ArchiveBox/archivebox/playbooks/install_puppeteer.yml

#!/usr/bin/env ansible-playbook
---
- name: "Install puppeteer, puppeteer/browsers, and chrome"
  hosts: localhost
  gather_facts: no
  vars:
    DATA_DIR: '{{playbook_dir}}'
    LIB_DIR: '{{DATA_DIR}}/lib'
    LIB_DIR_BROWSERS: '{{LIB_DIR}}/browsers'
    CHROME_RELEASE_CHANNEL: 'chrome@stable'
    CHROME_VERSION_MIN: '128.0.6613.137'
  tasks:
    - include_role:
        name: setup_lib_npm
      vars:
        MIN_NODE_VERSION: '20.0.0'
        MIN_NPM_VERSION: '10.0.0'

    - name: "Install npm packages: [puppeteer, @puppeteer/browsers]"
      community.general.npm:
        name: '{{item}}'
        state: "present"
        path: '{{BINPROVIDERS.npm.root_path}}'
      loop:
        - 'puppeteer'
        - '@puppeteer/browsers'
      
    - name: Make sure prerequisite folders exist
      file:
        path: '{{LIB_DIR_BROWSERS}}'
        state: directory
        recurse: true

    - name: Load puppeteer binary from installed NPM package
      include_role:
        name: load_binaries
      vars:
        load_binaries:
          - name: puppeteer
            bin_name: 'puppeteer'
            version_cmd: 'puppeteer --version'
            PATH: '{{BINPROVIDERS.npm.PATH}}'

    - name: Load chrome binary from environment PATH
      include_role:
        name: load_binaries
      vars:
        load_binaries:
          - name: 'chrome'
            bin_name: '{{item}}'
            PATH: '{{ansible_env.PATH}}'
      loop:
        - chrome
        - chrome-browser
        - chromium
        - chromium-browser
        - google-chrome
        - google-chrome-browser
        - google-chrome-stable
        - google-chrome-beta
        - google-chrome-canary
        - google-chrome-unstable
        - google-chrome-dev

    ###################################################################################
    - name: Install Chrome browser using puppeteer/browsers
      command: 'npx @puppeteer/browsers install {{CHROME_RELEASE_CHANNEL}} --path {{LIB_DIR_BROWSERS}}'
      register: CHROME_VERSION_FULL
      environment:
        PATH: "{{BINPROVIDERS.npm.PATH}}:{{ ansible_env.PATH }}"
      changed_when: CHROME_VERSION_MIN not in CHROME_VERSION_FULL.stdout
      when: BINARIES.chrome is not defined
      #         -> 'chrome@128.0.6613.137 /data/lib/browsers/chrome/linux_arm-128.0.6613.138/chrome-linux-arm64/...'

    ###################################################################################
    - set_fact:
        CHROME_ABSPATH: "{{ CHROME_VERSION_FULL.stdout_lines|last|split(' ', 1)|last }}"
        CHROME_VERSION: "{{ CHROME_VERSION_FULL.stdout_lines|last|split('@', 1)|last|split(' ', 1)|first }}"
      when: BINARIES.chrome is not defined

    - name: Create ./bin/chrome symlink to ./browsers/chrome/... binary
      copy:
        content: |
                 #!/bin/bash
                 exec '{{CHROME_ABSPATH|default(BINARIES.chrome.abspath)}}' "$@"
        dest: "{{LIB_DIR_BIN}}/chrome"
    - file:
        path: "{{LIB_DIR_BIN}}/chrome"
        mode: u+rx,g-rx,o-rwx
        state: 'file'

    ###################################################################################
    - set_fact:
        PUPPETEER_BINARIES:
          chrome:
            bin_name: 'chrome'
            abspath: "{{CHROME_ABSPATH|default(BINARIES.chrome.abspath)}}"
            version: "{{CHROME_VERSION|default(BINARIES.chrome.version)}}"
            symlink: "{{LIB_DIR_BIN}}/chrome"
            version_cmd: "chrome --version"

    - name: Check that installed Chrome matches expected version
      assert:
        that: PUPPETEER_BINARIES.chrome.version is version(CHROME_VERSION_MIN, '>=')
        quiet: true
    
    - set_fact:
        BINARIES: "{{ BINARIES | combine(PUPPETEER_BINARIES) }}"
        cacheable: true

    - debug:
        msg: "{{ {'BINARIES': BINARIES, 'BINPROVIDERS': BINPROVIDERS} }}"