[issuu] fix extraction
This commit is contained in:
@@ -29,8 +29,9 @@ class IssuuPublicationExtractor(IssuuBase, GalleryExtractor):
|
|||||||
example = "https://issuu.com/issuu/docs/TITLE/"
|
example = "https://issuu.com/issuu/docs/TITLE/"
|
||||||
|
|
||||||
def metadata(self, page):
|
def metadata(self, page):
|
||||||
|
pos = page.rindex('id="initial-data"')
|
||||||
data = util.json_loads(text.rextract(
|
data = util.json_loads(text.rextract(
|
||||||
page, '<script data-json="', '"')[0].replace(""", '"'))
|
page, '<script data-json="', '"', pos)[0].replace(""", '"'))
|
||||||
|
|
||||||
doc = data["initialDocumentData"]["document"]
|
doc = data["initialDocumentData"]["document"]
|
||||||
doc["date"] = text.parse_datetime(
|
doc["date"] = text.parse_datetime(
|
||||||
|
|||||||
Reference in New Issue
Block a user