[bellazon] add 'quoted' option (#8247)

ignore quoted content by default

docs/configuration.rst

@@ -1600,6 +1600,16 @@ Description
     ``image``, ``video``, ``mediacollection``, ``embed``, ``text``.
 
 
+extractor.bellazon.quoted
+-------------------------
+Type
+    ``bool``
+Default
+    ``false``
+Description
+    Extract files from quoted content.
+
+
 extractor.[blogger].api-key
 ---------------------------
 Type

docs/gallery-dl.conf

@@ -154,6 +154,10 @@
 
             "modules": ["image", "video", "mediacollection", "embed"]
         },
+        "bellazon":
+        {
+            "quoted": false
+        },
         "bilibili":
         {
             "sleep-request": "3.0-6.0"

gallery_dl/extractor/bellazon.py

@@ -29,8 +29,17 @@ class BellazonExtractor(Extractor):
             r'(?s)<((?:video .*?<source src|a [^>]*?href)="([^"]+).*?)</a>'
         ).findall
 
+        if self.config("quoted", False):
+            strip_quoted = None
+        else:
+            strip_quoted = text.re(r"(?s)<blockquote .*?</blockquote>").sub
+
         for post in self.posts():
-            urls = extract_urls(post["content"])
+            if strip_quoted is None:
+                urls = extract_urls(post["content"])
+            else:
+                urls = extract_urls(strip_quoted("", post["content"]))
+
             data = {"post": post}
             post["count"] = data["count"] = len(urls)