[artstation] support video clips (#2566, #3309, #3911)

- add 'videos' and 'previews' options
- fix 403 errors for video previews

docs/configuration.rst

@@ -1124,6 +1124,26 @@ Description
     Limit the number of posts/projects to download.
 
 
+extractor.artstation.previews
+-----------------------------
+Type
+    ``bool``
+Default
+    ``false``
+Description
+    Download video previews.
+
+
+extractor.artstation.videos
+---------------------------
+Type
+    ``bool``
+Default
+    ``true``
+Description
+    Download video clips.
+
+
 extractor.artstation.search.pro-first
 -------------------------------------
 Type

gallery_dl/extractor/artstation.py

@@ -29,11 +29,13 @@ class ArtstationExtractor(Extractor):
         self.user = match.group(1) or match.group(2)
 
     def items(self):
-        data = self.metadata()
-
-        projects = self.projects()
+        videos = self.config("videos", True)
+        previews = self.config("previews", False)
         external = self.config("external", False)
         max_posts = self.config("max-posts")
+
+        data = self.metadata()
+        projects = self.projects()
         if max_posts:
             projects = itertools.islice(projects, max_posts)
 
@@ -45,13 +47,29 @@ class ArtstationExtractor(Extractor):
                 asset["num"] = num
                 yield Message.Directory, asset
 
-                if adict["has_embedded_player"] and external:
+                if adict["has_embedded_player"]:
                     player = adict["player_embedded"]
                     url = (text.extr(player, 'src="', '"') or
                            text.extr(player, "src='", "'"))
-                    if url and not url.startswith(self.root):
-                        asset["extension"] = None
-                        yield Message.Url, "ytdl:" + url, asset
+                    if url.startswith(self.root):
+                        # video clip hosted on artstation
+                        if videos:
+                            page = self.request(url).text
+                            url = text.extr(page, ' src="', '"')
+                            text.nameext_from_url(url, asset)
+                            yield Message.Url, url, asset
+                    elif url:
+                        # external URL
+                        if external:
+                            asset["extension"] = "mp4"
+                            yield Message.Url, "ytdl:" + url, asset
+                    else:
+                        self.log.debug(player)
+                        self.log.warning(
+                            "Failed to extract embedded player URL (%s)",
+                            adict.get("id"))
+
+                    if not previews:
                         continue
 
                 if adict["has_image"]:
@@ -59,10 +77,11 @@ class ArtstationExtractor(Extractor):
                     text.nameext_from_url(url, asset)
 
                     url = self._no_cache(url)
-                    lhs, _, rhs = url.partition("/large/")
-                    if rhs:
-                        url = lhs + "/4k/" + rhs
-                        asset["_fallback"] = self._image_fallback(lhs, rhs)
+                    if "/video_clips/" not in url:
+                        lhs, _, rhs = url.partition("/large/")
+                        if rhs:
+                            url = lhs + "/4k/" + rhs
+                            asset["_fallback"] = self._image_fallback(lhs, rhs)
 
                     yield Message.Url, url, asset
 

test/results/artstation.py

@@ -71,14 +71,14 @@ __tests__ = (
     "#count"   : 10,
 
     "collection": {
-        "active_projects_count": 3,
         "id"            : 2647023,
         "is_private"    : False,
-        "micro_square_image_url": "https://cdna.artstation.com/p/assets/images/images/005/131/434/micro_square/gaeri-kim-cat-front.jpg?1488720625",
         "name"          : "テスト",
         "projects_count": 3,
-        "small_square_image_url": "https://cdna.artstation.com/p/assets/images/images/005/131/434/small_square/gaeri-kim-cat-front.jpg?1488720625",
         "user_id"       : 697975,
+        "active_projects_count" : 3,
+        "micro_square_image_url": "https://cdna.artstation.com/p/assets/images/images/005/131/434/micro_square/gaeri-kim-cat-front.jpg?1488720625",
+        "small_square_image_url": "https://cdna.artstation.com/p/assets/images/images/005/131/434/small_square/gaeri-kim-cat-front.jpg?1488720625",
     },
     "user": "mikf",
 },
@@ -92,14 +92,14 @@ __tests__ = (
         "https://www.artstation.com/mikf/collections/2647719",
     ),
 
-    "active_projects_count": int,
     "id"            : range(2647023, 2647719),
     "is_private"    : False,
-    "micro_square_image_url": str,
     "name"          : r"re:テスト|empty",
     "projects_count": int,
-    "small_square_image_url": str,
     "user_id"       : 697975,
+    "active_projects_count" : int,
+    "micro_square_image_url": str,
+    "small_square_image_url": str,
 },
 
 {
@@ -156,6 +156,19 @@ __tests__ = (
     "#count"   : 4,
 },
 
+{
+    "#url"     : "https://www.artstation.com/artwork/lR8b5k",
+    "#comment" : "artstation video clips (#2566)",
+    "#category": ("", "artstation", "image"),
+    "#class"   : artstation.ArtstationImageExtractor,
+    "#options" : {"videos": True},
+    "#range"   : "2-3",
+    "#urls"    : (
+        "https://cdn.artstation.com/p/video_sources/000/819/843/infection-4.mp4",
+        "https://cdn.artstation.com/p/video_sources/000/819/725/infection-veinonly-2.mp4",
+    ),
+},
+
 {
     "#url"     : "https://www.artstation.com/artwork/g4WPK",
     "#comment" : "embedded youtube video",