onebox 2.1.4 → 2.1.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: cba7774ebf124a2d110630be26869977c6acf71d1a07603e890d3a1231430834
-  data.tar.gz: fe5de86e2abdc20fee6139c596da13e1249a6d38d9c40a7759adbbaa5065c8f0
+  metadata.gz: d555fa4be5db2d4437afe2b951d7af3efff8e399fc30a2bc425d5f6a791f8ea2
+  data.tar.gz: d70a7b81ee90f534faa251b5b73af40077aec16f1450a08ef6adf195f5e725ef
 SHA512:
-  metadata.gz: e9d0e1df1c6d14c4984b30c7b3392d519b0eeec63ee7a34656e3089d9c1bc81e1d7e801660ceb2d4da0ea89a779e64a0cb7cd332485b584a9d6873866dbe8819
-  data.tar.gz: 4c3587d5445c97480c3d5dd4ee514ea9df427d26fb1fed1ace681b95aa6012d79c84323bd915803c4409b21695f4f2f28a6baf6016a8f50dc0f76a65d79be04d
+  metadata.gz: 30bc63c0960c76fd6ed876a43554502346cd5b31bf101075307aeb3c4341c32202d9700ef5a5097a46512745cb6659c152a4c9266a7b79769957a33ebf2db92b
+  data.tar.gz: 55f73e9b974ce8d5ad272335beaf1a72843fcfcd6b7be2b79988550c7cb33aa5ba6e0d3f419d4341a270d8f0232a849181332e3a3cc7ad07b93f5f432da2a7d1

data/lib/onebox/engine.rb

@@ -30,6 +30,7 @@ module Onebox
 
     attr_reader :url, :uri
     attr_reader :timeout
+    attr :errors
 
     DEFAULT = {}
     def options
@@ -44,6 +45,7 @@ module Onebox
     end
 
     def initialize(link, timeout = nil)
+      @errors = {}
       @options = DEFAULT
       class_name = self.class.name.split("::").last.to_s
 
@@ -150,6 +152,7 @@ require_relative "engine/amazon_onebox"
 require_relative "engine/github_issue_onebox"
 require_relative "engine/github_blob_onebox"
 require_relative "engine/github_commit_onebox"
+require_relative "engine/github_folder_onebox"
 require_relative "engine/github_gist_onebox"
 require_relative "engine/github_pullrequest_onebox"
 require_relative "engine/google_calendar_onebox"

data/lib/onebox/engine/allowlisted_generic_onebox.rb

@@ -256,6 +256,15 @@ module Onebox
             d[:data_1] = Onebox::Helpers.truncate("#{d[:price_currency].strip} #{d[:price_amount].strip}")
           end
 
+          skip_missing_tags = [:video]
+          d.each do |k, v|
+            next if skip_missing_tags.include?(k)
+            if v == nil || v == ''
+              errors[k] ||= []
+              errors[k] << 'is blank'
+            end
+          end
+
           d
         end
       end
@@ -316,7 +325,7 @@ module Onebox
         return true if AllowlistedGenericOnebox.html_providers.include?(data[:provider_name])
         return false unless data[:html]["iframe"]
 
-        fragment = Nokogiri::HTML::fragment(data[:html])
+        fragment = Nokogiri::HTML5::fragment(data[:html])
         src = fragment.at_css('iframe')&.[]("src")
         options[:allowed_iframe_regexes]&.any? { |r| src =~ r }
       end
@@ -367,7 +376,7 @@ module Onebox
       end
 
       def embedded_html
-        fragment = Nokogiri::HTML::fragment(data[:html])
+        fragment = Nokogiri::HTML5::fragment(data[:html])
         fragment.css("img").each { |img| img["class"] = "thumbnail" }
         if iframe = fragment.at_css("iframe")
           iframe.remove_attribute("style")

