[issuu] unescape HTML entities

2025-02-02 18:33:18 +01:00
parent 6c9b20fe45
commit 5807daa19a
2 changed files with 33 additions and 2 deletions
--- a/gallery_dl/extractor/issuu.py
+++ b/gallery_dl/extractor/issuu.py
@@ -30,8 +30,8 @@ class IssuuPublicationExtractor(IssuuBase, GalleryExtractor):

    def metadata(self, page):
        pos = page.rindex('id="initial-data"')
-        data = util.json_loads(text.rextract(
-            page, '<script data-json="', '"', pos)[0].replace("&quot;", '"'))
+        data = util.json_loads(text.unescape(text.rextract(
+            page, '<script data-json="', '"', pos)[0]))

        doc = data["initialDocumentData"]["document"]
        doc["date"] = text.parse_datetime(