[issuu] unescape HTML entities
This commit is contained in:
@@ -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(""", '"'))
|
||||
data = util.json_loads(text.unescape(text.rextract(
|
||||
page, '<script data-json="', '"', pos)[0]))
|
||||
|
||||
doc = data["initialDocumentData"]["document"]
|
||||
doc["date"] = text.parse_datetime(
|
||||
|
||||
@@ -34,6 +34,37 @@ __tests__ = (
|
||||
"num" : int,
|
||||
},
|
||||
|
||||
{
|
||||
"#url" : "https://issuu.com/foodhome1955/docs/fh_winter2025-issuu-011625",
|
||||
"#comment" : "HTML escapes",
|
||||
"#class" : issuu.IssuuPublicationExtractor,
|
||||
"#count" : 84,
|
||||
|
||||
"document": {
|
||||
"access" : "PUBLIC",
|
||||
"date" : "dt:2025-01-17 00:00:00",
|
||||
"description" : "Santa Barbara's Lifestyle Magazine",
|
||||
"documentName" : "fh_winter2025-issuu-011625",
|
||||
"isDocumentGated" : False,
|
||||
"originalPublishDateInISOString": "2025-01-17T00:00:00.000Z",
|
||||
"pageCount" : 84,
|
||||
"publicationId" : "b89e35d4bd2201c7ecd871160fe000fa",
|
||||
"revisionId" : "250117005419",
|
||||
"title" : "Food & Home Winter 2025",
|
||||
"username" : "foodhome1955",
|
||||
"contentRating" : {
|
||||
"isAdsafe" : True,
|
||||
"isExplicit" : False,
|
||||
"isReviewed" : True,
|
||||
},
|
||||
"path" : {
|
||||
"documentName": "fh_winter2025-issuu-011625",
|
||||
"type" : "user",
|
||||
"username" : "foodhome1955",
|
||||
},
|
||||
},
|
||||
},
|
||||
|
||||
{
|
||||
"#url" : "https://issuu.com/issuu",
|
||||
"#class" : issuu.IssuuUserExtractor,
|
||||
|
||||
Reference in New Issue
Block a user