tinami: get the full image.

Support grabbing the full image for Tinami uploads, rather than the sample.

Getting the full image requires making a request like this:

    curl -X POST \
    -H 'Referer: https://www.tinami.com/' \
    -H 'Content-Type: application/x-www-form-urlencoded' \
    -H 'Cookie: Tinami2SESSID=<redacted>;' \
    --data-raw 'action_view_original=true&cont_id=1087268&ethna_csrf=<redacted>' \
    https://www.tinami.com/view/1087268

Then scraping the <img> tag from the resulting HTML page.

If the post has multiple images, then we need to scrape and pass the
`sub_id` of the image too.

Fixes #2818.

.github/workflows/test.yaml

@@ -56,6 +56,7 @@ jobs:
       DANBOORU_BARAAG_CLIENT_ID: ${{ secrets.DANBOORU_BARAAG_CLIENT_ID }}
       DANBOORU_BARAAG_CLIENT_SECRET: ${{ secrets.DANBOORU_BARAAG_CLIENT_SECRET }}
       DANBOORU_FANTIA_SESSION_ID: ${{ secrets.DANBOORU_FANTIA_SESSION_ID }}
+      DANBOORU_TINAMI_SESSION_ID: ${{ secrets.DANBOORU_TINAMI_SESSION_ID }}
       DANBOORU_DISCORD_WEBHOOK_ID: ${{ secrets.DANBOORU_DISCORD_WEBHOOK_ID }}
       DANBOORU_DISCORD_WEBHOOK_SECRET: ${{ secrets.DANBOORU_DISCORD_WEBHOOK_SECRET }}
       DANBOORU_RAKISMET_KEY: ${{ secrets.DANBOORU_RAKISMET_KEY }}

app/logical/sources/strategies/tinami.rb

@@ -12,18 +12,35 @@ module Sources
       def image_urls
         if parsed_url.image_url?
           [url]
+
+        # http://www.tinami.com/view/1087268 (single image)
+        elsif page&.css("img.captify")&.size.to_i == 1
+          [full_image_url]
+
+        # http://www.tinami.com/view/1087270 (multiple images)
+        elsif image_sub_ids.present?
+          image_sub_ids.map { |sub_id| full_image_url(sub_id) }
+
+        # http://www.tinami.com/view/1087271 (multiple images)
+        elsif nv_body_image_urls.present?
+          nv_body_image_urls
+
+        # http://www.tinami.com/view/1087267 (no images, text only)
         else
-          # Page type 1: http://www.tinami.com/view/1087268
+          []
-          # Page type 2: http://www.tinami.com/view/1087271
-          # Page type 3: http://www.tinami.com/view/1087270
-          # Page type 4: http://www.tinami.com/view/1087267 (no images, text only)
-          page&.css(".viewbody img.captify, .viewbody .nv_body img").to_a.map do |img|
-            # img[:src] == "//img.tinami.com/illust2/img/619/6234b647da609.jpg"
-            "https:#{img[:src]}"
-          end
         end
       end
 
+      def nv_body_image_urls
+        page&.css(".viewbody .nv_body img").to_a.map do |img|
+          "https:#{img[:src]}" # img[:src] == "//img.tinami.com/illust2/img/619/6234b647da609.jpg"
+        end
+      end
+
+      def image_sub_ids
+        page&.css(".viewbody #controller_model .thumbnail_list").to_a.map { |td| td.attr("sub_id") }
+      end
+
       def page_url
         parsed_url.page_url || parsed_referer&.page_url
       end
@@ -59,6 +76,24 @@ module Sources
         Source::URL.parse(url)&.user_id
       end
 
+      def work_id
+        parsed_url.work_id || parsed_referer&.work_id
+      end
+
+      def ethna_csrf
+        page&.at("#open_original_content input[name=ethna_csrf]")&.attr("value")
+      end
+
+      def full_image_url(sub_id = nil)
+        return nil unless work_id.present? && ethna_csrf.present?
+
+        # Note that we have to spoof the Referer here.
+        response = http.post(page_url, form: { action_view_original: true, cont_id: work_id, sub_id: sub_id, ethna_csrf: ethna_csrf })
+        return nil unless response.status == 200
+
+        response.parse.at("body > div > a > img[src^='//img.tinami.com']")&.attr("src")&.prepend("https:")
+      end
+
       def page
         return nil if page_url.blank?
 
@@ -68,7 +103,11 @@ module Sources
         response.parse
       end
 
-      memoize :page, :user_id
+      def http
+        super.cookies(Tinami2SESSID: Danbooru.config.tinami_session_id).use(:spoof_referrer)
+      end
+
+      memoize :http, :page, :user_id, :work_id, :ethna_csrf, :image_urls, :image_sub_ids, :nv_body_image_urls
     end
   end
 end

config/danbooru_default_config.rb

@@ -305,6 +305,11 @@ module Danbooru
       nil
     end
 
+    # Your Tinami "Tinami2SESSID" cookie. Login to Tinami then use the devtools to find the "Tinami2SESSID" cookie.
+    def tinami_session_id
+      nil
+    end
+
     # 1. Register app at https://www.tumblr.com/oauth/register.
     # 2. Copy "OAuth Consumer Key" from https://www.tumblr.com/oauth/apps.
     def tumblr_consumer_key

test/unit/sources/tinami_test.rb

@@ -9,7 +9,7 @@ module Sources
 
           assert_equal("みぐめ", source.artist_name)
           assert_equal("https://www.tinami.com/view/1087268", source.page_url)
-          assert_equal(["https://img.tinami.com/illust2/img/647/6234fe5588e97.jpg"], source.image_urls)
+          assert_equal(["https://img.tinami.com/illust2/img/49/6234fe552348b.jpg"], source.image_urls)
           assert_equal("https://www.tinami.com/creator/profile/66493", source.profile_url)
           assert_equal(%w[横顔 アナログ ボールペン SP], source.tags.map(&:first))
           assert_equal("横顔", source.artist_commentary_title)
@@ -42,13 +42,13 @@ module Sources
           assert_equal("セラ箱", source.artist_name)
           assert_equal("https://www.tinami.com/view/1087270", source.page_url)
           assert_equal(%w[
-            https://img.tinami.com/illust2/img/934/623503bb9891b.jpg
+            https://img.tinami.com/illust2/img/399/623503bb2c686.jpg
-            https://img.tinami.com/illust2/img/398/623503bd481bb.jpg
+            https://img.tinami.com/illust2/img/505/623503bdd064e.jpg
-            https://img.tinami.com/illust2/img/698/623503bec2105.jpg
+            https://img.tinami.com/illust2/img/140/623503bf50d20.jpg
-            https://img.tinami.com/illust2/img/981/623503c029fbf.jpg
+            https://img.tinami.com/illust2/img/986/623503c0940f5.jpg
-            https://img.tinami.com/illust2/img/769/623503c187eab.jpg
+            https://img.tinami.com/illust2/img/954/623503c219ee9.jpg
-            https://img.tinami.com/illust2/img/847/623503c2dd8d6.jpg
+            https://img.tinami.com/illust2/img/655/623503c3646c0.jpg
-            https://img.tinami.com/illust2/img/252/623503c434204.jpg
+            https://img.tinami.com/illust2/img/401/623503c4b8171.jpg
           ], source.image_urls)
           assert_equal("https://www.tinami.com/creator/profile/38168", source.profile_url)
           assert_equal(%w[Re:ゼロから始める異世界生活 レム リゼロ セラ箱 rizero フィギュア リペイント], source.tags.map(&:first))