usm/spm/pacman/usm-spm-pacman

#!/bin/sh
# USM system-package-manager helper for pacman-based systems (Arch).
#
# Implements the USM SPM contract (see usm/README.md, "System package
# manager integration") for Arch-style systems:
#
#     usm-spm-pacman query <usm-ref>...        -> contract JSON on STDOUT
#     usm-spm-pacman install <native-name>...  -> contract JSONL events on STDOUT
#
# API choice: pure POSIX shell on purpose -- the arch base group ships
# neither python3 nor jq, and the SPM bootstrap must stay at zero extra
# packages. bsdtar (libarchive, a hard pacman dependency) decompresses
# the repository databases under /var/lib/pacman/sync so provides and
# file lists are parsed directly instead of shelling pattern queries
# through pacman:
#
#     lib: X  ->  the sync dbs' %PROVIDES% entries; makepkg auto-
#                 generates soname provides ("libfoo.so=2-64"), so the
#                 "=major-arch" suffix is stripped before matching;
#                 file ownership of /usr/lib/X is the fallback
#     everything file-backed ->
#                 the opt-in files databases (<repo>.files, fetched by
#                 `pacman -Fy` -- the arch analogue of apt-file update)
#                 are parsed into a path -> package map; without them,
#                 only files of INSTALLED packages resolve (via
#                 `pacman -Qo`), the same offline stance as the emerge
#                 helper. No package-name heuristics are used.
#
# dependency-count / installed-dependency-count model "a solo install on
# a from-scratch machine": the closure comes from `pacman -S --print`
# run against a throwaway --dbpath whose local database is empty and
# whose sync databases symlink the system's, so every dependency
# resolves as an install; the overlap with the installed set
# (`pacman -Qq`) is installed-dependency-count. If that run cannot
# resolve, the counts degrade to a (1, n) estimate with a warning.
#
# pacman exposes no per-download progress on a non-tty stdout; it names
# each package once ("(k/N) installing pkg" in older versions,
# "installing pkg..." since pacman 7), so "package" events are emitted
# at package granularity (progress 1.0) and begin.total comes from a
# `pacman -S --print` pass against the current system. Install runs
# with --needed, making repeated installs idempotent.
#
# No INSTALL_COMPANIONS table is needed: arch does not split toolchain
# packages, glibc already ships the crt objects.
#
# The query subcommand never modifies system state: databases are read
# from their normal locations, and the only writes land in a throwaway
# temp directory. The install subcommand must run as root; --noconfirm
# guarantees pacman never prompts interactively: unresolved targets or
# bad signatures fail instead.
#
# Arch path table (merged-usr layout, no multiarch): /usr/bin (sbin is
# a symlink), /usr/lib (lib64 is a symlink), /usr/include, /usr/share
# (+/usr/share/pkgconfig), /usr/share/vala*/vapi.

set -eu

export LC_ALL=C

PROG=usm-spm-pacman

EXIT_OK=0
EXIT_FAILURE=1
EXIT_USAGE=2
EXIT_RESOLVE=3
EXIT_DOWNLOAD=4
EXIT_TRANSACTION=5

WORK=""

SYNC_DIR=/var/lib/pacman/sync

REPO_INDEXED=0

cleanup() {
    if [ -n "$WORK" ]; then
        rm -rf "$WORK"
    fi
}

warn() {
    printf '%s: %s\n' "$PROG" "$*" >&2
}

die_usage() {
    printf 'usage: %s query <usm-ref>...\n' "$PROG" >&2
    printf '       %s plan <native-name>...\n' "$PROG" >&2
    printf '       %s install <native-name>...\n' "$PROG" >&2
    exit "$EXIT_USAGE"
}

emit() {
    printf '%s\n' "$1"
}

json_escape() {
    printf '%s' "$1" | awk '
        {
            s = $0
            gsub(/\\/, "\\\\", s)
            gsub(/"/, "\\\"", s)
            gsub(/\t/, "\\t", s)
            gsub(/[[:cntrl:]]/, "", s)
            print s
        }'
}

