[pixiv] retrieve 'caption' from AJAX API when empty (#4327, #5191)

gallery_dl/extractor/pixiv.py

@@ -76,6 +76,9 @@ class PixivExtractor(Extractor):
                 detail = self.api.illust_bookmark_detail(work["id"])
                 work["tags_bookmark"] = [tag["name"] for tag in detail["tags"]
                                          if tag["is_registered"]]
+            if self.sanity_workaround and not work.get("caption"):
+                body = self._request_ajax("/illust/" + str(work["id"]))
+                work["caption"] = text.unescape(body["illustComment"])
 
             if transform_tags:
                 transform_tags(work)
@@ -180,7 +183,6 @@ class PixivExtractor(Extractor):
         for key_app, key_ajax in (
             ("title"            , "illustTitle"),
             ("image_urls"       , "urls"),
-            ("caption"          , "illustComment"),
             ("create_date"      , "createDate"),
             ("width"            , "width"),
             ("height"           , "height"),
@@ -212,6 +214,7 @@ class PixivExtractor(Extractor):
                 translated_name = None
             tags.append({"name": name, "translated_name": translated_name})
 
+        work["caption"] = text.unescape(body["illustComment"])
         work["page_count"] = count = body["pageCount"]
         if count == 1:
             return ({"url": url},)

test/results/pixiv.py

@@ -266,6 +266,14 @@ __tests__ = (
     ],
 },
 
+{
+    "#url"     : "https://www.pixiv.net/en/artworks/103983466",
+    "#comment" : "empty 'caption' in App API response (#4327, #5191)",
+    "#class"   : pixiv.PixivWorkExtractor,
+
+    "caption": r"re:Either she doesn't know how to pose or she can't move with that much clothing on her, in any case she's very well dressed for a holiday trip around town. Lots of stuff to see and a perfect day to grab some sweet pastries at the bakery.<br />...",
+},
+
 {
     "#url"     : "https://www.pixiv.net/en/artworks/966412",
     "#category": ("", "pixiv", "work"),