[imagevenue] detect 404 image files (#7570)

gallery_dl/extractor/imagehosts.py

@@ -23,6 +23,7 @@ class ImagehostImageExtractor(Extractor):
     _params = None
     _cookies = None
     _encoding = None
+    _validate = None
 
     def __init__(self, match):
         Extractor.__init__(self, match)
@@ -57,6 +58,8 @@ class ImagehostImageExtractor(Extractor):
         data.update(self.metadata(page))
         if self._https and url.startswith("http:"):
             url = "https:" + url[5:]
+        if self._validate is not None:
+            data["_http_validate"] = self._validate
 
         yield Message.Directory, data
         yield Message.Url, url, data
@@ -164,6 +167,14 @@ class ImagevenueImageExtractor(ImagehostImageExtractor):
         filename, pos = text.extract(page, 'alt="', '"', pos)
         return url, text.unescape(filename)
 
+    def _validate(self, response):
+        hget = response.headers.get
+        return not (
+            hget("content-length") == "14396" and
+            hget("content-type") == "image/jpeg" and
+            hget("last-modified") == "Mon, 04 May 2020 07:19:52 GMT"
+        )
+
 
 class ImagetwistImageExtractor(ImagehostImageExtractor):
     """Extractor for single images from imagetwist.com"""