support accents in html files

This commit is contained in:
eliottp1089 2023-11-07 10:02:34 -06:00 committed by Puyodead1
parent 5d01e56756
commit 4880f7bfef
No known key found for this signature in database
GPG Key ID: A4FA4FEC0DD353FC

View File

@ -1574,7 +1574,7 @@ def parse_new(udemy: Udemy, udemy_object: dict):
if extension == "html": if extension == "html":
# if the html content is None or an empty string, skip it so we dont save empty html files # if the html content is None or an empty string, skip it so we dont save empty html files
if parsed_lecture.get("html_content") != None and parsed_lecture.get("html_content") != "": if parsed_lecture.get("html_content") != None and parsed_lecture.get("html_content") != "":
html_content = parsed_lecture.get("html_content").encode("ascii", "ignore").decode("ascii") html_content = parsed_lecture.get("html_content").encode("utf8", "ignore").decode("utf8")
lecture_path = os.path.join(chapter_dir, "{}.html".format(sanitize_filename(lecture_title))) lecture_path = os.path.join(chapter_dir, "{}.html".format(sanitize_filename(lecture_title)))
try: try:
with open(lecture_path, encoding="utf8", mode="w") as f: with open(lecture_path, encoding="utf8", mode="w") as f: