sources: factor out Source::URL::TwitPic.

2022-02-23 18:19:04 -06:00
parent 7ed8f95a8e
commit 043c08eb05
9 changed files with 117 additions and 30 deletions
--- a/app/logical/source/url.rb
+++ b/app/logical/source/url.rb
@@ -19,6 +19,7 @@ module Source
  class URL < Danbooru::URL
    SUBCLASSES = [
      Source::URL::Twitter,
+      Source::URL::TwitPic,
    ]

    # Parse a URL into a subclass of Source::URL, or raise an exception if the URL is not a valid HTTP or HTTPS URL.
--- a/app/logical/source/url/twit_pic.rb
+++ b/app/logical/source/url/twit_pic.rb
@@ -0,0 +1,72 @@
+# frozen_string_literal: true
+#
+# Page URLs:
+#
+# * https://twitpic.com/dvitq3 (live)
+# * https://twitpic.com/9tgik8 (dead)
+#
+# Image URLs:
+#
+# # Live. Image for https://twitpic.com/dvitq3. `dvitq3` is base36 for 839006715.
+# * https://dn3pm25xmtlyu.cloudfront.net/photos/large/839006715.jpg?Expires=1646850828&Signature=d60CmLlmNqZJvOTteTOan13QWZ8gY3C4rUWCkh-IUoRr012vYtUYtip74GslGwCG0dxV5mpUpVFkaVZf16PiY7CsTdpAlA8Pmu2tN98D2dmC5FuW9KhhygDv6eFC8faoaGEyj~ArLuwz-8lC6Y05TVf0FgweeWwsRxFOfD5JHgCeIB0iZqzUx1t~eb6UMAWvbaKpfgvcp2oaDuCdZlMNi9T5OUBFoTh2DfnGy8t5COys1nOYYfZ9l69TDvVb2PKBaV8lsKK9xMwjoJNaWa1HL5S4MgODS5hiNDvycoBpu9KUvQ7q~rhC8cV6ZNctB5H9u~MmvBPoTKfy4w37cSc5uw__&Key-Pair-Id=APKAJROXZ7FN26MABHYA
+#
+# # Live. Image for https://twitpic.com/dks0tb.
+# * http://d3j5vwomefv46c.cloudfront.net/photos/large/820960031.jpg?1384107199
+#
+# # Dead. Old images for http://twitpic/dvitq3.
+# * https://twitpic.com/show/large/dvitq3.jpg
+# * https://o.twimg.com/2/proxy.jpg?t=HBgpaHR0cHM6Ly90d2l0cGljLmNvbS9zaG93L2xhcmdlL2R2aXRxMy5qcGcUsAkUsg0AFgASAA&s=NeY89zVAEpDjLcxZ_8KOoF7VGr2dm1Vc3HIozPy__Ng
+#
+# Profile URLs:
+#
+# * http://twitpic.com/photos/Type10TK (dead)
+#
+class Source::URL::TwitPic < Source::URL
+  attr_reader :base36_id
+
+  def self.match?(url)
+    url.host.in?(%w[twitpic.com o.twimg.com dn3pm25xmtlyu.cloudfront.net d3j5vwomefv46c.cloudfront.net])
+  end
+
+  def parse
+    case [domain, *path_segments]
+
+    # https://twitpic.com/carwkf
+    in "twitpic.com", base36_id
+      @base36_id = base36_id
+
+    # https://twitpic.com/show/large/carwkf.jpg
+    in "twitpic.com", "show", size, filename
+      @filename, @file_ext = filename.split(".")
+      @base36_id = @filename
+
+    # https://o.twimg.com/1/proxy.jpg?t=FQQVBBgpaHR0cHM6Ly90d2l0cGljLmNvbS9zaG93L2xhcmdlL2NhcndrZi5qcGcUBBYAEgA&s=y8haxddqxJYpWql9uVnP3aoFFS7rA10vOGPdTO5HXvk
+    # https://o.twimg.com/2/proxy.jpg?t=HBgpaHR0cHM6Ly90d2l0cGljLmNvbS9zaG93L2xhcmdlL2R0bnVydS5qcGcUsAkU0ggAFgASAA&s=dnN4DHCdnojC-iCJWdvZ-UZinrlWqAP7k7lmll2fTxs
+    in "twimg.com", subdir, "proxy.jpg" if params[:t].present?
+      # FQQVBBgpaHR0cHM6Ly90d2l0cGljLmNvbS9zaG93L2xhcmdlL2NhcndrZi5qcGcUBBYAEgA
+      @base64_id = params[:t]
+
+      # "\x15\x04\x15\x04\x18)https://twitpic.com/show/large/carwkf.jpg\x14\x04\x16\x00\x12\x00"
+      @decoded_base64_id = Base64.decode64(@base64_id)
+
+      # https://twitpic.com/show/large/carwkf.jpg
+      @decoded_url = URI.extract(@decoded_base64_id, %w[http https]).first
+
+      # carwkf
+      @base36_id = Source::URL.parse(@decoded_url).base36_id if @decoded_url.present?
+
+    # http://d3j5vwomefv46c.cloudfront.net/photos/large/820960031.jpg?1384107199
+    # https://dn3pm25xmtlyu.cloudfront.net/photos/large/839006715.jpg?Expires=1646850828&Signature=d60CmLlmNqZJvOTteTOan13QWZ8gY3C4rUWCkh-IUoRr012vYtUYtip74GslGwCG0dxV5mpUpVFkaVZf16PiY7CsTdpAlA8Pmu2tN98D2dmC5FuW9KhhygDv6eFC8faoaGEyj~ArLuwz-8lC6Y05TVf0FgweeWwsRxFOfD5JHgCeIB0iZqzUx1t~eb6UMAWvbaKpfgvcp2oaDuCdZlMNi9T5OUBFoTh2DfnGy8t5COys1nOYYfZ9l69TDvVb2PKBaV8lsKK9xMwjoJNaWa1HL5S4MgODS5hiNDvycoBpu9KUvQ7q~rhC8cV6ZNctB5H9u~MmvBPoTKfy4w37cSc5uw__&Key-Pair-Id=APKAJROXZ7FN26MABHYA
+    in /cloudfront\.net/, "photos", size, filename
+      @filename, @file_ext = filename.split(".")
+      @base36_id = @filename.to_i.to_s(36)
+
+    else
+    end
+  end
+
+  def page_url
+    return nil unless base36_id.present?
+    "https://twitpic.com/#{base36_id}"
+  end
+end
--- a/app/logical/sources/strategies.rb
+++ b/app/logical/sources/strategies.rb
@@ -22,6 +22,7 @@ module Sources
        Strategies::Lofter,
        Strategies::Foundation,
        Strategies::Plurk,
+        Strategies::TwitPic,
      ]
    end

