[rule34xyz] add 'format' option (#1078)

docs/configuration.rst

@@ -3859,6 +3859,21 @@ Description
     restrict it to only one possible format.
 
 
+extractor.rule34xyz.format
+---------------------------
+Type
+    * ``string``
+    * ``list`` of ``strings``
+Default
+    ``["10", "40", "41", "2"]``
+Example
+    ``"33,34,4"``
+Description
+    Selects the file format to extract.
+
+    When more than one format is given, the first available one is selected.
+
+
 extractor.sankaku.id-format
 ---------------------------
 Type

gallery_dl/extractor/rule34xyz.py

@@ -29,17 +29,31 @@ class Rule34xyzExtractor(BooruExtractor):
         3: "artist",
     }
 
+    def _init(self):
+        formats = self.config("format")
+        if formats:
+            if isinstance(formats, str):
+                formats = formats.split(",")
+            self.formats = formats
+        else:
+            self.formats = ("10", "40", "41", "2")
+
     def _file_url(self, post):
         post["files"] = files = {
             str(link["type"]): link["url"]
             for link in post.pop("imageLinks")
         }
-        post["file_url"] = url = (
+
-            files.get("10") or  # mov
+        for fmt in self.formats:
-            files.get("40") or  # mov720
+            if fmt in files:
-            files.get("41") or  # mov480
+                break
-            files["2"]          # pic
+        else:
-        )
+            fmt = "2"
+            self.log.warning("%s: Requested format not available", post["id"])
+
+        post["file_url"] = url = files[fmt]
+        post["format_id"] = fmt
+        post["format"] = url.rsplit(".", 2)[1]
         return url
 
     def _prepare(self, post):

test/results/rule34xyz.py

@@ -44,6 +44,8 @@ __tests__ = (
     "extension" : "jpg",
     "file_url"  : "https://rule34xyz.b-cdn.net/posts/3613/3613851/3613851.pic.jpg",
     "filename"  : "3613851.pic",
+    "format"    : "pic",
+    "format_id" : "2",
     "id"        : 3613851,
     "likes"     : range(3, 100),
     "posted"    : "2023-03-29T06:01:07.900161",
@@ -121,6 +123,20 @@ __tests__ = (
     "#class"  : rule34xyz.Rule34xyzPostExtractor,
     "#urls"        : "https://rule34xyz.b-cdn.net/posts/3571/3571567/3571567.mov720.mp4",
     "#sha1_content": "c0a5e7e887774f91527f00e6142c435a3c482c1f",
+
+    "format"    : "mov720",
+    "format_id" : "40",
+},
+
+{
+    "#url"    : "https://rule34.xyz/post/3571567",
+    "#comment": "'format' option",
+    "#class"  : rule34xyz.Rule34xyzPostExtractor,
+    "#options": {"format": "10,33"},
+    "#urls"   : "https://rule34xyz.b-cdn.net/posts/3571/3571567/3571567.pic256avif.avif",
+
+    "format"    : "pic256avif",
+    "format_id" : "33",
 },
 
 )