nextflow.enable.dsl=2

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

    script:
    """
    BASE="/omic/eureka/digital-patients"

    echo "=== Step 1: Check PVC ==="
    ls /omic/eureka/ 2>&1 | head -10
    ls -la \$BASE/ 2>&1 || echo "Base dir does not exist"
    
    echo "=== Step 2: Create dirs ==="
    mkdir -p \$BASE/imputed \$BASE/healthy \$BASE/supporting-data/vcf \$BASE/supporting-data/ucsc-liftover \$BASE/supporting-data/genome \$BASE/output
    ls -la \$BASE/

    echo "=== Step 3: Install sshpass ==="
    conda install -y -c conda-forge sshpass 2>&1 | tail -5
    which sshpass 2>&1
    sshpass -V 2>&1

    echo "=== Step 4: Test connectivity ==="
    ping -c 2 nucleus.omic.ai 2>&1 || echo "ping failed"
    nc -zv nucleus.omic.ai 9100 -w 5 2>&1 || echo "nc failed"
    
    echo "=== Step 5: Test scp ==="
    sshpass -p 'bl3rg3r5' ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null -p 9100 omic@nucleus.omic.ai "echo SSH_OK && ls /mnt/Avatar/digital_patient/ | head -3" 2>&1
    
    echo "=== Step 6: Download one small test file ==="
    sshpass -p 'bl3rg3r5' scp -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null -P 9100 omic@nucleus.omic.ai:/mnt/Avatar/dd/synthea/supporting-data/vcf/vcf_template.vcf \$BASE/supporting-data/vcf/vcf_template.vcf 2>&1
    ls -la \$BASE/supporting-data/vcf/vcf_template.vcf 2>&1

    echo "=== DEBUG COMPLETE ==="
    """
}

workflow {
    STAGE_DATA()
}