[rule34] fix file downloads (#7697)

replace 'api-cdn' subdomain of image files with 'wimg'

gallery_dl/extractor/gelbooru_v02.py

@@ -21,6 +21,9 @@ class GelbooruV02Extractor(booru.BooruExtractor):
         self.user_id = self.config("user-id")
         self.root_api = self.config_instance("root-api") or self.root
 
+        if self.category == "rule34":
+            self._file_url = self._file_url_rule34
+
     def _api_request(self, params):
         url = self.root_api + "/index.php?page=dapi&s=post&q=index"
         return self.request_xml(url, params=params)
@@ -91,6 +94,16 @@ class GelbooruV02Extractor(booru.BooruExtractor):
                 return
             params["pid"] += self.per_page
 
+    def _file_url_rule34(self, post):
+        url = post["file_url"]
+
+        if text.ext_from_url(url) not in util.EXTS_VIDEO:
+            path = url.partition(".")[2]
+            post["_fallback"] = (url,)
+            post["file_url"] = url = "https://wimg." + path
+
+        return url
+
     def _prepare(self, post):
         post["tags"] = post["tags"].strip()
         post["date"] = text.parse_datetime(

test/results/rule34.py

@@ -12,8 +12,10 @@ __tests__ = (
     "#url"     : "https://rule34.xxx/index.php?page=post&s=list&tags=danraku",
     "#category": ("gelbooru_v02", "rule34", "tag"),
     "#class"   : gelbooru_v02.GelbooruV02TagExtractor,
-    "#pattern"     : r"https?://.*rule34\.xxx/images/\d+/[0-9a-f]+\.jpg",
-    "#count"       : 2,
+    "#results"     : (
+        "https://wimg.rule34.xxx/images/4615/00722987a1e8b5617a15a20c19a0915157048d3b.jpg",
+        "https://wimg.rule34.xxx/images/1845/04981deeac105a9c5fedc34a6ff017789e74f2a8.jpg",
+    ),
     "#sha1_content": [
         "5c6ae9ee13e6d4bc9cb8bdce224c84e67fbfa36c",
         "622e80be3f496672c44aab5c47fbc6941c61bc79",
@@ -53,7 +55,7 @@ __tests__ = (
         "tags" : True,
         "notes": True,
     },
-    "#pattern"     : r"https://api-cdn\.rule34\.xxx/images/1/6aafbdb3e22f3f3b412ea2cf53321317a37063f3\.jpg",
+    "#results"     : r"https://wimg.rule34.xxx/images/1/6aafbdb3e22f3f3b412ea2cf53321317a37063f3.jpg",
     "#sha1_content": [
         "a43f418aa350039af0d11cae501396a33bbe2201",
         "67b516295950867e1c1ab6bc13b35d3b762ed2a3",
@@ -84,4 +86,15 @@ __tests__ = (
     ],
 },
 
+{
+    "#url"     : "https://rule34.xxx/index.php?page=post&s=view&id=13853212",
+    "#comment" : "HTML response with 'api-cdn.' subdomain (#7697)",
+    "#category": ("gelbooru_v02", "rule34", "post"),
+    "#class"   : gelbooru_v02.GelbooruV02PostExtractor,
+    "#results" : "https://wimg.rule34.xxx/images/2164/60d61d06f3cd51be5152852d9c642d80.jpeg",
+    "#sha1_content": [
+        "0a07eb9e871589a012ec922c71eb4640fce09bb2",
+    ],
+},
+
 )