[zerochan] download webp and gif files, add 'extensions' option (#6576)

mikf · Dec 5, 2024 · 34e157e · 34e157e
1 parent 624dc7f
commit 34e157e
Show file tree

Hide file tree

Showing 2 changed files with 32 additions and 5 deletions.
diff --git a/docs/configuration.rst b/docs/configuration.rst
@@ -5190,6 +5190,22 @@ Description
     `youtube-dl's docstrings <https://github.com/ytdl-org/youtube-dl/blob/0153b387e57e0bb8e580f1869f85596d2767fb0d/youtube_dl/YoutubeDL.py#L157>`__
 
 
+extractor.zerochan.extensions
+-----------------------------
+Type
+    * ``string``
+    * ``list`` of ``strings``
+Default
+    ``["jpg", "png", "webp", "gif"]``
+Example
+    * ``"gif"``
+    * ``["webp", "gif", "jpg"}``
+Description
+    List of filename extensions to try when dynamically building download URLs
+    (`"pagination": "api" <extractor.zerochan.pagination_>`__ +
+    `"metadata": false <extractor.zerochan.metadata_>`__)
+
+
 extractor.zerochan.metadata
 ---------------------------
 Type

diff --git a/gallery_dl/extractor/zerochan.py b/gallery_dl/extractor/zerochan.py
@@ -145,6 +145,14 @@ def _init(self):
             self.posts = self.posts_api
             self.session.headers["User-Agent"] = util.USERAGENT
 
+        exts = self.config("extensions")
+        if exts:
+            if isinstance(exts, str):
+                exts = exts.split(",")
+            self.exts = exts
+        else:
+            self.exts = ("jpg", "png", "webp", "gif")
+
     def metadata(self):
         return {"search_tags": text.unquote(
             self.search_tag.replace("+", " "))}
@@ -194,8 +202,6 @@ def posts_api(self):
             "p"   : self.page_start,
         }
 
-        static = "https://static.zerochan.net/.full."
-
         while True:
             response = self.request(url, params=params, allow_redirects=False)
 
@@ -221,15 +227,20 @@ def posts_api(self):
                     yield post
             else:
                 for post in posts:
-                    base = static + str(post["id"])
-                    post["file_url"] = base + ".jpg"
-                    post["_fallback"] = (base + ".png",)
+                    urls = self._urls(post)
+                    post["file_url"] = next(urls)
+                    post["_fallback"] = urls
                     yield post
 
             if not data.get("next"):
                 return
             params["p"] += 1
 
+    def _urls(self, post, static="https://static.zerochan.net/.full."):
+        base = static + str(post["id"]) + "."
+        for ext in self.exts:
+            yield base + ext
+
 
 class ZerochanImageExtractor(ZerochanExtractor):
     subcategory = "image"