fix(NLTK): download NLTK data if NLTK is enabled

molecule/alternative_installation/converge.yml

@@ -18,3 +18,4 @@
  paperless_ngx_conf_port: 8001
  paperless_ngx_system_user_additional_groups:
  - root
+ paperless_ngx_conf_enable_nltk: 1

molecule/alternative_installation/roles/test_alternative_installation/tasks/main.yml

@@ -1,4 +1,8 @@
 ---
+- name: Test NTLK
+ ansible.builtin.include_tasks:
+ file: test_nltk.yml
+
 - name: Check the activated settings
  block:
  - name: Read the active settings

molecule/alternative_installation/roles/test_alternative_installation/tasks/test_nltk.yml

@@ -0,0 +1,9 @@
+---
+- name: Check that NLTK installation was successful by executing the document classifier
+ become: true
+ become_user: "{{ paperless_ngx_system_user }}"
+ ansible.builtin.shell: |
+ {{ paperless_ngx_dir_virtualenv }}/bin/python3 {{ paperless_ngx_dir_installation }}/src/manage.py document_create_classifier
+ register: _classifier_output
+ changed_when: false
+ failed_when: '"Classifier error" in _classifier_output.stderr'

tasks/paperless_ngx/main.yml

@@ -34,6 +34,13 @@
  tags: venv
  tags: venv
 
+ - name: Prepare NLTK environment
+ ansible.builtin.include_tasks:
+ file: nltk.yml
+ apply:
+ tags: nltk
+ tags: nltk
+
 - name: Configure paperless-ngx and dependencies
  ansible.builtin.include_tasks:
  file: configuration.yml

tasks/paperless_ngx/nltk.yml

@@ -0,0 +1,21 @@
+---
+- name: Create NLTK dir
+ become: true
+ ansible.builtin.file:
+ path: "{{ paperless_ngx_conf_nltk_dir }}"
+ state: directory
+ owner: "{{ paperless_ngx_system_user }}"
+ group: "{{ paperless_ngx_system_group }}"
+ mode: "750"
+
+- name: Download NLTK data to NLTK dir
+ become: true
+ become_user: "{{ paperless_ngx_system_user }}"
+ ansible.builtin.shell: |
+ {{ paperless_ngx_dir_virtualenv }}/bin/python3 -W ignore::RuntimeWarning -m nltk.downloader -d "{{ paperless_ngx_conf_nltk_dir }}" {{ item }}
+ register: _nltk_download
+ changed_when: '"already up-to-date" not in _nltk_download.stderr'
+ loop:
+ - snowball_data
+ - stopwords
+ - punkt