From 9343f7c912d74162b58e1efb4416d0bdd59702a8 Mon Sep 17 00:00:00 2001
From: evazion <noizave@gmail.com>
Date: Sat, 12 Mar 2022 03:08:06 -0600
Subject: [PATCH] Source::URL: add profile_url method.

Add a method for converting a source URL into a profile URL. This will
be used for normalizing profile URLs in artist entries.

Also add the ability to parse a few more profile URL formats.
---
 app/logical/source/url.rb                | 21 +++++++++++++++++++++
 app/logical/source/url/art_station.rb    |  4 ++++
 app/logical/source/url/deviant_art.rb    |  4 ++++
 app/logical/source/url/fanbox.rb         | 15 ++++++++++++++-
 app/logical/source/url/fantia.rb         |  4 ++++
 app/logical/source/url/foundation.rb     |  8 ++++++++
 app/logical/source/url/hentai_foundry.rb |  4 ++++
 app/logical/source/url/lofter.rb         |  4 ++++
 app/logical/source/url/mastodon.rb       |  8 ++++++++
 app/logical/source/url/newgrounds.rb     |  4 ++++
 app/logical/source/url/nico_seiga.rb     | 19 ++++++++++++++++++-
 app/logical/source/url/nijie.rb          |  4 ++++
 app/logical/source/url/pixiv.rb          | 21 ++++++++++++++-------
 app/logical/source/url/pixiv_sketch.rb   |  4 ++++
 app/logical/source/url/plurk.rb          |  4 ++++
 app/logical/source/url/skeb.rb           |  4 ++++
 app/logical/source/url/tumblr.rb         |  3 +--
 app/logical/source/url/twitter.rb        | 23 ++++++++++++++++++++++-
 app/logical/source/url/weibo.rb          | 20 +++++++-------------
 app/logical/sources/strategies/weibo.rb  |  2 +-
 20 files changed, 154 insertions(+), 26 deletions(-)

diff --git a/app/logical/source/url.rb b/app/logical/source/url.rb
index 1ea615cc6..d5025ad61 100644
--- a/app/logical/source/url.rb
+++ b/app/logical/source/url.rb
@@ -110,6 +110,27 @@ module Source
       end
     end
 
+    # Convert the current URL into a profile URL, or return nil if it's not
+    # possible to get the profile URL from the current URL.
+    #
+    # URLs in artist entries will be normalized into this form.
+    #
+    # Some sites may have multiple profile URLs, for example if the site has
+    # both usernames and user IDs. This may return different profile URLs,
+    # depending on whether the current URL has the username or the user ID.
+    #
+    # Examples:
+    #
+    # * https://www.pixiv.net/member.php?id=9948
+    # * https://www.pixiv.net/stacc/bkubb
+    # * https://twitter.com/bkub_comic
+    # * https://twitter.com/intent/user?user_id=889592953
+    #
+    # @return [String, nil]
+    def profile_url
+      nil
+    end
+
     protected def initialize(...)
       super(...)
       parse
diff --git a/app/logical/source/url/art_station.rb b/app/logical/source/url/art_station.rb
index 9dd094283..54fb76296 100644
--- a/app/logical/source/url/art_station.rb
+++ b/app/logical/source/url/art_station.rb
@@ -65,4 +65,8 @@ class Source::URL::ArtStation < Source::URL
       "https://cdn.artstation.com/p/assets/#{@asset_type}/images/#{@asset_subdir}/#{size}/#{@file}"
     end
   end
+
+  def profile_url
+    "https://www.artstation.com/#{username}" if username.present?
+  end
 end
diff --git a/app/logical/source/url/deviant_art.rb b/app/logical/source/url/deviant_art.rb
index 0b32d338f..769f0c434 100644
--- a/app/logical/source/url/deviant_art.rb
+++ b/app/logical/source/url/deviant_art.rb
@@ -142,6 +142,10 @@ module Source
       end
     end
 