--- a/app/logical/sources/strategies/null.rb
+++ b/app/logical/sources/strategies/null.rb
@@ -17,11 +17,6 @@ module Sources

      def normalize_for_source
        case url
-        when %r{\Ahttps?://(?:d3j5vwomefv46c|dn3pm25xmtlyu)\.cloudfront\.net/photos/large/(\d+)\.}i
-          base_10_id = $1.to_i
-          base_36_id = base_10_id.to_s(36)
-          "https://twitpic.com/#{base_36_id}"
-
        when %r{\Ahttp://www\.karabako\.net/images(?:ub)?/karabako_(\d+)(?:_\d+)?\.}i
          "http://www.karabako.net/post/view/#{$1}"

--- a/app/logical/sources/strategies/twit_pic.rb
+++ b/app/logical/sources/strategies/twit_pic.rb
@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+# @see Source::URL::TwitPic
+module Sources::Strategies
+  class TwitPic < Base
+    def match?
+      parsed_url&.site_name == "TwitPic"
+    end
+
+    def site_name
+      parsed_url.site_name
+    end
+
+    def normalize_for_source
+      parsed_url.page_url || url
+    end
+  end
+end
--- a/app/logical/sources/strategies/twitter.rb
+++ b/app/logical/sources/strategies/twitter.rb
@@ -23,12 +23,12 @@ module Sources::Strategies
      Danbooru.config.twitter_api_key.present? && Danbooru.config.twitter_api_secret.present?
    end

-    def domains
-      ["twitter.com", "twimg.com"]
+    def match?
+      parsed_url&.site_name == "Twitter"
    end

    def site_name
-      "Twitter"
+      parsed_url.site_name
    end

    def image_urls
@@ -118,18 +118,6 @@ module Sources::Strategies
        "https://twitter.com/#{tag_name_from_url}/status/#{status_id}"
      elsif status_id.present?
        "https://twitter.com/i/web/status/#{status_id}"
-      elsif url =~ %r{\Ahttps?://(?:o|image-proxy-origin)\.twimg\.com/\d/proxy\.jpg\?t=(\w+)&}i
-        str = Base64.decode64($1)
-        source = URI.extract(str, %w[http https])
-        if source.any?
-          source = source[0]
-          if source =~ %r{^https?://twitpic.com/show/large/[a-z0-9]+}i
-            source.gsub!(%r{show/large/}, "")
-            index = source.rindex(".")
-            source = source[0..index - 1]
-          end
-          source
-        end
      end
    end