data/lib/onebox/engine/github_folder_onebox.rb

@@ -0,0 +1,78 @@
+# frozen_string_literal: true
+
+module Onebox
+  module Engine
+    class GithubFolderOnebox
+      include Engine
+      include StandardEmbed
+      include LayoutSupport
+
+      matches_regexp Regexp.new(/^https?:\/\/(?:www\.)?(?:(?:\w)+\.)?(github)\.com[\:\d]*(\/[^\/]+){2}/)
+      always_https
+
+      def self.priority
+        # This engine should have lower priority than the other Github engines
+        150
+      end
+
+      private
+
+      def data
+        og = get_opengraph
+
+        max_length = 250
+
+        display_path = extract_path(og.url, max_length)
+        display_description = clean_description(og.description, og.title, max_length)
+
+        title = og.title
+
+        fragment = Addressable::URI.parse(url).fragment
+        if fragment
+          fragment = Addressable::URI.unencode(fragment)
+
+          if html_doc.css('.Box.md')
+            # For links to markdown docs
+            node = html_doc.css('a.anchor').find { |n| n['href'] == "##{fragment}" }
+            subtitle = node&.parent&.text
+          elsif html_doc.css('.Box.rdoc')
+            # For links to rdoc docs
+            node = html_doc.css('h3').find { |n| n['id'] == "user-content-#{fragment.downcase}" }
+            subtitle = node&.css('text()')&.first&.text
+          end
+
+          title = "#{title} - #{subtitle}" if subtitle
+        end
+
+        {
+          link: url,
+          image: og.image,
+          title: Onebox::Helpers.truncate(title, 250),
+          path: display_path,
+          description: display_description,
+          favicon: get_favicon
+        }
+      end
+
+      def extract_path(root, max_length)
+        path = url.split('#')[0].split('?')[0]
+        path = path["#{root}/tree/".length..-1]
+
+        return unless path
+
+        path.length > max_length ? path[-max_length..-1] : path
+      end
+
+      def clean_description(description, title, max_length)
+        return unless description
+
+        desc_end = " - #{title}"
+        if description[-desc_end.length..-1] == desc_end
+          description = description[0...-desc_end.length]
+        end
+
+        Onebox::Helpers.truncate(description, max_length)
+      end
+    end
+  end
+end

data/lib/onebox/engine/instagram_onebox.rb

@@ -15,21 +15,32 @@ module Onebox
       end
 
       def data
-        og = get_opengraph
-        author_name = og.description.match(/\(@(\w+)\) on Instagram/)[1]
-        permalink = clean_url.gsub("/#{author_name}/", "/")
+        oembed = get_oembed
+        raise "No oEmbed data found. Ensure 'facebook_app_access_token' is valid" if oembed.data.empty?
+
+        permalink = clean_url.gsub("/#{oembed.author_name}/", "/")
 
         { link: permalink,
-          title: "@#{author_name}",
-          image: "#{permalink}/media/?size=l",
-          description: Onebox::Helpers.truncate(og.title, 250)
+          title: "@#{oembed.author_name}",
+          image: oembed.thumbnail_url,
+          description: Onebox::Helpers.truncate(oembed.title, 250),
         }
+
       end
 
       protected
 
+      def access_token
+        (options[:facebook_app_access_token] || Onebox.options.facebook_app_access_token).to_s
+      end
+
       def get_oembed_url
-        oembed_url = "https://api.instagram.com/oembed/?url=#{clean_url}"
+        if access_token != ''
+          oembed_url = "https://graph.facebook.com/v9.0/instagram_oembed?url=#{clean_url}&access_token=#{access_token}"
+        else
+          # The following is officially deprecated by Instagram, but works in some limited circumstances.
+          oembed_url = "https://api.instagram.com/oembed/?url=#{clean_url}"
+        end
       end
     end
   end

data/lib/onebox/engine/standard_embed.rb

@@ -91,7 +91,7 @@ module Onebox
         html_doc.css('meta').each do |m|
           if (m["property"] && m["property"][/^twitter:(.+)$/i]) || (m["name"] && m["name"][/^twitter:(.+)$/i])
             value = (m["content"] || m["value"]).to_s
-            twitter[$1.tr('-:' , '_').to_sym] ||= value unless Onebox::Helpers::blank?(value)
+            twitter[$1.tr('-:' , '_').to_sym] ||= value unless (Onebox::Helpers::blank?(value) || value == "0 minutes")
           end
         end
 

data/lib/onebox/helpers.rb

@@ -27,7 +27,7 @@ module Onebox
     def self.fetch_html_doc(url, headers = nil)
       response = (fetch_response(url, nil, nil, headers) rescue nil)
       doc = Nokogiri::HTML(response)
-      uri = URI(url)
+      uri = Addressable::URI.parse(url)
 
