[vk:album] extract more metadata (#8029)

… when possible / allowed by VK

- album[name]
- album[count]
- user[name]
- user[nick]
- user[group]

gallery_dl/extractor/vk.py

@@ -187,17 +187,40 @@ class VkAlbumExtractor(VkExtractor):
     pattern = BASE_PATTERN + r"/album(-?\d+)_(\d+)$"
     example = "https://vk.com/album12345_00"
 
-    def __init__(self, match):
-        VkExtractor.__init__(self, match)
-        self.user_id, self.album_id = match.groups()
-
     def photos(self):
-        return self._pagination(f"album{self.user_id}_{self.album_id}")
+        user_id, album_id = self.groups
+        return self._pagination(f"album{user_id}_{album_id}")
 
     def metadata(self):
+        user_id, album_id = self.groups
+
+        url = f"{self.root}/album{user_id}_{album_id}"
+        page = self.request(url).text
+        desc = text.extr(page, 'name="og:description" value="', '"')
+        try:
+            album_name, user_name, photos = desc.rsplit(" - ", 2)
+        except ValueError:
+            if msg := text.extr(
+                    page, '<div class="message_page_title">Error</div>',
+                    "</div>"):
+                msg = f" ('{text.remove_html(msg)[:-5]}')"
+            self.log.warning("%s_%s: Failed to extract metadata%s",
+                             user_id, album_id, msg)
+            return {"user": {"id": user_id}, "album": {"id": album_id}}
+
         return {
-            "user": {"id": self.user_id},
+            "user": {
-            "album": {"id": self.album_id},
+                "id"   : user_id,
+                "nick" : text.unescape(user_name),
+                "name" : text.unescape(text.extr(
+                    page, 'class="ui_crumb" href="/', '"')),
+                "group": user_id[0] == "-",
+            },
+            "album": {
+                "id"   : album_id,
+                "name" : text.unescape(album_name),
+                "count": text.parse_int(photos[:-7])
+            },
         }
 
 

test/results/vk.py

@@ -79,6 +79,20 @@ __tests__ = (
     "#category": ("", "vk", "album"),
     "#class"   : vk.VkAlbumExtractor,
     "#count"   : 12,
+    "#log"     : "-165740836_281339889: Failed to extract metadata ('Access denied')",
+
+    "album": {
+        "id"    : "281339889",
+        "!name" : str,
+        "!count": int,
+    },
+    "user": {
+        "id": "-165740836",
+        "!name" : str,
+        "!nick" : str,
+        "!group": bool,
+
+    },
 },
 
 {
@@ -96,6 +110,36 @@ __tests__ = (
     "#exception": exception.AuthorizationError,
 },
 
+{
+    "#url"     : "https://vk.com/album-205150448_00",
+    "#class"   : vk.VkAlbumExtractor,
+    "#range"   : "1-25",
+    "#count"   : 25,
+
+    "id"           : r"re:\d+",
+    "width"        : range(100, 8_000),
+    "height"       : range(100, 8_000),
+    "filename"     : str,
+    "extension"    : {"jpg", "png"},
+    "date"         : "type:datetime",
+    "count"        : 826,
+    "num"          : range(1, 25),
+    "description"  : str,
+    "hash"         : r"re:[0-9a-f]{18}",
+    "likes"        : int,
+    "album"        : {
+        "id"   : "00",
+        "name" : "Community wall photos",
+        "count": 826,
+    },
+    "user"         : {
+        "id"   : "-205150448",
+        "name" : "otjareniy",
+        "nick" : "Отжареный Овощ(16+)",
+        "group": True,
+    },
+},
+
 {
     "#url"     : "https://vk.com/tag304303884",
     "#category": ("", "vk", "tagged"),