+    def profile_url
+      "https://www.deviantart.com/#{username}" if username.present?
+    end
+
     def pretty_title
       title.titleize.strip.squeeze(" ").tr(" ", "-") if title.present?
     end
diff --git a/app/logical/source/url/fanbox.rb b/app/logical/source/url/fanbox.rb
index 5e2557a2f..c3f344003 100644
--- a/app/logical/source/url/fanbox.rb
+++ b/app/logical/source/url/fanbox.rb
@@ -41,7 +41,12 @@ class Source::URL::Fanbox < Source::URL
       @work_id = work_id
 
     # https://www.pixiv.net/fanbox/creator/1566167
-    in "www.pixiv.net", "fanbox", "creator", user_id
+    # http://www.pixiv.net/fanbox/user/3410642
+    in "www.pixiv.net", "fanbox", ("creator" | "user"), user_id
+      @user_id = user_id
+
+    # http://www.pixiv.net/fanbox/member.php?user_id=3410642
+    in "www.pixiv.net", "fanbox", "member.php" if params[:user_id].present?
       @user_id = user_id
 
     # https://omu001.fanbox.cc/posts/39714
@@ -71,4 +76,12 @@ class Source::URL::Fanbox < Source::URL
     # https://pixiv.pximg.net/c/400x400_90_a2_g5/fanbox/public/images/creator/1566167/profile/BtxSp9MImFhnEZtjEZs2RPqL.jpeg
     to_s.gsub(%r{/[cw]/\w+/}, "/") if image_url?
   end
+
+  def profile_url
+    if username.present?
+      "https://#{username}.fanbox.cc"
+    elsif user_id.present?
+      "https://www.pixiv.net/fanbox/creator/#{user_id}"
+    end
+  end
 end
diff --git a/app/logical/source/url/fantia.rb b/app/logical/source/url/fantia.rb
index 1dc1540be..c60c817d7 100644
--- a/app/logical/source/url/fantia.rb
+++ b/app/logical/source/url/fantia.rb
@@ -76,6 +76,10 @@ class Source::URL::Fantia < Source::URL
     end
   end
 
+  def profile_url
+    "https://fantia.jp/fanclubs/#{fanclub_id}" if fanclub_id.present?
+  end
+
   def work_id
     @post_id || @product_id
   end
diff --git a/app/logical/source/url/foundation.rb b/app/logical/source/url/foundation.rb
index 4ae172eb6..76c3a8c0b 100644
--- a/app/logical/source/url/foundation.rb
+++ b/app/logical/source/url/foundation.rb
@@ -63,6 +63,14 @@ class Source::URL::Foundation < Source::URL
     end
   end
 
