[vk] add 'wall-post' extractor (#474 #6378 #8159)

* [vk] Added extractor for VK wall posts
* update
    - fix flake8
    - rename to 'wall-post'
    - remove __init__() / use self.groups
    - simplify 'description' extraction
* add test
* add to docs/supportedsites

---------

Co-authored-by: Mike Fährmann <mike_faehrmann@web.de>

docs/supportedsites.md

@@ -1042,7 +1042,7 @@ Consider all listed sites to potentially be NSFW.
 <tr id="vk" title="vk">
     <td>VK</td>
     <td>https://vk.com/</td>
-    <td>Albums, Photos, Tagged Photos</td>
+    <td>Albums, Photos, Tagged Photos, individual Wall Posts</td>
     <td></td>
 </tr>
 <tr id="vsco" title="vsco">

gallery_dl/extractor/vk.py

@@ -108,6 +108,10 @@ class VkExtractor(Extractor):
             total = payload[1]
             photos = payload[3]
 
+            for i in range(len(photos)):
+                photos[i]["num"] = self.offset + i + 1
+                photos[i]["count"] = total
+
             offset_next = self.offset + len(photos)
             if offset_next >= total:
                 # the last chunk of photos also contains the first few photos
@@ -128,7 +132,7 @@ class VkPhotosExtractor(VkExtractor):
     subcategory = "photos"
     pattern = (BASE_PATTERN + r"/(?:"
                r"(?:albums|photos|id)(-?\d+)"
-               r"|(?!(?:album|tag)-?\d+_?)([^/?#]+))")
+               r"|(?!(?:album|tag|wall)-?\d+_?)([^/?#]+))")
     example = "https://vk.com/id12345"
 
     def __init__(self, match):
@@ -209,3 +213,35 @@ class VkTaggedExtractor(VkExtractor):
 
     def metadata(self):
         return {"user": {"id": self.user_id}}
+
+
+class VkWallPostExtractor(VkExtractor):
+    """Extractor for a vk wall post"""
+    subcategory = "wall-post"
+    directory_fmt = ("{category}", "{user[id]}", "wall")
+    filename_fmt = "{wall[id]}_{num}.{extension}"
+    pattern = BASE_PATTERN + r"/wall(-?\d+)_(\d+)"
+    example = "https://vk.com/wall12345_123"
+
+    def photos(self):
+        user_id, wall_id = self.groups
+        return self._pagination(f"wall{user_id}_{wall_id}")
+
+    def metadata(self):
+        user_id, wall_id = self.groups
+
+        url = f"{self.root}/wall{user_id}_{wall_id}"
+        page = self.request(url).text
+        desc = text.unescape(
+            text.extr(page, 'data-testid="post_description">', "</div>") or
+            text.extr(page, 'name="description" content="', '"'))
+
+        return {
+            "user": {
+                "id": user_id,
+            },
+            "wall": {
+                "id": wall_id,
+                "description": desc,
+            },
+        }

scripts/supportedsites.py

@@ -403,6 +403,7 @@ SUBCATEGORY_MAP = {
     },
     "vk": {
         "tagged": "Tagged Photos",
+        "wall-post": "individual Wall Posts",
     },
     "vsco": {
         "spaces": "",

test/results/vk.py

@@ -99,4 +99,31 @@ __tests__ = (
     "#count"   : 44,
 },
 
+{
+    "#url"     : "https://vk.com/wall-213352498_2115",
+    "#class"   : vk.VkWallPostExtractor,
+    "#results" : (
+        "https://sun9-42.userapi.com/s/v1/ig2/53qxcL7M8408L2HNDTHdHz-HXbprXBn1BLbE5HTuj-OsZD4I483jtZb8yMk9Mr4zzfPJhqBIJlAprWVhIqlk4Fn4.jpg?quality=95&as=32x57,48x85,72x128,108x192,160x284,240x427,360x640,480x853,540x960,640x1138,720x1280&from=bu&cs=720x0",
+        "https://sun9-49.userapi.com/s/v1/ig2/FnvT8T3mC2yQWc5yJTOe25Kj864ohqvTgOcTudqrE4sPfCMexS1mzNmgUndgxUbqhht-YmIVKW_edDFtzCLXzf7h.jpg?quality=95&as=32x57,48x85,72x128,108x192,160x284,240x427,360x640,480x853,540x960,640x1138,720x1280&from=bu&cs=720x0",
+        "https://sun9-78.userapi.com/s/v1/ig2/6VB0Cnmdtb9rDNFd5iHv5QJAJ-y-xSVELEoCLlOf_ej2BWVf61G3DSXbnXgmx-QFtQkOOnHIhCLFFLTIFKeVBR5Q.jpg?quality=95&as=32x57,48x85,72x128,108x192,160x284,240x427,360x640,480x853,540x960,640x1138,720x1280&from=bu&cs=720x0",
+        "https://sun9-60.userapi.com/s/v1/ig2/KO5SzdRUHjZRKlHii4oJ4BrTo5nbdyP3CCpf6_RfHhrEIx6jiVPlWH1R--fpoK5-0rigqXuaG68q39m5VQVy6YFo.jpg?quality=95&as=32x57,48x85,72x128,108x192,160x284,240x427,360x640,480x853,540x960,640x1138,720x1280&from=bu&cs=720x0",
+        "https://sun9-33.userapi.com/s/v1/ig2/IAN1ZHmVVtjRj0U7wGAfnMc5Xp83EFFYZAVqNgMKpfthLHOe6wh0bodM_xDwIALvVl4pcZ66Fv3bOROG4sUTwY21.jpg?quality=95&as=32x57,48x85,72x128,108x192,160x284,240x427,360x640,480x853,540x960,640x1138,720x1280&from=bu&cs=720x0",
+        "https://sun9-44.userapi.com/s/v1/ig2/RLzDGnlmu7C0sLh2YI2R4L9RBgZ061QLOsxogjEtC0cBZJ9HvhNwe1V16QX0tNLkTOLELAp8JDHwOo6dMvoWydeh.jpg?quality=95&as=32x57,48x85,72x128,108x192,160x284,240x427,360x640,480x853,540x960,640x1138,720x1280&from=bu&cs=720x0",
+    ),
+
+    "id"           : r"re:^\d+$",
+    "width"        : 720,
+    "height"       : 1280,
+    "count"        : 6,
+    "num"          : range(1, 6),
+    "likes"        : int,
+    "user"         : {
+        "id": "-213352498",
+    },
+    "wall"         : {
+        "description": "🎄 Обновляем не только аватарки, но и обои на телефоне",
+        "id"         : "2115",
+    },
+},
+
 )