Add google seraching

app/assets/stylesheets/search_results.scss

@@ -4,33 +4,38 @@
 @use "typography";
 
 .search-results {
-	@extend .masonry;
+    @extend .masonry;
 
-	flex-flow: column-reverse wrap-reverse;
+    flex-flow: column-reverse wrap-reverse;
 
-	margin-bottom: clamp(1rem, 5vw, 5rem);
-	&:last-child {
-		margin-bottom: 0;
-	}
+    margin-bottom: clamp(1rem, 5vw, 5rem);
+    &:last-child {
+        margin-bottom: 0;
+    }
+
+    &.media {
+        flex-direction: column;
+        margin-right: 1rem;
+    }
 }
 
 .search-results__title {
-	@extend .h4;
+    @extend .h4;
 }
 
 .search-result {
-	@extend .box;
-	text-decoration: none;
-	width: 100%;
-	transition-property: transform;
-	transition-duration: 100ms;
-
-	&:hover {
-		transform: scale(1.03);
-	}
+    @extend .box;
+    text-decoration: none;
+    width: 100%;
+    transition-property: transform;
+    transition-duration: 100ms;
+
+    &:hover {
+        transform: scale(1.03);
+    }
 }
 
 .search-input-media {
-	max-width: shared.$screen-xs-ceiling;
-	margin: 0 auto;
+    max-width: shared.$screen-xs-ceiling;
+    margin: 0 auto;
 }

app/controllers/media_vault/search_controller.rb

@@ -6,12 +6,15 @@ class SearchParams < T::Struct
     const :msid, T.nilable(String) # Media Search ID
     const :private, T.nilable(T::Boolean)
   end
+
   class MediaSearchParams < T::Struct
     const :media, ActionDispatch::Http::UploadedFile
     const :q, T.nilable(String)
     const :private, T.nilable(T::Boolean)
   end
 
+  # Search for
+
   sig { void }
   def index
     typed_params = TypedParams[SearchParams].new.extract!(params)
@@ -67,12 +70,19 @@ def search_by_media(private: false)
     end
   end
 
+  def google_image_link
+    google_image_id = params[:google_image_id]
+    google_image = GoogleSearchResult.find(google_image_id)
+
+    send_data google_image.public_image_url
+  end
+
 private
 
   sig { params(id: String, private: T::Boolean).void }
   def search_by_media_search_id(id, private: false)
     @media_search = ImageSearch.find(id)
-    @results = @media_search.run
+    @results, @google_results = @media_search.run
     @post_results = @results.filter_map { |result|
       result.has_key?(:image) ? result[:image] : (
         result.has_key?(:video) ? result[:video] : nil

app/models/google_search_result.rb

@@ -0,0 +1,28 @@
+class GoogleSearchResult < ApplicationRecord
+  def public_image_url
+    Rails.cache.fetch("#{cache_key_with_version}/#{self.id}", expires_in: 24.hours) do
+      hydra = Typhoeus::Hydra.hydra
+      request = Typhoeus::Request.new(self.image_url, followlocation: true)
+      hydra.queue(request)
+      hydra.run
+
+      request.response.body
+    end
+  end
+
+  def self.from_api_response(api_response)
+    response = self.new
+    response.text = api_response["text"]
+    response.claimant = api_response["claimant"]
+    response.claim_date = DateTime.parse(api_response["claimDate"]) unless api_response["claimDate"].nil?
+    response.url = api_response["claimReview"].first["url"]
+    response.review_date = DateTime.parse(api_response["claimReview"].first["reviewDate"]) unless api_response["claimReview"].first["reviewDate"].nil?
+    response.rating = api_response["claimReview"].first["textualRating"]
+    response.title = api_response["claimReview"].first["title"]
+    response.language_code = api_response["claimReview"].first["languageCode"]
+    response.publisher_name = api_response["claimReview"].first["publisher"]["name"]
+    response.publisher_site = api_response["claimReview"].first["publisher"]["site"]
+
+    response
+  end
+end

app/models/image_search.rb

@@ -7,6 +7,38 @@
 # may speed this up since we can throw a huge instance at a very short-lived process?
 