mkwork() {
    WORK=$(mktemp -d "${TMPDIR:-/tmp}/usm-spm-pacman.XXXXXX" 2>/dev/null) || {
        WORK="${TMPDIR:-/tmp}/usm-spm-pacman.$$"
        (umask 077 && mkdir "$WORK")
    }
    trap cleanup EXIT
}

not_found_add() {
    printf '%s\n' "$1" >>"$WORK/nf"
}

map_add() {
    printf '%s\t%s\n' "$1" "$2" >>"$WORK/map"
}

# Parse the sync/files databases ONCE per invocation into flat lookup
# tables: $WORK/provides maps "soname-or-virtual<TAB>package" (the
# "=major-arch" suffix is stripped from soname provides so lib: refs
# carrying the full soname match exactly), $WORK/filemap maps
# "path-without-leading-slash<TAB>package" from the %FILES% sections of
# the .files databases. Both stay empty (and the lookups degrade to the
# installed database) when the databases are missing.
build_repo_index() {
    [ "$REPO_INDEXED" = 0 ] || return 0
    REPO_INDEXED=1
    : >"$WORK/provides"
    : >"$WORK/filemap"
    if ! command -v bsdtar >/dev/null 2>&1; then
        warn "bsdtar not found in PATH; repository lookups are unavailable"
        return 0
    fi
    _ri_any=0
    for _ri_db in "$SYNC_DIR"/*.db; do
        [ -f "$_ri_db" ] || continue
        _ri_any=1
        bsdtar -xOf "$_ri_db" 2>>"$WORK/index.err" | awk '
            /^%NAME%$/     { mode = "name"; next }
            /^%PROVIDES%$/ { mode = "prov";  next }
            /^%/           { mode = "";      next }
            NF == 0        { next }
            mode == "name" { pkg = $1; mode = ""; next }
            mode == "prov" && pkg != "" {
                p = $1
                sub(/=.*$/, "", p)
                print pkg "\t" p
            }
        ' >>"$WORK/provides" || true
    done
    for _ri_db in "$SYNC_DIR"/*.files; do
        [ -f "$_ri_db" ] || continue
        _ri_any=1
        bsdtar -xOf "$_ri_db" 2>>"$WORK/index.err" | awk '
            /^%NAME%$/      { mode = "name";  next }
            /^%FILES%$/     { mode = "files"; next }
            /^%/            { mode = "";      next }
            NF == 0         { next }
            mode == "name"  { pkg = $1; mode = ""; next }
            mode == "files" && pkg != "" { print pkg "\t" $1 }
        ' >>"$WORK/filemap" || true
    done
    if [ "$_ri_any" != 1 ]; then
        warn "no databases under $SYNC_DIR (run pacman -Sy as root); repository lookups are unavailable"
    elif [ ! -s "$WORK/filemap" ]; then
        warn "no files databases under $SYNC_DIR (run pacman -Fy as root); only installed files resolve"
    fi
}

# makepkg turns soname "libc.so.6" into the provide "libc.so=6-64", so a
# lib: ref's trailing ".<major>" is optionally stripped before matching:
# "lib:libc.so.6" matches "libc.so" exactly as an unversioned
# "lib:libc.so" would.
repo_provider_owners() {
    awk -F'\t' -v p="$1" '
        {
            r = p
            sub(/\.[0-9]+$/, "", r)
            if ($2 == p || ($2 == r && r != p)) { print $1 }
        }' "$WORK/provides" 2>/dev/null
}

repo_file_owners() {
    awk -F'\t' -v p="$1" '$2 == p { print $1 }' "$WORK/filemap" 2>/dev/null
}

repo_file_prefix_owners() {
    # exact file OR any file below the directory prefix
    awk -F'\t' -v p="$1" '$2 == p || index($2, p "/") == 1 { print $1 }' \
        "$WORK/filemap" 2>/dev/null
}

