#!/usr/bin/env nextflow
nextflow.enable.dsl=2

params.base_dir = '/omic/eureka/digital-patients'

process REASSEMBLE {
    container 'harbor.cluster.omic.ai/omic/digital-patients/corto:latest'
    memory '8 GB'

    script:
    """
    BASE="${params.base_dir}"

    echo "=== Listing _parts directories ==="
    find \$BASE -name "*_parts" -type d 2>/dev/null

    echo "=== Reassembling all chunked files ==="
    for parts_dir in \$(find \$BASE -name "*_parts" -type d 2>/dev/null); do
        # Derive the output filename by stripping _parts suffix
        out_file=\$(echo "\$parts_dir" | sed 's/_parts\$//')
        echo "Reassembling: \$parts_dir -> \$out_file"

        # Count parts
        n_parts=\$(ls \$parts_dir/part_* 2>/dev/null | wc -l)
        echo "  Found \$n_parts parts"

        if [ "\$n_parts" -gt 0 ]; then
            cat \$parts_dir/part_* > "\$out_file"
            out_size=\$(du -h "\$out_file" | cut -f1)
            echo "  Created: \$out_file (\$out_size)"
        else
            echo "  WARNING: No parts found in \$parts_dir"
        fi
    done

    echo "=== Verification ==="
    echo "Imputed files:"
    ls -lh \$BASE/imputed/*.bgz 2>/dev/null || echo "  None"
    echo "Healthy files:"
    ls -lh \$BASE/healthy/*.txt 2>/dev/null || echo "  None"
    echo "Genome files:"
    ls -lh \$BASE/supporting-data/genome/hg38.* 2>/dev/null || echo "  None"
    echo "Small files:"
    ls -lh \$BASE/MANE* \$BASE/regulon* \$BASE/LM22* \$BASE/ukbb* 2>/dev/null || echo "  None"

    echo "=== REASSEMBLY COMPLETE ==="
    """
}

workflow {
    REASSEMBLE()
}