Bootstrap: docker
From: debian:trixie

%labels
    Author ATGC

%post
    export DEBIAN_FRONTEND=noninteractive
    apt-get update && apt-get install -y --no-install-recommends \
        build-essential wget git autoconf automake libtool openmpi-bin libopenmpi-dev \
        bc dos2unix r-base-core ca-certificates unzip
    cd /opt
    wget https://github.com/stephaneguindon/phyml/archive/refs/tags/v3.3.20250515.tar.gz
    tar xvf v3.3.20250515.tar.gz
    cd phyml-3.3.20250515
    sh ./autogen.sh && ./configure --enable-phyml-mpi && make clean && make V=1
    cp src/phyml-mpi /usr/local/bin/
    cd /opt
    git clone --depth 1 https://gite.lirmm.fr/atgc/sms.git
    cd sms
    ZIP_FILE="$(find tarball -maxdepth 1 -name 'sms-*.zip' -print -quit)"
    if [[ -z "${ZIP_FILE}" ]]; then
        echo "Unable to locate SMS tarball in sms/tarball" >&2
        exit 1
    fi
    unzip -q "${ZIP_FILE}"
    ZIP_DIR="$(basename "${ZIP_FILE}" .zip)"
    rm -rf sources/phyml-src
    mv "${ZIP_DIR}/phyml-src" sources/
    rm -rf "${ZIP_DIR}"
    cd sources
    make
    sed -i 's|^PHYMLBIN=.*|PHYMLBIN="/usr/local/bin/phyml-mpi"|' sms.cfg
    chmod +x *.sh
    cat <<'EOF' >/usr/local/bin/phyml-wrapper
#!/bin/bash
set -euo pipefail

SMS_ROOT="/opt/sms/sources"
SMS_GAMMA_CATEGORIES=4
SMS_FREE_RATE_CATEGORIES=4
PHYML_BIN="phyml-mpi"

SMS_MODE=0
SMS_CRITERION=""
declare -a PHYML_ARGS=()

while [[ $# -gt 0 ]]; do
    case "$1" in
        --sms)
            SMS_MODE=1
            shift
            if [[ $# -gt 0 ]]; then
                SMS_CRITERION="$(echo "$1" | tr '[:lower:]' '[:upper:]')"
                shift
            fi
            ;;
        *)
            PHYML_ARGS+=("$1")
            shift
            ;;
    esac
done

if [[ ${SMS_MODE} -eq 0 ]]; then
    exec "${PHYML_BIN}" "${PHYML_ARGS[@]}"
fi

if [[ -z "${SMS_CRITERION}" ]]; then
    SMS_CRITERION="BIC"
fi

INPUT_FILE=""
DATA_TYPE=""
for (( idx=0; idx<${#PHYML_ARGS[@]}; idx++ )); do
    arg="${PHYML_ARGS[$idx]}"
    case "${arg}" in
        -i)
            if (( idx + 1 < ${#PHYML_ARGS[@]} )); then
                INPUT_FILE="${PHYML_ARGS[$((idx + 1))]}"
            fi
            ;;
        -d)
            if (( idx + 1 < ${#PHYML_ARGS[@]} )); then
                DATA_TYPE="${PHYML_ARGS[$((idx + 1))]}"
            fi
            ;;
    esac
done

if [[ -z "${INPUT_FILE}" || -z "${DATA_TYPE}" ]]; then
    echo "SMS wrapper error: missing alignment (-i) or data type (-d)." >&2
    exit 1
fi

SMS_DIR="sms_workspace"
SMS_CSV="${SMS_DIR}/sms_results.csv"

rm -rf "${SMS_DIR}"
mkdir -p "${SMS_DIR}"

echo "[phyml-wrapper] Running SMS (criterion=${SMS_CRITERION})" >&2
if ! "${SMS_ROOT}/sms.sh" -i "${INPUT_FILE}" -d "${DATA_TYPE}" -c "${SMS_CRITERION}" -o "${SMS_DIR}" -p "${SMS_CSV}"; then
    echo "SMS execution failed." >&2
    exit 1
fi

if [[ ! -f "${SMS_CSV}" ]]; then
    echo "SMS results file not found at ${SMS_CSV}" >&2
    exit 1
fi

cp "${SMS_CSV}" "sms_results.csv"
if [[ -f "${SMS_DIR}/sms.log" ]]; then
    cp "${SMS_DIR}/sms.log" "sms.log"
fi