# Map every owner name read from STDIN onto _om_ref; succeeds when at
# least one name arrived.
owners_to_map() {
    _om_ref=$1
    _om_any=0
    while IFS= read -r _om_nm; do
        [ -n "$_om_nm" ] || continue
        map_add "$_om_nm" "$_om_ref"
        _om_any=1
    done
    [ "$_om_any" = 1 ]
}

pacman_diag_line() {
    [ -r "$1" ] || return 0
    sed -n 's/^[[:space:]]*error:[[:space:]]*//p' "$1" | head -n 1
}

record_providers() {
    _rp_ref=$1
    _rp_prov=$2
    build_repo_index
    if repo_provider_owners "$_rp_prov" | sort -u | owners_to_map "$_rp_ref"; then
        return 0
    fi
    not_found_add "$_rp_ref"
}

record_file_owner() {
    _fo_ref=$1
    shift
    build_repo_index
    _fo_hit=0
    for _fo_p in "$@"; do
        _fo_names=
        if [ -s "$WORK/filemap" ]; then
            _fo_names=$(repo_file_owners "${_fo_p#/}" | sort -u)
        fi
        if [ -z "$_fo_names" ]; then
            # installed-database fallback, exact for present files only
            _fo_f=
            if [ -d "$_fo_p" ]; then
                _fo_f=$(find "$_fo_p" -type f 2>/dev/null | head -n 1) || _fo_f=
            elif [ -f "$_fo_p" ]; then
                _fo_f=$_fo_p
            fi
            if [ -n "$_fo_f" ]; then
                _fo_names=$(pacman -Qo -- "$_fo_f" 2>/dev/null \
                    | awk '{ print $(NF-1) }') || _fo_names=
            fi
        fi
        if [ -n "$_fo_names" ]; then
            printf '%s\n' "$_fo_names" | owners_to_map "$_fo_ref" && _fo_hit=1
        fi
    done
    if [ "$_fo_hit" != 1 ]; then
        not_found_add "$_fo_ref"
    fi
}

record_dir_owner() {
    # Like record_file_owner, but the ref names a directory (inc:), so
    # any file at or below the prefix satisfies it.
    _do_ref=$1
    _do_p=$2
    build_repo_index
    if [ -s "$WORK/filemap" ]; then
        if repo_file_prefix_owners "${_do_p#/}" | sort -u | owners_to_map "$_do_ref"; then
            return 0
        fi
    fi
    record_file_owner "$_do_ref" "$_do_p"
}

handle_ref() {
    _hr_ref=$1
    _hr_type=${_hr_ref%%:*}
    _hr_res=${_hr_ref#*:}
    if [ "$_hr_type" = "$_hr_ref" ] || [ -z "$_hr_res" ]; then
        warn "resource type of \"$_hr_ref\" has no system-package-manager translation"
        not_found_add "$_hr_ref"
        return
    fi
    case $_hr_res in
        *[!A-Za-z0-9._+/-]*)
            warn "resource name of \"$_hr_ref\" has no system-package-manager translation"
            not_found_add "$_hr_ref"
            return
            ;;
    esac
    case $_hr_type in
        bin)
            record_file_owner "$_hr_ref" "/usr/bin/$_hr_res"
            ;;
        sbin)
            record_file_owner "$_hr_ref" "/usr/bin/$_hr_res" "/usr/sbin/$_hr_res"
            ;;
        lib|libres)
            # soname provides first (exact, suffix-stripped), file second
            build_repo_index
            if ! repo_provider_owners "$_hr_res" | sort -u | owners_to_map "$_hr_ref"; then
                record_file_owner "$_hr_ref" "/usr/lib/$_hr_res"
            fi
            ;;
        pc)
            record_file_owner "$_hr_ref" \
                "/usr/lib/pkgconfig/$_hr_res" "/usr/share/pkgconfig/$_hr_res"
            ;;
        libexec)
            record_file_owner "$_hr_ref" "/usr/libexec/$_hr_res"
            ;;
        gir)
            record_file_owner "$_hr_ref" "/usr/share/gir-1.0/$_hr_res"
            ;;
        typelib)
            record_file_owner "$_hr_ref" "/usr/lib/girepository-1.0/$_hr_res"
            ;;
        gio)
            record_file_owner "$_hr_ref" "/usr/lib/gio/modules/$_hr_res"
            ;;
        res)
            record_file_owner "$_hr_ref" "/usr/share/$_hr_res"
            ;;
        cfg)
            record_file_owner "$_hr_ref" "/etc/$_hr_res"
            ;;
        man)
            record_file_owner "$_hr_ref" "/usr/share/man/$_hr_res"
            ;;
        info)
            record_file_owner "$_hr_ref" "/usr/share/info/$_hr_res"
            ;;
        locale)
            record_file_owner "$_hr_ref" "/usr/share/locale/$_hr_res"
            ;;
        inc)
            record_dir_owner "$_hr_ref" "/usr/include/$_hr_res"
            ;;
        vapi)
            build_repo_index
            if [ -s "$WORK/filemap" ]; then
                : >"$WORK/vown"
                repo_file_owners "usr/share/vala/vapi/$_hr_res" >>"$WORK/vown"
                awk -F'\t' -v suf="/vapi/$_hr_res" \
                    'substr($2, 1, 15) == "usr/share/vala" &&
                     substr($2, length($2) - length(suf) + 1) == suf { print $1 }' \
                    "$WORK/filemap" >>"$WORK/vown"
                if sort -u "$WORK/vown" | owners_to_map "$_hr_ref"; then
                    return 0
                fi
            fi
            set -- "/usr/share/vala/vapi/$_hr_res"
            for _hr_d in /usr/share/vala-*/vapi/"$_hr_res"; do
                if [ -f "$_hr_d" ]; then
                    set -- "$@" "$_hr_d"
                fi
            done
            record_file_owner "$_hr_ref" "$@"
            ;;
        rootpath)
            record_file_owner "$_hr_ref" "/$_hr_res"
            ;;
        tag)
            _hr_t=${_hr_res%.tag}
            record_file_owner "$_hr_ref" \
                "/usr/share/usm-tags/$(printf '%s' "$_hr_t" | tr '.' '/').tag"
            ;;
        *)
            warn "resource type of \"$_hr_ref\" has no system-package-manager translation"
            not_found_add "$_hr_ref"
            return
            ;;
    esac
}

