playbook-ubuntu-setup.yml

---
# Define variables for this playbook:
- hosts: all
  tasks:
    - set_fact:
        crawler_dir: iframe-crawler


- name: Ensure ansible's 'apt' module can be used with this host
  hosts: all
  become: yes
  tasks:
    - name: Install ansible's apt module's remote prerequisites
      # Unfornutanely, using 'command' means that these steps will always have
      # the state 'changed' regardless of whether any changes were made.
      # Maybe install python-apt too?
      command: "{{ item }}"
      loop:
        # DigitalOcean's apt needs an update first to find packages (aptitude)
        - apt update
        - apt upgrade
        - apt install -y python3-apt aptitude


- name: Ensure unzip in installed for the 'unarchive' module
  hosts: all
  become: yes
  gather_facts: no
  tasks:
    - name: Install unzip
      apt:
        name: unzip
        state: latest


- name: Ensure we have the crawler repository
  hosts: all
  gather_facts: no
  tasks:
    - name: Clone/update the crawler repository
      git:
        repo: https://github.com/jwatt/iframe-crawler.git
        dest: iframe-crawler


# I've used some JavaScript language features that don't work in Node.js v10,
# such as `for await`.  Ubuntu 19.10 Server has Node.js v10 (originally
# released on 2018-04-24) installed, and Ubuntu 18.04 only has Node.js v8 (and
# it's not installed by default), so we download our own more up to date Node.js
# here.
#
# TODO Consiber using `vars_prompt` to ask whether we can install a newer
# version system wide.
# https://stackoverflow.com/questions/25466675/ansible-to-conditionally-prompt-for-a-variable
#
# nodejs.org docs say to use the NodeSource distribution on Debian/Ubuntu if
# installing from a package manager:
# https://nodejs.org/en/download/package-manager/#debian-and-ubuntu-based-linux-distributions-enterprise-linux-fedora-and-snap-packages
#
- name: Install a local up to date version of Node.js
  hosts: all
  gather_facts: no
  tasks:
    - name: Get version number of Node.js, if installed
      ignore_errors: yes
      command: node --version
      register: installed_nodejs_info
      # To get the major version as an integer, we could use:
      # int(installed_nodejs_info.stdout[1:].split(".")[0])

    #- name: Skip the rest of this play if Node.js is already installed
    #  when: not installed_nodejs_info.failed
    #  meta: end_host

    - name: Get the version number of the latest Node.js LTS
      uri:
        url: "https://api.github.com/repos/nodejs/node/releases/latest"
        method: GET
        return_content: yes
        status_code: 200
        headers:
          Content-Type: "application/json"
        body_format: json
      register: latest_nodejs_info

    - name: Download latest Node.js
      shell: "curl -L https://nodejs.org/dist/latest/node-{{ node_version_str }}-linux-x64.tar.gz | tar xz --transform 's/node-{{ node_version_str }}-linux-x64/node/'"
      args:
        chdir: "{{ crawler_dir }}"
        creates: "{{ crawler_dir }}/node/bin/node"
        warn: false
      vars:
        node_version_str: "{{ latest_nodejs_info.json.tag_name }}"
      # Once https://github.com/ansible/ansible/issues/64612 is fixed, we could
      # use the 'unarchive' module, but the below doesn't yet use
      # --strip-components=1 etc. like the 'shell' line above.
      #unarchive:
      #  src: "https://nodejs.org/dist/latest/node-{{ node_version_str }}-linux-x64.tar.gz"
      #  remote_src: yes
      #  dest: "{{ crawler_dir }}"
      #  creates: "{{ crawler_dir }}/node-{{ node_version_str }}-linux-x64"
      #vars:
      #  node_version_str: "{{ latest_nodejs_info.json.tag_name }}"

    - name: Create a symbolic link for nodejs
      file:
        src: "node"
        dest: "{{ crawler_dir }}/node/bin/nodejs"
        state: link
        mode: "744"

    # If installing using the apt package manager, we'd do something like the
    # following.
    #
    #- name: Download NodeSource install script
    #  get_url:
    #    url: https://deb.nodesource.com/setup_{{ latest_node_lts_major_version_num }}.x
    #    dest: "{{ crawler_dir }}/nodesource-setup-script.bash"
    #    mode: "+x"
    #  vars:
    #    node_version_str: "{{ latest_nodejs_info.json.tag_name }}"
    #    # LTS versions have even version numbers
    #    latest_node_lts_major_version_num: "{{ round(int(latest_nodejs_info.json.tag_name[1:].split('.')[0]) / 2) * 2 }}"
    #
    #- name: Run NodeSource install script
    #  become: yes
    #  # Using recursive sudo here in addition to 'become: yes' for the -E flag.
    #  shell: cat nodesource-setup-script.bash | sudo -E bash -
    #  args:
    #    chdir: "{{ crawler_dir }}"
    #    executable: /bin/bash
    #
    #- name: Install Node.js
    #  apt:
    #    name: nodejs
    #    state: latest


