-to-date output is left untouched, so a restored build cache means
# the compile step does nothing — which is exactly what assertion (a)
# observes (USM always re-runs the build exec; reuse is the build
# directory being restored beforehand, not the exec being skipped).
cat > "$PKG/usm-scripts/build.sh" <<'EOF'
#!/bin/bash
set -e
if [ ! -f "$1/hello.txt" ]; then
    echo "built locally ${DCBA_E2E_FINGERPRINT:-unset}" > "$1/hello.txt"
fi
EOF
chmod +x "$PKG/usm-scripts/build.sh"

printf '== Root A: packaging the fixture\n'
(cd "$PKG" && USM_CONFIGDIR="$ROOT_A_CONFIG" USM_DESTDIR="$WORK/rootA/dest" \
    DCBA_E2E_FINGERPRINT="$FINGERPRINT_A" \
    "$USM_BIN" manifest package >/dev/null 2>&1) || { echo "usm manifest package failed"; exit 1; }
USMC="$WORK/dcba-e2e-demo-1.0.0.usmc"
[ -f "$USMC" ] || { echo "no .usmc produced"; exit 1; }

SRC_REPO="$WORK/source-repo"
mkdir -p "$SRC_REPO"
printf '== Root A: source repository\n'
(cd "$SRC_REPO" && USM_CONFIGDIR="$ROOT_A_CONFIG" USM_DESTDIR="$WORK/rootA/dest" \
    "$USM_BIN" repository init demo >/dev/null 2>&1) || { echo "usm repository init failed"; exit 1; }
(cd "$SRC_REPO" && USM_CONFIGDIR="$ROOT_A_CONFIG" USM_DESTDIR="$WORK/rootA/dest" \
    "$USM_BIN" repository add "$USMC" >/dev/null 2>&1) || { echo "usm repository add failed"; exit 1; }

usm_a add-repo -y "$SRC_REPO/demo.usmr" >/dev/null 2>&1 || { echo "usm add-repo failed"; exit 1; }
DCBA_E2E_FINGERPRINT="$FINGERPRINT_A" \
    usm_a install dcba-e2e-demo -y >"$WORK/rootA-install.log" 2>&1 \
    || { echo "root A install failed"; cat "$WORK/rootA-install.log"; exit 1; }

CACHE_A="$ROOT_A_STATE/packages/dcba-e2e-demo-1.0.0"
[ -f "$CACHE_A/build.tar.xz" ] || { echo "root A produced no build.tar.xz"; exit 1; }
[ -f "$CACHE_A/build.tar.xz.digest" ] || { echo "root A produced no build.tar.xz.digest"; exit 1; }

# ---------------------------------------------------------------------------
# 3. Root A: produce the .dcba and run the DCBA repository tooling
# ---------------------------------------------------------------------------

DCBA_REPO="$WORK/dcba-repo"
mkdir -p "$DCBA_REPO"
printf '== Producer: DCBA repository\n'
(cd "$DCBA_REPO" && USM_CONFIGDIR="$ROOT_A_CONFIG" USM_DESTDIR="$WORK/rootA/dest" \
    "$USM_BIN" dcba repository init demoart >/dev/null 2>&1) || { echo "usm dcba repository init failed"; exit 1; }

# The manual step: the cache's build.tar.xz IS the artifact, renamed
cp "$CACHE_A/build.tar.xz" "$DCBA_REPO/dcba-e2e-demo-1.0.0.dcba"
(cd "$DCBA_REPO" && USM_CONFIGDIR="$ROOT_A_CONFIG" USM_DESTDIR="$WORK/rootA/dest" \
    "$USM_BIN" dcba repository add dcba-e2e-demo-1.0.0.dcba >/dev/null 2>&1) \
    || { echo "usm dcba repository add failed"; exit 1; }
[ -f "$DCBA_REPO/public/ARTIFACTS.usml" ] || { echo "no ARTIFACTS.usml built"; exit 1; }

# The listing and the artifacts it references must verify
(cd "$DCBA_REPO" && USM_CONFIGDIR="$ROOT_A_CONFIG" USM_DESTDIR="$WORK/rootA/dest" \
    "$USM_BIN" dcba repository verify demoart.dcbar >/dev/null 2>&1)
report $? "dcba repository verify (signature + checksums)"

# ---------------------------------------------------------------------------
# 4. Root B: consume both repositories
# ---------------------------------------------------------------------------

printf '== Root B: install from scratch with the DCBA repository configured\n'
usm_b add-repo -y "$SRC_REPO/demo.usmr" >/dev/null 2>&1 || { echo "root B add-repo failed"; exit 1; }
usm_b dcba add-repo -y "$DCBA_REPO/demoart.dcbar" >"$WORK/rootB-addrepo.log" 2>&1 \
    || { echo "root B dcba add-repo failed"; cat "$WORK/rootB-addrepo.log"; exit 1; }

usm_b install dcba-e2e-demo -y >"$WORK/rootB-install.log" 2>&1
INSTALL_EXIT=$?
report $INSTALL_EXIT "(c) usm install succeeds with a DCBA repository configured"

