Compare commits

..

1 Commits

Author SHA1 Message Date
03efa17cf1 Print mtime before updating file metadata 2023-09-23 13:01:58 +02:00
7 changed files with 139 additions and 190 deletions

View File

@ -22,20 +22,6 @@ ambiguous situations.
## Unreleased
## 3.5.2 - 2024-04-14
### Fixed
- Crawling of personal desktop with ILIAS 8
- Crawling of empty personal desktops
## 3.5.1 - 2024-04-09
### Added
- Support for ILIAS 8
### Fixed
- Video name deduplication
## 3.5.0 - 2023-09-13
### Added

View File

@ -17,7 +17,7 @@ TargetType = Union[str, int]
class IliasElementType(Enum):
EXERCISE = "exercise"
EXERCISE_FILES = "exercise_files" # own submitted files
TEST = "test" # an online test. Will be ignored currently.
TEST = "test" # an online test. Will be ignored currently.
FILE = "file"
FOLDER = "folder"
FORUM = "forum"
@ -95,9 +95,13 @@ class IliasPage:
@staticmethod
def is_root_page(soup: BeautifulSoup) -> bool:
if permalink := IliasPage.get_soup_permalink(soup):
return "goto.php?target=root_" in permalink
return False
permalink = soup.find(id="current_perma_link")
if permalink is None:
return False
value = permalink.attrs.get("value")
if value is None:
return False
return "goto.php?target=root_" in value
def get_child_elements(self) -> List[IliasPageElement]:
"""
@ -275,14 +279,16 @@ class IliasPage:
return self._soup.find("a", attrs={"href": lambda x: x and "block_type=pditems" in x})
def _is_content_page(self) -> bool:
if link := self.get_permalink():
return "target=copa_" in link
return False
link = self._soup.find(id="current_perma_link")
if not link:
return False
return "target=copa_" in link.get("value")
def _is_learning_module_page(self) -> bool:
if link := self.get_permalink():
return "target=pg_" in link
return False
link = self._soup.find(id="current_perma_link")
if not link:
return False
return "target=pg_" in link.get("value")
def _contains_collapsed_future_meetings(self) -> bool:
return self._uncollapse_future_meetings_url() is not None
@ -378,10 +384,6 @@ class IliasPage:
name = _sanitize_path_name(link.text.strip())
url = self._abs_url_from_link(link)
if "cmd=manage" in url and "cmdClass=ilPDSelectedItemsBlockGUI" in url:
# Configure button/link does not have anything interesting
continue
type = self._find_type_from_link(name, link, url)
if not type:
_unexpected_html_warning()
@ -511,9 +513,12 @@ class IliasPage:
modification_string = link.parent.parent.parent.select_one(
f"td.std:nth-child({index})"
).getText().strip()
if match := re.search(r"\d+\.\d+.\d+ \d+:\d+", modification_string):
modification_time = datetime.strptime(match.group(0), "%d.%m.%Y %H:%M")
if re.search(r"\d+\.\d+.\d+ - \d+:\d+", modification_string):
log.explain(f"Converting {modification_string!r}")
modification_time = datetime.strptime(modification_string, "%d.%m.%Y - %H:%M")
break
else:
log.explain(f"Date has wrong format: {modification_string!r}")
if modification_time is None:
log.warn(f"Could not determine upload time for {link}")
@ -611,7 +616,7 @@ class IliasPage:
file_listings: List[Tag] = container.findAll(
name="a",
# download links contain the given command class
attrs={"href": lambda x: x and "cmdclass=ilexsubmissionfilegui" in x.lower()}
attrs={"href": lambda x: x and "cmdClass=ilexsubmissionfilegui" in x}
)
# Add each listing as a new
@ -915,9 +920,9 @@ class IliasPage:
@staticmethod
def _find_type_from_link(
element_name: str,
link_element: Tag,
url: str
element_name: str,
link_element: Tag,
url: str
) -> Optional[IliasElementType]:
"""
Decides which sub crawler to use for a given top level element.
@ -1065,45 +1070,6 @@ class IliasPage:
rest_of_name = split_delimiter.join(meeting_name.split(split_delimiter)[1:])
return datetime.strftime(date_portion, "%Y-%m-%d") + split_delimiter + rest_of_name
@staticmethod
def is_logged_in(soup: BeautifulSoup) -> bool:
# Normal ILIAS pages
mainbar: Optional[Tag] = soup.find(class_="il-maincontrols-metabar")
if mainbar is not None:
login_button = mainbar.find(attrs={"href": lambda x: x and "login.php" in x})
shib_login = soup.find(id="button_shib_login")
return not login_button and not shib_login
# Personal Desktop
if soup.find("a", attrs={"href": lambda x: x and "block_type=pditems" in x}):
return True
# Empty personal desktop has zero (0) markers. Match on the text...
if alert := soup.select_one(".alert-info"):
text = alert.getText().lower()
if "you have not yet selected any favourites" in text:
return True
if "sie haben aktuell noch keine favoriten ausgewählt" in text:
return True
# Video listing embeds do not have complete ILIAS html. Try to match them by
# their video listing table
video_table = soup.find(
recursive=True,
name="table",
attrs={"id": lambda x: x is not None and x.startswith("tbl_xoct")}
)
if video_table is not None:
return True
# The individual video player wrapper page has nothing of the above.
# Match it by its playerContainer.
if soup.select_one("#playerContainer") is not None:
return True
return False
def get_permalink(self) -> Optional[str]:
return IliasPage.get_soup_permalink(self._soup)
def _abs_url_from_link(self, link_tag: Tag) -> str:
"""
Create an absolute url from an <a> tag.
@ -1116,13 +1082,6 @@ class IliasPage:
"""
return urljoin(self._page_url, relative_url)
@staticmethod
def get_soup_permalink(soup: BeautifulSoup) -> Optional[str]:
perma_link_element: Tag = soup.select_one(".il-footer-permanent-url > a")
if not perma_link_element or not perma_link_element.get("href"):
return None
return perma_link_element.get("href")
def _unexpected_html_warning() -> None:
log.warn("Encountered unexpected HTML structure, ignoring element.")
@ -1146,7 +1105,7 @@ def demangle_date(date_str: str, fail_silently: bool = False) -> Optional[dateti
date_str = re.sub("Gestern|Yesterday", _format_date_english(_yesterday()), date_str, re.I)
date_str = re.sub("Heute|Today", _format_date_english(date.today()), date_str, re.I)
date_str = re.sub("Morgen|Tomorrow", _format_date_english(_tomorrow()), date_str, re.I)
date_str = re.sub("Morgen|Tomorrow", _format_date_english(_tomorrow()), date_str, re.I)
date_str = date_str.strip()
for german, english in zip(german_months, english_months):
date_str = date_str.replace(german, english)