BEST_LINE="$(tail -n +2 "${SMS_CSV}" | head -n 1)"
if [[ -z "${BEST_LINE}" ]]; then
    echo "SMS results are empty." >&2
    exit 1
fi

BEST_MODEL="$(echo "${BEST_LINE}" | cut -d ';' -f 1 | tr -d '[:space:]\r')"
BEST_DECORATION_RAW="$(echo "${BEST_LINE}" | cut -d ';' -f 2 | tr -d '\r')"
BEST_DECORATION="$(echo "${BEST_DECORATION_RAW}" | tr -d ' \"')"

if [[ -z "${BEST_MODEL}" ]]; then
    echo "Unable to parse best model from SMS results." >&2
    exit 1
fi

echo "[phyml-wrapper] SMS selected model: ${BEST_MODEL} ${BEST_DECORATION}" >&2

declare -a SMS_MODEL_ARGS=()
case "${BEST_MODEL}" in
    HKY85|JC69|K80|K2P|F81|F84|TN93|GTR)
        SMS_MODEL_ARGS+=("-m" "${BEST_MODEL}")
        ;;
    SYM)
        SMS_MODEL_ARGS+=("-m" "012345")
        ;;
    *)
        MATRIX_PATH="${SMS_ROOT}/matrix/${BEST_MODEL}.mt"
        if [[ -f "${MATRIX_PATH}" ]]; then
            SMS_MODEL_ARGS+=("-m" "custom" "--aa_rate_file" "${MATRIX_PATH}")
        else
            SMS_MODEL_ARGS+=("-m" "${BEST_MODEL}")
        fi
        ;;
esac

declare -a SMS_FREQ_ARGS=()
if [[ "${DATA_TYPE}" == "nt" ]]; then
    SMS_FREQ_ARGS+=("-f" "e")
else
    if [[ "${BEST_DECORATION}" == *"+F"* ]]; then
        SMS_FREQ_ARGS+=("-f" "e")
    else
        SMS_FREQ_ARGS+=("-f" "m")
    fi
fi

DECO_SANITIZED="${BEST_DECORATION//\'/}"
DECO_SANITIZED="${DECO_SANITIZED// /}"
DECO_SANITIZED="${DECO_SANITIZED//$'\r'/}"
DECO_SANITIZED="${DECO_SANITIZED//$'\n'/}"
DECO_SANITIZED="${DECO_SANITIZED//+F/}"

declare -a SMS_VAR_ARGS=()
case "${DECO_SANITIZED}" in
    "")
        ;;
    "+G")
        SMS_VAR_ARGS+=("-c" "${SMS_GAMMA_CATEGORIES}" "-a" "e")
        ;;
    "+I")
        SMS_VAR_ARGS+=("-v" "e")
        ;;
    "+R")
        SMS_VAR_ARGS+=("--freerates" "-c" "${SMS_FREE_RATE_CATEGORIES}")
        ;;
    "+G+I"|"+I+G")
        SMS_VAR_ARGS+=("-c" "${SMS_GAMMA_CATEGORIES}" "-a" "e" "-v" "e")
        ;;
    "+R+I"|"+I+R")
        SMS_VAR_ARGS+=("--freerates" "-c" "${SMS_FREE_RATE_CATEGORIES}" "-v" "e")
        ;;
    *)
        echo "[phyml-wrapper] Warning: unhandled SMS decoration '${DECO_SANITIZED}', proceeding without extra rate options." >&2
        ;;
esac

FINAL_ARGS=("${PHYML_ARGS[@]}" "${SMS_MODEL_ARGS[@]}" "${SMS_FREQ_ARGS[@]}" "${SMS_VAR_ARGS[@]}")

exec "${PHYML_BIN}" "${FINAL_ARGS[@]}"
EOF
    chmod +x /usr/local/bin/phyml-wrapper
    apt-get clean && rm -rf /var/lib/apt/lists/*

%environment
    export PATH=/usr/local/bin:$PATH

%runscript
    exec /usr/local/bin/phyml-wrapper "$@"