 class ImageSearch < ApplicationRecord
+  # class GoogleSearchResult
+  #   # TODO: Move this to a active record model instead
+  #   attr_reader :text, :claimant, :claim_date, :url, :review_date, :rating, :language_code, :publisher_name, :publisher_site, :title
+  #   attr_accessor :image_url
+
+  #   def initialize(google_object)
+  #     @text = google_object["text"]
+  #     @claimant = google_object["claimant"]
+  #     @claim_date = DateTime.parse(google_object["claimDate"]) unless google_object["claimDate"].nil?
+  #     @url = google_object["claimReview"].first["url"]
+  #     @review_date = DateTime.parse(google_object["claimReview"].first["reviewDate"]) unless google_object["claimReview"].first["reviewDate"].nil?
+  #     @rating = google_object["claimReview"].first["textualRating"]
+  #     @title = google_object["claimReview"].first["title"]
+  #     @language_code = google_object["claimReview"].first["languageCode"]
+  #     @publisher_name = google_object["claimReview"].first["publisher"]["name"]
+  #     @publisher_site = google_object["claimReview"].first["publisher"]["site"]
+  #     @image_url = nil
+  #   end
+
+  #   def image_file
+  #     Rails.cache.fetch("#{@url}/#{@image_url}", expires_in: 30.seconds) do
+  #       hydra = Typhoeus::Hydra.hydra
+  #       request = Typhoeus::Request.new(@image_url, followlocation: true) }
+  #       hydra.queue(request)
+  #       hydra.run
+
+  #       request.response.body
+  #     end
+  #   end
+  # end
+
+
   include ImageUploader::Attachment(:image) # adds an `image` virtual attribute
   include VideoUploader::Attachment(:video)
 
@@ -20,6 +52,9 @@ class ImageSearch < ApplicationRecord
     self.video_derivatives! unless self.video.nil?
   end
 
+  before_save do
+  end
+
   # Implemented for Dhashable to find so that only certain number of frames of video are hashed
   # when searching
   sig { returns(Integer) }
@@ -64,7 +99,8 @@ def run
         { image: archive_item, distance: archive_items_raw[index]["levenshtein"] }
       end
 
-      images
+      archive_final_items = images
+      google_items = search_google_by_media(self.image_url)
     else
       # For videos we have to loop
       archive_items_raw = self.dhashes.flat_map do |dhash|
@@ -81,9 +117,11 @@ def run
       # Videos are now sorted by distance, but we want to only keep the shortest distance
       # Probably can get this into the sql above
       videos.uniq! { |video_hash| video_hash[:video] }
-
-      videos
+      archive_final_items = videos
+      google_items = search_google_by_media(self.video_derivatives[:preview].url)
     end
+
+    [archive_final_items, google_items]
   end
 
 
@@ -110,4 +148,40 @@ def raw_sql(dhash)
 
     sql
   end
+
+private
+
+  sig { params(image_url: String).returns(T::Array[GoogleSearchResult]) }
+  def search_google_by_media(image_url)
+    request = build_google_search_request(image_url, image: true)
+    response = request.run
+    response_body = JSON.parse(response.response_body)
+
+    results = response_body["results"].map { |google_object| GoogleSearchResult.from_api_response(google_object["claim"]) }
+
+    hydra = Typhoeus::Hydra.hydra
+    requests = results.map { |result| Typhoeus::Request.new(result.url, followlocation: true) }
+    requests.each { |request| hydra.queue(request) }
+    hydra.run
+
+    requests.each_with_index.map do |request, index|
+      image_elements = Nokogiri::HTML.parse(request.response.body).xpath('//meta[@property="og:image"]')
+      results[index].update(image_url: image_elements.first.values[1])
+    end
+
+    results
+  end
+
+  sig { params(query: String, image: T::Boolean).returns(Typhoeus::Request) }
+  def build_google_search_request(query, image: false)
+    url = "https://factchecktools.googleapis.com/v1alpha1/claims:imageSearch"
+    params = { imageUri: query, key: Figaro.env.FACTCHECK_TOOLS_API_KEY }
+
+    Typhoeus::Request.new(
+      url,
+      method: :get,
+      params: params,
+      headers: { Accept: "text/html" }
+    )
+  end
 end