View File

@ -81,7 +81,7 @@ class KitIliasWebCrawlerSection(HttpCrawlerSection):
return self.s.getboolean("forums", fallback=False)
_DIRECTORY_PAGES: Set[IliasElementType] = {
_DIRECTORY_PAGES: Set[IliasElementType] = set([
IliasElementType.EXERCISE,
IliasElementType.EXERCISE_FILES,
IliasElementType.FOLDER,
@ -90,16 +90,16 @@ _DIRECTORY_PAGES: Set[IliasElementType] = {
IliasElementType.MEDIACAST_VIDEO_FOLDER,
IliasElementType.OPENCAST_VIDEO_FOLDER,
IliasElementType.OPENCAST_VIDEO_FOLDER_MAYBE_PAGINATED,
}
])
_VIDEO_ELEMENTS: Set[IliasElementType] = {
_VIDEO_ELEMENTS: Set[IliasElementType] = set([
IliasElementType.MEDIACAST_VIDEO_FOLDER,
IliasElementType.MEDIACAST_VIDEO,
IliasElementType.OPENCAST_VIDEO,
IliasElementType.OPENCAST_VIDEO_PLAYER,
IliasElementType.OPENCAST_VIDEO_FOLDER,
IliasElementType.OPENCAST_VIDEO_FOLDER_MAYBE_PAGINATED,
}
])
def _iorepeat(attempts: int, name: str, failure_is_error: bool = False) -> Callable[[AWrapped], AWrapped]:
@ -130,7 +130,6 @@ def _iorepeat(attempts: int, name: str, failure_is_error: bool = False) -> Calla
raise CrawlError("Impossible return in ilias _iorepeat")
return wrapper # type: ignore
return decorator
@ -141,10 +140,6 @@ def _wrap_io_in_warning(name: str) -> Callable[[AWrapped], AWrapped]:
return _iorepeat(1, name)
def _get_video_cache_key(element: IliasPageElement) -> str:
return f"ilias-video-cache-{element.id()}"
# Crawler control flow:
#
# crawl_desktop -+
@ -178,11 +173,11 @@ def _get_video_cache_key(element: IliasPageElement) -> str:
class KitIliasWebCrawler(HttpCrawler):
def __init__(
self,
name: str,
section: KitIliasWebCrawlerSection,
config: Config,
authenticators: Dict[str, Authenticator]
self,
name: str,
section: KitIliasWebCrawlerSection,
config: Config,
authenticators: Dict[str, Authenticator]
):
# Setting a main authenticator for cookie sharing
auth = section.auth(authenticators)
@ -228,7 +223,7 @@ instance's greatest bottleneck.
await self._crawl_url(root_url, expected_id=course_id)
async def _crawl_desktop(self) -> None:
appendix = r"ILIAS\Repository\Provider\RepositoryMainBarProvider|mm_pd_sel_items"
appendix = r"ILIAS\PersonalDesktop\PDMainBarProvider|mm_pd_sel_items"
appendix = appendix.encode("ASCII").hex()
await self._crawl_url(self._base_url + "/gs_content.php?item=" + appendix)
@ -254,8 +249,8 @@ instance's greatest bottleneck.
soup = await self._get_page(next_stage_url, root_page_allowed=True)
if current_parent is None and expected_id is not None:
perma_link = IliasPage.get_soup_permalink(soup)
if not perma_link or "crs_" not in perma_link:
perma_link_element: Tag = soup.find(id="current_perma_link")
if not perma_link_element or "crs_" not in perma_link_element.get("value"):
raise CrawlError("Invalid course id? Didn't find anything looking like a course")
log.explain_topic(f"Parsing HTML page for {fmt_path(cl.path)}")
@ -552,8 +547,8 @@ instance's greatest bottleneck.
# Copy old mapping as it is likely still relevant
if self.prev_report:
self.report.add_custom_value(
_get_video_cache_key(element),
self.prev_report.get_custom_value(_get_video_cache_key(element))
str(element_path),
self.prev_report.get_custom_value(str(element_path))
)
# A video might contain other videos, so let's "crawl" the video first
@ -563,69 +558,58 @@ instance's greatest bottleneck.
# to ensure backwards compatibility.
maybe_dl = await self.download(element_path, mtime=element.mtime, redownload=Redownload.ALWAYS)
# If we do not want to crawl it (user filter), we can move on
if not maybe_dl:
return None
# If we have every file from the cached mapping already, we can ignore this and bail
if self._all_opencast_videos_locally_present(element, maybe_dl.path):