CACHE_B="$ROOT_B_STATE/packages/dcba-e2e-demo-1.0.0"
INSTALLED_B="$WORK/rootB/dest/usr/share/dcba-e2e-demo/hello.txt"

# (a) the artifact landed as the build cache, without a local build
[ -f "$CACHE_B/build.tar.xz" ]
report $? "(a) build.tar.xz exists in the cache directory after the download"

if tar -xJOf "$CACHE_B/build.tar.xz" ./hello.txt 2>/dev/null | grep -q "$FINGERPRINT_A"; then
    report 0 "(a) artifact is the producer's (root A fingerprint present)"
else
    report 1 "(a) artifact is the producer's (root A fingerprint present)"
fi
if tar -xJOf "$CACHE_B/build.tar.xz" ./hello.txt 2>/dev/null | grep -q "$FINGERPRINT_B"; then
    report 1 "(a) artifact was reused, not rebuilt locally (root B fingerprint absent)"
else
    report 0 "(a) artifact was reused, not rebuilt locally (root B fingerprint absent)"
fi

# (b) provenance
grep -q '"dcba"[[:space:]]*:[[:space:]]*"demoart"' "$CACHE_B/origin-info"
report $? "(b) origin-info records the DCBA repository in \"dcba\""

# (c) the install actually installed the resource
[ -f "$INSTALLED_B" ]
report $? "(c) the package's resource was installed"

# The download log must show the artifact retrieval
tr '\r' '\n' < "$WORK/rootB-install.log" | grep -q "Retrieved pre-built artifact"
report $? "(evidence) install log reports the retrieved pre-built artifact"

# ---------------------------------------------------------------------------
# 5. (d) usm clean builds clears the artifact cache
# ---------------------------------------------------------------------------

usm_b clean builds >/dev/null 2>&1
CLEAN_OK=$?
[ ! -e "$CACHE_B/build.tar.xz" ] && [ ! -e "$CACHE_B/build.tar.xz.digest" ]
NO_CACHE=$?
report $CLEAN_OK "(d) usm clean builds succeeds"
report $NO_CACHE "(d) usm clean builds removed build.tar.xz and build.tar.xz.digest"

# ---------------------------------------------------------------------------
# 6. (e) a clean install re-downloads the artifact
# ---------------------------------------------------------------------------

# Remove the package entirely so the .usmc transfers again — the DCBA
# fetch only rides along a package download
rm -rf "$CACHE_B" "$ROOT_B_STATE/installed/dcba-e2e-demo-1.0.0" "$INSTALLED_B"
usm_b install dcba-e2e-demo -y >"$WORK/rootB-reinstall.log" 2>&1
REINSTALL_EXIT=$?
report $REINSTALL_EXIT "(e) reinstall after cache removal succeeds"
tr '\r' '\n' < "$WORK/rootB-reinstall.log" | grep -q "Retrieved pre-built artifact"
report $? "(e) reinstall re-downloaded the pre-built artifact"
[ -f "$CACHE_B/build.tar.xz" ]
report $? "(e) build.tar.xz is back in the cache directory"

# ---------------------------------------------------------------------------
# 7. (f) a tampered artifact falls back to a source build
# ---------------------------------------------------------------------------

# Corrupt the published artifact WITHOUT rebuilding the listing, so the
# signed listing still carries the honest checksum and the client must
# reject the download by checksum mismatch — the non-fatal path
printf 'tampered, not an archive' > "$DCBA_REPO/public/dcba-e2e-demo-1.0.0.dcba"
rm -rf "$ROOT_B_STATE/lists/demoart"

rm -rf "$CACHE_B" "$ROOT_B_STATE/installed/dcba-e2e-demo-1.0.0" "$INSTALLED_B"
usm_b install dcba-e2e-demo -y >"$WORK/rootB-tamper.log" 2>&1
TAMPER_EXIT=$?
report $TAMPER_EXIT "(f) install with a tampered artifact still succeeds"

tr '\r' '\n' < "$WORK/rootB-tamper.log" \
    | grep -q 'Warning: DCBA artifact.*could not be used.*building from source instead'
report $? "(f) a warning explains the rejected artifact"

[ -f "$INSTALLED_B" ] && grep -q "$FINGERPRINT_B" "$INSTALLED_B"
report $? "(f) the package was built from source (root B fingerprint present)"

# ---------------------------------------------------------------------------

printf '\n%d passed, %d failed\n' "$PASS_COUNT" "$FAIL_COUNT"
if [ "$FAIL_COUNT" -gt 0 ]; then
    exit 1
fi
exit 0