Class: NewsScraper::Trainer::UrlTrainer

Inherits:
Object
  • Object
show all
Defined in:
lib/news_scraper/trainer/url_trainer.rb

Instance Method Summary collapse

Constructor Details

#initialize(url) ⇒ UrlTrainer

Returns a new instance of UrlTrainer.



4
5
6
7
8
 
# File 'lib/news_scraper/trainer/url_trainer.rb', line 4

def initialize(url)
  @url = url
  @root_domain = URIParser.new(@url).host
  @payload = Extractors::Article.new(url: @url).extract
end

Instance Method Details

#train ⇒ Object 



10
11
12
13
14
15
16
17
18
19
20
21
22
23
 
# File 'lib/news_scraper/trainer/url_trainer.rb', line 10

def train
  return if NewsScraper.configuration.scrape_patterns['domains'].key?(@root_domain)

  CLI.put_header(@root_domain)
  CLI.log("There is no scrape pattern defined for #{@root_domain}")
  CLI.log "Fetching information..."
  CLI.put_footer

  selected_presets = {}
  NewsScraper.configuration.scrape_patterns['data_types'].each do |data_type|
    selected_presets[data_type] = selected_pattern(data_type)
  end
  save_selected_presets(selected_presets)
end