---
- name: Analyze Disk Space and Identify Large Directories
  hosts: all
  gather_facts: true
  vars:
    scan_paths:
      - "/"
      - "/var"
      - "/home"
      - "/opt"
      - "/usr"
      - "/tmp"
    max_depth: 5
    size_threshold_gb: 1
    output_file: "/tmp/disk_space_report_{{ ansible_date_time.iso8601_basic_short }}.json"

  tasks:
    - name: Get overall disk usage
      command: df -h
      register: df_output
      changed_when: false

    - name: Get inode usage
      command: df -i
      register: df_inode_output
      changed_when: false

    - name: Analyze directory sizes
      shell: >-
        du -h -d{{ max_depth }} {{ item }} 2>/dev/null | grep -E '^[0-9]+\.?[0-9]*G' | awk '{print $1 "\t" $2}' | sort -hr
      loop: "{{ scan_paths }}"
      register: dir_sizes
      changed_when: false
      failed_when: false

    - name: Find files larger than threshold
      find:
        paths: "{{ item }}"
        size: "{{ (size_threshold_gb * 1024 * 1024 * 1024) | int }}"
        recurse: true
      loop: "{{ scan_paths }}"
      register: large_files
      failed_when: false

    - name: Generate disk space report
      copy:
        dest: "{{ output_file }}"
        content: >-
          {
            "hostname": "{{ ansible_hostname }}",
            "ip_address": "{{ ansible_default_ipv4.address | default(ansible_ssh_host | default('unknown')) }}",
            "os": "{{ ansible_distribution }} {{ ansible_distribution_version }}",
            "analysis_date": "{{ ansible_date_time.iso8601 }}",
            "disk_usage_output": "{{ df_output.stdout | default('') }}",
            "inode_usage_output": "{{ df_inode_output.stdout | default('') }}",
            "scan_parameters": {
              "paths": {{ scan_paths | to_json }},
              "max_depth": {{ max_depth }},
              "size_threshold_gb": {{ size_threshold_gb }}
            },
            "summary": {
              "scan_paths_count": {{ scan_paths | length }},
              "large_files_count": {{ large_files.results | sum(attribute='matched') | default(0) }}
            }
          }
        mode: '0600'

    - name: Display disk space summary
      debug:
        msg:
          - "Disk space analysis completed on {{ ansible_hostname }}"
          - "Large files found: {{ large_files.results | sum(attribute='matched') | default(0) }}"
          - "Report saved to: {{ output_file }}"

    - name: Return disk space findings
      set_fact:
        disk_space_report:
          hostname: ansible_hostname
          ip_address: ansible_default_ipv4.address | default(ansible_ssh_host | default('unknown'))
          os: ansible_distribution + ' ' + ansible_distribution_version
          analysis_date: ansible_date_time.iso8601
          report_file: output_file