nextflow.enable.dsl=2

process STAGE_DATA {
    container 'harbor.cluster.omic.ai/omic/digital-patients/synthea:cudf'
    memory '4 GB'
    cpus 2

    script:
    """
    set +e
    BASE="/omic/eureka/digital-patients"

    echo "=== Cleaning up old _parts directories ==="
    find \$BASE -name "*_parts" -type d -exec rm -rf {} + 2>/dev/null
    echo "Cleanup done"

    echo "=== Creating directory structure ==="
    mkdir -p \$BASE/imputed \$BASE/healthy \$BASE/supporting-data/vcf \$BASE/supporting-data/ucsc-liftover \$BASE/supporting-data/genome \$BASE/output
    echo "Dirs created"

    echo "=== Current PVC state ==="
    ls -la \$BASE/ 2>/dev/null || echo "Base dir issue"

    echo "=== Installing sshpass ==="
    conda install -y -c conda-forge sshpass 2>&1 | tail -3
    which sshpass && echo "sshpass OK" || echo "sshpass NOT FOUND"

    download() {
        remote="\$1"
        local_path="\$2"
        if [ -f "\$local_path" ]; then
            echo "SKIP (exists): \$local_path"
            return 0
        fi
        echo "Downloading: \$remote -> \$local_path"
        sshpass -p 'bl3rg3r5' scp -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null -P 9100 "omic@nucleus.omic.ai:\$remote" "\$local_path" 2>&1
        if [ -f "\$local_path" ]; then
            ls -lh "\$local_path"
            return 0
        else
            echo "FAILED: \$local_path"
            return 1
        fi
    }

    echo "=== Downloading imputed files ==="
    download "/mnt/Avatar/imputed/ukbb/imputed/F5_SCHIZO.gwas.imputed_v3.both_sexes.tsv.bgz" "\$BASE/imputed/F5_SCHIZO.gwas.imputed_v3.both_sexes.tsv.bgz"
    download "/mnt/Avatar/imputed/ukbb/imputed/F5_SCHIZO.gwas.imputed_v3.female.tsv.bgz" "\$BASE/imputed/F5_SCHIZO.gwas.imputed_v3.female.tsv.bgz"
    download "/mnt/Avatar/imputed/ukbb/imputed/F5_SCHIZO.gwas.imputed_v3.male.tsv.bgz" "\$BASE/imputed/F5_SCHIZO.gwas.imputed_v3.male.tsv.bgz"

    echo "=== Downloading gnomad files ==="
    download "/mnt/Avatar/digital_patient/gnomad.genomes.v4.1.sites.female.txt" "\$BASE/healthy/gnomad.genomes.v4.1.sites.female.txt"
    download "/mnt/Avatar/digital_patient/gnomad.genomes.v4.1.sites.male.txt" "\$BASE/healthy/gnomad.genomes.v4.1.sites.male.txt"

    echo "=== Downloading supporting-data files ==="
    download "/mnt/Avatar/dd/synthea/supporting-data/genome/hg38.fa" "\$BASE/supporting-data/genome/hg38.fa"
    download "/mnt/Avatar/dd/synthea/supporting-data/genome/hg38.dict" "\$BASE/supporting-data/genome/hg38.dict"
    download "/mnt/Avatar/dd/synthea/supporting-data/ucsc-liftover/hg19ToHg38.over.chain.gz" "\$BASE/supporting-data/ucsc-liftover/hg19ToHg38.over.chain.gz"
    download "/mnt/Avatar/dd/synthea/supporting-data/vcf/vcf_template.vcf" "\$BASE/supporting-data/vcf/vcf_template.vcf"

    echo "=== Downloading phenotype file ==="
    download "/mnt/Avatar/imputed/ukbb/metadata/ukbb_phenotypes.csv" "\$BASE/ukbb_phenotypes_filtered.csv"

    echo "=== Final verification ==="
    echo "Imputed:"
    ls -lh \$BASE/imputed/*.bgz 2>/dev/null || echo "  NONE"
    echo "Healthy:"
    ls -lh \$BASE/healthy/*.txt 2>/dev/null || echo "  NONE"
    echo "Genome:"
    ls -lh \$BASE/supporting-data/genome/* 2>/dev/null || echo "  NONE"
    echo "Support:"
    ls -lh \$BASE/supporting-data/vcf/* \$BASE/supporting-data/ucsc-liftover/* 2>/dev/null || echo "  NONE"
    echo "Small:"
    ls -lh \$BASE/MANE* \$BASE/regulon* \$BASE/LM22* \$BASE/ukbb* 2>/dev/null || echo "  NONE"

    echo "=== STAGING COMPLETE ==="
    """
}

workflow {
    STAGE_DATA()
}