# Mark all existing videos as known to ensure they do not get deleted during cleanup.
# We "downloaded" them, just without actually making a network request as we assumed
# they did not change.
contained = self._previous_contained_opencast_videos(element, maybe_dl.path)
if len(contained) > 1:
# Only do this if we threw away the original dl token,
# to not download single-stream videos twice
for video in contained:
await self.download(video)
# If we do not want to crawl it (user filter) or we have every file
# from the cached mapping already, we can ignore this and bail
if not maybe_dl or self._all_opencast_videos_locally_present(element_path):
# Mark all existing cideos as known so they do not get deleted
# during dleanup. We "downloaded" them, just without actually making
# a network request as we assumed they did not change.
for video in self._previous_contained_opencast_videos(element_path):
await self.download(video)
return None
return self._download_opencast_video(element, maybe_dl)
return self._download_opencast_video(element_path, element, maybe_dl)
def _previous_contained_opencast_videos(
self, element: IliasPageElement, element_path: PurePath
) -> List[PurePath]:
def _previous_contained_opencast_videos(self, video_path: PurePath) -> List[PurePath]:
if not self.prev_report:
return []
custom_value = self.prev_report.get_custom_value(_get_video_cache_key(element))
custom_value = self.prev_report.get_custom_value(str(video_path))
if not custom_value:
return []
cached_value = cast(dict[str, Any], custom_value)
if "known_paths" not in cached_value or "own_path" not in cached_value:
log.explain(f"'known_paths' or 'own_path' missing from cached value: {cached_value}")
return []
transformed_own_path = self._transformer.transform(element_path)
if cached_value["own_path"] != str(transformed_own_path):
log.explain(
f"own_path '{transformed_own_path}' does not match cached value: '{cached_value['own_path']}"
)
return []
return [PurePath(name) for name in cached_value["known_paths"]]
names = cast(List[str], custom_value)
folder = video_path.parent
return [PurePath(folder, name) for name in names]
def _all_opencast_videos_locally_present(self, element: IliasPageElement, element_path: PurePath) -> bool:
log.explain_topic(f"Checking local cache for video {fmt_path(element_path)}")
if contained_videos := self._previous_contained_opencast_videos(element, element_path):
log.explain(
f"The following contained videos are known: {','.join(map(fmt_path, contained_videos))}"
)
if all(self._output_dir.resolve(path).exists() for path in contained_videos):
log.explain("Found all known videos locally, skipping enumeration request")
def _all_opencast_videos_locally_present(self, video_path: PurePath) -> bool:
if contained_videos := self._previous_contained_opencast_videos(video_path):
log.explain_topic(f"Checking local cache for video {video_path.name}")
all_found_locally = True
for video in contained_videos:
transformed_path = self._to_local_opencast_video_path(video)
if transformed_path:
exists_locally = self._output_dir.resolve(transformed_path).exists()
all_found_locally = all_found_locally and exists_locally
if all_found_locally:
log.explain("Found all videos locally, skipping enumeration request")
return True
log.explain("Missing at least one video, continuing with requests!")
else:
log.explain("No local cache present")
return False
def _to_local_opencast_video_path(self, path: PurePath) -> Optional[PurePath]:
if transformed := self._transformer.transform(path):
return self._deduplicator.fixup_path(transformed)
return None
@anoncritical
@_iorepeat(3, "downloading video")
async def _download_opencast_video(self, element: IliasPageElement, dl: DownloadToken) -> None:
def add_to_report(paths: list[str]) -> None:
self.report.add_custom_value(
_get_video_cache_key(element),
{"known_paths": paths, "own_path": str(self._transformer.transform(dl.path))}
)
async def _download_opencast_video(
self,
original_path: PurePath,
element: IliasPageElement,
dl: DownloadToken
) -> None:
stream_elements: List[IliasPageElement] = []
async with dl as (bar, sink):
page = IliasPage(await self._get_page(element.url), element.url, element)
stream_elements = page.get_child_elements()
@ -636,25 +620,32 @@ instance's greatest bottleneck.
log.explain(f"Using single video mode for {element.name}")
stream_element = stream_elements[0]
transformed_path = self._to_local_opencast_video_path(original_path)
if not transformed_path:
raise CrawlError(f"Download returned a path but transform did not for {original_path}")
# We do not have a local cache yet
await self._stream_from_url(stream_element.url, sink, bar, is_video=True)
add_to_report([str(self._transformer.transform(dl.path))])
if self._output_dir.resolve(transformed_path).exists():
log.explain(f"Video for {element.name} existed locally")
else:
await self._stream_from_url(stream_element.url, sink, bar, is_video=True)
self.report.add_custom_value(str(original_path), [original_path.name])
return
contained_video_paths: List[str] = []
for stream_element in stream_elements:
video_path = dl.path.parent / stream_element.name
video_path = original_path.parent / stream_element.name
contained_video_paths.append(str(video_path))
maybe_dl = await self.download(video_path, mtime=element.mtime, redownload=Redownload.NEVER)
if not maybe_dl:
continue
async with maybe_dl as (bar, sink):
log.explain(f"Streaming video from real url {stream_element.url}")
contained_video_paths.append(str(self._transformer.transform(maybe_dl.path)))
await self._stream_from_url(stream_element.url, sink, bar, is_video=True)
add_to_report(contained_video_paths)
self.report.add_custom_value(str(original_path), contained_video_paths)
async def _handle_file(
self,
@ -666,8 +657,8 @@ instance's greatest bottleneck.
return None
return self._download_file(element, maybe_dl)
@_iorepeat(3, "downloading file")
@anoncritical
@_iorepeat(3, "downloading file")
async def _download_file(self, element: IliasPageElement, dl: DownloadToken) -> None:
assert dl # The function is only reached when dl is not None
async with dl as (bar, sink):
@ -675,28 +666,12 @@ instance's greatest bottleneck.
async def _stream_from_url(self, url: str, sink: FileSink, bar: ProgressBar, is_video: bool) -> None:
async def try_stream() -> bool:
next_url = url
# Normal files redirect to the magazine if we are not authenticated. As files could be HTML,
# we can not match on the content type here. Instead, we disallow redirects and inspect the
# new location. If we are redirected anywhere but the ILIAS 8 "sendfile" command, we assume
# our authentication expired.
if not is_video:
async with self.session.get(url, allow_redirects=False) as resp:
# Redirect to anything except a "sendfile" means we weren't authenticated
async with self.session.get(url, allow_redirects=is_video) as resp:
if not is_video:
# Redirect means we weren't authenticated
if hdrs.LOCATION in resp.headers:
if "&cmd=sendfile" not in resp.headers[hdrs.LOCATION]:
return False
# Directly follow the redirect to not make a second, unnecessary request
next_url = resp.headers[hdrs.LOCATION]
# Let's try this again and follow redirects
return await fetch_follow_redirects(next_url)
async def fetch_follow_redirects(file_url: str) -> bool:
async with self.session.get(file_url) as resp:
# We wanted a video but got HTML => Forbidden, auth expired. Logging in won't really
# solve that depending on the setup, but it is better than nothing.
return False
