osmedeus/test/testdata/workflows/test-complex-docker-workflow.yaml

name: test-complex-docker-workflow
kind: module
description: Complex workflow demonstrating bash, function steps with docker step_runner
tags: test,docker,comprehensive

params:
  - name: target
    required: true
  - name: output_dir
    default: /tmp/osm-complex-test
  - name: threads
    default: "5"

steps:
  # Step 1: Setup - Create directories using function
  - name: setup-workspace
    type: function
    log: "Setting up workspace for {{target}}"
    function: createDir("{{output_dir}}")
    exports:
      workspace_created: "output"

  # Step 2: Create input file with bash
  - name: create-target-list
    type: bash
    log: "Creating target list for {{target}}"
    commands:
      - mkdir -p {{output_dir}}/targets
      - |
        cat > {{output_dir}}/targets/hosts.txt << 'EOF'
        sub1.{{target}}
        sub2.{{target}}
        api.{{target}}
        www.{{target}}
        admin.{{target}}
        EOF
    exports:
      target_file: "{{output_dir}}/targets/hosts.txt"

  # Step 3: Docker-based DNS resolution simulation
  - name: dns-resolve
    type: remote-bash
    log: "Resolving DNS for targets in Docker"
    timeout: 60
    step_runner: docker
    step_runner_config:
      image: alpine:latest
      env:
        TARGET_DOMAIN: "{{target}}"
      volumes:
        - "{{output_dir}}:/workspace"
      workdir: /workspace
    command: |
      echo "Resolving DNS for $TARGET_DOMAIN"
      cat /workspace/targets/hosts.txt | while read host; do
        echo "$host -> 127.0.0.1" >> /workspace/dns-resolved.txt
      done
      echo "DNS resolution complete"
    exports:
      dns_output: "{{output_dir}}/dns-resolved.txt"

  # Step 4: Parallel docker commands - simulating port scanning
  - name: parallel-port-scan
    type: remote-bash
    log: "Running parallel port scans in Docker"
    timeout: 120
    step_runner: docker
    step_runner_config:
      image: alpine:latest
      volumes:
        - "{{output_dir}}:/workspace"
    parallel_commands:
      - 'echo "Scanning ports 1-1000 on {{target}}" && sleep 1 && echo "Port 80 open" > /workspace/ports-1.txt'
      - 'echo "Scanning ports 1001-2000 on {{target}}" && sleep 1 && echo "Port 443 open" > /workspace/ports-2.txt'
      - 'echo "Scanning ports 2001-3000 on {{target}}" && sleep 1 && echo "Port 8080 open" > /workspace/ports-3.txt'
      - 'echo "Scanning ports 3001-4000 on {{target}}" && sleep 1 && echo "Port 3306 open" > /workspace/ports-4.txt'

  # Step 5: Merge port scan results
  - name: merge-port-results
    type: bash
    log: "Merging port scan results"
    command: cat {{output_dir}}/ports-*.txt > {{output_dir}}/all-ports.txt
    exports:
      ports_file: "{{output_dir}}/all-ports.txt"

  # Step 6: Function to check file existence
  - name: verify-ports-file
    type: function
    log: "Verifying ports file exists"
    function: file_exists("{{ports_file}}")
    exports:
      ports_verified: "output"

  # Step 7: Docker-based HTTP probing with parallel steps
  - name: http-probe-parallel
    type: parallel-steps
    log: "Running parallel HTTP probes"
    parallel_steps:
      - name: probe-http
        type: remote-bash
        step_runner: docker
        step_runner_config:
          image: alpine:latest
          volumes:
            - "{{output_dir}}:/workspace"
        command: |
          echo "Probing HTTP on port 80"
          echo "http://{{target}}:80 [200]" > /workspace/http-80.txt
      - name: probe-https
        type: remote-bash
        step_runner: docker
        step_runner_config:
          image: alpine:latest
          volumes:
            - "{{output_dir}}:/workspace"
        command: |
          echo "Probing HTTPS on port 443"
          echo "https://{{target}}:443 [200]" > /workspace/https-443.txt
      - name: probe-alt
        type: remote-bash
        step_runner: docker
        step_runner_config:
          image: alpine:latest
          volumes:
            - "{{output_dir}}:/workspace"
        command: |
          echo "Probing alternate port 8080"
          echo "http://{{target}}:8080 [404]" > /workspace/http-8080.txt

  # Step 8: Foreach loop with docker - process each subdomain
  - name: process-subdomains
    type: foreach
    log: "Processing each subdomain"
    input: "{{output_dir}}/targets/hosts.txt"
    variable: subdomain
    threads: 3
    step:
      name: scan-subdomain
      type: remote-bash
      step_runner: docker
      step_runner_config:
        image: alpine:latest
        volumes:
          - "{{output_dir}}:/workspace"
      command: |
        echo "Scanning [[subdomain]]..."
        echo "[[subdomain]]: status=200, title=Example" >> /workspace/subdomain-results.txt

  # Step 9: Read results with function
  - name: read-subdomain-results
    type: function
    log: "Reading subdomain scan results"
    function: read_file("{{output_dir}}/subdomain-results.txt")
    exports:
      scan_results: "output"

  # Step 10: Decision based routing
  - name: check-results
    type: bash
    log: "Checking scan results"
    command: wc -l < {{output_dir}}/subdomain-results.txt
    exports:
      result_count: "output"
    decision:
      switch: "{{result_count}}"
      cases:
        "0":
          goto: _end
      default:
        goto: generate-report

  # Step 11: Generate final report in docker
  - name: generate-report
    type: remote-bash
    log: "Generating final report"
    timeout: 30
    step_runner: docker
    step_runner_config:
      image: alpine:latest
      volumes:
        - "{{output_dir}}:/workspace"
    commands:
      - echo "=== Scan Report for {{target}} ===" > /workspace/report.txt
      - echo "" >> /workspace/report.txt
      - echo "--- DNS Results ---" >> /workspace/report.txt
      - cat /workspace/dns-resolved.txt >> /workspace/report.txt 2>/dev/null || echo "No DNS results" >> /workspace/report.txt
      - echo "" >> /workspace/report.txt
      - echo "--- Open Ports ---" >> /workspace/report.txt
      - cat /workspace/all-ports.txt >> /workspace/report.txt 2>/dev/null || echo "No ports found" >> /workspace/report.txt
      - echo "" >> /workspace/report.txt
      - echo "--- Subdomain Results ---" >> /workspace/report.txt
      - cat /workspace/subdomain-results.txt >> /workspace/report.txt 2>/dev/null || echo "No subdomain results" >> /workspace/report.txt
      - echo "" >> /workspace/report.txt
      - echo "Report generated at $(date)" >> /workspace/report.txt
    exports:
      report_file: "{{output_dir}}/report.txt"

  # Step 12: Parallel functions to get file stats
  - name: get-file-stats
    type: function
    log: "Getting file statistics"
    parallel_functions:
      - file_length("{{output_dir}}/report.txt")
      - file_exists("{{output_dir}}/all-ports.txt")
      - trim("  {{target}}  ")
    exports:
      file_stats: "output"

  # Step 13: Cleanup (optional - controlled by pre_condition)
  - name: cleanup-temp-files
    type: bash
    log: "Cleaning up temporary files"
    pre_condition: "false"
    command: rm -rf {{output_dir}}/ports-*.txt
    on_error:
      - action: log
        message: "Cleanup failed but continuing"
      - action: continue