+  def profile_url
+    if username.present?
+      "https://foundation.app/@#{username}"
+    elsif user_id.present?
+      "https://foundation.app/#{user_id}"
+    end
+  end
+
   def page_url
     return nil unless work_id.present?
     return nil if host == "f8n-production-collection-assets.imgix.net" && @hash.blank?
diff --git a/app/logical/source/url/hentai_foundry.rb b/app/logical/source/url/hentai_foundry.rb
index 6cd7e8d20..e5c37e687 100644
--- a/app/logical/source/url/hentai_foundry.rb
+++ b/app/logical/source/url/hentai_foundry.rb
@@ -62,4 +62,8 @@ class Source::URL::HentaiFoundry < Source::URL
     else
     end
   end
+
+  def profile_url
+    "https://www.hentai-foundry.com/user/#{username}" if username.present?
+  end
 end
diff --git a/app/logical/source/url/lofter.rb b/app/logical/source/url/lofter.rb
index 4445722a8..8767ec9eb 100644
--- a/app/logical/source/url/lofter.rb
+++ b/app/logical/source/url/lofter.rb
@@ -42,4 +42,8 @@ class Source::URL::Lofter < Source::URL
   def full_image_url
     "#{site}#{path}" if image_url?
   end
+
+  def profile_url
+    "https://#{username}.lofter.com" if username.present?
+  end
 end
diff --git a/app/logical/source/url/mastodon.rb b/app/logical/source/url/mastodon.rb
index ce40b3b58..4d656ec9b 100644
--- a/app/logical/source/url/mastodon.rb
+++ b/app/logical/source/url/mastodon.rb
@@ -70,4 +70,12 @@ class Source::URL::Mastodon < Source::URL
   def image_url?
     full_image_url.present?
   end
+
+  def profile_url
+    if username.present?
+      "https://#{host}/@#{username}"
+    elsif user_id.present?
+      "https://#{host}/web/accounts/#{user_id}"
+    end
+  end
 end
diff --git a/app/logical/source/url/newgrounds.rb b/app/logical/source/url/newgrounds.rb
index 57375abff..fbab609c0 100644
--- a/app/logical/source/url/newgrounds.rb
+++ b/app/logical/source/url/newgrounds.rb
@@ -58,4 +58,8 @@ class Source::URL::Newgrounds < Source::URL
   def image_url?
     url.host == "art.ngfiles.com"
   end
+
+  def profile_url
+    "https://#{username}.newgrounds.com" if username.present?
+  end
 end
diff --git a/app/logical/source/url/nico_seiga.rb b/app/logical/source/url/nico_seiga.rb
index 8e7079e19..4e4101dad 100644
--- a/app/logical/source/url/nico_seiga.rb
+++ b/app/logical/source/url/nico_seiga.rb
@@ -88,7 +88,20 @@ module Source
         # unhandled
 
       # https://seiga.nicovideo.jp/user/illust/456831
-      in "seiga.nicovideo.jp", "user", "illust", user_id
+      # https://sp.seiga.nicovideo.jp/user/illust/20542122
+      # https://ext.seiga.nicovideo.jp/user/illust/20542122
+      in /seiga\.nicovideo\.jp$/, "user", "illust", user_id
+        @user_id = user_id
+
+      # http://seiga.nicovideo.jp/manga/list?user_id=23839737
+      # http://sp.seiga.nicovideo.jp/manga/list?user_id=23839737
+      in /seiga\.nicovideo\.jp$/, "manga", "list" if params[:user_id].present?
+        @user_id = params[:user_id]
+
+      # https://www.nicovideo.jp/user/4572975
+      # https://www.nicovideo.jp/user/20446930/mylist/28674289
+      # https://commons.nicovideo.jp/user/696839
+      in ("commons.nicovideo.jp" | "www.nicovideo.jp"), "user", /^\d+$/ => user_id, *rest
         @user_id = user_id
 
       else
@@ -104,5 +117,9 @@ module Source
         "https://seiga.nicovideo.jp/image/source/#{image_id}"
       end
     end
+
+    def profile_url
+      "https://seiga.nicovideo.jp/user/illust/#{user_id}" if user_id.present?
+    end
   end
 end
diff --git a/app/logical/source/url/nijie.rb b/app/logical/source/url/nijie.rb
index dfec1ca59..658577b69 100644
--- a/app/logical/source/url/nijie.rb
+++ b/app/logical/source/url/nijie.rb
@@ -93,4 +93,8 @@ class Source::URL::Nijie < Source::URL
   def full_image_url
     to_s.remove(%r{__rs_\w+/}i).gsub("http:", "https:") if image_url?
   end
+
+  def profile_url
+    "https://nijie.info/members.php?id=#{user_id}" if user_id.present?
+  end
 end
diff --git a/app/logical/source/url/pixiv.rb b/app/logical/source/url/pixiv.rb
index a47c04f36..2887d78ec 100644
--- a/app/logical/source/url/pixiv.rb
+++ b/app/logical/source/url/pixiv.rb
@@ -54,13 +54,19 @@ module Source
         @work_id = params[:illust_id]
 
