dumper task for ml image classification work

This commit is contained in:
Kyle Drake 2019-07-02 13:46:39 -07:00
parent 7fd75170e2
commit e183ac32f6

View file

@ -465,3 +465,17 @@ task :dedupe_site_blocks => [:environment] do
duped_block.destroy
end
end
desc 'ml_screenshots_list_dump'
task :ml_screenshots_list_dump => [:environment] do
['phishing', 'spam', 'ham', nil].each do |classifier|
File.open("./files/screenshot-urls-#{classifier.to_s}.txt", 'w') do |fp|
SiteFile.where(classifier: classifier).where(path: 'index.html').each do |site_file|
begin
fp.write "#{site_file.site.screenshot_url('index.html', Site::SCREENSHOT_RESOLUTIONS.first)}\n"
rescue NoMethodError
end
end
end
end
end