# we wanted a video but got HTML
if is_video and "html" in resp.content_type:
return False
@ -753,6 +728,7 @@ instance's greatest bottleneck.
raise CrawlWarning("Failed to extract forum data")
if download_data.empty:
log.explain("Forum had no threads")
elements = []
return
html = await self._post_authenticated(download_data.url, download_data.form_data)
elements = parse_ilias_forum_export(soupify(html))
@ -918,7 +894,7 @@ instance's greatest bottleneck.
auth_id = await self._current_auth_id()
async with self.session.get(url) as request:
soup = soupify(await request.read())
if IliasPage.is_logged_in(soup):
if self._is_logged_in(soup):
return self._verify_page(soup, url, root_page_allowed)
# We weren't authenticated, so try to do that
@ -927,12 +903,11 @@ instance's greatest bottleneck.
# Retry once after authenticating. If this fails, we will die.
async with self.session.get(url) as request:
soup = soupify(await request.read())
if IliasPage.is_logged_in(soup):
if self._is_logged_in(soup):
return self._verify_page(soup, url, root_page_allowed)
raise CrawlError(f"get_page failed even after authenticating on {url!r}")
@staticmethod
def _verify_page(soup: BeautifulSoup, url: str, root_page_allowed: bool) -> BeautifulSoup:
def _verify_page(self, soup: BeautifulSoup, url: str, root_page_allowed: bool) -> BeautifulSoup:
if IliasPage.is_root_page(soup) and not root_page_allowed:
raise CrawlError(
"Unexpectedly encountered ILIAS root page. "
@ -986,10 +961,38 @@ instance's greatest bottleneck.
# We repeat this as the login method in shibboleth doesn't handle I/O errors.
# Shibboleth is quite reliable as well, the repeat is likely not critical here.
@_iorepeat(3, "Login", failure_is_error=True)
@ _iorepeat(3, "Login", failure_is_error=True)
async def _authenticate(self) -> None:
await self._shibboleth_login.login(self.session)
@ staticmethod
def _is_logged_in(soup: BeautifulSoup) -> bool:
# Normal ILIAS pages
mainbar: Optional[Tag] = soup.find(class_="il-maincontrols-metabar")
if mainbar is not None:
login_button = mainbar.find(attrs={"href": lambda x: x and "login.php" in x})
shib_login = soup.find(id="button_shib_login")
return not login_button and not shib_login
# Personal Desktop
if soup.find("a", attrs={"href": lambda x: x and "block_type=pditems" in x}):
return True
# Video listing embeds do not have complete ILIAS html. Try to match them by
# their video listing table
video_table = soup.find(
recursive=True,
name="table",
attrs={"id": lambda x: x is not None and x.startswith("tbl_xoct")}
)
if video_table is not None:
return True
# The individual video player wrapper page has nothing of the above.
# Match it by its playerContainer.
if soup.select_one("#playerContainer") is not None:
return True
return False
class KitShibbolethLogin:
"""
@ -1069,9 +1072,9 @@ class KitShibbolethLogin:
await sess.post(url, data=data)
async def _authenticate_tfa(
self,
session: aiohttp.ClientSession,
soup: BeautifulSoup
self,
session: aiohttp.ClientSession,
soup: BeautifulSoup
) -> BeautifulSoup:
if not self._tfa_auth:
self._tfa_auth = TfaAuthenticator("ilias-anon-tfa")
@ -1136,7 +1139,7 @@ async def _shib_post(
async with session.get(correct_url, allow_redirects=False) as response:
location = response.headers.get("location")
log.explain(f"Redirected to {location!r} with status {response.status}")
# If shib still has a valid session, it will directly respond to the request
# If shib still still has a valid session, it will directly respond to the request
if location is None:
log.explain("Shib recognized us, returning its response directly")
return soupify(await response.read())

