[zerochan] fix 'source' extraction when not logged in

mikf · Dec 12, 2024 · d2c66ac · d2c66ac
1 parent 63008f7
commit d2c66ac
Show file tree

Hide file tree

Showing 2 changed files with 23 additions and 3 deletions.
diff --git a/gallery_dl/extractor/zerochan.py b/gallery_dl/extractor/zerochan.py
@@ -78,8 +78,8 @@ def _parse_entry_html(self, entry_id):
                 'class="breadcrumbs', '</nav>'))[2:],
             "uploader": extr('href="/user/', '"'),
             "tags"    : extr('<ul id="tags"', '</ul>'),
-            "source"  : text.unescape(text.extr(
-                extr('id="source-url"', '</a>'), 'href="', '"')),
+            "source"  : text.unescape(text.remove_html(extr(
+                'id="source-url"', '</p>').rpartition("</s>")[2])),
         }
 
         html = data["tags"]

diff --git a/test/results/zerochan.py b/test/results/zerochan.py
@@ -134,7 +134,7 @@
         "DRAGON BALL",
     ],
     "size"     : 136192,
-    "source": "https://x.com/Raydash30/status/1766012730769862774",
+    "source"   : "https://x.com/Raydash30/status/1766012730769862774",
     "tags"     : [
         "Mangaka:Raydash",
         "Series:DRAGON BALL",
@@ -177,6 +177,26 @@
     "width"    : 750,
 },
 
+{
+    "#url"     : "https://www.zerochan.net/4233756",
+    "#class"   : zerochan.ZerochanImageExtractor,
+    "#auth"    : False,
+    "#urls"    : "https://static.zerochan.net/DRAGON.BALL.full.4233756.jpg",
+
+    "source"   : "https://x.com/Raydash30/status/1766012730769862774",
+    "tags"     : [
+        "Mangaka:Raydash",
+        "Series:DRAGON BALL",
+        "Series:DRAGON BALL Z",
+        "Character:Piccolo",
+        "Character:Son Gohan",
+        "Theme:Green Skin",
+        "Source:Fanart",
+        "Source:Fanart from X (Twitter)",
+        "Source:X (Twitter)",
+    ],
+},
+
 {
     "#url"     : "https://www.zerochan.net/1395035",
     "#comment" : "Invalid control character '\r' in 'source' field (#5892)",