counts_for() {
    _cf_name=$1
    _cf_cached=
    if [ -s "$WORK/counts" ]; then
        _cf_cached=$(awk -F'\t' -v n="$_cf_name" '$1 == n { print $2 " " $3; exit }' "$WORK/counts") || _cf_cached=
    fi
    if [ -n "$_cf_cached" ]; then
        printf '%s\n' "$_cf_cached"
        return
    fi
    # Repo-only closure: a throwaway database whose empty local db makes
    # every package resolve as an install and whose sync symlink reuses
    # the system's repositories.
    _cf_root="$WORK/rootdb"
    if [ ! -d "$_cf_root" ]; then
        mkdir -p "$_cf_root/local"
        ln -s "$SYNC_DIR" "$_cf_root/sync" 2>/dev/null || true
    fi
    _cf_dep=0
    _cf_inst=0
    _cf_cl=$(pacman --dbpath "$_cf_root" -S --print --print-format '%n' \
        --noconfirm -- "$_cf_name" 2>"$WORK/cnt.err" | grep -E '^[A-Za-z0-9@._+-]+$' || true)
    if [ -n "$_cf_cl" ]; then
        printf '%s\n' "$_cf_cl" >"$WORK/cl"
        _cf_dep=$(awk 'END { print NR }' "$WORK/cl")
        if [ -s "$WORK/installed" ]; then
            _cf_inst=$(grep -x -F -f "$WORK/installed" "$WORK/cl" | awk 'END { print NR }') || _cf_inst=0
        fi
    else
        warn "could not resolve solo install of $_cf_name, estimating dependency counts"
        if grep -qx -F "$_cf_name" "$WORK/installed" 2>/dev/null; then
            _cf_dep=1
            _cf_inst=1
        else
            _cf_dep=1
            _cf_inst=0
        fi
    fi
    printf '%s\t%s\t%s\n' "$_cf_name" "$_cf_dep" "$_cf_inst" >>"$WORK/counts"
    printf '%s %s\n' "$_cf_dep" "$_cf_inst"
}