app/views/media_vault/search/_google_search_result.html.erb

@@ -0,0 +1,23 @@
+<div
+    class="archive-item"
+    data-publishing-platform="<%# publishing_platform_shortname %>"
+    data-controller="media-vault--archive"
+    data-media-vault--archive-caption-collapse-mode-value="<%# caption_is_collapsable ? 'collapsed' : 'static' %>"
+>
+  <div class="archive-item__metadatum__label archive-item__inner flex gap-4 flex-row">
+    <img class="w-24 h-24 rounded-full" src="<%= media_vault_google_image_link_url(search_result.id) %>" alt="Rounded avatar">
+    <div class="flex flex-col gap-2">
+        <div class="flex flex-row gap-4">
+            <div>
+                <div class="font-semibold">Claim By <%= search_result.claimant.nil? ? "A Post" : search_result.claimant %>:</div>
+                <div><%= search_result.text %></div>
+            </div>
+        </div>
+
+        <div class="font-semibold"><%= search_result.publisher_name %> rating: <span><%= search_result.rating %></span></div>
+        <div><%= link_to search_result.title, search_result.url %></div>
+        <% date_to_show = search_result.review_date.nil? ? search_result.claim_date : search_result.review_date %>
+        <div style="color: #777;" class="text-sm"><span class="font-semibold">Reviewed:</span> <%=  date_to_show.strftime("%-d %b %Y") %></div>
+    </div>
+  </div>
+</div>

app/views/media_vault/search/index.html.erb

@@ -15,32 +15,59 @@
 		<% end %>
 	</div>
 
-	<% if (@results || []).any? %>
-		<% if @author_results.try(:any?) %>
-			<h2 class="search-results__title">
-				<%= pluralize(@author_results.length, "matching author") %> found<% if @myvault %> in your MyVault&nbsp;<span style="font-size: 1.5rem; vertical-align: text-top; color: gray;" class="super">Beta</span><% end %>
-			</h2>
-			<div class="search-results search-results--authors">
-				<% @author_results.each do |author| %>
-					<%= link_to media_vault_author_path(author, platform: author.platform), class: "search-result" do %>
-						<%= render partial: "media_vault/authors/author_info/#{underscore_author_model(author)}", locals: { author: author } %>
-					<% end %>
-				<% end %>
+	<% if defined?(@media_search) %>
+	    <div class="grid grid-cols-2 divide-neutral-200 divide-x-2">
+			<div>
+            <% if @post_results.any? %>
+          		<h2 class="search-results__title">
+         			<%= pluralize(@post_results.length, "matching MediaVault post") %> found<% if @myvault %> in your MyVault&nbsp;<span style="font-size: 1.5rem; vertical-align: text-top; color: gray;" class="super">Beta</span><% end %>:
+          		</h2>
+          		<div class="search-results search-results--posts archive-items archive-items--boxed-items archive-items--masonry media">
+         			<% @post_results.each do |post| %>
+            				<%= render partial: "media_vault/media/archive_item", locals: { preview: true, **post.normalized_attrs_for_views } %>
+         			<% end %>
+          		</div>
+           	<% end %>
+	        </div>
+			<div class="pl-4">
+                <h2 class="search-results__title">
+         			<%= pluralize(@post_results.length, "matching Fact Check Explorer result") %> found:
+          		</h2>
+          		<div class="search-results search-results--posts archive-items archive-items--boxed-items archive-items--masonry media">
+         			<% @google_results.each do |post| %>
+          				<%= render partial: "media_vault/search/google_search_result", locals: { search_result: post } %>
+         			<% end %>
+          		</div>
 			</div>
-		<% end %>
+	    </div>
+	<% else %>
+    	<% if (@results || []).any? %>
+    		<% if @author_results.try(:any?) %>
+    			<h2 class="search-results__title">
+    				<%= pluralize(@author_results.length, "matching author") %> found<% if @myvault %> in your MyVault&nbsp;<span style="font-size: 1.5rem; vertical-align: text-top; color: gray;" class="super">Beta</span><% end %>
+    			</h2>
+    			<div class="search-results search-results--authors">
+    				<% @author_results.each do |author| %>
+    					<%= link_to media_vault_author_path(author, platform: author.platform), class: "search-result" do %>
+    						<%= render partial: "media_vault/authors/author_info/#{underscore_author_model(author)}", locals: { author: author } %>
+    					<% end %>
+    				<% end %>
+    			</div>
+    		<% end %>
 
