list-dereko-korap-release.yml

- name: List DeReKo-KorAP Releases
  hosts: worker-03
  run_once: true
  tasks:
    - name: Get current year
      set_fact:
        current_year: "{{ ansible_date_time.year }}"

    - name: Receive new relevant list from database
      community.mysql.mysql_query:
        login_host: klinux10
        login_db: corpora
        login_user: viewer
        query: >
          SELECT concat('/vol/corpora/DeReKo/incoming/KorAP/zip/', replace(name,'.i5.xml','.zip')) as cname
          FROM fileMeta{{ current_year }}I, basename
          WHERE (fileMeta{{ current_year }}I.name LIKE 'pp-%' or fileMeta{{ current_year }}I.rsr = 1 or
                fileMeta{{ current_year }}I.main = 1 or
                (fileMeta{{ current_year }}I.name like '%2_.i5.xml' and (basename.rsr=1 or basename.inkorap=1))) and
                fileMeta{{ current_year }}I.base=basename.id
          ORDER by name
      register: sqlresult

    - name: Find all changed from the last 200 days
      ansible.builtin.find:
        paths:
          - "/vol/corpora/DeReKo/incoming/KorAP/zip/"
        age: "-200d"
        age_stamp: "mtime"
        use_regex: true
        follow: true
        file_type: "any"
        patterns:
          - "^[^\\.]*\\.zip$"
      register: newfiles

    - name: Compare both lists and show intersection
      ansible.legacy.copy:
        content: "{{ newfiles.files | map(attribute='path') | sort() | intersect( sqlresult.query_result[0] | map(attribute='cname') ) | join('\n') }}"
        dest: dereko-ingestion.txt
      delegate_to: localhost