- name: Ensure webdriver.io in installed
  hosts: all
  gather_facts: no
  tasks:
    - name: Install webdriver.io
      npm:
        name: webdriverio
        global: no
        # Install into 'node_modules' in our crawler directory:
        path: "{{ crawler_dir }}"
      environment:
        # Do we actually need this to invoke our local 'npm'?
        PATH: "{{ ansible_env.HOME }}/{{ crawler_dir }}/node/bin:{{ ansible_env.PATH }}"


- name: Get geckodriver
  hosts: all
  gather_facts: no
  tasks:
    # Maybe someday the github redirector[*] will allow redirecting to the
    # latest assets even when they contain a version number in their name.
    # For now, we need to get the version number manually and insert it into
    # the URL for the .tar.gz that we want to download.
    #
    # * https://help.github.com/en/github/administering-a-repository/linking-to-releases
    #
    - name: Get latest geckodriver release tag
      uri:
        url: "https://api.github.com/repos/mozilla/geckodriver/releases/latest"
        method: GET
        return_content: yes
        status_code: 200
        headers:
          Content-Type: "application/json"
        body_format: json
      register: geckodriver_info

    - name: Fetch and extract geckodriver
      # Temporary method due to https://github.com/ansible/ansible/issues/64612
      shell: curl -L https://github.com/mozilla/geckodriver/releases/latest/download/geckodriver-{{ geckodriver_info.json.tag_name }}-linux64.tar.gz | tar xz
      args:
        chdir: "{{ crawler_dir }}"
        creates: geckodriver
        warn: false
      #unarchive:
      #  src: "https://github.com/mozilla/geckodriver/releases/latest/download/geckodriver-{{ geckodriver_info.json.tag_name }}-linux64.tar.gz"
      #  remote_src: yes
      #  dest: "{{ crawler_dir }}"
      #  creates: "{{ crawler_dir }}/geckodriver"

    - name: Allow localhost connections on port 4444
      ufw:
        rule: allow
        from_ip: 127.0.0.1
        from_port: "4444"
        to_ip: 127.0.0.1
        to_port: "4444"
        proto: tcp
      become: yes


- name: Ensure the Alexa and Cisco top sites lists are present
  hosts: all
  gather_facts: no
  tasks:
    - name: Get the Alexa top 1M list
      unarchive:
        src: https://s3.amazonaws.com/alexa-static/top-1m.csv.zip
        remote_src: yes
        dest: "{{ crawler_dir }}"
        # Kinda lying here:
        creates: "{{ crawler_dir }}/alexa-top-1m.csv"
      register: alexa1Mstate

    - name: Rename top-1m.csv to alexa-top-1m.csv
      command: mv top-1m.csv alexa-top-1m.csv
      args:
        chdir: "{{ crawler_dir }}"
      when: alexa1Mstate.changed

    - name: Create Alexa top 1k file
      shell: head -1000 alexa-top-1m.csv > alexa-top-1k.csv
      args:
        chdir: "{{ crawler_dir }}"
      when: alexa1Mstate.changed

    - name: Get the Cisco top 1M list
      unarchive:
        src: https://s3-us-west-1.amazonaws.com/umbrella-static/top-1m.csv.zip
        remote_src: yes
        dest: "{{ crawler_dir }}"
        # Kinda lying here:
        creates: "{{ crawler_dir }}/cisco-top-1m.csv"
      register: cisco1Mstate

    - name: Rename top-1m.csv to cisco-top-1m.csv
      command: mv top-1m.csv cisco-top-1m.csv
      args:
        chdir: "{{ crawler_dir }}"
      when: cisco1Mstate.changed

    - name: Create Cisco top 1k file
      shell: head -1000 cisco-top-1m.csv > cisco-top-1k.csv
      args:
        chdir: "{{ crawler_dir }}"
      when: cisco1Mstate.changed


- name: Ensure we have Nightly Firefox
  hosts: all
  gather_facts: no
  tasks:
    # We intall Firefox to get all the many lib dependencies (such as
    # libgtk-3-0) which we need even if we don't run this version.
    - name: Ensure Firefox is installed
      apt:
        name: firefox
        state: latest
      become: yes

    - name: Download Nightly Firefox
      # Temporary method due to https://github.com/ansible/ansible/issues/64612
      # Note: must quote URL due to "&".
      shell: curl -L "https://download.mozilla.org/?product=firefox-nightly-latest&os=linux64" | tar xj
      args:
        chdir: "{{ crawler_dir }}"
        creates: firefox
        warn: false
      #unarchive:
      #  # Discussion about version number in links: https://bugzilla.mozilla.org/show_bug.cgi?id=1595045
      #  src: https://download.mozilla.org/?product=firefox-nightly-latest&os=linux64
      #  remote_src: yes
      #  dest: "{{ crawler_dir }}"
      #  creates: "{{ crawler_dir }}/firefox"
  # Then to use, set:
  # export IFRAME_CRAWLER_FIREFOX_BIN="$PWD/firefox/firefox"