       ignore_canonical_tag = doc.at('meta[property="og:ignore_canonical"]')
       should_ignore_canonical = IGNORE_CANONICAL_DOMAINS.map { |hostname| uri.hostname.match?(hostname) }.any?
@@ -35,8 +35,9 @@ module Onebox
       unless (ignore_canonical_tag && ignore_canonical_tag['content'].to_s == 'true') || should_ignore_canonical
         # prefer canonical link
         canonical_link = doc.at('//link[@rel="canonical"]/@href')
-        if canonical_link && "#{URI(canonical_link).host}#{URI(canonical_link).path}" != "#{uri.host}#{uri.path}"
-          response = (fetch_response(canonical_link, nil, nil, headers) rescue nil)
+        canonical_uri = Addressable::URI.parse(canonical_link)
+        if canonical_link && "#{canonical_uri.host}#{canonical_uri.path}" != "#{uri.host}#{uri.path}"
+          response = (fetch_response(canonical_uri.to_s, nil, nil, headers) rescue nil)
           doc = Nokogiri::HTML(response) if response
         end
       end
@@ -156,6 +157,7 @@ module Onebox
     end
 
     def self.truncate(string, length = 50)
+      return string if string.nil?
       string.size > length ? string[0...(string.rindex(" ", length) || length)] + "..." : string
     end
 

data/lib/onebox/preview.rb

@@ -31,6 +31,16 @@ module Onebox
       ""
     end
 
+    def errors
+      return {} unless engine
+      engine.errors
+    end
+
+    def data
+      return {} unless engine
+      engine.data
+    end
+
     def options
       OpenStruct.new(@options)
     end
@@ -45,7 +55,7 @@ module Onebox
       return "" unless html
 
       if @options[:max_width]
-        doc = Nokogiri::HTML::fragment(html)
+        doc = Nokogiri::HTML5::fragment(html)
         if doc
           doc.css('[width]').each do |e|
             width = e['width'].to_i

data/lib/onebox/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Onebox
-  VERSION = "2.1.4"
+  VERSION = "2.1.9"
 end

data/onebox.gemspec

@@ -33,7 +33,7 @@ Gem::Specification.new do |spec|
   spec.add_development_dependency 'fakeweb', '~> 1.3'
   spec.add_development_dependency 'pry', '~> 0.10'
   spec.add_development_dependency 'mocha', '~> 1.1'
-  spec.add_development_dependency 'rubocop-discourse', '~> 2.1.2'
+  spec.add_development_dependency 'rubocop-discourse', '~> 2.4.0'
   spec.add_development_dependency 'twitter', '~> 4.8'
   spec.add_development_dependency 'guard-rspec', '~> 4.2.8'
   spec.add_development_dependency 'sinatra', '~> 1.4'

data/templates/githubfolder.mustache

@@ -0,0 +1,11 @@
+{{#image}}<img src="{{image}}" class="thumbnail"/>{{/image}}
+
+<h3><a href='{{link}}' target="_blank" rel="noopener">{{title}}</a></h3>
+
+{{#path}}
+<p><a href='{{link}}' target="_blank" rel="noopener">{{path}}</a></p>
+{{/path}}
+
+{{#description}}
+  <p><span class="label1">{{description}}</span></p>
+{{/description}}

data/templates/instagram.mustache

@@ -8,4 +8,6 @@
   </div>
 {{/image}}
 
-<div class="instagram-description">{{description}}</div>
+{{#description}}
+  <div class="instagram-description">{{description}}</div>
+{{/description}}

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: onebox
 version: !ruby/object:Gem::Version
-  version: 2.1.4
+  version: 2.1.9
 platform: ruby
 authors:
 - Joanna Zeta
@@ -10,7 +10,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-10-21 00:00:00.000000000 Z
+date: 2020-12-07 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: addressable
@@ -186,14 +186,14 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 2.1.2
+        version: 2.4.0
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 2.1.2
+        version: 2.4.0
 - !ruby/object:Gem::Dependency
   name: twitter
   requirement: !ruby/object:Gem::Requirement
@@ -316,6 +316,7 @@ files:
 - lib/onebox/engine/giphy_onebox.rb
 - lib/onebox/engine/github_blob_onebox.rb
 - lib/onebox/engine/github_commit_onebox.rb
+- lib/onebox/engine/github_folder_onebox.rb
 - lib/onebox/engine/github_gist_onebox.rb
 - lib/onebox/engine/github_issue_onebox.rb
 - lib/onebox/engine/github_pullrequest_onebox.rb
@@ -383,6 +384,7 @@ files:
 - templates/amazon.mustache
 - templates/githubblob.mustache
 - templates/githubcommit.mustache
+- templates/githubfolder.mustache
 - templates/githubgist.mustache
 - templates/githubissue.mustache
 - templates/githubpullrequest.mustache