OEmbed support for PreviewCard (#2337)

* OEmbed support for PreviewCard

* Improve ProviderDiscovery code failure treatment

* Do not crawl links if there is a content warning, since those
don't display a link card anyway

* Reset db schema

* Fresh migrate

* Fix rubocop style issues
Fix #1681 - return existing access token when applicable instead of creating new

* Fix test

* Extract http client to helper

* Improve oembed controller

app/assets/javascripts/components/actions/cards.jsx

@@ -13,7 +13,7 @@ export function fetchStatusCard(id) {
     dispatch(fetchStatusCardRequest(id));
 
     api(getState).get(`/api/v1/statuses/${id}/card`).then(response => {
-      if (!response.data.url || !response.data.title || !response.data.description) {
+      if (!response.data.url) {
         return;
       }
 

app/assets/javascripts/components/features/status/components/card.jsx

@@ -14,14 +14,11 @@ const getHostname = url => {
 
 class Card extends React.PureComponent {
 
-  render () {
+  renderLink () {
     const { card } = this.props;
 
-    if (card === null) {
-      return null;
-    }
-
-    let image = '';
+    let image    = '';
+    let provider = card.get('provider_name');
 
     if (card.get('image')) {
       image = (
@@ -31,18 +28,64 @@ class Card extends React.PureComponent {
       );
     }
 
+    if (provider.length < 1) {
+      provider = getHostname(card.get('url'))
+    }
+
     return (
       <a href={card.get('url')} className='status-card' target='_blank' rel='noopener'>
         {image}
 
         <div className='status-card__content'>
           <strong className='status-card__title' title={card.get('title')}>{card.get('title')}</strong>
-          <p className='status-card__description'>{card.get('description').substring(0, 50)}</p>
-          <span className='status-card__host' style={hostStyle}>{getHostname(card.get('url'))}</span>
+          <p className='status-card__description'>{(card.get('description') || '').substring(0, 50)}</p>
+          <span className='status-card__host' style={hostStyle}>{provider}</span>
         </div>
       </a>
     );
   }
+
+  renderPhoto () {
+    const { card } = this.props;
+
+    return (
+      <a href={card.get('url')} className='status-card-photo' target='_blank' rel='noopener'>
+        <img src={card.get('url')} alt={card.get('title')} width={card.get('width')} height={card.get('height')} />
+      </a>
+    );
+  }
+
+  renderVideo () {
+    const { card } = this.props;
+    const content  = { __html: card.get('html') };
+
+    return (
+      <div
+        className='status-card-video'
+        dangerouslySetInnerHTML={content}
+      />
+    );
+  }
+
+  render () {
+    const { card } = this.props;
+
+    if (card === null) {
+      return null;
+    }
+
+    switch(card.get('type')) {
+    case 'link':
+      return this.renderLink();
+    case 'photo':
+      return this.renderPhoto();
+    case 'video':
+      return this.renderVideo();
+    case 'rich':
+    default:
+      return null;
+    }
+  }
 }
 
 Card.propTypes = {

app/assets/stylesheets/components.scss

@@ -1734,6 +1734,28 @@ button.icon-button.active i.fa-retweet {
   }
 }
 
+.status-card-video, .status-card-rich, .status-card-photo {
+  margin-top: 14px;
+  overflow: hidden;
+
+  iframe {
+    width: 100%;
+    height: auto;
+  }
+}
+
+.status-card-photo {
+  display: block;
+  text-decoration: none;
+
+  img {
+    display: block;
+    width: 100%;
+    height: auto;
+    margin: 0;
+  }
+}
+
 .status-card__title {
   display: block;
   font-weight: 500;

app/controllers/api/oembed_controller.rb

@@ -14,8 +14,20 @@ class Api::OEmbedController < ApiController
   def stream_entry_from_url(url)
     params = Rails.application.routes.recognize_path(url)
 
-    raise ActiveRecord::RecordNotFound unless params[:controller] == 'stream_entries' && params[:action] == 'show'
+    raise ActiveRecord::RecordNotFound unless recognized_stream_entry_url?(params)
 
-    StreamEntry.find(params[:id])
+    stream_entry(params)
+  end
+
+  def recognized_stream_entry_url?(params)
+    %w(stream_entries statuses).include?(params[:controller]) && params[:action] == 'show'
+  end
+
+  def stream_entry(params)
+    if params[:controller] == 'stream_entries'
+      StreamEntry.find(params[:id])
+    else
+      Status.find(params[:id]).stream_entry
+    end
   end
 end

app/helpers/http_helper.rb

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module HttpHelper
+  USER_AGENT = "#{HTTP::Request::USER_AGENT} (Mastodon/#{Mastodon::VERSION}; +http://#{Rails.configuration.x.local_domain}/)"
+
+  def http_client(options = {})
+    timeout = { write: 10, connect: 10, read: 10 }.merge(options)
+
+    HTTP.headers(user_agent: USER_AGENT)
+        .timeout(:per_operation, timeout)
+        .follow
+  end
+end

