vlv-search/db/scrape.rb

require 'dotenv/load'
require 'sequel'

require_relative '../db/connect'
require_relative '../lib/auth'
require_relative '../lib/fetch'
require_relative '../lib/parse'
require_relative '../lib/models/post'
require_relative '../lib/models/thread'

class Scraper
  def initialize(first: 0, last: 0, log: false)
    @first = first
    @last = last
    @log = log

    authenticate!
  end

  def scrape
    (@first..@last).each_with_index do |page_number, page_index|
      page = Fetch.page(page_number, @cookie)
      threads = Parse.threads(page)

      threads.each do |t|
        next if page_index > 0 && t[:is_sticky]
        no_new_posts = scrape_thread(t)
        if no_new_posts
          next if t[:is_sticky]
          return
        end
      end
    end
  end

  private

  def authenticate!
    @cookie = login(ENV['VLV_USERNAME'], ENV['VLV_PASSWORD'])

    raise "Error logging into VLV. Check your credentials." if @cookie.nil?
  end

  def scrape_thread(t)
    log t[:title]

    page = Fetch.thread(t, @cookie)
    first_post = page.at_css('.postinfo:first-child')
    return false if first_post.nil?

    t[:created_at] = Parse.thread_created_at(first_post)

    thread = DB.from(:threads).first(remote_id: t[:remote_id])
    if thread.nil?
      thread = VLV::Thread.create(t.delete_if { |k| k == :is_sticky })
      log '  Inserting thread'
    end

    scrape_posts(thread, page)
  end

  def scrape_posts(thread, page)
    posts = Parse.posts(thread, page)
    last_post = posts.last
    unless DB.from(:posts).first(remote_id: last_post[:remote_id]).nil?
      log '  No new posts'
      return true
    end

    posts_count = posts.size
    posts.each_with_index do |p, index|
      msg = "  Inserting post #{index + 1}/#{posts_count}"
      print msg if @log
      if DB.from(:posts).first(remote_id: p[:remote_id]).nil?
        VLV::Post.create(p)
      end
      print "\b" * msg.size unless index == posts_count - 1 if @log
    end

    log ''

    false
  end

  def log(msg)
    puts msg if @log
  end
end