cmd_query() {
    [ $# -ge 1 ] || die_usage
    mkwork
    pacman -Qq >"$WORK/installed" 2>/dev/null || : >"$WORK/installed"
    : >"$WORK/map"
    : >"$WORK/nf"
    : >"$WORK/counts"
    : >"$WORK/seen"
    for _q_ref in "$@"; do
        if [ -n "$_q_ref" ] && grep -x -F -q "$_q_ref" "$WORK/seen" 2>/dev/null; then
            continue
        fi
        printf '%s\n' "$_q_ref" >>"$WORK/seen"
        handle_ref "$_q_ref"
    done
    _q_out='{"not-found":['
    _q_first=1
    while IFS= read -r _q_r; do
        [ -n "$_q_r" ] || continue
        if [ "$_q_first" = 1 ]; then _q_first=0; else _q_out="$_q_out,"; fi
        _q_out="$_q_out\"$(json_escape "$_q_r")\""
    done <"$WORK/nf"
    _q_out="$_q_out],\"packages\":["
    _q_first=1
    for _q_nm in $(awk -F'\t' '{ print $1 }' "$WORK/map" | sort -u); do
        if [ "$_q_first" = 1 ]; then _q_first=0; else _q_out="$_q_out,"; fi
        awk -F'\t' -v n="$_q_nm" '$1 == n { print $2 }' "$WORK/map" >"$WORK/tmpres"
        _q_res=""
        _q_rf=1
        while IFS= read -r _q_rr; do
            if [ "$_q_rf" = 1 ]; then _q_rf=0; else _q_res="$_q_res,"; fi
            _q_res="$_q_res\"$(json_escape "$_q_rr")\""
        done <"$WORK/tmpres"
        _q_cnts=$(counts_for "$_q_nm")
        _q_dep=${_q_cnts%% *}
        _q_inst=${_q_cnts##* }
        _q_out="$_q_out{\"name\":\"$(json_escape "$_q_nm")\",\"resources\":[${_q_res}],\"dependency-count\":$_q_dep,\"installed-dependency-count\":$_q_inst}"
    done
    _q_out="$_q_out]}"
    emit "$_q_out"
    exit "$EXIT_OK"
}

classify_failure() {
    if grep -Eqi 'target not found|could not satisfy dependencies|conflicting|unresolvable|failed to prepare transaction|hold onto' "$1" 2>/dev/null; then
        printf '%s\n' "$EXIT_RESOLVE"
    elif grep -Eqi 'network|download|retriev|signature|keyring|gpg|invalid or corrupted|checksum|integrity|connection|timed out|mirror' "$1" 2>/dev/null; then
        printf '%s\n' "$EXIT_DOWNLOAD"
    else
        printf '%s\n' "$EXIT_TRANSACTION"
    fi
}