app/lib/formatter.rb

@@ -1,13 +1,13 @@
 # frozen_string_literal: true
 
 require 'singleton'
+require_relative './sanitize_config'
 
 class Formatter
   include Singleton
   include RoutingHelper
 
   include ActionView::Helpers::TextHelper
-  include ActionView::Helpers::SanitizeHelper
 
   def format(status)
     return reformat(status.content) unless status.local?
@@ -23,7 +23,7 @@ class Formatter
   end
 
   def reformat(html)
-    sanitize(html, tags: %w(a br p span), attributes: %w(href rel class))
+    sanitize(html, Sanitize::Config::MASTODON_STRICT)
   end
 
   def plaintext(status)
@@ -43,6 +43,10 @@ class Formatter
     html.html_safe # rubocop:disable Rails/OutputSafety
   end
 
+  def sanitize(html, config)
+    Sanitize.fragment(html, config)
+  end
+
   private
 
   def encode(html)

app/lib/provider_discovery.rb

@@ -0,0 +1,36 @@
+# frozen_string_literal: true
+
+class ProviderDiscovery < OEmbed::ProviderDiscovery
+  include HttpHelper
+
+  class << self
+    def discover_provider(url, options = {})
+      res    = http_client.get(url)
+      format = options[:format]
+
+      raise OEmbed::NotFound, url if res.code != 200 || res.mime_type != 'text/html'
+
+      html = Nokogiri::HTML(res.to_s)
+
+      if format.nil? || format == :json
+        provider_endpoint ||= html.at_xpath('//link[@type="application/json+oembed"]')&.attribute('href')&.value
+        format ||= :json if provider_endpoint
+      end
+
+      if format.nil? || format == :xml
+        provider_endpoint ||= html.at_xpath('//link[@type="application/xml+oembed"]')&.attribute('href')&.value
+        format ||= :xml if provider_endpoint
+      end
+
+      begin
+        provider_endpoint = Addressable::URI.parse(provider_endpoint)
+        provider_endpoint.query = nil
+        provider_endpoint = provider_endpoint.to_s
+      rescue Addressable::URI::InvalidURIError
+        raise OEmbed::NotFound, url
+      end
+
+      OEmbed::Provider.new(provider_endpoint, format || OEmbed::Formatter.default)
+    end
+  end
+end

app/lib/sanitize_config.rb

@@ -0,0 +1,42 @@
+# frozen_string_literal: true
+
+class Sanitize
+  module Config
+    HTTP_PROTOCOLS ||= ['http', 'https', :relative].freeze
+
+    MASTODON_STRICT ||= freeze_config(
+      elements: %w(p br span a),
+
+      attributes: {
+        'a'    => %w(href),
+        'span' => %w(class),
+      },
+
+      protocols: {
+        'a' => { 'href' => HTTP_PROTOCOLS },
+      }
+    )
+
+    MASTODON_OEMBED ||= freeze_config merge(
+      RELAXED,
+      elements: RELAXED[:elements] + %w(audio embed iframe source video),
+
+      attributes: merge(
+        RELAXED[:attributes],
+        'audio'  => %w(controls),
+        'embed'  => %w(height src type width),
+        'iframe' => %w(allowfullscreen frameborder height scrolling src width),
+        'source' => %w(src type),
+        'video'  => %w(controls height loop width),
+        'div'    => [:data]
+      ),
+
+      protocols: merge(
+        RELAXED[:protocols],
+        'embed'  => { 'src' => HTTP_PROTOCOLS },
+        'iframe' => { 'src' => HTTP_PROTOCOLS },
+        'source' => { 'src' => HTTP_PROTOCOLS }
+      )
+    )
+  end
+end

app/models/preview_card.rb

@@ -3,6 +3,10 @@
 class PreviewCard < ApplicationRecord
   IMAGE_MIME_TYPES = ['image/jpeg', 'image/png', 'image/gif'].freeze
 
+  self.inheritance_column = false
+
+  enum type: [:link, :photo, :video, :rich]
+
   belongs_to :status
 
   has_attached_file :image, styles: { original: '120x120#' }, convert_options: { all: '-quality 80 -strip' }

app/services/fetch_atom_service.rb

@@ -1,6 +1,8 @@
 # frozen_string_literal: true
 
 class FetchAtomService < BaseService
+  include HttpHelper
+
   def call(url)
     return if url.blank?
 
@@ -45,8 +47,4 @@ class FetchAtomService < BaseService
   def fetch(url)
     http_client.get(url).to_s
   end
-
-  def http_client
-    HTTP.timeout(:per_operation, write: 10, connect: 10, read: 10).follow
-  end
 end

app/services/fetch_link_card_service.rb

@@ -1,8 +1,9 @@
 # frozen_string_literal: true
 
 class FetchLinkCardService < BaseService
+  include HttpHelper
+
   URL_PATTERN = %r{https?://\S+}
-  USER_AGENT = "#{HTTP::Request::USER_AGENT} (Mastodon/#{Mastodon::VERSION}; +http://#{Rails.configuration.x.local_domain}/)"
 
   def call(status)
     # Get first http/https URL that isn't local
@@ -10,13 +11,53 @@ class FetchLinkCardService < BaseService
 
     return if url.nil?
 
