[zerochan] parse API response manually when json.loads() fails (#6632)

gallery_dl/extractor/zerochan.py

@@ -93,14 +93,12 @@ class ZerochanExtractor(BooruExtractor):
 
     def _parse_entry_api(self, entry_id):
         url = "{}/{}?json".format(self.root, entry_id)
-        text = self.request(url).text
+        txt = self.request(url).text
         try:
-            item = util.json_loads(text)
-        except ValueError as exc:
-            if " control character " not in str(exc):
-                raise
-            text = re.sub(r"[\x00-\x1f\x7f]", "", text)
-            item = util.json_loads(text)
+            item = util.json_loads('"' + txt)
+        except ValueError:
+            item = self._parse_json(txt)
+            item["id"] = text.parse_int(entry_id)
 
         data = {
             "id"      : item["id"],
@@ -118,6 +116,27 @@ class ZerochanExtractor(BooruExtractor):
 
         return data
 
+    def _parse_json(self, txt):
+        txt = re.sub(r"[\x00-\x1f\x7f]", "", txt)
+        main, _, tags = txt.partition('tags": [')
+
+        item = {}
+        for line in main.split(',  "')[1:]:
+            key, _, value = line.partition('": ')
+            if value:
+                if value[0] == '"':
+                    value = value[1:-1]
+                else:
+                    value = text.parse_int(value)
+            if key:
+                item[key] = value
+
+        item["tags"] = tags = tags[5:].split('",    "')
+        if tags:
+            tags[-1] = tags[-1][:-5]
+
+        return item
+
     def _tags(self, post, page):
         tags = collections.defaultdict(list)
         for tag in post["tags"]:

test/results/zerochan.py

@@ -208,4 +208,46 @@ __tests__ = (
     "source": "http://www.youtube.com/watch?v=0vodqkGPxt8",
 },
 
+{
+    "#url"     : "https://www.zerochan.net/4354955",
+    "#comment" : "unescaped quotes in 'JSON' data (#6632)",
+    "#category": ("booru", "zerochan", "image"),
+    "#class"   : zerochan.ZerochanImageExtractor,
+    "#auth"    : False,
+    "#options" : {"metadata": True},
+
+    "author"  : "SEGA",
+    "date"    : "dt:2024-12-05 06:06:14",
+    "file_url": "https://static.zerochan.net/Miles.%22Tails%22.Prower.full.4354955.jpg",
+    "filename": "Miles.\"Tails\".Prower.full.4354955",
+    "height"  : 705,
+    "id"      : 4354955,
+    "name"    : "Miles \"Tails\" Prower",
+    "size"    : 252928,
+    "source"  : "https://x.com/kellanstover/status/1580237736874606597",
+    "uploader": "Anima-Chao",
+    "width"   : 4096,
+    "path"    : [
+        "Sonic the Hedgehog",
+        "Miles \"Tails\" Prower",
+    ],
+    "tags"    : [
+        "Male",
+        "Animal",
+        "Fox",
+        "Sonic the Hedgehog",
+        "Flying",
+        "Character Sheet",
+        "Airplane",
+        "SEGA",
+        "Miles \"Tails\" Prower",
+        "Official Art",
+        "Midair",
+        "X (Twitter)",
+        "Sonic Origins",
+        "Official Art from X",
+        "Tory Patterson",
+    ],
+},
+
 )