cmd_install() {
    [ $# -ge 1 ] || die_usage
    for _i_n in "$@"; do
        case $_i_n in
            -*) die_usage ;;
        esac
    done
    mkwork
    _i_simrc=0
    _i_sim=$(pacman -S --needed --noconfirm --print --print-format '%n' \
        -- "$@" 2>"$WORK/sim.err") || _i_simrc=$?
    if [ "$_i_simrc" != 0 ]; then
        _i_msg=$(pacman_diag_line "$WORK/sim.err")
        [ -n "$_i_msg" ] || _i_msg="pacman --print failed with status $_i_simrc"
        emit "{\"type\":\"error\",\"message\":\"$(json_escape "$_i_msg")\"}"
        exit "$(classify_failure "$WORK/sim.err")"
    fi
    _i_total=$(printf '%s\n' "$_i_sim" | grep -Ec '^[A-Za-z0-9@._+-]+$') || _i_total=0
    emit "{\"type\":\"begin\",\"total\":$_i_total}"
    {
        # || keeps set -e from aborting the pipeline group before the
        # status reaches the rc file
        _i_status=0
        pacman -S --needed --noconfirm --noprogressbar -- "$@" \
            2>"$WORK/inst.err" || _i_status=$?
        printf '%s\n' "$_i_status" >"$WORK/rc"
    } | tr '\r' '\n' | {
        _i_done=0
        _i_lastpkg=
        while IFS= read -r _i_line || [ -n "$_i_line" ]; do
            # pacman's transaction output names each package exactly
            # once: older versions redraw "(k/N) installing pkg" lines
            # (\r-flattened here), pacman 7 pipes plain "installing
            # pkg..." lines. Dedupe on the package name either way.
            case $_i_line in
                \(*\)\ *)
                    _i_kn=${_i_line#"("}
                    _i_kn=${_i_kn%%")"*}
                    _i_rest=${_i_line#*") "}
                    _i_verb=${_i_rest%% *}
                    _i_pkg=${_i_rest#* }
                    _i_pkg=${_i_pkg%% *}
                    _i_k=${_i_kn%%/*}
                    _i_n=${_i_kn##*/}
                    ;;
                installing\ *|upgrading\ *|downgrading\ *|reinstalling\ *)
                    _i_verb=${_i_line%% *}
                    _i_pkg=${_i_line#* }
                    _i_pkg=${_i_pkg%% *}
                    _i_pkg=${_i_pkg%"..."}
                    _i_k=$((_i_done + 1))
                    _i_n=$_i_total
                    ;;
                *)
                    continue
                    ;;
            esac
            case $_i_verb in
                installing|upgrading|downgrading|reinstalling)
                    if [ "$_i_pkg" != "$_i_lastpkg" ]; then
                        _i_lastpkg=$_i_pkg
                        emit "{\"type\":\"package\",\"name\":\"$(json_escape "$_i_pkg")\",\"current\":$_i_k,\"total\":$_i_n,\"progress\":1.0}"
                        emit "{\"type\":\"package-complete\",\"name\":\"$(json_escape "$_i_pkg")\"}"
                        _i_done=$((_i_done + 1))
                    fi
                    ;;
            esac
        done
        printf '%s\n' "$_i_done" >"$WORK/done"
    }
    _i_rc=$(cat "$WORK/rc" 2>/dev/null) || _i_rc=1
    if [ "$_i_rc" = 0 ]; then
        _i_done=$(cat "$WORK/done" 2>/dev/null) || _i_done=0
        emit "{\"type\":\"complete\",\"status\":\"ok\",\"installed\":$_i_done}"
        exit "$EXIT_OK"
    fi
    _i_msg=$(pacman_diag_line "$WORK/inst.err")
    [ -n "$_i_msg" ] || _i_msg="pacman -S failed with status $_i_rc"
    emit "{\"type\":\"error\",\"message\":\"$(json_escape "$_i_msg")\"}"
    exit "$(classify_failure "$WORK/inst.err")"
}

cmd_plan() {
    [ $# -ge 1 ] || die_usage
    for _p_n in "$@"; do
        case $_p_n in
            -*) die_usage ;;
        esac
    done
    mkwork
    _p_rc=0
    _p_names=$(pacman -S --needed --noconfirm --print --print-format '%n' \
        -- "$@" 2>"$WORK/plan.err") || _p_rc=$?
    if [ "$_p_rc" != 0 ]; then
        _p_msg=$(pacman_diag_line "$WORK/plan.err")
        [ -n "$_p_msg" ] || _p_msg="pacman --print failed with status $_p_rc"
        printf '%s: %s\n' "$PROG" "$_p_msg" >&2
        exit "$(classify_failure "$WORK/plan.err")"
    fi
    _p_json=$(printf '%s\n' "$_p_names" \
        | sed -n '/^[A-Za-z0-9@._+-]\{1,\}$/s/.*/"&"/p' \
        | awk 'NR>1{printf ","} {printf "%s", $0}')
    printf '{"packages":[%s]}\n' "$_p_json"
    exit "$EXIT_OK"
}

main() {
    if [ $# -lt 1 ]; then
        die_usage
    fi
    if ! command -v pacman >/dev/null 2>&1; then
        if [ "$1" = install ]; then
            emit '{"type":"error","message":"pacman not found in PATH"}'
            exit "$EXIT_FAILURE"
        fi
        warn "pacman not found in PATH"
        exit "$EXIT_RESOLVE"
    fi
    _m_cmd=$1
    shift
    case $_m_cmd in
        query)
            cmd_query "$@"
            ;;
        plan)
            cmd_plan "$@"
            ;;
        install)
            cmd_install "$@"
            ;;
        -h|--help|help)
            printf 'usage: %s query <usm-ref>...\n' "$PROG"
            printf '       %s plan <native-name>...\n' "$PROG"
            printf '       %s install <native-name>...\n' "$PROG"
            exit "$EXIT_OK"
            ;;
        *)
            die_usage
            ;;
    esac
}

main "$@"