vlv-search/spec/parser_spec.rb

require_relative '../lib/parser'

RSpec.describe Parser do
  subject { Parser.new }

  describe '#threads' do
    let(:html) do
      Nokogiri::HTML(<<~HTML)
        <div class="even" id="thread_12345">
          <ul class="list read">
            <li class="member">
              <span>Thread By: </span>
              <a href="/member/view/creator1/" class="memberlink">creator1</a>
            </li>
            <li class="subject">
              <span>Subject: </span>
              <a href="/thread/view/12345/&p=999">
                <strong>Sticky:</sticky> Thread title 1
              </a>
            </li>
            <li class="posts"><span>Posts: </span>999</li>
            <li class="lastpost">
              <span>Last Post By:</span>
              <a href="/member/view/lastposter1/" class="memberlink">lastposter1</a> on Fri&nbsp;Apr&nbsp;10&nbsp;2020&nbsp;01:23&nbsp;am</li>
          </ul>
        </div>
        <div class="even" id="thread_123456">
          <ul class="list read">
            <li class="member">
              <span>Thread By: </span>
              <a href="/member/view/creator2/" class="memberlink">creator2</a>
            </li>
            <li class="subject">
              <span>Subject: </span>
              <a href="/thread/view/123456/&p=999">Thread title 2</a>
            </li>
            <li class="posts"><span>Posts: </span>999</li>
            <li class="lastpost">
              <span>Last Post By:</span>
              <a href="/member/view/lastposter2/" class="memberlink">lastposter2</a> on Fri&nbsp;Apr&nbsp;10&nbsp;2020&nbsp;01:23&nbsp;am</li>
          </ul>
        </div>
      HTML
    end

    it 'parses threads' do
      expect(subject.threads(html)).to eq([
        {remote_id: '12345', title: 'Sticky: Thread title 1', creator: 'creator1', is_sticky: true},
        {remote_id: '123456', title: 'Thread title 2', creator: 'creator2', is_sticky: false},
      ])
    end
  end

  describe '#posts' do
    let(:html) do
      Nokogiri::HTML(<<~HTML)
        <div>
          <div class="post">
            <ul class="view" id="post_69">
              <li class="info">
                <div class="postinfo">
                  <a class="memberlink" href="/member/view/User1">User1</a>
                  posted this October 26th, 2021 @ 12:34:56 am
                </div>
              </li>
              <li class="postbody">
                This is the body of the first post
              </li>
            </ul>
          </div>
          <div class="post">
            <ul class="view" id="post_420">
              <li class="info">
                <div class="postinfo">
                  <a class="memberlink" href="/member/view/User2">User2</a>
                  posted this October 27th, 2021 @ 12:34:56 am
                </div>
              </li>
              <li class="postbody">
                This is the body of the second post
              </li>
            </ul>
          </div>
        </div>
      HTML
    end

    it 'parses posts' do
      expect(subject.posts({ id: 666 }, html)).to match_array([
        {
          remote_id: 69,
          creator: 'User1',
          thread_id: 666,
          created_at: Time.new(2021, 10, 26, 0, 34, 56),
          body: 'This is the body of the first post'
        },
        {
          remote_id: 420,
          creator: 'User2',
          thread_id: 666,
          created_at: Time.new(2021, 10, 27, 0, 34, 56),
          body: 'This is the body of the second post'
        }
      ])
    end
  end

  describe '#thread_created_at' do
    let(:html) do
      Nokogiri::HTML(<<~HTML)
      <div class="postinfo">
        <a class="memberlink" href="/member/view/User1">User1</a>
        posted this October 27th, 2021 @ 12:34:56 am
      </div>
      HTML
    end

    it 'parses the timestamp of the first post' do
      expect(subject.thread_created_at(html))
        .to eq(Time.new(2021, 10, 27, 00, 34, 56))
    end
  end
end