[bunkr] fix extraction (#4514, #4532, #4529, #4540)

gallery_dl/extractor/bunkr.py

@@ -38,36 +38,37 @@ class BunkrAlbumExtractor(LolisafeAlbumExtractor):
             page, "<h1", "</div>").partition(">")[2])
         count, _, size = info[1].split(None, 2)
 
-        # files
-        cdn = None
-        files = []
-        append = files.append
-
         pos = page.index('class="grid-images')
-        for url in text.extract_iter(page, '<a href="', '"', pos):
+        urls = list(text.extract_iter(page, '<a href="', '"', pos))
-            if url.startswith("/"):
-                if not cdn:
-                    # fetch cdn root from download page
-                    durl = "{}/d/{}".format(self.root, url[3:])
-                    cdn = text.extr(self.request(
-                        durl).text, 'link.href = "', '"')
-                    cdn = cdn[:cdn.index("/", 8)]
-                url = cdn + url[2:]
 
-            url = text.unescape(url)
+        return self._extract_files(urls), {
-            if url.lower().endswith(CDN_HOSTED_EXTENSIONS):
-                scheme, domain, path, query, fragment = urlsplit(url)
-                if domain in MEDIA_DOMAIN_OVERRIDES:
-                    domain = MEDIA_DOMAIN_OVERRIDES[domain]
-                else:
-                    domain = domain.replace("cdn", "media-files", 1)
-                url = urlunsplit((scheme, domain, path, query, fragment))
-            append({"file": url})
-
-        return files, {
             "album_id"   : self.album_id,
             "album_name" : text.unescape(info[0]),
             "album_size" : size[1:-1],
             "description": text.unescape(info[2]) if len(info) > 2 else "",
-            "count"      : len(files),
+            "count"      : len(urls),
         }
+
+    def _extract_files(self, urls):
+        for url in urls:
+            if url.startswith("/"):
+                try:
+                    page = self.request(self.root + text.unescape(url)).text
+                    if url[1] == "v":
+                        url = text.extr(page, '<source src="', '"')
+                    else:
+                        url = text.extr(page, '<img src="', '"')
+                except Exception as exc:
+                    self.log.error("%s: %s", exc.__class__.__name__, exc)
+                    continue
+
+            else:
+                if url.lower().endswith(CDN_HOSTED_EXTENSIONS):
+                    scheme, domain, path, query, fragment = urlsplit(url)
+                    if domain in MEDIA_DOMAIN_OVERRIDES:
+                        domain = MEDIA_DOMAIN_OVERRIDES[domain]
+                    else:
+                        domain = domain.replace("cdn", "media-files", 1)
+                    url = urlunsplit((scheme, domain, path, query, fragment))
+
+            yield {"file": text.unescape(url)}

test/results/bunkr.py

@@ -12,7 +12,7 @@ __tests__ = (
     "#url"     : "https://bunkrr.su/a/Lktg9Keq",
     "#category": ("lolisafe", "bunkr", "album"),
     "#class"   : bunkr.BunkrAlbumExtractor,
-    "#pattern"     : r"https://cdn\.bunkr\.ru/test-テスト-\"&>-QjgneIQv\.png",
+    "#urls"        : "https://i-burger.bunkr.ru/test-テスト-\"&>-QjgneIQv.png",
     "#sha1_content": "0c8768055e4e20e7c7259608b67799171b691140",
 
     "album_id"  : "Lktg9Keq",
@@ -29,7 +29,7 @@ __tests__ = (
     "#comment" : "mp4 (#2239)",
     "#category": ("lolisafe", "bunkr", "album"),
     "#class"   : bunkr.BunkrAlbumExtractor,
-    "#pattern"     : r"https://media-files\.bunkr\.ru/_-RnHoW69L\.mp4",
+    "#urls"        : "https://burger.bunkr.ru/_-RnHoW69L.mp4",
     "#sha1_content": "80e61d1dbc5896ae7ef9a28734c747b28b320471",
 },
 
@@ -38,14 +38,17 @@ __tests__ = (
     "#comment" : "cdn4",
     "#category": ("lolisafe", "bunkr", "album"),
     "#class"   : bunkr.BunkrAlbumExtractor,
-    "#pattern"     : r"https://(cdn|media-files)4\.bunkr\.ru/",
+    "#urls"        : (
+        "https://media-files4.bunkr.ru/video-wFO9FtxG.mp4",
+        "https://i4.bunkr.ru/image-sZrQUeOx.jpg",
+    ),
     "#sha1_content": "da29aae371b7adc8c5ef8e6991b66b69823791e8",
 
     "album_id"   : "iXTTc1o2",
     "album_name" : "test2",
     "album_size" : "691.1 KB",
     "count"      : 2,
-    "description": "072022",
+    "description": "",
     "filename"   : r"re:video-wFO9FtxG|image-sZrQUeOx",
     "id"         : r"re:wFO9FtxG|sZrQUeOx",
     "name"       : r"re:video|image",
@@ -57,8 +60,8 @@ __tests__ = (
     "#comment" : "cdn12 .ru TLD (#4147)",
     "#category": ("lolisafe", "bunkr", "album"),
     "#class"   : bunkr.BunkrAlbumExtractor,
-    "#pattern" : r"https://(cdn12.bunkr.ru|media-files12.bunkr.la)/\w+",
+    "#pattern" : r"https://(i-)?meatballs.bunkr.ru/\w+",
-    "#count"   : 8,
+    "#count"   : 10,
 },
 
 {