From 0b927290e33913ae055d7e05c4f3ab0f69b9204f Mon Sep 17 00:00:00 2001
From: Benjamin Neff <benjamin@coding4coffee.ch>
Date: Sat, 2 Sep 2017 04:18:31 +0200
Subject: [PATCH] Add DiasporaUrlParser to extract diaspora:// URLs from texts

---
 lib/diaspora_federation/federation.rb         |  1 +
 .../federation/diaspora_url_parser.rb         | 29 ++++++++
 .../federation/diaspora_url_parser_spec.rb    | 68 +++++++++++++++++++
 3 files changed, 98 insertions(+)
 create mode 100644 lib/diaspora_federation/federation/diaspora_url_parser.rb
 create mode 100644 spec/lib/diaspora_federation/federation/diaspora_url_parser_spec.rb

diff --git a/lib/diaspora_federation/federation.rb b/lib/diaspora_federation/federation.rb
index a55eed4..b279cfd 100644
--- a/lib/diaspora_federation/federation.rb
+++ b/lib/diaspora_federation/federation.rb
@@ -4,6 +4,7 @@ module DiasporaFederation
   end
 end
 
+require "diaspora_federation/federation/diaspora_url_parser"
 require "diaspora_federation/federation/fetcher"
 require "diaspora_federation/federation/receiver"
 require "diaspora_federation/federation/sender"
diff --git a/lib/diaspora_federation/federation/diaspora_url_parser.rb b/lib/diaspora_federation/federation/diaspora_url_parser.rb
new file mode 100644
index 0000000..3fb4300
--- /dev/null
+++ b/lib/diaspora_federation/federation/diaspora_url_parser.rb
@@ -0,0 +1,29 @@
+module DiasporaFederation
+  module Federation
+    # This module is for parsing and fetching linked entities.
+    module DiasporaUrlParser
+      include Logging
+
+      # Regex to find diaspora:// URLs
+      DIASPORA_URL_REGEX = %r{diaspora://(#{Entity::ENTITY_NAME_REGEX})/(#{Validation::Rule::Guid::VALID_CHARS})}
+
+      # Parses all diaspora:// URLs from the text and fetches the entities from
+      # the remote server if needed.
+      # @param [String] sender the diaspora* ID of the sender of the entity
+      # @param [String] text text with diaspora:// URLs to fetch
+      def self.fetch_linked_entities(sender, text)
+        text.scan(DIASPORA_URL_REGEX).each do |type, guid|
+          fetch_entity(sender, type, guid)
+        end
+      end
+
+      private_class_method def self.fetch_entity(sender, type, guid)
+        class_name = Entity.entity_class(type).to_s.rpartition("::").last
+        return if DiasporaFederation.callbacks.trigger(:fetch_related_entity, class_name, guid)
+        Fetcher.fetch_public(sender, type, guid)
+      rescue => e
+        logger.error "Failed to fetch linked entity #{type}:#{guid}: #{e.class}: #{e.message}"
+      end
+    end
+  end
+end
diff --git a/spec/lib/diaspora_federation/federation/diaspora_url_parser_spec.rb b/spec/lib/diaspora_federation/federation/diaspora_url_parser_spec.rb
new file mode 100644
index 0000000..bce2b2d
--- /dev/null
+++ b/spec/lib/diaspora_federation/federation/diaspora_url_parser_spec.rb
@@ -0,0 +1,68 @@
+module DiasporaFederation
+  describe Federation::DiasporaUrlParser do
+    let(:sender) { Fabricate.sequence(:diaspora_id) }
+    let(:guid) { Fabricate.sequence(:guid) }
+
+    describe ".fetch_linked_entities" do
+      it "parses linked posts from the text" do
+        guid2 = Fabricate.sequence(:guid)
+        guid3 = Fabricate.sequence(:guid)
+        expect_callback(:fetch_related_entity, "Post", guid).and_return(double)
+        expect_callback(:fetch_related_entity, "Post", guid2).and_return(double)
+        expect_callback(:fetch_related_entity, "Post", guid3).and_return(double)
+
+        text = "This is a [link to a post with markdown](diaspora://post/#{guid}) and one without " \
+               "diaspora://post/#{guid2} and finally a last one diaspora://post/#{guid3}."
+
+        Federation::DiasporaUrlParser.fetch_linked_entities(sender, text)
+      end
+
+      it "ignores invalid diaspora:// urls" do
+        expect(DiasporaFederation.callbacks).not_to receive(:trigger)
+
+        text = "This is an invalid link diaspora://Post/#{guid}) and another one: " \
+               "diaspora://post/abcd."
+
+        Federation::DiasporaUrlParser.fetch_linked_entities(sender, text)
+      end
+
+      it "allows to link other entities" do
+        expect_callback(:fetch_related_entity, "Event", guid).and_return(double)
+
+        text = "This is a link to an event diaspora://event/#{guid}."
+
+        Federation::DiasporaUrlParser.fetch_linked_entities(sender, text)
+      end
+
+      it "handles unknown entities gracefully" do
+        expect(DiasporaFederation.callbacks).not_to receive(:trigger)
+
+        text = "This is a link to an event diaspora://unknown/#{guid}."
+
+        Federation::DiasporaUrlParser.fetch_linked_entities(sender, text)
+      end
+
+      it "fetches entities from sender when not found locally" do
+        expect_callback(:fetch_related_entity, "Post", guid).and_return(nil)
+        expect(Federation::Fetcher).to receive(:fetch_public).with(sender, "post", guid)
+
+        text = "This is a link to a post: diaspora://post/#{guid}."
+
+        Federation::DiasporaUrlParser.fetch_linked_entities(sender, text)
+      end
+
+      it "handles fetch errors gracefully" do
+        expect_callback(:fetch_related_entity, "Post", guid).and_return(nil)
+        expect(Federation::Fetcher).to receive(:fetch_public).with(
+          sender, "post", guid
+        ).and_raise(Federation::Fetcher::NotFetchable, "Something went wrong!")
+
+        text = "This is a link to a post: diaspora://post/#{guid}."
+
+        expect {
+          Federation::DiasporaUrlParser.fetch_linked_entities(sender, text)
+        }.not_to raise_error
+      end
+    end
+  end
+end