View File

@ -415,6 +415,7 @@ class OutputDirectory:
def _update_metadata(self, info: DownloadInfo) -> None:
if mtime := info.heuristics.mtime:
log.explain(f"Setting mtime to {mtime}")
mtimestamp = mtime.timestamp()
os.utime(info.local_path, times=(mtimestamp, mtimestamp))

View File

@ -1,2 +1,2 @@
NAME = "PFERD"
VERSION = "3.5.2"
VERSION = "3.5.0"

8
flake.lock generated
View File

@ -2,16 +2,16 @@
"nodes": {
"nixpkgs": {
"locked": {
"lastModified": 1708979614,
"narHash": "sha256-FWLWmYojIg6TeqxSnHkKpHu5SGnFP5um1uUjH+wRV6g=",
"lastModified": 1694499547,
"narHash": "sha256-R7xMz1Iia6JthWRHDn36s/E248WB1/je62ovC/dUVKI=",
"owner": "NixOS",
"repo": "nixpkgs",
"rev": "b7ee09cf5614b02d289cd86fcfa6f24d4e078c2a",
"rev": "e5f018cf150e29aac26c61dac0790ea023c46b24",
"type": "github"
},
"original": {
"owner": "NixOS",
"ref": "nixos-23.11",
"ref": "nixos-23.05",
"repo": "nixpkgs",
"type": "github"
}

View File

@ -2,7 +2,7 @@
description = "Tool for downloading course-related files from ILIAS";
inputs = {
nixpkgs.url = "github:NixOS/nixpkgs/nixos-23.11";
nixpkgs.url = "github:NixOS/nixpkgs/nixos-23.05";
};
outputs = { self, nixpkgs }: