From 03d2a86ef1802fe8d50f5edd7636d7fe5f97d2ad Mon Sep 17 00:00:00 2001 From: evazion Date: Thu, 17 Mar 2022 19:40:16 -0500 Subject: [PATCH] artists: normalize fc2.com profile urls. --- app/logical/source/url.rb | 1 + app/logical/source/url/fc2.rb | 42 +++++++++++++++++++++++++++++++++++ app/models/artist_url.rb | 4 ---- test/unit/artist_url_test.rb | 9 ++------ 4 files changed, 45 insertions(+), 11 deletions(-) create mode 100644 app/logical/source/url/fc2.rb diff --git a/app/logical/source/url.rb b/app/logical/source/url.rb index d5025ad61..cf9e739ab 100644 --- a/app/logical/source/url.rb +++ b/app/logical/source/url.rb @@ -28,6 +28,7 @@ module Source Source::URL::DeviantArt, Source::URL::Fanbox, Source::URL::Fantia, + Source::URL::Fc2, Source::URL::Foundation, Source::URL::HentaiFoundry, Source::URL::Lofter, diff --git a/app/logical/source/url/fc2.rb b/app/logical/source/url/fc2.rb new file mode 100644 index 000000000..bc2229eae --- /dev/null +++ b/app/logical/source/url/fc2.rb @@ -0,0 +1,42 @@ +# frozen_string_literal: true + +class Source::URL::Fc2 < Source::URL + attr_reader :username, :profile_url + + def self.match?(url) + url.domain.in?(%w[fc2.com fc2blog.net fc2blog.us]) + end + + def site_name + "FC2" + end + + def parse + case [*host.split("."), *path_segments] + + # http://silencexs.blog.fc2.com + # http://silencexs.blog106.fc2.com + in username, /blog\d*/, "fc2", "com", *rest + @username = username + @profile_url = "http://#{username}.blog.fc2.com" + + # http://794ancientkyoto.web.fc2.com + # http://yorokobi.x.fc2.com + # https://lilish28.bbs.fc2.com + # http://jpmaid.h.fc2.com + # http://toritokaizoku.web.fc2.com/tori.html (404: http://toritokaizoku.web.fc2.com) + in username, ("bbs" | "web" | "h" | "x") => subsite, "fc2", "com", *rest + @username = username + @subsite = subsite + @profile_url = ["http://#{username}.#{subsite}.fc2.com", *rest].join("/") + + # http://swordsouls.blog131.fc2blog.net + # http://swordsouls.blog131.fc2blog.us + in username, /blog\d*/, "fc2blog", ("net" | "us") => tld, *rest + @username = username + @profile_url = "http://#{username}.blog.fc2blog.#{tld}" + + else + end + end +end diff --git a/app/models/artist_url.rb b/app/models/artist_url.rb index 445bab327..f17cf1b08 100644 --- a/app/models/artist_url.rb +++ b/app/models/artist_url.rb @@ -23,10 +23,6 @@ class ArtistURL < ApplicationRecord url = Source::URL.parse(url)&.profile_url || url url = url.sub(%r{^https://}, "http://") - url = url.sub(%r{^http://blog-imgs-\d+\.fc2}, "http://blog.fc2") - url = url.sub(%r{^http://blog-imgs-\d+-\w+\.fc2}, "http://blog.fc2") - url = url.sub(%r{^http://blog\d*\.fc2\.com/(?:\w/){,3}(\w+)}, "http://\\1.blog.fc2.com") - url = url.gsub(%r{/+\Z}, "") url + "/" end diff --git a/test/unit/artist_url_test.rb b/test/unit/artist_url_test.rb index e3593ad6f..64db6de79 100644 --- a/test/unit/artist_url_test.rb +++ b/test/unit/artist_url_test.rb @@ -71,13 +71,8 @@ class ArtistURLTest < ActiveSupport::TestCase end should "normalize fc2 urls" do - url = create(:artist_url, url: "http://blog55.fc2.com/monet") - assert_equal("http://blog55.fc2.com/monet", url.url) - assert_equal("http://monet.blog.fc2.com/", url.normalized_url) - - url = create(:artist_url, url: "http://blog-imgs-55.fc2.com/monet") - assert_equal("http://blog-imgs-55.fc2.com/monet", url.url) - assert_equal("http://monet.blog.fc2.com/", url.normalized_url) + url = create(:artist_url, url: "http://silencexs.blog106.fc2.com/") + assert_equal("http://silencexs.blog.fc2.com/", url.normalized_url) end should "normalize deviant art artist urls" do