-		<% if @post_results.any? %>
-			<h2 class="search-results__title">
-				<%= pluralize(@post_results.length, "matching post") %> found<% if @myvault %> in your MyVault&nbsp;<span style="font-size: 1.5rem; vertical-align: text-top; color: gray;" class="super">Beta</span><% end %>:
-			</h2>
-			<div class="search-results search-results--posts archive-items archive-items--boxed-items archive-items--masonry">
-				<% @post_results.each do |post| %>
-					<%= render partial: "media_vault/media/archive_item", locals: { preview: true, **post.normalized_attrs_for_views } %>
-				<% end %>
-			</div>
-		<% end %>
+    		<% if @post_results.any? %>
+    			<h2 class="search-results__title">
+    				<%= pluralize(@post_results.length, "matching post") %> found<% if @myvault %> in your MyVault&nbsp;<span style="font-size: 1.5rem; vertical-align: text-top; color: gray;" class="super">Beta</span><% end %>:
+    			</h2>
+    			<div class="search-results search-results--posts archive-items archive-items--boxed-items archive-items--masonry">
+    				<% @post_results.each do |post| %>
+    					<%= render partial: "media_vault/media/archive_item", locals: { preview: true, **post.normalized_attrs_for_views } %>
+    				<% end %>
+    			</div>
+    		<% end %>
 
-	<% else %>
-		<h1 class="text--center">No results found<%= " in your MyVault" if @myvault %>.</h1>
-	<% end %>
+    	<% else %>
+    		<h1 class="text--center">No results found<%= " in your MyVault" if @myvault %>.</h1>
+    	<% end %>
+    <% end %>
 </div>

config/initializers/figaro.rb

@@ -42,3 +42,4 @@
 
 Figaro.require_keys("OLLAMA_URL")
 Figaro.require_keys("OLLAMA_PASSWORD")
+Figaro.require_keys("FACTCHECK_TOOLS_API_KEY")

config/initializers/shrine.rb

@@ -53,3 +53,4 @@ def make_shrine_storage
 Shrine.plugin :keep_files if ENV["STAGING"]
 Shrine.plugin :remote_url, max_size: 20 * 1024 * 1024 # 20MB
 Shrine.plugin :download_endpoint, prefix: "media/vault", host: "https://#{Figaro.env.PUBLIC_LINK_HOST}"
+Shrine.plugin :remote_url

config/routes.rb

@@ -115,6 +115,7 @@
 
       get "search", to: "search#index", as: "search"
       post "search_by_media", to: "search#search_by_media", as: "search_by_media"
+      get "search_image/:google_image_id", to: "search#google_image_link", as: "google_image_link"
 
       get "authors/:platform/:id", to: "authors#show", as: "author"
       resources :media, only: [:show, :destroy]

db/migrate/20240928173524_create_google_search_results.rb

@@ -0,0 +1,19 @@
+class CreateGoogleSearchResults < ActiveRecord::Migration[7.0]
+  def change
+    create_table :google_search_results, id: :uuid do |t|
+      t.text      :text
+      t.string    :claimant
+      t.datetime  :claim_date
+      t.string    :url
+      t.datetime  :review_date
+      t.string    :rating
+      t.string    :title
+      t.string    :language_code
+      t.string    :publisher_name
+      t.string    :publisher_site
+      t.string    :image_url
+
+      t.timestamps
+    end
+  end
+end