--- a/test/unit/sources/null_test.rb
+++ b/test/unit/sources/null_test.rb
@@ -29,11 +29,6 @@ module Sources
    end

    context "normalizing for source" do
-      should "normalize twitpic links" do
-        source = "http://d3j5vwomefv46c.cloudfront.net/photos/large/820960031.jpg?1384107199"
-        assert_equal("https://twitpic.com/dks0tb", Sources::Strategies.normalize_source(source))
-      end
-
      should "normalize karabako links" do
        source = "http://www.karabako.net/images/karabako_38835.jpg"
        assert_equal("http://www.karabako.net/post/view/38835", Sources::Strategies.normalize_source(source))
--- a/test/unit/sources/twit_pic_test.rb
+++ b/test/unit/sources/twit_pic_test.rb
@@ -0,0 +1,22 @@
+require 'test_helper'
+
+module Sources
+  class TwitPicTest < ActiveSupport::TestCase
+    context "normalizing for source" do
+      should "normalize d3j5vwomefv46c.cloudfront.net links" do
+        source = "http://d3j5vwomefv46c.cloudfront.net/photos/large/820960031.jpg?1384107199"
+        assert_equal("https://twitpic.com/dks0tb", Sources::Strategies.normalize_source(source))
+      end
+
+      should "normalize dn3pm25xmtlyu.cloudfront.net links" do
+        source = "https://dn3pm25xmtlyu.cloudfront.net/photos/large/839006715.jpg?Expires=1646850828&Signature=d60CmLlmNqZJvOTteTOan13QWZ8gY3C4rUWCkh-IUoRr012vYtUYtip74GslGwCG0dxV5mpUpVFkaVZf16PiY7CsTdpAlA8Pmu2tN98D2dmC5FuW9KhhygDv6eFC8faoaGEyj~ArLuwz-8lC6Y05TVf0FgweeWwsRxFOfD5JHgCeIB0iZqzUx1t~eb6UMAWvbaKpfgvcp2oaDuCdZlMNi9T5OUBFoTh2DfnGy8t5COys1nOYYfZ9l69TDvVb2PKBaV8lsKK9xMwjoJNaWa1HL5S4MgODS5hiNDvycoBpu9KUvQ7q~rhC8cV6ZNctB5H9u~MmvBPoTKfy4w37cSc5uw__&Key-Pair-Id=APKAJROXZ7FN26MABHYA"
+        assert_equal("https://twitpic.com/dvitq3", Sources::Strategies.normalize_source(source))
+      end
+
+      should "normalize o.twimg.com links" do
+        source = "https://o.twimg.com/2/proxy.jpg?t=HBgpaHR0cHM6Ly90d2l0cGljLmNvbS9zaG93L2xhcmdlL2R0bnVydS5qcGcUsAkU0ggAFgASAA&s=dnN4DHCdnojC-iCJWdvZ-UZinrlWqAP7k7lmll2fTxs"
+        assert_equal("https://twitpic.com/dtnuru", Sources::Strategies.normalize_source(source))
+      end
+    end
+  end
+end
--- a/test/unit/sources/twitter_test.rb
+++ b/test/unit/sources/twitter_test.rb
@@ -321,11 +321,6 @@ module Sources
        assert_equal(source2, Sources::Strategies.normalize_source(source3))
        assert_equal(source2, Sources::Strategies.normalize_source(source4))
      end
-
-      should "normalize twimg twitpic correctly" do
-        source = "https://o.twimg.com/2/proxy.jpg?t=HBgpaHR0cHM6Ly90d2l0cGljLmNvbS9zaG93L2xhcmdlL2R0bnVydS5qcGcUsAkU0ggAFgASAA&s=dnN4DHCdnojC-iCJWdvZ-UZinrlWqAP7k7lmll2fTxs"
-        assert_equal("https://twitpic.com/dtnuru", Sources::Strategies.normalize_source(source))
-      end
    end
  end
 end