[archivedmoe] fix thebarchive webm URLs (#5116)

gallery_dl/extractor/foolfuuka.py

@@ -24,6 +24,8 @@ def __init__(self, match):
         BaseExtractor.__init__(self, match)
         if self.category == "b4k":
             self.remote = self._remote_direct
+        elif self.category == "archivedmoe":
+            self.referer = False
 
     def items(self):
         yield Message.Directory, self.metadata()
@@ -53,9 +55,12 @@ def posts(self):
 
     def remote(self, media):
         """Resolve a remote media link"""
-        needle = '<meta http-equiv="Refresh" content="0; url='
         page = self.request(media["remote_media_link"]).text
-        return text.extr(page, needle, '"')
+        url = text.extr(page, 'http-equiv="Refresh" content="0; url=', '"')
+        if url.endswith(".webm") and \
+                url.startswith("https://thebarchive.com/"):
+            return url[:-1]
+        return url
 
     @staticmethod
     def _remote_direct(media):

test/results/archivedmoe.py

@@ -23,6 +23,18 @@
     "#sha1_url": "ffec05a1a1b906b5ca85992513671c9155ee9e87",
 },
 
+{
+    "#url"     : "https://archived.moe/b/thread/912594917/",
+    "#comment" : "broken thebarchive .webm URLs (#5116)",
+    "#category": ("foolfuuka", "archivedmoe", "thread"),
+    "#class"   : foolfuuka.FoolfuukaThreadExtractor,
+    "#urls"    : (
+        "https://thebarchive.com/b/full_image/1705625299234839.gif",
+        "https://thebarchive.com/b/full_image/1705625431133806.web",
+        "https://thebarchive.com/b/full_image/1705626190307840.web",
+    ),
+},
+
 {
     "#url"     : "https://archived.moe/gd/",
     "#category": ("foolfuuka", "archivedmoe", "board"),