       # https://www.pixiv.net/member.php?id=339253
-      in "www.pixiv.net", "member.php" if params[:id].present?
+      # http://www.pixiv.net/novel/member.php?id=76567
+      in "www.pixiv.net", *, "member.php" if params[:id].present?
         @user_id = params[:id]
 
       # https://www.pixiv.net/u/9202877
       # https://www.pixiv.net/users/9202877
+      # https://www.pixiv.net/users/76567/novels
+      in "www.pixiv.net", ("u" | "users"), user_id, *rest
+        @user_id = user_id
+
       # https://www.pixiv.net/en/users/9202877
-      in "www.pixiv.net", *, ("u" | "users"), user_id
+      # https://www.pixiv.net/en/users/76567/novels
+      in "www.pixiv.net", _, ("u" | "users"), user_id, *rest
         @user_id = user_id
 
       # https://www.pixiv.net/stacc/noizave
@@ -123,11 +129,12 @@ module Source
     end
 
     def profile_url
-      "https://www.pixiv.net/users/#{user_id}" if user_id.present?
-    end
-
-    def stacc_url
-      "https://www.pixiv.net/stacc/#{username}" if username.present?
+      if user_id.present?
+        # "https://www.pixiv.net/users/#{user_id}"
+        "https://www.pixiv.net/member.php?id=#{user_id}"
+      elsif username.present?
+        "https://www.pixiv.net/stacc/#{username}"
+      end
     end
   end
 end
diff --git a/app/logical/source/url/pixiv_sketch.rb b/app/logical/source/url/pixiv_sketch.rb
index 526d31f06..06706dd6f 100644
--- a/app/logical/source/url/pixiv_sketch.rb
+++ b/app/logical/source/url/pixiv_sketch.rb
@@ -39,6 +39,10 @@ module Source
       "https://sketch.pixiv.net/items/#{work_id}" if work_id.present?
     end
 
+    def profile_url
+      "https://sketch.pixiv.net/@#{username}" if username.present?
+    end
+
     def api_url
       # https://sketch.pixiv.net/api/items/5835314698645024323.json (won't work in the browser; use curl)
       "https://sketch.pixiv.net/api/items/#{work_id}.json" if work_id.present?
diff --git a/app/logical/source/url/plurk.rb b/app/logical/source/url/plurk.rb
index f77105d78..b70fc21d7 100644
--- a/app/logical/source/url/plurk.rb
+++ b/app/logical/source/url/plurk.rb
@@ -38,4 +38,8 @@ class Source::URL::Plurk < Source::URL
   def image_url?
     host == "images.plurk.com"
   end
+
+  def profile_url
+    "https://www.plurk.com/#{username}" if username.present?
+  end
 end
diff --git a/app/logical/source/url/skeb.rb b/app/logical/source/url/skeb.rb
index 552a3982b..e4f0dbaf1 100644
--- a/app/logical/source/url/skeb.rb
+++ b/app/logical/source/url/skeb.rb
@@ -42,6 +42,10 @@ class Source::URL::Skeb < Source::URL
     end
   end
 
+  def profile_url
+    "https://skeb.jp/@#{username}" if username.present?
+  end
+
   def image_url?
     domain.in?(%[imgix.net amazonaws.com])
   end
diff --git a/app/logical/source/url/tumblr.rb b/app/logical/source/url/tumblr.rb
index 0f8c0e7c8..dab3e85ad 100644
--- a/app/logical/source/url/tumblr.rb
+++ b/app/logical/source/url/tumblr.rb
@@ -65,7 +65,6 @@ class Source::URL::Tumblr < Source::URL
   end
 
   def profile_url
-    return nil unless @blog_name.present?
-    "https://#{@blog_name}.tumblr.com"
+    "https://#{blog_name}.tumblr.com" if blog_name.present?
   end
 end
diff --git a/app/logical/source/url/twitter.rb b/app/logical/source/url/twitter.rb
index ef703e5f6..be3042a3c 100644
--- a/app/logical/source/url/twitter.rb
+++ b/app/logical/source/url/twitter.rb
@@ -26,7 +26,7 @@ class Source::URL::Twitter < Source::URL
   # https://developer.twitter.com/en/docs/developer-utilities/configuration/api-reference/get-help-configuration
   RESERVED_USERNAMES = %w[home i intent search]
 
