pferd/PFERD/pferd.py

132 lines
4.8 KiB
Python
Raw Normal View History

"""
Convenience functions for using PFERD.
"""
2020-04-23 19:55:37 +02:00
import logging
2020-04-23 11:44:13 +02:00
from pathlib import Path
2020-04-24 20:00:21 +02:00
from typing import List, Optional, Union
2020-04-23 11:44:13 +02:00
from .cookie_jar import CookieJar
from .ilias import (IliasAuthenticator, IliasCrawler, IliasDirectoryFilter,
IliasDownloader, IliasDownloadStrategy,
KitShibbolethAuthenticator, download_modified_or_new)
2020-04-23 19:38:28 +02:00
from .location import Location
2020-04-23 11:44:13 +02:00
from .organizer import Organizer
from .tmp_dir import TmpDir
2020-04-24 20:26:28 +02:00
from .transform import TF, Transform, apply_transform
2020-04-24 20:39:30 +02:00
from .utils import PrettyLogger, PathLike, to_path
2020-04-23 11:44:13 +02:00
2020-04-23 12:49:01 +02:00
# TODO save known-good cookies as soon as possible
2020-04-23 19:55:37 +02:00
LOGGER = logging.getLogger(__name__)
PRETTY = PrettyLogger(LOGGER)
2020-04-23 12:49:01 +02:00
2020-04-23 11:44:13 +02:00
class Pferd(Location):
# pylint: disable=too-many-arguments
"""
The main entrypoint in your Pferd usage: This class combines a number of
useful shortcuts for running synchronizers in a single interface.
"""
2020-04-23 11:44:13 +02:00
2020-04-24 20:00:21 +02:00
def __init__(
self,
base_dir: Path,
tmp_dir: Path = Path(".tmp"),
test_run: bool = False
):
2020-04-23 11:44:13 +02:00
super().__init__(Path(base_dir))
self._tmp_dir = TmpDir(self.resolve(tmp_dir))
2020-04-24 20:00:21 +02:00
self._test_run = test_run
2020-04-24 20:26:28 +02:00
@staticmethod
def _print_transformables(transformables: List[TF]) -> None:
2020-04-24 20:00:21 +02:00
LOGGER.info("")
LOGGER.info("Results of the test run:")
for transformable in transformables:
LOGGER.info(transformable.path)
2020-04-23 11:44:13 +02:00
def _ilias(
self,
2020-04-24 20:39:30 +02:00
target: PathLike,
2020-04-23 11:44:13 +02:00
base_url: str,
course_id: str,
authenticator: IliasAuthenticator,
2020-04-24 20:39:30 +02:00
cookies: Optional[PathLike],
dir_filter: IliasDirectoryFilter,
2020-04-23 11:44:13 +02:00
transform: Transform,
download_strategy: IliasDownloadStrategy,
2020-04-23 11:44:13 +02:00
) -> None:
# pylint: disable=too-many-locals
2020-04-24 20:39:30 +02:00
cookie_jar = CookieJar(to_path(cookies) if cookies else None)
2020-04-23 11:44:13 +02:00
session = cookie_jar.create_session()
tmp_dir = self._tmp_dir.new_subdir()
2020-04-24 20:39:30 +02:00
organizer = Organizer(self.resolve(to_path(target)))
2020-04-23 11:44:13 +02:00
crawler = IliasCrawler(base_url, course_id, session, authenticator, dir_filter)
downloader = IliasDownloader(tmp_dir, organizer, session, authenticator, download_strategy)
2020-04-23 11:44:13 +02:00
cookie_jar.load_cookies()
info = crawler.crawl()
cookie_jar.save_cookies()
2020-04-24 20:00:21 +02:00
transformed = apply_transform(transform, info)
if self._test_run:
self._print_transformables(transformed)
return
downloader.download_all(transformed)
2020-04-23 11:44:13 +02:00
cookie_jar.save_cookies()
2020-04-24 16:17:23 +02:00
organizer.cleanup()
2020-04-23 11:44:13 +02:00
def ilias_kit(
self,
2020-04-24 20:39:30 +02:00
target: PathLike,
2020-04-23 11:44:13 +02:00
course_id: str,
dir_filter: IliasDirectoryFilter = lambda x: True,
2020-04-23 11:44:13 +02:00
transform: Transform = lambda x: x,
2020-04-24 20:39:30 +02:00
cookies: Optional[PathLike] = None,
2020-04-23 11:44:13 +02:00
username: Optional[str] = None,
password: Optional[str] = None,
download_strategy: IliasDownloadStrategy = download_modified_or_new,
2020-04-23 11:44:13 +02:00
) -> None:
"""
Synchronizes a folder with the ILIAS instance of the KIT.
Arguments:
target {Path} -- the target path to write the data to
course_id {str} -- the id of the main course page (found in the URL after ref_id
when opening the course homepage)
Keyword Arguments:
dir_filter {IliasDirectoryFilter} -- A filter for directories. Will be applied on the
crawler level, these directories and all of their content is skipped.
(default: {lambdax:True})
transform {Transform} -- A transformation function for the output paths. Return None
to ignore a file. (default: {lambdax:x})
cookies {Optional[Path]} -- The path to store and load cookies from.
(default: {None})
username {Optional[str]} -- The SCC username. If none is given, it will prompt
the user. (default: {None})
password {Optional[str]} -- The SCC password. If none is given, it will prompt
the user. (default: {None})
download_strategy {DownloadStrategy} -- A function to determine which files need to
be downloaded. Can save bandwidth and reduce the number of requests.
(default: {download_modified_or_new})
"""
2020-04-23 11:44:13 +02:00
# This authenticator only works with the KIT ilias instance.
authenticator = KitShibbolethAuthenticator(username=username, password=password)
2020-04-23 19:55:37 +02:00
PRETTY.starting_synchronizer(target, "ILIAS", course_id)
2020-04-23 11:44:13 +02:00
self._ilias(
target=target,
base_url="https://ilias.studium.kit.edu/",
course_id=course_id,
authenticator=authenticator,
cookies=cookies,
dir_filter=dir_filter,
2020-04-23 11:44:13 +02:00
transform=transform,
download_strategy=download_strategy,
2020-04-23 11:44:13 +02:00
)