+    card = PreviewCard.where(status: status).first_or_initialize(status: status, url: url)
+    attempt_opengraph(card, url) unless attempt_oembed(card, url)
+  end
+
+  private
+
+  def attempt_oembed(card, url)
+    response = OEmbed::Providers.get(url)
+
+    card.type          = response.type
+    card.title         = response.respond_to?(:title)         ? response.title         : ''
+    card.author_name   = response.respond_to?(:author_name)   ? response.author_name   : ''
+    card.author_url    = response.respond_to?(:author_url)    ? response.author_url    : ''
+    card.provider_name = response.respond_to?(:provider_name) ? response.provider_name : ''
+    card.provider_url  = response.respond_to?(:provider_url)  ? response.provider_url  : ''
+    card.width         = 0
+    card.height        = 0
+
+    case card.type
+    when 'link'
+      card.image = URI.parse(response.thumbnail_url) if response.respond_to?(:thumbnail_url)
+    when 'photo'
+      card.url    = response.url
+      card.width  = response.width.presence  || 0
+      card.height = response.height.presence || 0
+    when 'video'
+      card.width  = response.width.presence  || 0
+      card.height = response.height.presence || 0
+      card.html   = Formatter.instance.sanitize(response.html, Sanitize::Config::MASTODON_OEMBED)
+    when 'rich'
+      # Most providers rely on <script> tags, which is a no-no
+      return false
+    end
+
+    card.save_with_optional_image!
+  rescue OEmbed::NotFound
+    false
+  end
+
+  def attempt_opengraph(card, url)
     response = http_client.get(url)
 
     return if response.code != 200 || response.mime_type != 'text/html'
 
     page = Nokogiri::HTML(response.to_s)
-    card = PreviewCard.where(status: status).first_or_initialize(status: status, url: url)
 
+    card.type        = :link
     card.title       = meta_property(page, 'og:title') || page.at_xpath('//title')&.content
     card.description = meta_property(page, 'og:description') || meta_property(page, 'description')
     card.image       = URI.parse(Addressable::URI.parse(meta_property(page, 'og:image')).normalize.to_s) if meta_property(page, 'og:image')
@@ -26,12 +67,6 @@ class FetchLinkCardService < BaseService
     card.save_with_optional_image!
   end
 
-  private
-
-  def http_client
-    HTTP.headers(user_agent: USER_AGENT).timeout(:per_operation, write: 10, connect: 10, read: 10).follow
-  end
-
   def meta_property(html, property)
     html.at_xpath("//meta[@property=\"#{property}\"]")&.attribute('content')&.value || html.at_xpath("//meta[@name=\"#{property}\"]")&.attribute('content')&.value
   end

app/services/follow_remote_account_service.rb

@@ -2,6 +2,7 @@
 
 class FollowRemoteAccountService < BaseService
   include OStatus2::MagicKey
+  include HttpHelper
 
   DFRN_NS = 'http://purl.org/macgirvin/dfrn/1.0'
 
@@ -73,7 +74,7 @@ class FollowRemoteAccountService < BaseService
   end
 
   def get_feed(url)
-    response = http_client.get(Addressable::URI.parse(url).normalize)
+    response = http_client(write: 20, connect: 20, read: 50).get(Addressable::URI.parse(url).normalize)
     [response.to_s, Nokogiri::XML(response)]
   end
 
@@ -98,8 +99,4 @@ class FollowRemoteAccountService < BaseService
   def get_profile(body, account)
     RemoteProfileUpdateWorker.perform_async(account.id, body.force_encoding('UTF-8'), false)
   end
-
-  def http_client
-    HTTP.timeout(:per_operation, write: 20, connect: 20, read: 50)
-  end
 end

app/services/post_status_service.rb

@@ -34,7 +34,7 @@ class PostStatusService < BaseService
     process_mentions_service.call(status)
     process_hashtags_service.call(status)
 
-    LinkCrawlWorker.perform_async(status.id)
+    LinkCrawlWorker.perform_async(status.id) unless status.spoiler_text.present?
     DistributionWorker.perform_async(status.id)
     Pubsubhubbub::DistributionWorker.perform_async(status.stream_entry.id)
 

app/views/api/v1/statuses/card.rabl

@@ -1,5 +1,7 @@
 object @card
 
-attributes :url, :title, :description
+attributes :url, :title, :description, :type,
+           :author_name, :author_url, :provider_name,
+           :provider_url, :html, :width, :height
 
 node(:image) { |card| card.image? ? full_asset_url(card.image.url(:original)) : nil }