def parse_content_page(soup) -> List[Union[SDoc, SFolder, SLecture]]:
# NOTE e.g. "course_id": "_306327_1", "content_id": "_1790226_1"
contentList = soup.find("ul", {"id": "content_listContainer"})
if contentList is None:
return []
children = [c for c in contentList.children if c.name is not None]
result: List[Union[SDoc, SFolder, SLecture]] = []
def is_content_folder(child, img):
return img is not None and img.get("alt") == "Content Folder"
def is_item(child, img):
# some items do not have the item icon
return img is not None and (img.get("alt") == "Item" or
(img.get("alt") == "" and child.find(
"div", {"class": "item clearfix"})))
def is_AcuStudio(child, img):
return img is not None and img.get("alt") == "AcuStudio"
def is_file(child, img):
return img is not None and img.get("alt") == "File"
for c in children:
img = c.find("img")
if is_content_folder(c, img):
hyperlink = c.find("a")
name = hyperlink.text
link = hyperlink.get("href")
details = c.find("div", {"class": "details"}).text
result.append(
SFolder(name.strip(), link.strip(), details.strip(), None))
elif is_item(c, img):
folder = __item_to_folder(c)
if folder:
result.append(folder)
elif is_AcuStudio(c, img):
# ignore that for now
hyperlink = c.find("a")
name = hyperlink.text
link = hyperlink.get("href")
result.append(SLecture(name.strip(), link))
elif is_file(c, img):
hyperlink = c.find("a")
# sometimes file link is broken, in that case no href tag is rendered
# see: https://github.com/leafgecko/NTULearn-Downloader/issues/8
if hyperlink is None:
continue
name = hyperlink.text
link = hyperlink.get("href")
result.append(SDoc(name.strip(), link))
else:
pass
return result
def __item_to_folder(item):
folder_name = item.find("h3").text
details = item.find("div", {"class", "details"})
dl_links = [
a for a in details.find_all("a") if is_download_link(a.get("href"))
]
children = []
for a_tag in dl_links:
link, name = a_tag.get("href"), a_tag.text
children.append(SDoc(name=name.strip(), link=link))
if children:
return SFolder(name=folder_name.strip(),
link=None,
details="",
children=children)
return None