-  attr_reader :status_id, :twitter_username
+  attr_reader :status_id, :twitter_username, :user_id
 
   def self.match?(url)
     url.host.in?(%w[twitter.com mobile.twitter.com pic.twitter.com pbs.twimg.com video.twimg.com t.co])
@@ -51,6 +51,18 @@ class Source::URL::Twitter < Source::URL
     in "twitter.com", username, *rest
       @twitter_username = username unless username.in?(RESERVED_USERNAMES)
 
+    # https://twitter.com/intent/user?user_id=1485229827984531457
+    in "twitter.com", "intent", "user" if params[:user_id].present?
+      @user_id = params[:user_id]
+
+    # https://twitter.com/intent/user?screen_name=ryuudog_NFT
+    in "twitter.com", "intent", "user" if params[:screen_name].present?
+      @twitter_username = params[:screen_name]
+
+    # https://twitter.com/i/user/889592953
+    in "twitter.com", "i", "user", user_id
+      @user_id = user_id
+
     # https://pbs.twimg.com/media/EBGbJe_U8AA4Ekb.jpg
     # https://pbs.twimg.com/media/EBGbJe_U8AA4Ekb.jpg:small
     # https://pbs.twimg.com/media/EBGbJe_U8AA4Ekb?format=jpg&name=900x900
@@ -87,4 +99,13 @@ class Source::URL::Twitter < Source::URL
     return nil unless @file_path.present?
     "#{site}/#{@file_path}:orig"
   end
+
+  def profile_url
+    if twitter_username.present?
+      "https://twitter.com/#{twitter_username}"
+    elsif user_id.present?
+      # "https://twitter.com/i/user/#{user_id}
+      "https://twitter.com/intent/user?user_id=#{user_id}"
+    end
+  end
 end
diff --git a/app/logical/source/url/weibo.rb b/app/logical/source/url/weibo.rb
index 1df039fbe..a2d465588 100644
--- a/app/logical/source/url/weibo.rb
+++ b/app/logical/source/url/weibo.rb
@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
 class Source::URL::Weibo < Source::URL
-  attr_reader :full_image_url
+  attr_reader :full_image_url, :artist_short_id, :artist_long_id
 
   def self.match?(url)
     url.domain.in?(["weibo.com", "weibo.cn", "sinaimg.cn"])
@@ -63,18 +63,12 @@ class Source::URL::Weibo < Source::URL
     full_image_url.present?
   end
 
-  def profile_urls
-    [profile_short_url, profile_long_url].compact
-  end
-
-  def profile_short_url
-    return if @artist_short_id.blank?
-    "https://www.weibo.com/u/#{@artist_short_id}"
-  end
-
-  def profile_long_url
-    return if @artist_long_id.blank?
-    "https://www.weibo.com/p/#{@artist_long_id}"
+  def profile_url
+    if artist_short_id.present?
+      "https://www.weibo.com/u/#{artist_short_id}"
+    elsif artist_long_id.present?
+      "https://www.weibo.com/p/#{artist_long_id}"
+    end
   end
 
   def mobile_url
diff --git a/app/logical/sources/strategies/weibo.rb b/app/logical/sources/strategies/weibo.rb
index e88858a74..4f723b4b0 100644
--- a/app/logical/sources/strategies/weibo.rb
+++ b/app/logical/sources/strategies/weibo.rb
@@ -49,7 +49,7 @@ module Sources
       end
 
       def profile_urls
-        (parsed_url.profile_urls + parsed_referer&.profile_urls.to_a).uniq
+        [parsed_url.profile_url, parsed_referer&.profile_url].compact.uniq
       end
 
       def profile_url